Merge pull request #1337 from lightpanda-io/ci-check

ci: try to fix tag

.github/actions/install/action.yml

@@ -13,7 +13,7 @@ inputs:
   zig-v8:
     description: 'zig v8 version to install'
     required: false
-    default: 'v0.3.1'
+    default: 'v0.1.35'
   v8:
     description: 'v8 version to install'
     required: false
@@ -22,10 +22,6 @@ inputs:
     description: 'cache dir to use'
     required: false
     default: '~/.cache'
-  debug:
-    description: 'enable v8 pre-built debug version, only available for linux x86_64'
-    required: false
-    default: 'false'
 
 runs:
   using: "composite"
@@ -36,14 +32,11 @@ runs:
       shell: bash
       run: |
         sudo apt-get update
-        sudo apt-get install -y wget xz-utils ca-certificates clang make git
+        sudo apt-get install -y wget xz-utils python3 ca-certificates git pkg-config libglib2.0-dev gperf libexpat1-dev cmake clang
 
     # Zig version used from the `minimum_zig_version` field in build.zig.zon
     - uses: mlugg/setup-zig@v2
 
-    # Rust Toolchain for html5ever
-    - uses: dtolnay/rust-toolchain@stable
-
     - name: Cache v8
       id: cache-v8
       uses: actions/cache@v4
@@ -51,17 +44,43 @@ runs:
         cache-name: cache-v8
       with:
         path: ${{ inputs.cache-dir }}/v8
-        key: libc_v8_${{ inputs.v8 }}_${{ inputs.os }}_${{ inputs.arch }}_${{ inputs.zig-v8 }}${{inputs.debug == 'true' && '_debug' || '' }}.a
+        key: libc_v8_${{ inputs.v8 }}_${{ inputs.os }}_${{ inputs.arch }}_${{ inputs.zig-v8 }}.a
 
     - if: ${{ steps.cache-v8.outputs.cache-hit != 'true' }}
       shell: bash
       run: |
         mkdir -p ${{ inputs.cache-dir }}/v8
 
-        wget -O ${{ inputs.cache-dir }}/v8/libc_v8.a https://github.com/lightpanda-io/zig-v8-fork/releases/download/${{ inputs.zig-v8 }}/libc_v8_${{ inputs.v8 }}_${{ inputs.os }}_${{ inputs.arch }}${{inputs.debug == 'true' && '_debug' || '' }}.a
+        wget -O ${{ inputs.cache-dir }}/v8/libc_v8.a https://github.com/lightpanda-io/zig-v8-fork/releases/download/${{ inputs.zig-v8 }}/libc_v8_${{ inputs.v8 }}_${{ inputs.os }}_${{ inputs.arch }}.a
 
     - name: install v8
       shell: bash
       run: |
         mkdir -p v8
-        ln -s ${{ inputs.cache-dir }}/v8/libc_v8.a v8/libc_v8${{inputs.debug == 'true' && '_debug' || '' }}.a
+        ln -s ${{ inputs.cache-dir }}/v8/libc_v8.a v8/libc_v8.a
+
+    - name: Cache libiconv
+      id: cache-libiconv
+      uses: actions/cache@v4
+      env:
+        cache-name: cache-libiconv
+      with:
+        path: ${{ inputs.cache-dir }}/libiconv
+        key: vendor/libiconv/libiconv-1.17
+
+    - name: download libiconv
+      if: ${{ steps.cache-libiconv.outputs.cache-hit != 'true' }}
+      shell: bash
+      run: make download-libiconv
+
+    - name: build libiconv
+      shell: bash
+      run: make build-libiconv
+
+    - name: build mimalloc
+      shell: bash
+      run: make install-mimalloc
+
+    - name: build netsurf
+      shell: bash
+      run: make install-netsurf

.github/workflows/build.yml

@@ -27,13 +27,12 @@ jobs:
       OS: linux
 
     runs-on: ubuntu-22.04
-    timeout-minutes: 20
+    timeout-minutes: 15
 
     steps:
       - uses: actions/checkout@v6
         with:
           fetch-depth: 0
-          # fetch submodules recusively, to get zig-js-runtime submodules also.
           submodules: recursive
 
       - uses: ./.github/actions/install
@@ -41,11 +40,8 @@ jobs:
           os: ${{env.OS}}
           arch: ${{env.ARCH}}
 
-      - name: v8 snapshot
-        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseFast snapshot_creator -- src/snapshot.bin
-
       - name: zig build
-        run: zig build -Dsnapshot_path=../../snapshot.bin -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseFast -Dcpu=x86_64 -Dgit_commit=$(git rev-parse --short ${{ github.sha }})
+        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseSafe -Dcpu=x86_64 -Dgit_commit=$(git rev-parse --short ${{ github.sha }})
 
       - name: Rename binary
         run: mv zig-out/bin/lightpanda lightpanda-${{ env.ARCH }}-${{ env.OS }}
@@ -61,7 +57,6 @@ jobs:
           allowUpdates: true
           artifacts: lightpanda-${{ env.ARCH }}-${{ env.OS }}
           tag: ${{ env.RELEASE }}
-          makeLatest: true
 
   build-linux-aarch64:
     env:
@@ -69,7 +64,7 @@ jobs:
       OS: linux
 
     runs-on: ubuntu-22.04-arm
-    timeout-minutes: 20
+    timeout-minutes: 15
 
     steps:
       - uses: actions/checkout@v4
@@ -83,11 +78,8 @@ jobs:
           os: ${{env.OS}}
           arch: ${{env.ARCH}}
 
-      - name: v8 snapshot
-        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseFast snapshot_creator -- src/snapshot.bin
-
       - name: zig build
-        run: zig build -Dsnapshot_path=../../snapshot.bin -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseFast -Dcpu=generic -Dgit_commit=$(git rev-parse --short ${{ github.sha }})
+        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseSafe -Dcpu=generic -Dgit_commit=$(git rev-parse --short ${{ github.sha }})
 
       - name: Rename binary
         run: mv zig-out/bin/lightpanda lightpanda-${{ env.ARCH }}-${{ env.OS }}
@@ -103,7 +95,6 @@ jobs:
           allowUpdates: true
           artifacts: lightpanda-${{ env.ARCH }}-${{ env.OS }}
           tag: ${{ env.RELEASE }}
-          makeLatest: true
 
   build-macos-aarch64:
     env:
@@ -113,7 +104,7 @@ jobs:
     # macos-14 runs on arm CPU. see
     # https://github.com/actions/runner-images?tab=readme-ov-file
     runs-on: macos-14
-    timeout-minutes: 20
+    timeout-minutes: 15
 
     steps:
       - uses: actions/checkout@v4
@@ -127,11 +118,8 @@ jobs:
           os: ${{env.OS}}
           arch: ${{env.ARCH}}
 
-      - name: v8 snapshot
-        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseFast snapshot_creator -- src/snapshot.bin
-
       - name: zig build
-        run: zig build -Dsnapshot_path=../../snapshot.bin -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseFast -Dgit_commit=$(git rev-parse --short ${{ github.sha }})
+        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseSafe -Dgit_commit=$(git rev-parse --short ${{ github.sha }})
 
       - name: Rename binary
         run: mv zig-out/bin/lightpanda lightpanda-${{ env.ARCH }}-${{ env.OS }}
@@ -147,7 +135,6 @@ jobs:
           allowUpdates: true
           artifacts: lightpanda-${{ env.ARCH }}-${{ env.OS }}
           tag: ${{ env.RELEASE }}
-          makeLatest: true
 
   build-macos-x86_64:
     env:
@@ -155,7 +142,7 @@ jobs:
       OS: macos
 
     runs-on: macos-14-large
-    timeout-minutes: 20
+    timeout-minutes: 15
 
     steps:
       - uses: actions/checkout@v4
@@ -169,11 +156,8 @@ jobs:
           os: ${{env.OS}}
           arch: ${{env.ARCH}}
 
-      - name: v8 snapshot
-        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseFast snapshot_creator -- src/snapshot.bin
-
       - name: zig build
-        run: zig build -Dsnapshot_path=../../snapshot.bin -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseFast -Dgit_commit=$(git rev-parse --short ${{ github.sha }})
+        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseSafe -Dgit_commit=$(git rev-parse --short ${{ github.sha }})
 
       - name: Rename binary
         run: mv zig-out/bin/lightpanda lightpanda-${{ env.ARCH }}-${{ env.OS }}
@@ -189,4 +173,3 @@ jobs:
           allowUpdates: true
           artifacts: lightpanda-${{ env.ARCH }}-${{ env.OS }}
           tag: ${{ env.RELEASE }}
-          makeLatest: true

.github/workflows/cla.yml

@@ -28,7 +28,7 @@ jobs:
           path-to-document: 'https://github.com/lightpanda-io/browser/blob/main/CLA.md'
           # branch should not be protected
           branch: 'main'
-          allowlist: krichprollsch,francisbouvier,katie-lpd,sjorsdonkers,bornlex
+          allowlist: krichprollsch,francisbouvier,katie-lpd
 
           remote-organization-name: lightpanda-io
           remote-repository-name: cla

.github/workflows/e2e-integration-test.yml

@@ -63,6 +63,6 @@ jobs:
 
       - name: run end to end integration tests
         run: |
-          ./lightpanda serve --log_level error & echo $! > LPD.pid
+          ./lightpanda serve & echo $! > LPD.pid
           go run integration/main.go
           kill `cat LPD.pid`

.github/workflows/e2e-test.yml

@@ -52,7 +52,6 @@ jobs:
       - uses: actions/checkout@v6
         with:
           fetch-depth: 0
-          # fetch submodules recusively, to get zig-js-runtime submodules also.
           submodules: recursive
 
       - uses: ./.github/actions/install
@@ -122,19 +121,10 @@ jobs:
     needs: zig-build-release
 
     env:
-      MAX_VmHWM: 28000 # 28MB (KB)
-      MAX_CG_PEAK: 8000 # 8MB (KB)
-      MAX_AVG_DURATION: 17
+      MAX_MEMORY: 28000
+      MAX_AVG_DURATION: 23
       LIGHTPANDA_DISABLE_TELEMETRY: true
 
-      # How to give cgroups access to the user actions-runner on the host:
-      # $ sudo apt install cgroup-tools
-      # $ sudo chmod o+w /sys/fs/cgroup/cgroup.procs
-      # $ sudo mkdir -p /sys/fs/cgroup/actions-runner
-      # $ sudo chown -R actions-runner:actions-runner /sys/fs/cgroup/actions-runner
-      CG_ROOT: /sys/fs/cgroup
-      CG: actions-runner/lpd_${{ github.run_id }}_${{ github.run_attempt }}
-
     # use a self host runner.
     runs-on: lpd-bench-hetzner
     timeout-minutes: 15
@@ -159,53 +149,22 @@ jobs:
           go run ws/main.go & echo $! > WS.pid
           sleep 2
 
-      - name: run lightpanda in cgroup
-        run: |
-          if [ ! -f /sys/fs/cgroup/cgroup.controllers ]; then
-            echo "cgroup v2 not available: /sys/fs/cgroup/cgroup.controllers missing"
-            exit 1
-          fi
-
-          mkdir -p $CG_ROOT/$CG
-          cgexec -g memory:$CG ./lightpanda serve & echo $! > LPD.pid
-
-          sleep 2
-
       - name: run puppeteer
         run: |
+          ./lightpanda serve & echo $! > LPD.pid
+          sleep 2
           RUNS=100 npm run bench-puppeteer-cdp > puppeteer.out || exit 1
           cat /proc/`cat LPD.pid`/status |grep VmHWM|grep -oP '\d+' > LPD.VmHWM
           kill `cat LPD.pid`
 
-          PID=$(cat LPD.pid)
-          while kill -0 $PID 2>/dev/null; do
-            sleep 1
-          done
-          if [ ! -f $CG_ROOT/$CG/memory.peak ]; then
-            echo "memory.peak not available in $CG"
-            exit 1
-          fi
-          cat $CG_ROOT/$CG/memory.peak > LPD.cg_mem_peak
-
       - name: puppeteer result
         run: cat puppeteer.out
 
-      - name: cgroup memory regression
-        run: |
-          PEAK_BYTES=$(cat LPD.cg_mem_peak)
-          PEAK_KB=$((PEAK_BYTES / 1024))
-          echo "memory.peak_bytes=$PEAK_BYTES"
-          echo "memory.peak_kb=$PEAK_KB"
-          test "$PEAK_KB" -le "$MAX_CG_PEAK"
-
-      - name: virtual memory regression
+      - name: memory regression
         run: |
           export LPD_VmHWM=`cat LPD.VmHWM`
           echo "Peak resident set size: $LPD_VmHWM"
-          test "$LPD_VmHWM" -le "$MAX_VmHWM"
-
-      - name: cleanup cgroup
-        run: rmdir $CG_ROOT/$CG
+          test "$LPD_VmHWM" -le "$MAX_MEMORY"
 
       - name: duration regression
         run: |
@@ -218,8 +177,7 @@ jobs:
           export AVG_DURATION=`cat puppeteer.out|grep 'avg run'|sed 's/avg run duration (ms) //'`
           export TOTAL_DURATION=`cat puppeteer.out|grep 'total duration'|sed 's/total duration (ms) //'`
           export LPD_VmHWM=`cat LPD.VmHWM`
-          export LPD_CG_PEAK_KB=$(( $(cat LPD.cg_mem_peak) / 1024 ))
-          echo "{\"duration_total\":${TOTAL_DURATION},\"duration_avg\":${AVG_DURATION},\"mem_peak\":${LPD_VmHWM},\"cg_mem_peak\":${LPD_CG_PEAK_KB}}" > bench.json
+          echo "{\"duration_total\":${TOTAL_DURATION},\"duration_avg\":${AVG_DURATION},\"mem_peak\":${LPD_VmHWM}}" > bench.json
           cat bench.json
 
       - name: run hyperfine
@@ -271,19 +229,3 @@ jobs:
 
       - name: format and send json result
         run: /perf-fmt hyperfine ${{ github.sha }} hyperfine.json
-
-  browser-fetch:
-    name: browser fetch
-    needs: zig-build-release
-
-    runs-on: ubuntu-latest
-
-    steps:
-      - name: download artifact
-        uses: actions/download-artifact@v4
-        with:
-          name: lightpanda-build-release
-
-      - run: chmod a+x ./lightpanda
-
-      - run: ./lightpanda fetch https://demo-browser.lightpanda.io/campfire-commerce/

.github/workflows/wpt.yml

@@ -15,100 +15,22 @@ on:
   workflow_dispatch:
 
 jobs:
-  wpt-build-release:
-    name: zig build release
+  wpt:
+    name: web platform tests json output
 
     runs-on: ubuntu-latest
-    timeout-minutes: 15
+    timeout-minutes: 90
 
     steps:
       - uses: actions/checkout@v6
         with:
           fetch-depth: 0
-          # fetch submodules recusively, to get zig-js-runtime submodules also.
           submodules: recursive
 
       - uses: ./.github/actions/install
 
-      - name: zig build release
-        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a -Doptimize=ReleaseFast -Dcpu=x86_64 -Dgit_commit=$(git rev-parse --short ${{ github.sha }})
-
-      - name: upload artifact
-        uses: actions/upload-artifact@v4
-        with:
-          name: lightpanda-build-release
-          path: |
-            zig-out/bin/lightpanda
-          retention-days: 1
-
-  wpt-build-runner:
-    name: build wpt runner
-
-    runs-on: ubuntu-latest
-    timeout-minutes: 15
-
-    steps:
-      - uses: actions/checkout@v6
-        with:
-          repository: 'lightpanda-io/demo'
-          fetch-depth: 0
-
-      - run: |
-          cd ./wptrunner
-          CGO_ENABLED=0 go build
-
-      - name: upload artifact
-        uses: actions/upload-artifact@v4
-        with:
-          name: wptrunner
-          path: |
-            wptrunner/wptrunner
-          retention-days: 1
-
-  run-wpt:
-    name: web platform tests json output
-    needs:
-      - wpt-build-release
-      - wpt-build-runner
-
-    # use a self host runner.
-    runs-on: lpd-bench-hetzner
-    timeout-minutes: 120
-
-    steps:
-      - uses: actions/checkout@v6
-        with:
-          ref: fork
-          repository: 'lightpanda-io/wpt'
-          fetch-depth: 0
-
-      # The hosts are configured manually on the self host runner.
-      # - name: create custom hosts
-      #   run: ./wpt make-hosts-file | sudo tee -a /etc/hosts
-
-      - name: generate manifest
-        run: ./wpt manifest
-
-      - name: download lightpanda release
-        uses: actions/download-artifact@v4
-        with:
-          name: lightpanda-build-release
-
-      - run: chmod a+x ./lightpanda
-
-      - name: download wptrunner
-        uses: actions/download-artifact@v4
-        with:
-          name: wptrunner
-
-      - run: chmod a+x ./wptrunner
-
-      - name: run test with json output
-        run: |
-          ./wpt serve 2> /dev/null & echo $! > WPT.pid
-          sleep 10s
-          ./wptrunner -lpd-path ./lightpanda -json -concurrency 3 > wpt.json
-          kill `cat WPT.pid`
+      - name: json output
+        run: zig build wpt -- --json > wpt.json
 
       - name: write commit
         run: |
@@ -125,7 +47,7 @@ jobs:
 
   perf-fmt:
     name: perf-fmt
-    needs: run-wpt
+    needs: wpt
 
     runs-on: ubuntu-latest
     timeout-minutes: 15

.github/workflows/zig-test.yml

@@ -12,7 +12,8 @@ on:
       - main
     paths:
       - "build.zig"
-      - "src/**"
+      - "src/**/*.zig"
+      - "src/*.zig"
       - "vendor/zig-js-runtime"
       - ".github/**"
       - "vendor/**"
@@ -37,25 +38,51 @@ on:
   workflow_dispatch:
 
 jobs:
-  zig-test-debug:
-    name: zig test using v8 in debug mode
-    timeout-minutes: 15
+  zig-build-dev:
+    name: zig build dev
+
+    # Don't run the CI with draft PR.
+    if: github.event.pull_request.draft == false
 
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
-          # fetch submodules recusively, to get zig-js-runtime submodules also.
           submodules: recursive
 
       - uses: ./.github/actions/install
-        with:
-          debug: true
 
-      - name: zig build test
-        run: zig build -Dprebuilt_v8_path=v8/libc_v8_debug.a -Dtsan=true test
+      - name: zig build debug
+        run: zig build -Dprebuilt_v8_path=v8/libc_v8.a
+
+      - name: upload artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: lightpanda-build-dev
+          path: |
+            zig-out/bin/lightpanda
+          retention-days: 1
+
+  browser-fetch:
+    name: browser fetch
+    needs: zig-build-dev
+
+    # Don't run the CI with draft PR.
+    if: github.event.pull_request.draft == false
+
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: download artifact
+        uses: actions/download-artifact@v4
+        with:
+          name: lightpanda-build-dev
+
+      - run: chmod a+x ./lightpanda
+
+      - run: ./lightpanda fetch https://httpbin.io/xhr/get
 
   zig-test:
     name: zig test
@@ -76,7 +103,7 @@ jobs:
       - uses: ./.github/actions/install
 
       - name: zig build test
-        run: METRICS=true zig build -Dprebuilt_v8_path=v8/libc_v8.a test > bench.json
+        run: zig build  -Dprebuilt_v8_path=v8/libc_v8.a test -- --json > bench.json
 
       - name: write commit
         run: |

.gitignore

@@ -1,6 +1,8 @@
+zig-cache
 /.zig-cache/
 /.lp-cache/
 zig-out
+/vendor/netsurf/out
+/vendor/libiconv/
 lightpanda.id
-/src/html5ever/target/
-src/snapshot.bin
+/v8/

.gitmodules

@@ -0,0 +1,33 @@
+[submodule "vendor/netsurf/libwapcaplet"]
+	path = vendor/netsurf/libwapcaplet
+	url = https://github.com/lightpanda-io/libwapcaplet.git/
+[submodule "vendor/netsurf/libparserutils"]
+	path = vendor/netsurf/libparserutils
+	url = https://github.com/lightpanda-io/libparserutils.git/
+[submodule "vendor/netsurf/libdom"]
+	path = vendor/netsurf/libdom
+	url = https://github.com/lightpanda-io/libdom.git/
+[submodule "vendor/netsurf/share/netsurf-buildsystem"]
+	path = vendor/netsurf/share/netsurf-buildsystem
+	url = https://github.com/lightpanda-io/netsurf-buildsystem.git
+[submodule "vendor/netsurf/libhubbub"]
+	path = vendor/netsurf/libhubbub
+	url = https://github.com/lightpanda-io/libhubbub.git/
+[submodule "tests/wpt"]
+	path = tests/wpt
+	url = https://github.com/lightpanda-io/wpt
+[submodule "vendor/mimalloc"]
+	path = vendor/mimalloc
+	url = https://github.com/microsoft/mimalloc.git/
+[submodule "vendor/nghttp2"]
+	path = vendor/nghttp2
+	url = https://github.com/nghttp2/nghttp2.git
+[submodule "vendor/zlib"]
+	path = vendor/zlib
+	url = https://github.com/madler/zlib.git
+[submodule "vendor/curl"]
+	path = vendor/curl
+	url = https://github.com/curl/curl.git
+[submodule "vendor/brotli"]
+	path = vendor/brotli
+	url = https://github.com/google/brotli

Dockerfile

@@ -3,17 +3,16 @@ FROM debian:stable-slim
 ARG MINISIG=0.12
 ARG ZIG_MINISIG=RWSGOq2NVecA2UPNdBUZykf1CCb147pkmdtYxgb3Ti+JO/wCYvhbAb/U
 ARG V8=14.0.365.4
-ARG ZIG_V8=v0.3.1
+ARG ZIG_V8=v0.1.34
 ARG TARGETPLATFORM
 
 RUN apt-get update -yq && \
-    apt-get install -yq xz-utils ca-certificates \
+    apt-get install -yq xz-utils \
+        python3 ca-certificates git \
         pkg-config libglib2.0-dev \
-        clang make curl git
-
-# Get Rust
-RUN curl https://sh.rustup.rs -sSf | sh -s -- --profile minimal -y
-ENV PATH="/root/.cargo/bin:${PATH}"
+        gperf libexpat1-dev \
+        cmake clang \
+        curl git
 
 # install minisig
 RUN curl --fail -L -O https://github.com/jedisct1/minisign/releases/download/${MINISIG}/minisign-${MINISIG}-linux.tar.gz && \
@@ -40,6 +39,10 @@ RUN ZIG=$(grep '\.minimum_zig_version = "' "build.zig.zon" | cut -d'"' -f2) && \
 RUN git submodule init && \
     git submodule update --recursive
 
+RUN make install-libiconv && \
+    make install-netsurf && \
+    make install-mimalloc
+
 # download and install v8
 RUN case $TARGETPLATFORM in \
     "linux/arm64") ARCH="aarch64" ;; \
@@ -49,16 +52,8 @@ RUN case $TARGETPLATFORM in \
     mkdir -p v8/ && \
     mv libc_v8.a v8/libc_v8.a
 
-# build v8 snapshot
-RUN zig build -Doptimize=ReleaseFast \
-    -Dprebuilt_v8_path=v8/libc_v8.a \
-    snapshot_creator -- src/snapshot.bin
-
 # build release
-RUN zig build -Doptimize=ReleaseFast \
-    -Dsnapshot_path=../../snapshot.bin \
-    -Dprebuilt_v8_path=v8/libc_v8.a \
-    -Dgit_commit=$(git rev-parse --short HEAD)
+RUN zig build -Doptimize=ReleaseSafe -Dprebuilt_v8_path=v8/libc_v8.a -Dgit_commit=$$(git rev-parse --short HEAD)
 
 FROM debian:stable-slim
 

LICENSING.md

@@ -5,6 +5,14 @@ List](https://spdx.org/licenses/).
 
 The default license for this project is [AGPL-3.0-only](LICENSE).
 
+## MIT
+
+The following files are licensed under MIT:
+
+```
+src/polyfill/fetch.js
+```
+
 The following directories and their subdirectories are licensed under their
 original upstream licenses:
 

Makefile

@@ -47,18 +47,12 @@ help:
 
 # $(ZIG) commands
 # ------------
-.PHONY: build build-v8-snapshot build-dev run run-release shell test bench data end2end
+.PHONY: build build-dev run run-release shell test bench wpt data end2end
 
-## Build v8 snapshot
-build-v8-snapshot:
-	@printf "\033[36mBuilding v8 snapshot (release safe)...\033[0m\n"
-	@$(ZIG) build -Doptimize=ReleaseFast snapshot_creator -- src/snapshot.bin || (printf "\033[33mBuild ERROR\033[0m\n"; exit 1;)
-	@printf "\033[33mBuild OK\033[0m\n"
-
-## Build in release-fast mode
-build: build-v8-snapshot
-	@printf "\033[36mBuilding (release fast)...\033[0m\n"
-	@$(ZIG) build -Doptimize=ReleaseFast -Dsnapshot_path=../../snapshot.bin -Dgit_commit=$$(git rev-parse --short HEAD) || (printf "\033[33mBuild ERROR\033[0m\n"; exit 1;)
+## Build in release-safe mode
+build:
+	@printf "\033[36mBuilding (release safe)...\033[0m\n"
+	$(ZIG) build -Doptimize=ReleaseSafe -Dgit_commit=$$(git rev-parse --short HEAD) || (printf "\033[33mBuild ERROR\033[0m\n"; exit 1;)
 	@printf "\033[33mBuild OK\033[0m\n"
 
 ## Build in debug mode
@@ -82,14 +76,23 @@ shell:
 	@printf "\033[36mBuilding shell...\033[0m\n"
 	@$(ZIG) build shell || (printf "\033[33mBuild ERROR\033[0m\n"; exit 1;)
 
+## Run WPT tests
+wpt:
+	@printf "\033[36mBuilding wpt...\033[0m\n"
+	@$(ZIG) build wpt -- $(filter-out $@,$(MAKECMDGOALS)) || (printf "\033[33mBuild ERROR\033[0m\n"; exit 1;)
+
+wpt-summary:
+	@printf "\033[36mBuilding wpt...\033[0m\n"
+	@$(ZIG) build wpt -- --summary $(filter-out $@,$(MAKECMDGOALS)) || (printf "\033[33mBuild ERROR\033[0m\n"; exit 1;)
+
 ## Test - `grep` is used to filter out the huge compile command on build
 ifeq ($(OS), macos)
 test:
-	@script -q /dev/null sh -c 'TEST_FILTER="${F}" $(ZIG) build test -freference-trace' 2>&1 \
+	@script -q /dev/null sh -c 'TEST_FILTER="${F}" $(ZIG) build test -freference-trace --summary all' 2>&1 \
 		| grep --line-buffered -v "^/.*zig test -freference-trace"
 else
 test:
-	@script -qec 'TEST_FILTER="${F}" $(ZIG) build test -freference-trace' /dev/null 2>&1 \
+	@script -qec 'TEST_FILTER="${F}" $(ZIG) build test -freference-trace --summary all' /dev/null 2>&1 \
 		| grep --line-buffered -v "^/.*zig test -freference-trace"
 endif
 
@@ -100,10 +103,129 @@ end2end:
 
 # Install and build required dependencies commands
 # ------------
-.PHONY: install
+.PHONY: install-submodule
+.PHONY: install-libiconv
+.PHONY: _install-netsurf install-netsurf clean-netsurf test-netsurf install-netsurf-dev
+.PHONY: install-mimalloc install-mimalloc-dev clean-mimalloc
+.PHONY: install-dev install
 
-install: build
+## Install and build dependencies for release
+install: install-submodule install-libiconv install-netsurf install-mimalloc
+
+## Install and build dependencies for dev
+install-dev: install-submodule install-libiconv install-netsurf-dev install-mimalloc-dev
+
+install-netsurf-dev: _install-netsurf
+install-netsurf-dev: OPTCFLAGS := -O0 -g -DNDEBUG
+
+install-netsurf: _install-netsurf
+install-netsurf: OPTCFLAGS := -DNDEBUG
+
+BC_NS := $(BC)vendor/netsurf/out/$(OS)-$(ARCH)
+ICONV := $(BC)vendor/libiconv/out/$(OS)-$(ARCH)
+# TODO: add Linux iconv path (I guess it depends on the distro)
+# TODO: this way of linking libiconv is not ideal. We should have a more generic way
+# and stick to a specif version. Maybe build from source. Anyway not now.
+_install-netsurf: clean-netsurf
+	@printf "\033[36mInstalling NetSurf...\033[0m\n" && \
+	ls $(ICONV)/lib/libiconv.a 1> /dev/null || (printf "\033[33mERROR: you need to execute 'make install-libiconv'\033[0m\n"; exit 1;) && \
+	mkdir -p $(BC_NS) && \
+	cp -R vendor/netsurf/share $(BC_NS) && \
+	export PREFIX=$(BC_NS) && \
+	export OPTLDFLAGS="-L$(ICONV)/lib" && \
+	export OPTCFLAGS="$(OPTCFLAGS) -I$(ICONV)/include" && \
+	printf "\033[33mInstalling libwapcaplet...\033[0m\n" && \
+	cd vendor/netsurf/libwapcaplet && \
+	BUILDDIR=$(BC_NS)/build/libwapcaplet make install && \
+	cd ../libparserutils && \
+	printf "\033[33mInstalling libparserutils...\033[0m\n" && \
+	BUILDDIR=$(BC_NS)/build/libparserutils make install && \
+	cd ../libhubbub && \
+	printf "\033[33mInstalling libhubbub...\033[0m\n" && \
+	BUILDDIR=$(BC_NS)/build/libhubbub make install && \
+	rm src/treebuilder/autogenerated-element-type.c && \
+	cd ../libdom && \
+	printf "\033[33mInstalling libdom...\033[0m\n" && \
+	BUILDDIR=$(BC_NS)/build/libdom make install && \
+	printf "\033[33mRunning libdom example...\033[0m\n" && \
+	cd examples && \
+	$(ZIG) cc \
+	-I$(ICONV)/include \
+	-I$(BC_NS)/include \
+	-L$(ICONV)/lib \
+	-L$(BC_NS)/lib \
+	-liconv \
+	-ldom \
+	-lhubbub \
+	-lparserutils \
+	-lwapcaplet \
+	-o a.out \
+	dom-structure-dump.c \
+	$(ICONV)/lib/libiconv.a && \
+	./a.out > /dev/null && \
+	rm a.out && \
+	printf "\033[36mDone NetSurf $(OS)\033[0m\n"
+
+clean-netsurf:
+	@printf "\033[36mCleaning NetSurf build...\033[0m\n" && \
+	rm -Rf $(BC_NS)
+
+test-netsurf:
+	@printf "\033[36mTesting NetSurf...\033[0m\n" && \
+	export PREFIX=$(BC_NS) && \
+	export LDFLAGS="-L$(ICONV)/lib -L$(BC_NS)/lib" && \
+	export CFLAGS="-I$(ICONV)/include -I$(BC_NS)/include" && \
+	cd vendor/netsurf/libdom && \
+	BUILDDIR=$(BC_NS)/build/libdom make test
+
+download-libiconv:
+ifeq ("$(wildcard vendor/libiconv/libiconv-1.17)","")
+	@mkdir -p vendor/libiconv
+	@cd vendor/libiconv && \
+	curl -L https://github.com/lightpanda-io/libiconv/releases/download/1.17/libiconv-1.17.tar.gz | tar -xvzf -
+endif
+
+build-libiconv: clean-libiconv
+	@cd vendor/libiconv/libiconv-1.17 && \
+	./configure --prefix=$(ICONV) --enable-static && \
+	make && make install
+
+install-libiconv: download-libiconv build-libiconv
+
+clean-libiconv:
+ifneq ("$(wildcard vendor/libiconv/libiconv-1.17/Makefile)","")
+	@cd vendor/libiconv/libiconv-1.17 && \
+	make clean
+endif
 
 data:
 	cd src/data && go run public_suffix_list_gen.go > public_suffix_list.zig
 
+.PHONY: _build_mimalloc
+
+MIMALLOC := $(BC)vendor/mimalloc/out/$(OS)-$(ARCH)
+_build_mimalloc: clean-mimalloc
+	@mkdir -p $(MIMALLOC)/build && \
+	cd $(MIMALLOC)/build && \
+	cmake -DMI_BUILD_SHARED=OFF -DMI_BUILD_OBJECT=OFF -DMI_BUILD_TESTS=OFF -DMI_OVERRIDE=OFF $(OPTS) ../../.. && \
+	make && \
+	mkdir -p $(MIMALLOC)/lib
+
+install-mimalloc-dev: _build_mimalloc
+install-mimalloc-dev: OPTS=-DCMAKE_BUILD_TYPE=Debug
+install-mimalloc-dev:
+	@cd $(MIMALLOC) && \
+	mv build/libmimalloc-debug.a lib/libmimalloc.a
+
+install-mimalloc: _build_mimalloc
+install-mimalloc:
+	@cd $(MIMALLOC) && \
+	mv build/libmimalloc.a lib/libmimalloc.a
+
+clean-mimalloc:
+	@rm -Rf $(MIMALLOC)/build
+
+## Init and update git submodule
+install-submodule:
+	@git submodule init && \
+	git submodule update

README.md

@@ -78,49 +78,23 @@ docker run -d --name lightpanda -p 9222:9222 lightpanda/browser:nightly
 ### Dump a URL
 
 ```console
-./lightpanda fetch --obey_robots --log_format pretty  --log_level info https://demo-browser.lightpanda.io/campfire-commerce/
+./lightpanda fetch --dump https://lightpanda.io
 ```
 ```console
-INFO  telemetry : telemetry status . . . . . . . . . . . . .  [+0ms]
-      disabled = false
-
-INFO  page : navigate . . . . . . . . . . . . . . . . . . . . [+6ms]
-      url = https://demo-browser.lightpanda.io/campfire-commerce/
-      method = GET
-      reason = address_bar
-      body = false
-      req_id = 1
-
-INFO  browser : executing script . . . . . . . . . . . . . .  [+118ms]
-      src = https://demo-browser.lightpanda.io/campfire-commerce/script.js
-      kind = javascript
-      cacheable = true
-
-INFO  http : request complete . . . . . . . . . . . . . . . . [+140ms]
-      source = xhr
-      url = https://demo-browser.lightpanda.io/campfire-commerce/json/product.json
-      status = 200
-      len = 4770
-
-INFO  http : request complete . . . . . . . . . . . . . . . . [+141ms]
-      source = fetch
-      url = https://demo-browser.lightpanda.io/campfire-commerce/json/reviews.json
-      status = 200
-      len = 1615
+info(browser): GET https://lightpanda.io/ http.Status.ok
+info(browser): fetch script https://api.website.lightpanda.io/js/script.js: http.Status.ok
+info(browser): eval remote https://api.website.lightpanda.io/js/script.js: TypeError: Cannot read properties of undefined (reading 'pushState')
 <!DOCTYPE html>
 ```
 
 ### Start a CDP server
 
 ```console
-./lightpanda serve --obey_robots --log_format pretty  --log_level info --host 127.0.0.1 --port 9222
+./lightpanda serve --host 127.0.0.1 --port 9222
 ```
 ```console
-INFO  telemetry : telemetry status . . . . . . . . . . . . .  [+0ms]
-      disabled = false
-
-INFO  app : server running . . . . . . . . . . . . . . . . .  [+0ms]
-      address = 127.0.0.1:9222
+info(websocket): starting blocking worker to listen on 127.0.0.1:9222
+info(server): accepting new conn...
 ```
 
 Once the CDP server started, you can run a Puppeteer script by configuring the
@@ -141,7 +115,7 @@ const context = await browser.createBrowserContext();
 const page = await context.newPage();
 
 // Dump all the links from the page.
-await page.goto('https://demo-browser.lightpanda.io/amiibo/', {waitUntil: "networkidle0"});
+await page.goto('https://wikipedia.com/');
 
 const links = await page.evaluate(() => {
   return Array.from(document.querySelectorAll('a')).map(row => {
@@ -166,14 +140,13 @@ You may still encounter errors or crashes. Please open an issue with specifics i
 
 Here are the key features we have implemented:
 
-- [x] HTTP loader ([Libcurl](https://curl.se/libcurl/))
-- [x] HTML parser ([html5ever](https://github.com/servo/html5ever))
-- [x] DOM tree
-- [x] Javascript support ([v8](https://v8.dev/))
+- [x] HTTP loader (based on Libcurl)
+- [x] HTML parser and DOM tree (based on Netsurf libs)
+- [x] Javascript support (v8)
 - [x] DOM APIs
 - [x] Ajax
   - [x] XHR API
-  - [x] Fetch API
+  - [x] Fetch API (polyfill)
 - [x] DOM dump
 - [x] CDP/websockets server
 - [x] Click
@@ -182,12 +155,9 @@ Here are the key features we have implemented:
 - [x] Custom HTTP headers
 - [x] Proxy support
 - [x] Network interception
-- [x] Respect `robots.txt` with option `--obey_robots`
 
 NOTE: There are hundreds of Web APIs. Developing a browser (even just for headless mode) is a huge task. Coverage will increase over time.
 
-You can also follow the progress of our Javascript support in our dedicated [zig-js-runtime](https://github.com/lightpanda-io/zig-js-runtime#development) project.
-
 ## Build from sources
 
 ### Prerequisites
@@ -196,53 +166,84 @@ Lightpanda is written with [Zig](https://ziglang.org/) `0.15.2`. You have to
 install it with the right version in order to build the project.
 
 Lightpanda also depends on
-[zig-js-runtime](https://github.com/lightpanda-io/zig-js-runtime/) (with v8),
-[Libcurl](https://curl.se/libcurl/) and [html5ever](https://github.com/servo/html5ever).
+[zig-v8-fork](https://github.com/lightpanda-io/zig-v8-fork/),
+[Libcurl](https://curl.se/libcurl/),
+[Brotli](https://github.com/google/brotli),
+[Netsurf libs](https://www.netsurf-browser.org/) and
+[Mimalloc](https://microsoft.github.io/mimalloc).
 
-To be able to build the v8 engine for zig-js-runtime, you have to install some libs:
+To be able to build the v8 engine, you have to install some libs:
 
-For **Debian/Ubuntu based Linux**:
+For Debian/Ubuntu based Linux:
 
 ```
-sudo apt install xz-utils ca-certificates \
+sudo apt install xz-utils \
+    python3 ca-certificates git \
     pkg-config libglib2.0-dev \
-    clang make curl git
+    gperf libexpat1-dev unzip rsync \
+    cmake clang
 ```
-You also need to [install Rust](https://rust-lang.org/tools/install/).
 
-For systems with [**Nix**](https://nixos.org/download/), you can use the devShell:
+For systems with [Nix](https://nixos.org/download/), you can use the devShell:
 ```
 nix develop
 ```
 
-For **MacOS**, you need cmake and [Rust](https://rust-lang.org/tools/install/).
+For MacOS, you need [Xcode](https://developer.apple.com/xcode/) and the following pacakges from homebrew:
+
 ```
-brew install cmake
+brew install cmake pkgconf
 ```
 
-### Build and run
+### Install and build dependencies
 
-You an build the entire browser with `make build` or `make build-dev` for debug
-env.
+#### All in one build
 
-But you can directly use the zig command: `zig build run`.
+You can run `make install` to install deps all in one (or `make install-dev` if you need the development versions).
 
-#### Embed v8 snapshot
+Be aware that the build task is very long and cpu consuming, as you will build from sources all dependencies, including the v8 Javascript engine.
 
-Lighpanda uses v8 snapshot. By default, it is created on startup but you can
-embed it by using the following commands:
+#### Step by step build dependency
+
+The project uses git submodules for dependencies.
+
+To init or update the submodules in the `vendor/` directory:
 
-Generate the snapshot.
 ```
-zig build snapshot_creator -- src/snapshot.bin
+make install-submodule
 ```
 
-Build using the snapshot binary.
+**iconv**
+
+libiconv is an internationalization library used by Netsurf.
+
 ```
-zig build -Dsnapshot_path=../../snapshot.bin
+make install-libiconv
 ```
 
-See [#1279](https://github.com/lightpanda-io/browser/pull/1279) for more details.
+**Netsurf libs**
+
+Netsurf libs are used for HTML parsing and DOM tree generation.
+
+```
+make install-netsurf
+```
+
+For dev env, use `make install-netsurf-dev`.
+
+**Mimalloc**
+
+Mimalloc is used as a C memory allocator.
+
+```
+make install-mimalloc
+```
+
+For dev env, use `make install-mimalloc-dev`.
+
+Note: when Mimalloc is built in dev mode, you can dump memory stats with the
+env var `MIMALLOC_SHOW_STATS=1`. See
+[https://microsoft.github.io/mimalloc/environment.html](https://microsoft.github.io/mimalloc/environment.html).
 
 ## Test
 
@@ -269,75 +270,35 @@ make end2end
 Lightpanda is tested against the standardized [Web Platform
 Tests](https://web-platform-tests.org/).
 
-We use [a fork](https://github.com/lightpanda-io/wpt/tree/fork) including a custom
-[`testharnessreport.js`](https://github.com/lightpanda-io/wpt/commit/01a3115c076a3ad0c84849dbbf77a6e3d199c56f).
+The relevant tests cases are committed in a [dedicated repository](https://github.com/lightpanda-io/wpt) which is fetched by the `make install-submodule` command.
+
+All the tests cases executed are located in the `tests/wpt` sub-directory.
 
 For reference, you can easily execute a WPT test case with your browser via
 [wpt.live](https://wpt.live).
 
-#### Configure WPT HTTP server
-
-To run the test, you must clone the repository, configure the custom hosts and generate the
-`MANIFEST.json` file.
-
-Clone the repository with the `fork` branch.
-```
-git clone -b fork --depth=1 git@github.com:lightpanda-io/wpt.git
-```
-
-Enter into the `wpt/` dir.
-
-Install custom domains in your `/etc/hosts`
-```
-./wpt make-hosts-file | sudo tee -a /etc/hosts
-```
-
-Generate `MANIFEST.json`
-```
-./wpt manifest
-```
-Use the [WPT's setup
-guide](https://web-platform-tests.org/running-tests/from-local-system.html) for
-details.
-
 #### Run WPT test suite
 
-An external [Go](https://go.dev) runner is provided by
-[github.com/lightpanda-io/demo/](https://github.com/lightpanda-io/demo/)
-repository, located into `wptrunner/` dir.
-You need to clone the project first.
-
-First start the WPT's HTTP server from your `wpt/` clone dir.
-```
-./wpt serve
-```
-
-Run a Lightpanda browser
+To run all the tests:
 
 ```
-zig build run -- --insecure_disable_tls_host_verification
-```
-
-Then you can start the wptrunner from the Demo's clone dir:
-```
-cd wptrunner && go run .
+make wpt
 ```
 
 Or one specific test:
 
 ```
-cd wptrunner && go run . Node-childNodes.html
+make wpt Node-childNodes.html
 ```
 
-`wptrunner` command accepts `--summary` and `--json` options modifying output.
-Also `--concurrency` define the concurrency limit.
+#### Add a new WPT test case
 
-:warning: Running the whole test suite will take a long time. In this case,
-it's useful to build in `releaseFast` mode to make tests faster.
+We add new relevant tests cases files when we implemented changes in Lightpanda.
 
-```
-zig build -Doptimize=ReleaseFast run
-```
+To add a new test, copy the file you want from the [WPT
+repo](https://github.com/web-platform-tests/wpt) into the `tests/wpt` directory.
+
+:warning: Please keep the original directory tree structure of `tests/wpt`.
 
 ## Contributing
 

build.zig.zon

@@ -5,32 +5,18 @@
     .minimum_zig_version = "0.15.2",
     .dependencies = .{
         .v8 = .{
-            .url = "https://github.com/lightpanda-io/zig-v8-fork/archive/refs/tags/v0.3.1.tar.gz",
-            .hash = "v8-0.0.0-xddH64J7BAC81mkf6G9RbEJxS-W3TIRl5iFnShwbqCqy",
-
+            .url = "https://github.com/lightpanda-io/zig-v8-fork/archive/e047d2a4d5af5783763f0f6a652fab8982a08603.tar.gz",
+            .hash = "v8-0.0.0-xddH65gMBACRBQMM7EwmVgfi94FJyyX-0jpe5KhXYhfv",
         },
-        //.v8 = .{ .path = "../zig-v8-fork" },
-        .brotli = .{
-            // v1.2.0
-            .url = "https://github.com/google/brotli/archive/028fb5a23661f123017c060daa546b55cf4bde29.tar.gz",
-            .hash = "N-V-__8AAJudKgCQCuIiH6MJjAiIJHfg_tT_Ew-0vZwVkCo_",
-        },
-        .zlib = .{
-            .url = "https://github.com/madler/zlib/releases/download/v1.3.2/zlib-1.3.2.tar.gz",
-            .hash = "N-V-__8AAJ2cNgAgfBtAw33Bxfu1IWISDeKKSr3DAqoAysIJ",
-        },
-        .nghttp2 = .{
-            .url = "https://github.com/nghttp2/nghttp2/releases/download/v1.68.0/nghttp2-1.68.0.tar.gz",
-            .hash = "N-V-__8AAL15vQCI63ZL6Zaz5hJg6JTEgYXGbLnMFSnf7FT3",
+        // .v8 = .{ .path = "../zig-v8-fork" },
+        .@"ada-singleheader" = .{
+            .url = "https://github.com/ada-url/ada/releases/download/v3.3.0/singleheader.zip",
+            .hash = "N-V-__8AAPmhFAAw64ALjlzd5YMtzpSrmZ6KymsT84BKfB4s",
         },
         .@"boringssl-zig" = .{
             .url = "git+https://github.com/Syndica/boringssl-zig.git#c53df00d06b02b755ad88bbf4d1202ed9687b096",
             .hash = "boringssl-0.1.0-VtJeWehMAAA4RNnwRnzEvKcS9rjsR1QVRw1uJrwXxmVK",
         },
-        .curl = .{
-            .url = "https://github.com/curl/curl/releases/download/curl-8_18_0/curl-8.18.0.tar.gz",
-            .hash = "N-V-__8AALp9QAGn6CCHZ6fK_FfMyGtG824LSHYHHasM3w-y",
-        },
     },
     .paths = .{""},
 }

flake.lock

@@ -1,26 +1,5 @@
 {
   "nodes": {
-    "fenix": {
-      "inputs": {
-        "nixpkgs": [
-          "nixpkgs"
-        ],
-        "rust-analyzer-src": "rust-analyzer-src"
-      },
-      "locked": {
-        "lastModified": 1770708269,
-        "narHash": "sha256-OnZW86app7hHJJoB5lC9GNXY5QBBIESJB+sIdwEyld0=",
-        "owner": "nix-community",
-        "repo": "fenix",
-        "rev": "6b5325a017a9a9fe7e6252ccac3680cc7181cd63",
-        "type": "github"
-      },
-      "original": {
-        "owner": "nix-community",
-        "repo": "fenix",
-        "type": "github"
-      }
-    },
     "flake-compat": {
       "flake": false,
       "locked": {
@@ -96,11 +75,11 @@
     },
     "nixpkgs": {
       "locked": {
-        "lastModified": 1768649915,
-        "narHash": "sha256-jc21hKogFnxU7KXSVTRmxC7u5D4RHwm9BAvDf5/Z1Uo=",
+        "lastModified": 1760968520,
+        "narHash": "sha256-EjGslHDzCBKOVr+dnDB1CAD7wiQSHfUt3suOpFj9O1Q=",
         "owner": "nixos",
         "repo": "nixpkgs",
-        "rev": "3e3f3c7f9977dc123c23ee21e8085ed63daf8c37",
+        "rev": "e755547441a0413942a37692f7bf7fc6315bb7f6",
         "type": "github"
       },
       "original": {
@@ -112,30 +91,12 @@
     },
     "root": {
       "inputs": {
-        "fenix": "fenix",
         "flake-utils": "flake-utils",
         "nixpkgs": "nixpkgs",
         "zigPkgs": "zigPkgs",
         "zlsPkg": "zlsPkg"
       }
     },
-    "rust-analyzer-src": {
-      "flake": false,
-      "locked": {
-        "lastModified": 1770668050,
-        "narHash": "sha256-Q05yaIZtQrBKHpyWaPmyJmDRj0lojnVf8nUFE0vydcY=",
-        "owner": "rust-lang",
-        "repo": "rust-analyzer",
-        "rev": "9efc1f709f3c8134c3acac5d3592a8e4c184a0c6",
-        "type": "github"
-      },
-      "original": {
-        "owner": "rust-lang",
-        "ref": "nightly",
-        "repo": "rust-analyzer",
-        "type": "github"
-      }
-    },
     "systems": {
       "locked": {
         "lastModified": 1681028828,
@@ -175,11 +136,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1770598090,
-        "narHash": "sha256-k+82IDgTd9o5sxHIqGlvfwseKln3Ejx1edGtDltuPXo=",
+        "lastModified": 1760747435,
+        "narHash": "sha256-wNB/W3x+or4mdNxFPNOH5/WFckNpKgFRZk7OnOsLtm0=",
         "owner": "mitchellh",
         "repo": "zig-overlay",
-        "rev": "142495696982c88edddc8e17e4da90d8164acadf",
+        "rev": "d0f239b887b1ac736c0f3dde91bf5bf2ecf3a420",
         "type": "github"
       },
       "original": {

flake.nix

@@ -11,11 +11,6 @@
     zlsPkg.inputs.zig-overlay.follows = "zigPkgs";
     zlsPkg.inputs.nixpkgs.follows = "nixpkgs";
 
-    fenix = {
-      url = "github:nix-community/fenix";
-      inputs.nixpkgs.follows = "nixpkgs";
-    };
-
     flake-utils.url = "github:numtide/flake-utils";
   };
 
@@ -24,7 +19,6 @@
       nixpkgs,
       zigPkgs,
       zlsPkg,
-      fenix,
       flake-utils,
       ...
     }:
@@ -42,8 +36,6 @@
           inherit system overlays;
         };
 
-        rustToolchain = fenix.packages.${system}.stable.toolchain;
-
         # We need crtbeginS.o for building.
         crtFiles = pkgs.runCommand "crt-files" { } ''
           mkdir -p $out/lib
@@ -59,7 +51,6 @@
               # Build Tools
               zigpkgs."0.15.2"
               zls
-              rustToolchain
               python3
               pkg-config
               cmake
@@ -75,6 +66,7 @@
               glib.dev
               glibc.dev
               zlib
+              zlib.dev
             ];
         };
       in

src/App.zig

@@ -1,113 +0,0 @@
-// Copyright (C) 2023-2026  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-
-const Allocator = std.mem.Allocator;
-
-const log = @import("log.zig");
-const Config = @import("Config.zig");
-const Snapshot = @import("browser/js/Snapshot.zig");
-const Platform = @import("browser/js/Platform.zig");
-const Telemetry = @import("telemetry/telemetry.zig").Telemetry;
-const RobotStore = @import("browser/Robots.zig").RobotStore;
-
-pub const Http = @import("http/Http.zig");
-pub const ArenaPool = @import("ArenaPool.zig");
-
-const App = @This();
-
-http: Http,
-config: *const Config,
-platform: Platform,
-snapshot: Snapshot,
-telemetry: Telemetry,
-allocator: Allocator,
-arena_pool: ArenaPool,
-robots: RobotStore,
-app_dir_path: ?[]const u8,
-shutdown: bool = false,
-
-pub fn init(allocator: Allocator, config: *const Config) !*App {
-    const app = try allocator.create(App);
-    errdefer allocator.destroy(app);
-
-    app.config = config;
-    app.allocator = allocator;
-
-    app.robots = RobotStore.init(allocator);
-
-    app.http = try Http.init(allocator, &app.robots, config);
-    errdefer app.http.deinit();
-
-    app.platform = try Platform.init();
-    errdefer app.platform.deinit();
-
-    app.snapshot = try Snapshot.load();
-    errdefer app.snapshot.deinit();
-
-    app.app_dir_path = getAndMakeAppDir(allocator);
-
-    app.telemetry = try Telemetry.init(app, config.mode);
-    errdefer app.telemetry.deinit();
-
-    app.arena_pool = ArenaPool.init(allocator, 512, 1024 * 16);
-    errdefer app.arena_pool.deinit();
-
-    return app;
-}
-
-pub fn deinit(self: *App) void {
-    if (@atomicRmw(bool, &self.shutdown, .Xchg, true, .monotonic)) {
-        return;
-    }
-
-    const allocator = self.allocator;
-    if (self.app_dir_path) |app_dir_path| {
-        allocator.free(app_dir_path);
-        self.app_dir_path = null;
-    }
-    self.telemetry.deinit();
-    self.robots.deinit();
-    self.http.deinit();
-    self.snapshot.deinit();
-    self.platform.deinit();
-    self.arena_pool.deinit();
-
-    allocator.destroy(self);
-}
-
-fn getAndMakeAppDir(allocator: Allocator) ?[]const u8 {
-    if (@import("builtin").is_test) {
-        return allocator.dupe(u8, "/tmp") catch unreachable;
-    }
-    const app_dir_path = std.fs.getAppDataDir(allocator, "lightpanda") catch |err| {
-        log.warn(.app, "get data dir", .{ .err = err });
-        return null;
-    };
-
-    std.fs.cwd().makePath(app_dir_path) catch |err| switch (err) {
-        error.PathAlreadyExists => return app_dir_path,
-        else => {
-            allocator.free(app_dir_path);
-            log.warn(.app, "create data dir", .{ .err = err, .path = app_dir_path });
-            return null;
-        },
-    };
-    return app_dir_path;
-}

src/ArenaPool.zig

@@ -1,212 +0,0 @@
-// Copyright (C) 2023-2026  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-
-const Allocator = std.mem.Allocator;
-const ArenaAllocator = std.heap.ArenaAllocator;
-
-const ArenaPool = @This();
-
-allocator: Allocator,
-retain_bytes: usize,
-free_list_len: u16 = 0,
-free_list: ?*Entry = null,
-free_list_max: u16,
-entry_pool: std.heap.MemoryPool(Entry),
-mutex: std.Thread.Mutex = .{},
-
-const Entry = struct {
-    next: ?*Entry,
-    arena: ArenaAllocator,
-};
-
-pub fn init(allocator: Allocator, free_list_max: u16, retain_bytes: usize) ArenaPool {
-    return .{
-        .allocator = allocator,
-        .free_list_max = free_list_max,
-        .retain_bytes = retain_bytes,
-        .entry_pool = .init(allocator),
-    };
-}
-
-pub fn deinit(self: *ArenaPool) void {
-    var entry = self.free_list;
-    while (entry) |e| {
-        entry = e.next;
-        e.arena.deinit();
-    }
-    self.entry_pool.deinit();
-}
-
-pub fn acquire(self: *ArenaPool) !Allocator {
-    self.mutex.lock();
-    defer self.mutex.unlock();
-
-    if (self.free_list) |entry| {
-        self.free_list = entry.next;
-        self.free_list_len -= 1;
-        return entry.arena.allocator();
-    }
-
-    const entry = try self.entry_pool.create();
-    entry.* = .{
-        .next = null,
-        .arena = ArenaAllocator.init(self.allocator),
-    };
-
-    return entry.arena.allocator();
-}
-
-pub fn release(self: *ArenaPool, allocator: Allocator) void {
-    const arena: *std.heap.ArenaAllocator = @ptrCast(@alignCast(allocator.ptr));
-    const entry: *Entry = @fieldParentPtr("arena", arena);
-
-    // Reset the arena before acquiring the lock to minimize lock hold time
-    _ = arena.reset(.{ .retain_with_limit = self.retain_bytes });
-
-    self.mutex.lock();
-    defer self.mutex.unlock();
-
-    const free_list_len = self.free_list_len;
-    if (free_list_len == self.free_list_max) {
-        arena.deinit();
-        self.entry_pool.destroy(entry);
-        return;
-    }
-
-    entry.next = self.free_list;
-    self.free_list_len = free_list_len + 1;
-    self.free_list = entry;
-}
-
-pub fn reset(_: *const ArenaPool, allocator: Allocator, retain: usize) void {
-    const arena: *std.heap.ArenaAllocator = @ptrCast(@alignCast(allocator.ptr));
-    _ = arena.reset(.{ .retain_with_limit = retain });
-}
-
-const testing = std.testing;
-
-test "arena pool - basic acquire and use" {
-    var pool = ArenaPool.init(testing.allocator, 512, 1024 * 16);
-    defer pool.deinit();
-
-    const alloc = try pool.acquire();
-    const buf = try alloc.alloc(u8, 64);
-    @memset(buf, 0xAB);
-    try testing.expectEqual(@as(u8, 0xAB), buf[0]);
-
-    pool.release(alloc);
-}
-
-test "arena pool - reuse entry after release" {
-    var pool = ArenaPool.init(testing.allocator, 512, 1024 * 16);
-    defer pool.deinit();
-
-    const alloc1 = try pool.acquire();
-    try testing.expectEqual(@as(u16, 0), pool.free_list_len);
-
-    pool.release(alloc1);
-    try testing.expectEqual(@as(u16, 1), pool.free_list_len);
-
-    // The same entry should be returned from the free list.
-    const alloc2 = try pool.acquire();
-    try testing.expectEqual(@as(u16, 0), pool.free_list_len);
-    try testing.expectEqual(alloc1.ptr, alloc2.ptr);
-
-    pool.release(alloc2);
-}
-
-test "arena pool - multiple concurrent arenas" {
-    var pool = ArenaPool.init(testing.allocator, 512, 1024 * 16);
-    defer pool.deinit();
-
-    const a1 = try pool.acquire();
-    const a2 = try pool.acquire();
-    const a3 = try pool.acquire();
-
-    // All three must be distinct arenas.
-    try testing.expect(a1.ptr != a2.ptr);
-    try testing.expect(a2.ptr != a3.ptr);
-    try testing.expect(a1.ptr != a3.ptr);
-
-    _ = try a1.alloc(u8, 16);
-    _ = try a2.alloc(u8, 32);
-    _ = try a3.alloc(u8, 48);
-
-    pool.release(a1);
-    pool.release(a2);
-    pool.release(a3);
-
-    try testing.expectEqual(@as(u16, 3), pool.free_list_len);
-}
-
-test "arena pool - free list respects max limit" {
-    // Cap the free list at 1 so the second release discards its arena.
-    var pool = ArenaPool.init(testing.allocator, 1, 1024 * 16);
-    defer pool.deinit();
-
-    const a1 = try pool.acquire();
-    const a2 = try pool.acquire();
-
-    pool.release(a1);
-    try testing.expectEqual(@as(u16, 1), pool.free_list_len);
-
-    // The free list is full; a2's arena should be destroyed, not queued.
-    pool.release(a2);
-    try testing.expectEqual(@as(u16, 1), pool.free_list_len);
-}
-
-test "arena pool - reset clears memory without releasing" {
-    var pool = ArenaPool.init(testing.allocator, 512, 1024 * 16);
-    defer pool.deinit();
-
-    const alloc = try pool.acquire();
-
-    const buf = try alloc.alloc(u8, 128);
-    @memset(buf, 0xFF);
-
-    // reset() frees arena memory but keeps the allocator in-flight.
-    pool.reset(alloc, 0);
-
-    // The free list must stay empty; the allocator was not released.
-    try testing.expectEqual(@as(u16, 0), pool.free_list_len);
-
-    // Allocating again through the same arena must still work.
-    const buf2 = try alloc.alloc(u8, 64);
-    @memset(buf2, 0x00);
-    try testing.expectEqual(@as(u8, 0x00), buf2[0]);
-
-    pool.release(alloc);
-}
-
-test "arena pool - deinit with entries in free list" {
-    // Verifies that deinit properly cleans up free-listed arenas (no leaks
-    // detected by the test allocator).
-    var pool = ArenaPool.init(testing.allocator, 512, 1024 * 16);
-
-    const a1 = try pool.acquire();
-    const a2 = try pool.acquire();
-    _ = try a1.alloc(u8, 256);
-    _ = try a2.alloc(u8, 512);
-    pool.release(a1);
-    pool.release(a2);
-    try testing.expectEqual(@as(u16, 2), pool.free_list_len);
-
-    pool.deinit();
-}

src/Config.zig

@@ -1,849 +0,0 @@
-// Copyright (C) 2023-2026  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-const builtin = @import("builtin");
-const Allocator = std.mem.Allocator;
-
-const log = @import("log.zig");
-const dump = @import("browser/dump.zig");
-
-pub const RunMode = enum {
-    help,
-    fetch,
-    serve,
-    version,
-    mcp,
-};
-
-pub const CDP_MAX_HTTP_REQUEST_SIZE = 4096;
-
-// max message size
-// +14 for max websocket payload overhead
-// +140 for the max control packet that might be interleaved in a message
-pub const CDP_MAX_MESSAGE_SIZE = 512 * 1024 + 14 + 140;
-
-mode: Mode,
-exec_name: []const u8,
-http_headers: HttpHeaders,
-
-const Config = @This();
-
-pub fn init(allocator: Allocator, exec_name: []const u8, mode: Mode) !Config {
-    var config = Config{
-        .mode = mode,
-        .exec_name = exec_name,
-        .http_headers = undefined,
-    };
-    config.http_headers = try HttpHeaders.init(allocator, &config);
-    return config;
-}
-
-pub fn deinit(self: *const Config, allocator: Allocator) void {
-    self.http_headers.deinit(allocator);
-}
-
-pub fn tlsVerifyHost(self: *const Config) bool {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.tls_verify_host,
-        else => unreachable,
-    };
-}
-
-pub fn obeyRobots(self: *const Config) bool {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.obey_robots,
-        else => unreachable,
-    };
-}
-
-pub fn httpProxy(self: *const Config) ?[:0]const u8 {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.http_proxy,
-        else => unreachable,
-    };
-}
-
-pub fn proxyBearerToken(self: *const Config) ?[:0]const u8 {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.proxy_bearer_token,
-        .help, .version => null,
-    };
-}
-
-pub fn httpMaxConcurrent(self: *const Config) u8 {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.http_max_concurrent orelse 10,
-        else => unreachable,
-    };
-}
-
-pub fn httpMaxHostOpen(self: *const Config) u8 {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.http_max_host_open orelse 4,
-        else => unreachable,
-    };
-}
-
-pub fn httpConnectTimeout(self: *const Config) u31 {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.http_connect_timeout orelse 0,
-        else => unreachable,
-    };
-}
-
-pub fn httpTimeout(self: *const Config) u31 {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.http_timeout orelse 5000,
-        else => unreachable,
-    };
-}
-
-pub fn httpMaxRedirects(_: *const Config) u8 {
-    return 10;
-}
-
-pub fn httpMaxResponseSize(self: *const Config) ?usize {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.http_max_response_size,
-        else => unreachable,
-    };
-}
-
-pub fn logLevel(self: *const Config) ?log.Level {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.log_level,
-        else => unreachable,
-    };
-}
-
-pub fn logFormat(self: *const Config) ?log.Format {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.log_format,
-        else => unreachable,
-    };
-}
-
-pub fn logFilterScopes(self: *const Config) ?[]const log.Scope {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.log_filter_scopes,
-        else => unreachable,
-    };
-}
-
-pub fn userAgentSuffix(self: *const Config) ?[]const u8 {
-    return switch (self.mode) {
-        inline .serve, .fetch, .mcp => |opts| opts.common.user_agent_suffix,
-        .help, .version => null,
-    };
-}
-
-pub fn maxConnections(self: *const Config) u16 {
-    return switch (self.mode) {
-        .serve => |opts| opts.cdp_max_connections,
-        else => unreachable,
-    };
-}
-
-pub fn maxPendingConnections(self: *const Config) u31 {
-    return switch (self.mode) {
-        .serve => |opts| opts.cdp_max_pending_connections,
-        else => unreachable,
-    };
-}
-
-pub const Mode = union(RunMode) {
-    help: bool, // false when being printed because of an error
-    fetch: Fetch,
-    serve: Serve,
-    version: void,
-    mcp: Mcp,
-};
-
-pub const Serve = struct {
-    host: []const u8 = "127.0.0.1",
-    port: u16 = 9222,
-    timeout: u31 = 10,
-    cdp_max_connections: u16 = 16,
-    cdp_max_pending_connections: u16 = 128,
-    common: Common = .{},
-};
-
-pub const Mcp = struct {
-    common: Common = .{},
-};
-
-pub const DumpFormat = enum {
-    html,
-    markdown,
-    wpt,
-};
-
-pub const Fetch = struct {
-    url: [:0]const u8,
-    dump_mode: ?DumpFormat = null,
-    common: Common = .{},
-    with_base: bool = false,
-    with_frames: bool = false,
-    strip: dump.Opts.Strip = .{},
-};
-
-pub const Common = struct {
-    obey_robots: bool = false,
-    proxy_bearer_token: ?[:0]const u8 = null,
-    http_proxy: ?[:0]const u8 = null,
-    http_max_concurrent: ?u8 = null,
-    http_max_host_open: ?u8 = null,
-    http_timeout: ?u31 = null,
-    http_connect_timeout: ?u31 = null,
-    http_max_response_size: ?usize = null,
-    tls_verify_host: bool = true,
-    log_level: ?log.Level = null,
-    log_format: ?log.Format = null,
-    log_filter_scopes: ?[]log.Scope = null,
-    user_agent_suffix: ?[]const u8 = null,
-};
-
-/// Pre-formatted HTTP headers for reuse across Http and Client.
-/// Must be initialized with an allocator that outlives all HTTP connections.
-pub const HttpHeaders = struct {
-    const user_agent_base: [:0]const u8 = "Lightpanda/1.0";
-
-    user_agent: [:0]const u8, // User agent value (e.g. "Lightpanda/1.0")
-    user_agent_header: [:0]const u8,
-
-    proxy_bearer_header: ?[:0]const u8,
-
-    pub fn init(allocator: Allocator, config: *const Config) !HttpHeaders {
-        const user_agent: [:0]const u8 = if (config.userAgentSuffix()) |suffix|
-            try std.fmt.allocPrintSentinel(allocator, "{s} {s}", .{ user_agent_base, suffix }, 0)
-        else
-            user_agent_base;
-        errdefer if (config.userAgentSuffix() != null) allocator.free(user_agent);
-
-        const user_agent_header = try std.fmt.allocPrintSentinel(allocator, "User-Agent: {s}", .{user_agent}, 0);
-        errdefer allocator.free(user_agent_header);
-
-        const proxy_bearer_header: ?[:0]const u8 = if (config.proxyBearerToken()) |token|
-            try std.fmt.allocPrintSentinel(allocator, "Proxy-Authorization: Bearer {s}", .{token}, 0)
-        else
-            null;
-
-        return .{
-            .user_agent = user_agent,
-            .user_agent_header = user_agent_header,
-            .proxy_bearer_header = proxy_bearer_header,
-        };
-    }
-
-    pub fn deinit(self: *const HttpHeaders, allocator: Allocator) void {
-        if (self.proxy_bearer_header) |hdr| {
-            allocator.free(hdr);
-        }
-        allocator.free(self.user_agent_header);
-        if (self.user_agent.ptr != user_agent_base.ptr) {
-            allocator.free(self.user_agent);
-        }
-    }
-};
-
-pub fn printUsageAndExit(self: *const Config, success: bool) void {
-    //                                                                     MAX_HELP_LEN|
-    const common_options =
-        \\
-        \\--insecure_disable_tls_host_verification
-        \\                Disables host verification on all HTTP requests. This is an
-        \\                advanced option which should only be set if you understand
-        \\                and accept the risk of disabling host verification.
-        \\
-        \\--obey_robots
-        \\                Fetches and obeys the robots.txt (if available) of the web pages
-        \\                we make requests towards.
-        \\                Defaults to false.
-        \\
-        \\--http_proxy    The HTTP proxy to use for all HTTP requests.
-        \\                A username:password can be included for basic authentication.
-        \\                Defaults to none.
-        \\
-        \\--proxy_bearer_token
-        \\                The <token> to send for bearer authentication with the proxy
-        \\                Proxy-Authorization: Bearer <token>
-        \\
-        \\--http_max_concurrent
-        \\                The maximum number of concurrent HTTP requests.
-        \\                Defaults to 10.
-        \\
-        \\--http_max_host_open
-        \\                The maximum number of open connection to a given host:port.
-        \\                Defaults to 4.
-        \\
-        \\--http_connect_timeout
-        \\                The time, in milliseconds, for establishing an HTTP connection
-        \\                before timing out. 0 means it never times out.
-        \\                Defaults to 0.
-        \\
-        \\--http_timeout
-        \\                The maximum time, in milliseconds, the transfer is allowed
-        \\                to complete. 0 means it never times out.
-        \\                Defaults to 10000.
-        \\
-        \\--http_max_response_size
-        \\                Limits the acceptable response size for any request
-        \\                (e.g. XHR, fetch, script loading, ...).
-        \\                Defaults to no limit.
-        \\
-        \\--log_level     The log level: debug, info, warn, error or fatal.
-        \\                Defaults to
-    ++ (if (builtin.mode == .Debug) " info." else "warn.") ++
-        \\
-        \\
-        \\--log_format    The log format: pretty or logfmt.
-        \\                Defaults to
-    ++ (if (builtin.mode == .Debug) " pretty." else " logfmt.") ++
-        \\
-        \\
-        \\--log_filter_scopes
-        \\                Filter out too verbose logs per scope:
-        \\                http, unknown_prop, event, ...
-        \\
-        \\--user_agent_suffix
-        \\                Suffix to append to the Lightpanda/X.Y User-Agent
-        \\
-    ;
-
-    //                                                                     MAX_HELP_LEN|
-    const usage =
-        \\usage: {s} command [options] [URL]
-        \\
-        \\Command can be either 'fetch', 'serve', 'mcp' or 'help'
-        \\
-        \\fetch command
-        \\Fetches the specified URL
-        \\Example: {s} fetch --dump html https://lightpanda.io/
-        \\
-        \\Options:
-        \\--dump          Dumps document to stdout.
-        \\                Argument must be 'html' or 'markdown'.
-        \\                Defaults to no dump.
-        \\
-        \\--strip_mode    Comma separated list of tag groups to remove from dump
-        \\                the dump. e.g. --strip_mode js,css
-        \\                  - "js" script and link[as=script, rel=preload]
-        \\                  - "ui" includes img, picture, video, css and svg
-        \\                  - "css" includes style and link[rel=stylesheet]
-        \\                  - "full" includes js, ui and css
-        \\
-        \\--with_base     Add a <base> tag in dump. Defaults to false.
-        \\
-        \\--with_frames   Includes the contents of iframes. Defaults to false.
-        \\
-    ++ common_options ++
-        \\
-        \\serve command
-        \\Starts a websocket CDP server
-        \\Example: {s} serve --host 127.0.0.1 --port 9222
-        \\
-        \\Options:
-        \\--host          Host of the CDP server
-        \\                Defaults to "127.0.0.1"
-        \\
-        \\--port          Port of the CDP server
-        \\                Defaults to 9222
-        \\
-        \\--timeout       Inactivity timeout in seconds before disconnecting clients
-        \\                Defaults to 10 (seconds). Limited to 604800 (1 week).
-        \\
-        \\--cdp_max_connections
-        \\                Maximum number of simultaneous CDP connections.
-        \\                Defaults to 16.
-        \\
-        \\--cdp_max_pending_connections
-        \\                Maximum pending connections in the accept queue.
-        \\                Defaults to 128.
-        \\
-    ++ common_options ++
-        \\
-        \\mcp command
-        \\Starts an MCP (Model Context Protocol) server over stdio
-        \\Example: {s} mcp
-        \\
-    ++ common_options ++
-        \\
-        \\version command
-        \\Displays the version of {s}
-        \\
-        \\help command
-        \\Displays this message
-        \\
-    ;
-    std.debug.print(usage, .{ self.exec_name, self.exec_name, self.exec_name, self.exec_name, self.exec_name });
-    if (success) {
-        return std.process.cleanExit();
-    }
-    std.process.exit(1);
-}
-
-pub fn parseArgs(allocator: Allocator) !Config {
-    var args = try std.process.argsWithAllocator(allocator);
-    defer args.deinit();
-
-    const exec_name = try allocator.dupe(u8, std.fs.path.basename(args.next().?));
-
-    const mode_string = args.next() orelse "";
-    const run_mode = std.meta.stringToEnum(RunMode, mode_string) orelse blk: {
-        const inferred_mode = inferMode(mode_string) orelse
-            return init(allocator, exec_name, .{ .help = false });
-        // "command" wasn't a command but an option. We can't reset args, but
-        // we can create a new one. Not great, but this fallback is temporary
-        // as we transition to this command mode approach.
-        args.deinit();
-
-        args = try std.process.argsWithAllocator(allocator);
-        // skip the exec_name
-        _ = args.skip();
-
-        break :blk inferred_mode;
-    };
-
-    const mode: Mode = switch (run_mode) {
-        .help => .{ .help = true },
-        .serve => .{ .serve = parseServeArgs(allocator, &args) catch
-            return init(allocator, exec_name, .{ .help = false }) },
-        .fetch => .{ .fetch = parseFetchArgs(allocator, &args) catch
-            return init(allocator, exec_name, .{ .help = false }) },
-        .mcp => .{ .mcp = parseMcpArgs(allocator, &args) catch
-            return init(allocator, exec_name, .{ .help = false }) },
-        .version => .{ .version = {} },
-    };
-    return init(allocator, exec_name, mode);
-}
-
-fn inferMode(opt: []const u8) ?RunMode {
-    if (opt.len == 0) {
-        return .serve;
-    }
-
-    if (std.mem.startsWith(u8, opt, "--") == false) {
-        return .fetch;
-    }
-
-    if (std.mem.eql(u8, opt, "--dump")) {
-        return .fetch;
-    }
-
-    if (std.mem.eql(u8, opt, "--noscript")) {
-        return .fetch;
-    }
-
-    if (std.mem.eql(u8, opt, "--strip_mode")) {
-        return .fetch;
-    }
-
-    if (std.mem.eql(u8, opt, "--with_base")) {
-        return .fetch;
-    }
-
-    if (std.mem.eql(u8, opt, "--with_frames")) {
-        return .fetch;
-    }
-
-    if (std.mem.eql(u8, opt, "--host")) {
-        return .serve;
-    }
-
-    if (std.mem.eql(u8, opt, "--port")) {
-        return .serve;
-    }
-
-    if (std.mem.eql(u8, opt, "--timeout")) {
-        return .serve;
-    }
-
-    return null;
-}
-
-fn parseServeArgs(
-    allocator: Allocator,
-    args: *std.process.ArgIterator,
-) !Serve {
-    var serve: Serve = .{};
-
-    while (args.next()) |opt| {
-        if (std.mem.eql(u8, "--host", opt)) {
-            const str = args.next() orelse {
-                log.fatal(.app, "missing argument value", .{ .arg = "--host" });
-                return error.InvalidArgument;
-            };
-            serve.host = try allocator.dupe(u8, str);
-            continue;
-        }
-
-        if (std.mem.eql(u8, "--port", opt)) {
-            const str = args.next() orelse {
-                log.fatal(.app, "missing argument value", .{ .arg = "--port" });
-                return error.InvalidArgument;
-            };
-
-            serve.port = std.fmt.parseInt(u16, str, 10) catch |err| {
-                log.fatal(.app, "invalid argument value", .{ .arg = "--port", .err = err });
-                return error.InvalidArgument;
-            };
-            continue;
-        }
-
-        if (std.mem.eql(u8, "--timeout", opt)) {
-            const str = args.next() orelse {
-                log.fatal(.app, "missing argument value", .{ .arg = "--timeout" });
-                return error.InvalidArgument;
-            };
-
-            serve.timeout = std.fmt.parseInt(u31, str, 10) catch |err| {
-                log.fatal(.app, "invalid argument value", .{ .arg = "--timeout", .err = err });
-                return error.InvalidArgument;
-            };
-            continue;
-        }
-
-        if (std.mem.eql(u8, "--cdp_max_connections", opt)) {
-            const str = args.next() orelse {
-                log.fatal(.app, "missing argument value", .{ .arg = "--cdp_max_connections" });
-                return error.InvalidArgument;
-            };
-
-            serve.cdp_max_connections = std.fmt.parseInt(u16, str, 10) catch |err| {
-                log.fatal(.app, "invalid argument value", .{ .arg = "--cdp_max_connections", .err = err });
-                return error.InvalidArgument;
-            };
-            continue;
-        }
-
-        if (std.mem.eql(u8, "--cdp_max_pending_connections", opt)) {
-            const str = args.next() orelse {
-                log.fatal(.app, "missing argument value", .{ .arg = "--cdp_max_pending_connections" });
-                return error.InvalidArgument;
-            };
-
-            serve.cdp_max_pending_connections = std.fmt.parseInt(u16, str, 10) catch |err| {
-                log.fatal(.app, "invalid argument value", .{ .arg = "--cdp_max_pending_connections", .err = err });
-                return error.InvalidArgument;
-            };
-            continue;
-        }
-
-        if (try parseCommonArg(allocator, opt, args, &serve.common)) {
-            continue;
-        }
-
-        log.fatal(.app, "unknown argument", .{ .mode = "serve", .arg = opt });
-        return error.UnkownOption;
-    }
-
-    return serve;
-}
-
-fn parseMcpArgs(
-    allocator: Allocator,
-    args: *std.process.ArgIterator,
-) !Mcp {
-    var mcp: Mcp = .{};
-
-    while (args.next()) |opt| {
-        if (try parseCommonArg(allocator, opt, args, &mcp.common)) {
-            continue;
-        }
-
-        log.fatal(.mcp, "unknown argument", .{ .mode = "mcp", .arg = opt });
-        return error.UnkownOption;
-    }
-
-    return mcp;
-}
-
-fn parseFetchArgs(
-    allocator: Allocator,
-    args: *std.process.ArgIterator,
-) !Fetch {
-    var dump_mode: ?DumpFormat = null;
-    var with_base: bool = false;
-    var with_frames: bool = false;
-    var url: ?[:0]const u8 = null;
-    var common: Common = .{};
-    var strip: dump.Opts.Strip = .{};
-
-    while (args.next()) |opt| {
-        if (std.mem.eql(u8, "--dump", opt)) {
-            var peek_args = args.*;
-            if (peek_args.next()) |next_arg| {
-                if (std.meta.stringToEnum(DumpFormat, next_arg)) |mode| {
-                    dump_mode = mode;
-                    _ = args.next();
-                } else {
-                    dump_mode = .html;
-                }
-            } else {
-                dump_mode = .html;
-            }
-            continue;
-        }
-
-        if (std.mem.eql(u8, "--noscript", opt)) {
-            log.warn(.app, "deprecation warning", .{
-                .feature = "--noscript argument",
-                .hint = "use '--strip_mode js' instead",
-            });
-            strip.js = true;
-            continue;
-        }
-
-        if (std.mem.eql(u8, "--with_base", opt)) {
-            with_base = true;
-            continue;
-        }
-
-        if (std.mem.eql(u8, "--with_frames", opt)) {
-            with_frames = true;
-            continue;
-        }
-
-        if (std.mem.eql(u8, "--strip_mode", opt)) {
-            const str = args.next() orelse {
-                log.fatal(.app, "missing argument value", .{ .arg = "--strip_mode" });
-                return error.InvalidArgument;
-            };
-
-            var it = std.mem.splitScalar(u8, str, ',');
-            while (it.next()) |part| {
-                const trimmed = std.mem.trim(u8, part, &std.ascii.whitespace);
-                if (std.mem.eql(u8, trimmed, "js")) {
-                    strip.js = true;
-                } else if (std.mem.eql(u8, trimmed, "ui")) {
-                    strip.ui = true;
-                } else if (std.mem.eql(u8, trimmed, "css")) {
-                    strip.css = true;
-                } else if (std.mem.eql(u8, trimmed, "full")) {
-                    strip.js = true;
-                    strip.ui = true;
-                    strip.css = true;
-                } else {
-                    log.fatal(.app, "invalid option choice", .{ .arg = "--strip_mode", .value = trimmed });
-                }
-            }
-            continue;
-        }
-
-        if (try parseCommonArg(allocator, opt, args, &common)) {
-            continue;
-        }
-
-        if (std.mem.startsWith(u8, opt, "--")) {
-            log.fatal(.app, "unknown argument", .{ .mode = "fetch", .arg = opt });
-            return error.UnkownOption;
-        }
-
-        if (url != null) {
-            log.fatal(.app, "duplicate fetch url", .{ .help = "only 1 URL can be specified" });
-            return error.TooManyURLs;
-        }
-        url = try allocator.dupeZ(u8, opt);
-    }
-
-    if (url == null) {
-        log.fatal(.app, "missing fetch url", .{ .help = "URL to fetch must be provided" });
-        return error.MissingURL;
-    }
-
-    return .{
-        .url = url.?,
-        .dump_mode = dump_mode,
-        .strip = strip,
-        .common = common,
-        .with_base = with_base,
-        .with_frames = with_frames,
-    };
-}
-
-fn parseCommonArg(
-    allocator: Allocator,
-    opt: []const u8,
-    args: *std.process.ArgIterator,
-    common: *Common,
-) !bool {
-    if (std.mem.eql(u8, "--insecure_disable_tls_host_verification", opt)) {
-        common.tls_verify_host = false;
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--obey_robots", opt)) {
-        common.obey_robots = true;
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--http_proxy", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--http_proxy" });
-            return error.InvalidArgument;
-        };
-        common.http_proxy = try allocator.dupeZ(u8, str);
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--proxy_bearer_token", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--proxy_bearer_token" });
-            return error.InvalidArgument;
-        };
-        common.proxy_bearer_token = try allocator.dupeZ(u8, str);
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--http_max_concurrent", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--http_max_concurrent" });
-            return error.InvalidArgument;
-        };
-
-        common.http_max_concurrent = std.fmt.parseInt(u8, str, 10) catch |err| {
-            log.fatal(.app, "invalid argument value", .{ .arg = "--http_max_concurrent", .err = err });
-            return error.InvalidArgument;
-        };
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--http_max_host_open", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--http_max_host_open" });
-            return error.InvalidArgument;
-        };
-
-        common.http_max_host_open = std.fmt.parseInt(u8, str, 10) catch |err| {
-            log.fatal(.app, "invalid argument value", .{ .arg = "--http_max_host_open", .err = err });
-            return error.InvalidArgument;
-        };
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--http_connect_timeout", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--http_connect_timeout" });
-            return error.InvalidArgument;
-        };
-
-        common.http_connect_timeout = std.fmt.parseInt(u31, str, 10) catch |err| {
-            log.fatal(.app, "invalid argument value", .{ .arg = "--http_connect_timeout", .err = err });
-            return error.InvalidArgument;
-        };
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--http_timeout", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--http_timeout" });
-            return error.InvalidArgument;
-        };
-
-        common.http_timeout = std.fmt.parseInt(u31, str, 10) catch |err| {
-            log.fatal(.app, "invalid argument value", .{ .arg = "--http_timeout", .err = err });
-            return error.InvalidArgument;
-        };
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--http_max_response_size", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--http_max_response_size" });
-            return error.InvalidArgument;
-        };
-
-        common.http_max_response_size = std.fmt.parseInt(usize, str, 10) catch |err| {
-            log.fatal(.app, "invalid argument value", .{ .arg = "--http_max_response_size", .err = err });
-            return error.InvalidArgument;
-        };
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--log_level", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--log_level" });
-            return error.InvalidArgument;
-        };
-
-        common.log_level = std.meta.stringToEnum(log.Level, str) orelse blk: {
-            if (std.mem.eql(u8, str, "error")) {
-                break :blk .err;
-            }
-            log.fatal(.app, "invalid option choice", .{ .arg = "--log_level", .value = str });
-            return error.InvalidArgument;
-        };
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--log_format", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--log_format" });
-            return error.InvalidArgument;
-        };
-
-        common.log_format = std.meta.stringToEnum(log.Format, str) orelse {
-            log.fatal(.app, "invalid option choice", .{ .arg = "--log_format", .value = str });
-            return error.InvalidArgument;
-        };
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--log_filter_scopes", opt)) {
-        if (builtin.mode != .Debug) {
-            log.fatal(.app, "experimental", .{ .help = "log scope filtering is only available in debug builds" });
-            return false;
-        }
-
-        const str = args.next() orelse {
-            // disables the default filters
-            common.log_filter_scopes = &.{};
-            return true;
-        };
-
-        var arr: std.ArrayList(log.Scope) = .empty;
-
-        var it = std.mem.splitScalar(u8, str, ',');
-        while (it.next()) |part| {
-            try arr.append(allocator, std.meta.stringToEnum(log.Scope, part) orelse {
-                log.fatal(.app, "invalid option choice", .{ .arg = "--log_filter_scopes", .value = part });
-                return false;
-            });
-        }
-        common.log_filter_scopes = arr.items;
-        return true;
-    }
-
-    if (std.mem.eql(u8, "--user_agent_suffix", opt)) {
-        const str = args.next() orelse {
-            log.fatal(.app, "missing argument value", .{ .arg = "--user_agent_suffix" });
-            return error.InvalidArgument;
-        };
-        for (str) |c| {
-            if (!std.ascii.isPrint(c)) {
-                log.fatal(.app, "not printable character", .{ .arg = "--user_agent_suffix" });
-                return error.InvalidArgument;
-            }
-        }
-        common.user_agent_suffix = try allocator.dupe(u8, str);
-        return true;
-    }
-
-    return false;
-}

src/Notification.zig

@@ -1,419 +0,0 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-const lp = @import("lightpanda");
-
-const log = @import("log.zig");
-const Page = @import("browser/Page.zig");
-const Transfer = @import("http/Client.zig").Transfer;
-
-const Allocator = std.mem.Allocator;
-
-const List = std.DoublyLinkedList;
-
-// Allows code to register for and emit events.
-// Keeps two lists
-// 1 - for a given event type, a linked list of all the listeners
-// 2 - for a given listener, a list of all it's registration
-// The 2nd one is so that a listener can unregister all of it's listeners
-// (there's currently no need for a listener to unregister only 1 or more
-// specific listener).
-//
-// Scoping is important. Imagine we created a global singleton registry, and our
-// CDP code registers for the "network_bytes_sent" event, because it needs to
-// send messages to the client when this happens. Our HTTP client could then
-// emit a "network_bytes_sent" message. It would be easy, and it would work.
-// That is, it would work until multiple CDP clients connect, and because
-// everything's just one big global, events from one CDP session would be sent
-// to all CDP clients.
-//
-// To avoid this, one way or another, we need scoping. We could still have
-// a global registry but every "register" and every "emit" has some type of
-// "scope". This would have a run-time cost and still require some coordination
-// between components to share a common scope.
-//
-// Instead, the approach that we take is to have a notification instance per
-// CDP connection (BrowserContext). Each CDP connection has its own notification
-// that is shared across all Sessions (tabs) within that connection. This ensures
-// proper isolation between different CDP clients while allowing a single client
-// to receive events from all its tabs.
-const Notification = @This();
-// Every event type (which are hard-coded), has a list of Listeners.
-// When the event happens, we dispatch to those listener.
-event_listeners: EventListeners,
-
-// list of listeners for a specified receiver
-// @intFromPtr(receiver) -> [listener1, listener2, ...]
-// Used when `unregisterAll` is called.
-listeners: std.AutoHashMapUnmanaged(usize, std.ArrayList(*Listener)),
-
-allocator: Allocator,
-mem_pool: std.heap.MemoryPool(Listener),
-
-const EventListeners = struct {
-    page_remove: List = .{},
-    page_created: List = .{},
-    page_navigate: List = .{},
-    page_navigated: List = .{},
-    page_network_idle: List = .{},
-    page_network_almost_idle: List = .{},
-    page_frame_created: List = .{},
-    http_request_fail: List = .{},
-    http_request_start: List = .{},
-    http_request_intercept: List = .{},
-    http_request_done: List = .{},
-    http_request_auth_required: List = .{},
-    http_response_data: List = .{},
-    http_response_header_done: List = .{},
-};
-
-const Events = union(enum) {
-    page_remove: PageRemove,
-    page_created: *Page,
-    page_navigate: *const PageNavigate,
-    page_navigated: *const PageNavigated,
-    page_network_idle: *const PageNetworkIdle,
-    page_network_almost_idle: *const PageNetworkAlmostIdle,
-    page_frame_created: *const PageFrameCreated,
-    http_request_fail: *const RequestFail,
-    http_request_start: *const RequestStart,
-    http_request_intercept: *const RequestIntercept,
-    http_request_auth_required: *const RequestAuthRequired,
-    http_request_done: *const RequestDone,
-    http_response_data: *const ResponseData,
-    http_response_header_done: *const ResponseHeaderDone,
-};
-const EventType = std.meta.FieldEnum(Events);
-
-pub const PageRemove = struct {};
-
-pub const PageNavigate = struct {
-    req_id: u32,
-    frame_id: u32,
-    timestamp: u64,
-    url: [:0]const u8,
-    opts: Page.NavigateOpts,
-};
-
-pub const PageNavigated = struct {
-    req_id: u32,
-    frame_id: u32,
-    timestamp: u64,
-    url: [:0]const u8,
-    opts: Page.NavigatedOpts,
-};
-
-pub const PageNetworkIdle = struct {
-    req_id: u32,
-    frame_id: u32,
-    timestamp: u64,
-};
-
-pub const PageNetworkAlmostIdle = struct {
-    req_id: u32,
-    frame_id: u32,
-    timestamp: u64,
-};
-
-pub const PageFrameCreated = struct {
-    frame_id: u32,
-    parent_id: u32,
-    timestamp: u64,
-};
-
-pub const RequestStart = struct {
-    transfer: *Transfer,
-};
-
-pub const RequestIntercept = struct {
-    transfer: *Transfer,
-    wait_for_interception: *bool,
-};
-
-pub const RequestAuthRequired = struct {
-    transfer: *Transfer,
-    wait_for_interception: *bool,
-};
-
-pub const ResponseData = struct {
-    data: []const u8,
-    transfer: *Transfer,
-};
-
-pub const ResponseHeaderDone = struct {
-    transfer: *Transfer,
-};
-
-pub const RequestDone = struct {
-    transfer: *Transfer,
-};
-
-pub const RequestFail = struct {
-    transfer: *Transfer,
-    err: anyerror,
-};
-
-pub fn init(allocator: Allocator) !*Notification {
-    const notification = try allocator.create(Notification);
-    errdefer allocator.destroy(notification);
-
-    notification.* = .{
-        .listeners = .{},
-        .event_listeners = .{},
-        .allocator = allocator,
-        .mem_pool = std.heap.MemoryPool(Listener).init(allocator),
-    };
-
-    return notification;
-}
-
-pub fn deinit(self: *Notification) void {
-    const allocator = self.allocator;
-
-    var it = self.listeners.valueIterator();
-    while (it.next()) |listener| {
-        listener.deinit(allocator);
-    }
-    self.listeners.deinit(allocator);
-    self.mem_pool.deinit();
-    allocator.destroy(self);
-}
-
-pub fn register(self: *Notification, comptime event: EventType, receiver: anytype, func: EventFunc(event)) !void {
-    var list = &@field(self.event_listeners, @tagName(event));
-
-    var listener = try self.mem_pool.create();
-    errdefer self.mem_pool.destroy(listener);
-
-    listener.* = .{
-        .node = .{},
-        .list = list,
-        .receiver = receiver,
-        .event = event,
-        .func = @ptrCast(func),
-        .struct_name = @typeName(@typeInfo(@TypeOf(receiver)).pointer.child),
-    };
-
-    const allocator = self.allocator;
-    const gop = try self.listeners.getOrPut(allocator, @intFromPtr(receiver));
-    if (gop.found_existing == false) {
-        gop.value_ptr.* = .{};
-    }
-    try gop.value_ptr.append(allocator, listener);
-
-    // we don't add this until we've successfully added the entry to
-    // self.listeners
-    list.append(&listener.node);
-}
-
-pub fn unregister(self: *Notification, comptime event: EventType, receiver: anytype) void {
-    var listeners = self.listeners.getPtr(@intFromPtr(receiver)) orelse return;
-
-    var i: usize = 0;
-    while (i < listeners.items.len) {
-        const listener = listeners.items[i];
-        if (listener.event != event) {
-            i += 1;
-            continue;
-        }
-        listener.list.remove(&listener.node);
-        self.mem_pool.destroy(listener);
-        _ = listeners.swapRemove(i);
-    }
-
-    if (listeners.items.len == 0) {
-        listeners.deinit(self.allocator);
-        const removed = self.listeners.remove(@intFromPtr(receiver));
-        lp.assert(removed == true, "Notification.unregister", .{ .type = event });
-    }
-}
-
-pub fn unregisterAll(self: *Notification, receiver: *anyopaque) void {
-    var kv = self.listeners.fetchRemove(@intFromPtr(receiver)) orelse return;
-    for (kv.value.items) |listener| {
-        listener.list.remove(&listener.node);
-        self.mem_pool.destroy(listener);
-    }
-    kv.value.deinit(self.allocator);
-}
-
-pub fn dispatch(self: *Notification, comptime event: EventType, data: ArgType(event)) void {
-    if (self.listeners.count() == 0) {
-        return;
-    }
-    const list = &@field(self.event_listeners, @tagName(event));
-
-    var node = list.first;
-    while (node) |n| {
-        const listener: *Listener = @fieldParentPtr("node", n);
-        const func: EventFunc(event) = @ptrCast(@alignCast(listener.func));
-        func(listener.receiver, data) catch |err| {
-            log.err(.app, "dispatch error", .{
-                .err = err,
-                .event = event,
-                .source = "notification",
-                .listener = listener.struct_name,
-            });
-        };
-        node = n.next;
-    }
-}
-
-// Given an event type enum, returns the type of arg the event emits
-fn ArgType(comptime event: Notification.EventType) type {
-    inline for (std.meta.fields(Notification.Events)) |f| {
-        if (std.mem.eql(u8, f.name, @tagName(event))) {
-            return f.type;
-        }
-    }
-    unreachable;
-}
-
-// Given an event type enum, returns the listening function type
-fn EventFunc(comptime event: Notification.EventType) type {
-    return *const fn (*anyopaque, ArgType(event)) anyerror!void;
-}
-
-// A listener. This is 1 receiver, with its function, and the linked list
-// node that goes in the appropriate EventListeners list.
-const Listener = struct {
-    // the receiver of the event, i.e. the self parameter to `func`
-    receiver: *anyopaque,
-
-    // the function to call
-    func: *const anyopaque,
-
-    // For logging slightly better error
-    struct_name: []const u8,
-
-    event: Notification.EventType,
-
-    // intrusive linked list node
-    node: List.Node,
-
-    // The event list this listener belongs to.
-    // We need this in order to be able to remove the node from the list
-    list: *List,
-};
-
-const testing = std.testing;
-test "Notification" {
-    var notifier = try Notification.init(testing.allocator);
-    defer notifier.deinit();
-
-    // noop
-    notifier.dispatch(.page_navigate, &.{
-        .frame_id = 0,
-        .req_id = 1,
-        .timestamp = 4,
-        .url = undefined,
-        .opts = .{},
-    });
-
-    var tc = TestClient{};
-
-    try notifier.register(.page_navigate, &tc, TestClient.pageNavigate);
-    notifier.dispatch(.page_navigate, &.{
-        .frame_id = 0,
-        .req_id = 1,
-        .timestamp = 4,
-        .url = undefined,
-        .opts = .{},
-    });
-    try testing.expectEqual(4, tc.page_navigate);
-
-    notifier.unregisterAll(&tc);
-    notifier.dispatch(.page_navigate, &.{
-        .frame_id = 0,
-        .req_id = 1,
-        .timestamp = 10,
-        .url = undefined,
-        .opts = .{},
-    });
-    try testing.expectEqual(4, tc.page_navigate);
-
-    try notifier.register(.page_navigate, &tc, TestClient.pageNavigate);
-    try notifier.register(.page_navigated, &tc, TestClient.pageNavigated);
-    notifier.dispatch(.page_navigate, &.{
-        .frame_id = 0,
-        .req_id = 1,
-        .timestamp = 10,
-        .url = undefined,
-        .opts = .{},
-    });
-    notifier.dispatch(.page_navigated, &.{ .frame_id = 0, .req_id = 1, .timestamp = 6, .url = undefined, .opts = .{} });
-    try testing.expectEqual(14, tc.page_navigate);
-    try testing.expectEqual(6, tc.page_navigated);
-
-    notifier.unregisterAll(&tc);
-    notifier.dispatch(.page_navigate, &.{
-        .frame_id = 0,
-        .req_id = 1,
-        .timestamp = 100,
-        .url = undefined,
-        .opts = .{},
-    });
-    notifier.dispatch(.page_navigated, &.{ .frame_id = 0, .req_id = 1, .timestamp = 100, .url = undefined, .opts = .{} });
-    try testing.expectEqual(14, tc.page_navigate);
-    try testing.expectEqual(6, tc.page_navigated);
-
-    {
-        // unregister
-        try notifier.register(.page_navigate, &tc, TestClient.pageNavigate);
-        try notifier.register(.page_navigated, &tc, TestClient.pageNavigated);
-        notifier.dispatch(.page_navigate, &.{ .frame_id = 0, .req_id = 1, .timestamp = 100, .url = undefined, .opts = .{} });
-        notifier.dispatch(.page_navigated, &.{ .frame_id = 0, .req_id = 1, .timestamp = 1000, .url = undefined, .opts = .{} });
-        try testing.expectEqual(114, tc.page_navigate);
-        try testing.expectEqual(1006, tc.page_navigated);
-
-        notifier.unregister(.page_navigate, &tc);
-        notifier.dispatch(.page_navigate, &.{ .frame_id = 0, .req_id = 1, .timestamp = 100, .url = undefined, .opts = .{} });
-        notifier.dispatch(.page_navigated, &.{ .frame_id = 0, .req_id = 1, .timestamp = 1000, .url = undefined, .opts = .{} });
-        try testing.expectEqual(114, tc.page_navigate);
-        try testing.expectEqual(2006, tc.page_navigated);
-
-        notifier.unregister(.page_navigated, &tc);
-        notifier.dispatch(.page_navigate, &.{ .frame_id = 0, .req_id = 1, .timestamp = 100, .url = undefined, .opts = .{} });
-        notifier.dispatch(.page_navigated, &.{ .frame_id = 0, .req_id = 1, .timestamp = 1000, .url = undefined, .opts = .{} });
-        try testing.expectEqual(114, tc.page_navigate);
-        try testing.expectEqual(2006, tc.page_navigated);
-
-        // already unregistered, try anyways
-        notifier.unregister(.page_navigated, &tc);
-        notifier.dispatch(.page_navigate, &.{ .frame_id = 0, .req_id = 1, .timestamp = 100, .url = undefined, .opts = .{} });
-        notifier.dispatch(.page_navigated, &.{ .frame_id = 0, .req_id = 1, .timestamp = 1000, .url = undefined, .opts = .{} });
-        try testing.expectEqual(114, tc.page_navigate);
-        try testing.expectEqual(2006, tc.page_navigated);
-    }
-}
-
-const TestClient = struct {
-    page_navigate: u64 = 0,
-    page_navigated: u64 = 0,
-
-    fn pageNavigate(ptr: *anyopaque, data: *const Notification.PageNavigate) !void {
-        const self: *TestClient = @ptrCast(@alignCast(ptr));
-        self.page_navigate += data.timestamp;
-    }
-
-    fn pageNavigated(ptr: *anyopaque, data: *const Notification.PageNavigated) !void {
-        const self: *TestClient = @ptrCast(@alignCast(ptr));
-        self.page_navigated += data.timestamp;
-    }
-};

src/Sighandler.zig

@@ -1,107 +0,0 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-//! This structure processes operating system signals (SIGINT, SIGTERM)
-//! and runs callbacks to clean up the system gracefully.
-//!
-//! The structure does not clear the memory allocated in the arena,
-//! clear the entire arena when exiting the program.
-const std = @import("std");
-const assert = std.debug.assert;
-const Allocator = std.mem.Allocator;
-const lp = @import("lightpanda");
-
-const log = lp.log;
-
-const SigHandler = @This();
-
-arena: Allocator,
-
-sigset: std.posix.sigset_t = undefined,
-handle_thread: ?std.Thread = null,
-
-attempt: u32 = 0,
-listeners: std.ArrayList(Listener) = .empty,
-
-pub const Listener = struct {
-    args: []const u8,
-    start: *const fn (context: *const anyopaque) void,
-};
-
-pub fn install(self: *SigHandler) !void {
-    // Block SIGINT and SIGTERM for the current thread and all created from it
-    self.sigset = std.posix.sigemptyset();
-    std.posix.sigaddset(&self.sigset, std.posix.SIG.INT);
-    std.posix.sigaddset(&self.sigset, std.posix.SIG.TERM);
-    std.posix.sigaddset(&self.sigset, std.posix.SIG.QUIT);
-    std.posix.sigprocmask(std.posix.SIG.BLOCK, &self.sigset, null);
-
-    self.handle_thread = try std.Thread.spawn(.{ .allocator = self.arena }, SigHandler.sighandle, .{self});
-    self.handle_thread.?.detach();
-}
-
-pub fn on(self: *SigHandler, func: anytype, args: std.meta.ArgsTuple(@TypeOf(func))) !void {
-    assert(@typeInfo(@TypeOf(func)).@"fn".return_type.? == void);
-
-    const Args = @TypeOf(args);
-    const TypeErased = struct {
-        fn start(context: *const anyopaque) void {
-            const args_casted: *const Args = @ptrCast(@alignCast(context));
-            @call(.auto, func, args_casted.*);
-        }
-    };
-
-    const buffer = try self.arena.alignedAlloc(u8, .of(Args), @sizeOf(Args));
-    errdefer self.arena.free(buffer);
-
-    const bytes: []const u8 = @ptrCast((&args)[0..1]);
-    @memcpy(buffer, bytes);
-
-    try self.listeners.append(self.arena, .{
-        .args = buffer,
-        .start = TypeErased.start,
-    });
-}
-
-fn sighandle(self: *SigHandler) noreturn {
-    while (true) {
-        var sig: c_int = 0;
-
-        const rc = std.c.sigwait(&self.sigset, &sig);
-        if (rc != 0) {
-            log.err(.app, "Unable to process signal {}", .{rc});
-            std.process.exit(1);
-        }
-
-        switch (sig) {
-            std.posix.SIG.INT, std.posix.SIG.TERM => {
-                if (self.attempt > 1) {
-                    std.process.exit(1);
-                }
-                self.attempt += 1;
-
-                log.info(.app, "Received termination signal...", .{});
-                for (self.listeners.items) |*item| {
-                    item.start(item.args.ptr);
-                }
-                continue;
-            },
-            else => continue,
-        }
-    }
-}

src/TestHTTPServer.zig

@@ -1,27 +1,8 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
 const std = @import("std");
-const URL = @import("browser/URL.zig");
 
 const TestHTTPServer = @This();
 
-shutdown: std.atomic.Value(bool),
+shutdown: bool,
 listener: ?std.net.Server,
 handler: Handler,
 
@@ -29,23 +10,16 @@ const Handler = *const fn (req: *std.http.Server.Request) anyerror!void;
 
 pub fn init(handler: Handler) TestHTTPServer {
     return .{
-        .shutdown = .init(true),
+        .shutdown = true,
         .listener = null,
         .handler = handler,
     };
 }
 
 pub fn deinit(self: *TestHTTPServer) void {
-    self.listener = null;
-}
-
-pub fn stop(self: *TestHTTPServer) void {
-    self.shutdown.store(true, .release);
+    self.shutdown = true;
     if (self.listener) |*listener| {
-        switch (@import("builtin").target.os.tag) {
-            .linux => std.posix.shutdown(listener.stream.handle, .recv) catch {},
-            else => std.posix.close(listener.stream.handle),
-        }
+        listener.deinit();
     }
 }
 
@@ -54,13 +28,12 @@ pub fn run(self: *TestHTTPServer, wg: *std.Thread.WaitGroup) !void {
 
     self.listener = try address.listen(.{ .reuse_address = true });
     var listener = &self.listener.?;
-    self.shutdown.store(false, .release);
 
     wg.finish();
 
     while (true) {
         const conn = listener.accept() catch |err| {
-            if (self.shutdown.load(.acquire) or err == error.SocketNotListening) {
+            if (self.shutdown) {
                 return;
             }
             return err;
@@ -88,7 +61,6 @@ fn handleConnection(self: *TestHTTPServer, conn: std.net.Server.Connection) !voi
                 return err;
             },
         };
-
         self.handler(&req) catch |err| {
             std.debug.print("test http error '{s}': {}\n", .{ req.head.target, err });
             try req.respond("server error", .{ .status = .internal_server_error });
@@ -98,10 +70,7 @@ fn handleConnection(self: *TestHTTPServer, conn: std.net.Server.Connection) !voi
 }
 
 pub fn sendFile(req: *std.http.Server.Request, file_path: []const u8) !void {
-    var url_buf: [1024]u8 = undefined;
-    var fba = std.heap.FixedBufferAllocator.init(&url_buf);
-    const unescaped_file_path = try URL.unescape(fba.allocator(), file_path);
-    var file = std.fs.cwd().openFile(unescaped_file_path, .{}) catch |err| switch (err) {
+    var file = std.fs.cwd().openFile(file_path, .{}) catch |err| switch (err) {
         error.FileNotFound => return req.respond("server error", .{ .status = .not_found }),
         else => return err,
     };
@@ -131,6 +100,11 @@ fn getContentType(file_path: []const u8) []const u8 {
         return "application/json";
     }
 
+    if (std.mem.endsWith(u8, file_path, ".mjs")) {
+        // mjs are ECMAScript modules
+        return "application/json";
+    }
+
     if (std.mem.endsWith(u8, file_path, ".html")) {
         return "text/html";
     }
@@ -144,11 +118,6 @@ fn getContentType(file_path: []const u8) []const u8 {
         return "text/xml";
     }
 
-    if (std.mem.endsWith(u8, file_path, ".mjs")) {
-        // mjs are ECMAScript modules
-        return "application/json";
-    }
-
     std.debug.print("TestHTTPServer asked to serve an unknown file type: {s}\n", .{file_path});
     return "text/html";
 }

src/app.zig

@@ -0,0 +1,121 @@
+const std = @import("std");
+
+const Allocator = std.mem.Allocator;
+
+const log = @import("log.zig");
+const Http = @import("http/Http.zig");
+const Platform = @import("browser/js/Platform.zig");
+
+const Telemetry = @import("telemetry/telemetry.zig").Telemetry;
+const Notification = @import("notification.zig").Notification;
+
+// Container for global state / objects that various parts of the system
+// might need.
+pub const App = struct {
+    http: Http,
+    config: Config,
+    platform: Platform,
+    allocator: Allocator,
+    telemetry: Telemetry,
+    app_dir_path: ?[]const u8,
+    notification: *Notification,
+    shutdown: bool = false,
+
+    pub const RunMode = enum {
+        help,
+        fetch,
+        serve,
+        version,
+    };
+
+    pub const Config = struct {
+        run_mode: RunMode,
+        tls_verify_host: bool = true,
+        http_proxy: ?[:0]const u8 = null,
+        proxy_bearer_token: ?[:0]const u8 = null,
+        http_timeout_ms: ?u31 = null,
+        http_connect_timeout_ms: ?u31 = null,
+        http_max_host_open: ?u8 = null,
+        http_max_concurrent: ?u8 = null,
+        user_agent: [:0]const u8,
+    };
+
+    pub fn init(allocator: Allocator, config: Config) !*App {
+        const app = try allocator.create(App);
+        errdefer allocator.destroy(app);
+
+        const notification = try Notification.init(allocator, null);
+        errdefer notification.deinit();
+
+        var http = try Http.init(allocator, .{
+            .max_host_open = config.http_max_host_open orelse 4,
+            .max_concurrent = config.http_max_concurrent orelse 10,
+            .timeout_ms = config.http_timeout_ms orelse 5000,
+            .connect_timeout_ms = config.http_connect_timeout_ms orelse 0,
+            .http_proxy = config.http_proxy,
+            .tls_verify_host = config.tls_verify_host,
+            .proxy_bearer_token = config.proxy_bearer_token,
+            .user_agent = config.user_agent,
+        });
+        errdefer http.deinit();
+
+        const platform = try Platform.init();
+        errdefer platform.deinit();
+
+        const app_dir_path = getAndMakeAppDir(allocator);
+
+        app.* = .{
+            .http = http,
+            .allocator = allocator,
+            .telemetry = undefined,
+            .platform = platform,
+            .app_dir_path = app_dir_path,
+            .notification = notification,
+            .config = config,
+        };
+
+        app.telemetry = try Telemetry.init(app, config.run_mode);
+        errdefer app.telemetry.deinit();
+
+        try app.telemetry.register(app.notification);
+
+        return app;
+    }
+
+    pub fn deinit(self: *App) void {
+        if (@atomicRmw(bool, &self.shutdown, .Xchg, true, .monotonic)) {
+            return;
+        }
+
+        const allocator = self.allocator;
+        if (self.app_dir_path) |app_dir_path| {
+            allocator.free(app_dir_path);
+            self.app_dir_path = null;
+        }
+        self.telemetry.deinit();
+        self.notification.deinit();
+        self.http.deinit();
+        self.platform.deinit();
+        allocator.destroy(self);
+    }
+};
+
+fn getAndMakeAppDir(allocator: Allocator) ?[]const u8 {
+    if (@import("builtin").is_test) {
+        return allocator.dupe(u8, "/tmp") catch unreachable;
+    }
+    const app_dir_path = std.fs.getAppDataDir(allocator, "lightpanda") catch |err| {
+        log.warn(.app, "get data dir", .{ .err = err });
+        return null;
+    };
+
+    std.fs.cwd().makePath(app_dir_path) catch |err| switch (err) {
+        error.PathAlreadyExists => return app_dir_path,
+        else => {
+            allocator.free(app_dir_path);
+            log.warn(.app, "create data dir", .{ .err = err, .path = app_dir_path });
+            return null;
+        },
+    };
+    return app_dir_path;
+}

src/browser/Browser.zig

@@ -1,115 +0,0 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-
-const Allocator = std.mem.Allocator;
-const ArenaAllocator = std.heap.ArenaAllocator;
-
-const js = @import("js/js.zig");
-const log = @import("../log.zig");
-const App = @import("../App.zig");
-const HttpClient = @import("../http/Client.zig");
-
-const ArenaPool = App.ArenaPool;
-
-const IS_DEBUG = @import("builtin").mode == .Debug;
-
-const Session = @import("Session.zig");
-const Notification = @import("../Notification.zig");
-
-// Browser is an instance of the browser.
-// You can create multiple browser instances.
-// A browser contains only one session.
-const Browser = @This();
-
-env: js.Env,
-app: *App,
-session: ?Session,
-allocator: Allocator,
-arena_pool: *ArenaPool,
-http_client: *HttpClient,
-
-const InitOpts = struct {
-    env: js.Env.InitOpts = .{},
-    http_client: *HttpClient,
-};
-
-pub fn init(app: *App, opts: InitOpts) !Browser {
-    const allocator = app.allocator;
-
-    var env = try js.Env.init(app, opts.env);
-    errdefer env.deinit();
-
-    return .{
-        .app = app,
-        .env = env,
-        .session = null,
-        .allocator = allocator,
-        .arena_pool = &app.arena_pool,
-        .http_client = opts.http_client,
-    };
-}
-
-pub fn deinit(self: *Browser) void {
-    self.closeSession();
-    self.env.deinit();
-}
-
-pub fn newSession(self: *Browser, notification: *Notification) !*Session {
-    self.closeSession();
-    self.session = @as(Session, undefined);
-    const session = &self.session.?;
-    try Session.init(session, self, notification);
-    return session;
-}
-
-pub fn closeSession(self: *Browser) void {
-    if (self.session) |*session| {
-        session.deinit();
-        self.session = null;
-        self.env.memoryPressureNotification(.critical);
-    }
-}
-
-pub fn runMicrotasks(self: *Browser) void {
-    self.env.runMicrotasks();
-}
-
-pub fn runMacrotasks(self: *Browser) !?u64 {
-    const env = &self.env;
-
-    const time_to_next = try self.env.runMacrotasks();
-    env.pumpMessageLoop();
-
-    // either of the above could have queued more microtasks
-    env.runMicrotasks();
-
-    return time_to_next;
-}
-
-pub fn hasBackgroundTasks(self: *Browser) bool {
-    return self.env.hasBackgroundTasks();
-}
-pub fn waitForBackgroundTasks(self: *Browser) void {
-    self.env.waitForBackgroundTasks();
-}
-
-pub fn runIdleTasks(self: *const Browser) void {
-    self.env.runIdleTasks();
-}

src/browser/DataURI.zig

@@ -0,0 +1,52 @@
+const std = @import("std");
+const Allocator = std.mem.Allocator;
+
+// Parses data:[<media-type>][;base64],<data>
+pub fn parse(allocator: Allocator, src: []const u8) !?[]const u8 {
+    if (!std.mem.startsWith(u8, src, "data:")) {
+        return null;
+    }
+
+    const uri = src[5..];
+    const data_starts = std.mem.indexOfScalar(u8, uri, ',') orelse return null;
+
+    var data = uri[data_starts + 1 ..];
+
+    // Extract the encoding.
+    const metadata = uri[0..data_starts];
+    if (std.mem.endsWith(u8, metadata, ";base64")) {
+        const decoder = std.base64.standard.Decoder;
+        const decoded_size = try decoder.calcSizeForSlice(data);
+
+        const buffer = try allocator.alloc(u8, decoded_size);
+        errdefer allocator.free(buffer);
+
+        try decoder.decode(buffer, data);
+        data = buffer;
+    }
+
+    return data;
+}
+
+const testing = @import("../testing.zig");
+test "DataURI: parse valid" {
+    try test_valid("data:text/javascript; charset=utf-8;base64,Zm9v", "foo");
+    try test_valid("data:text/javascript; charset=utf-8;,foo", "foo");
+    try test_valid("data:,foo", "foo");
+}
+
+test "DataURI: parse invalid" {
+    try test_cannot_parse("atad:,foo");
+    try test_cannot_parse("data:foo");
+    try test_cannot_parse("data:");
+}
+
+fn test_valid(uri: []const u8, expected: []const u8) !void {
+    defer testing.reset();
+    const data_uri = try parse(testing.arena_allocator, uri) orelse return error.TestFailed;
+    try testing.expectEqual(expected, data_uri);
+}
+
+fn test_cannot_parse(uri: []const u8) !void {
+    try testing.expectEqual(null, parse(undefined, uri));
+}

src/browser/EventManager.zig

@@ -1,907 +0,0 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-const builtin = @import("builtin");
-
-const log = @import("../log.zig");
-const String = @import("../string.zig").String;
-
-const js = @import("js/js.zig");
-const Page = @import("Page.zig");
-
-const Node = @import("webapi/Node.zig");
-const Event = @import("webapi/Event.zig");
-const EventTarget = @import("webapi/EventTarget.zig");
-const Element = @import("webapi/Element.zig");
-
-const Allocator = std.mem.Allocator;
-
-const IS_DEBUG = builtin.mode == .Debug;
-
-const EventKey = struct {
-    event_target: usize,
-    type_string: String,
-};
-
-const EventKeyContext = struct {
-    pub fn hash(_: @This(), key: EventKey) u64 {
-        var hasher = std.hash.Wyhash.init(0);
-        hasher.update(std.mem.asBytes(&key.event_target));
-        hasher.update(key.type_string.str());
-        return hasher.final();
-    }
-
-    pub fn eql(_: @This(), a: EventKey, b: EventKey) bool {
-        return a.event_target == b.event_target and a.type_string.eql(b.type_string);
-    }
-};
-
-pub const EventManager = @This();
-
-page: *Page,
-arena: Allocator,
-// Used as an optimization in Page._documentIsComplete. If we know there are no
-// 'load' listeners in the document, we can skip dispatching the per-resource
-// 'load' event (e.g. amazon product page has no listener and ~350 resources)
-has_dom_load_listener: bool,
-listener_pool: std.heap.MemoryPool(Listener),
-ignore_list: std.ArrayList(*Listener),
-list_pool: std.heap.MemoryPool(std.DoublyLinkedList),
-lookup: std.HashMapUnmanaged(
-    EventKey,
-    *std.DoublyLinkedList,
-    EventKeyContext,
-    std.hash_map.default_max_load_percentage,
-),
-dispatch_depth: usize,
-deferred_removals: std.ArrayList(struct { list: *std.DoublyLinkedList, listener: *Listener }),
-
-pub fn init(arena: Allocator, page: *Page) EventManager {
-    return .{
-        .page = page,
-        .lookup = .{},
-        .arena = arena,
-        .ignore_list = .{},
-        .list_pool = .init(arena),
-        .listener_pool = .init(arena),
-        .dispatch_depth = 0,
-        .deferred_removals = .{},
-        .has_dom_load_listener = false,
-    };
-}
-
-pub const RegisterOptions = struct {
-    once: bool = false,
-    capture: bool = false,
-    passive: bool = false,
-    signal: ?*@import("webapi/AbortSignal.zig") = null,
-};
-
-pub const Callback = union(enum) {
-    function: js.Function,
-    object: js.Object,
-};
-
-pub fn register(self: *EventManager, target: *EventTarget, typ: []const u8, callback: Callback, opts: RegisterOptions) !void {
-    if (comptime IS_DEBUG) {
-        log.debug(.event, "eventManager.register", .{ .type = typ, .capture = opts.capture, .once = opts.once, .target = target.toString() });
-    }
-
-    // If a signal is provided and already aborted, don't register the listener
-    if (opts.signal) |signal| {
-        if (signal.getAborted()) {
-            return;
-        }
-    }
-
-    // Allocate the type string we'll use in both listener and key
-    const type_string = try String.init(self.arena, typ, .{});
-
-    if (type_string.eql(comptime .wrap("load")) and target._type == .node) {
-        self.has_dom_load_listener = true;
-    }
-
-    const gop = try self.lookup.getOrPut(self.arena, .{
-        .type_string = type_string,
-        .event_target = @intFromPtr(target),
-    });
-    if (gop.found_existing) {
-        // check for duplicate callbacks already registered
-        var node = gop.value_ptr.*.first;
-        while (node) |n| {
-            const listener: *Listener = @alignCast(@fieldParentPtr("node", n));
-            const is_duplicate = switch (callback) {
-                .object => |obj| listener.function.eqlObject(obj),
-                .function => |func| listener.function.eqlFunction(func),
-            };
-            if (is_duplicate and listener.capture == opts.capture) {
-                return;
-            }
-            node = n.next;
-        }
-    } else {
-        gop.value_ptr.* = try self.list_pool.create();
-        gop.value_ptr.*.* = .{};
-    }
-
-    const func = switch (callback) {
-        .function => |f| Function{ .value = try f.persist() },
-        .object => |o| Function{ .object = try o.persist() },
-    };
-
-    const listener = try self.listener_pool.create();
-    listener.* = .{
-        .node = .{},
-        .once = opts.once,
-        .capture = opts.capture,
-        .passive = opts.passive,
-        .function = func,
-        .signal = opts.signal,
-        .typ = type_string,
-    };
-    // append the listener to the list of listeners for this target
-    gop.value_ptr.*.append(&listener.node);
-
-    // Track load listeners for script execution ignore list
-    if (type_string.eql(comptime .wrap("load"))) {
-        try self.ignore_list.append(self.arena, listener);
-    }
-}
-
-pub fn remove(self: *EventManager, target: *EventTarget, typ: []const u8, callback: Callback, use_capture: bool) void {
-    const list = self.lookup.get(.{
-        .type_string = .wrap(typ),
-        .event_target = @intFromPtr(target),
-    }) orelse return;
-    if (findListener(list, callback, use_capture)) |listener| {
-        self.removeListener(list, listener);
-    }
-}
-
-pub fn clearIgnoreList(self: *EventManager) void {
-    self.ignore_list.clearRetainingCapacity();
-}
-
-// Dispatching can be recursive from the compiler's point of view, so we need to
-// give it an explicit error set so that other parts of the code can use and
-// inferred error.
-const DispatchError = error{
-    OutOfMemory,
-    StringTooLarge,
-    JSExecCallback,
-    CompilationError,
-    ExecutionError,
-    JsException,
-};
-
-pub const DispatchOpts = struct {
-    // A "load" event triggered by a script (in ScriptManager) should not trigger
-    // a "load" listener added within that script. Therefore, any "load" listener
-    // that we add go into an ignore list until after the script finishes executing.
-    // The ignore list is only checked when apply_ignore  == true, which is only
-    // set by the ScriptManager when raising the script's "load" event.
-    apply_ignore: bool = false,
-};
-
-pub fn dispatch(self: *EventManager, target: *EventTarget, event: *Event) DispatchError!void {
-    return self.dispatchOpts(target, event, .{});
-}
-
-pub fn dispatchOpts(self: *EventManager, target: *EventTarget, event: *Event, comptime opts: DispatchOpts) DispatchError!void {
-    event.acquireRef();
-    defer event.deinit(false, self.page);
-
-    if (comptime IS_DEBUG) {
-        log.debug(.event, "eventManager.dispatch", .{ .type = event._type_string.str(), .bubbles = event._bubbles });
-    }
-
-    switch (target._type) {
-        .node => |node| try self.dispatchNode(node, event, opts),
-        else => try self.dispatchDirect(target, event, null, .{ .context = "dispatch" }),
-    }
-}
-
-// There are a lot of events that can be attached via addEventListener or as
-// a property, like the XHR events, or window.onload. You might think that the
-// property is just a shortcut for calling addEventListener, but they are distinct.
-// An event set via property cannot be removed by removeEventListener. If you
-// set both the property and add a listener, they both execute.
-const DispatchDirectOptions = struct {
-    context: []const u8,
-    inject_target: bool = true,
-};
-
-// Direct dispatch for non-DOM targets (Window, XHR, AbortSignal) or DOM nodes with
-// property handlers. No propagation - just calls the handler and registered listeners.
-// Handler can be: null, ?js.Function.Global, ?js.Function.Temp, or js.Function
-pub fn dispatchDirect(self: *EventManager, target: *EventTarget, event: *Event, handler: anytype, comptime opts: DispatchDirectOptions) !void {
-    const page = self.page;
-
-    event.acquireRef();
-    defer event.deinit(false, page);
-
-    if (comptime IS_DEBUG) {
-        log.debug(.event, "dispatchDirect", .{ .type = event._type_string, .context = opts.context });
-    }
-
-    if (comptime opts.inject_target) {
-        event._target = target;
-        event._dispatch_target = target; // Store original target for composedPath()
-    }
-
-    var was_dispatched = false;
-
-    var ls: js.Local.Scope = undefined;
-    page.js.localScope(&ls);
-    defer {
-        ls.local.runMicrotasks();
-        ls.deinit();
-    }
-
-    if (getFunction(handler, &ls.local)) |func| {
-        event._current_target = target;
-        if (func.callWithThis(void, target, .{event})) {
-            was_dispatched = true;
-        } else |err| {
-            // a non-JS error
-            log.warn(.event, opts.context, .{ .err = err });
-        }
-    }
-
-    // listeners reigstered via addEventListener
-    const list = self.lookup.get(.{
-        .event_target = @intFromPtr(target),
-        .type_string = event._type_string,
-    }) orelse return;
-
-    // This is a slightly simplified version of what you'll find in dispatchPhase
-    // It is simpler because, for direct dispatching, we know there's no ancestors
-    // and only the single target phase.
-
-    // Track dispatch depth for deferred removal
-    self.dispatch_depth += 1;
-    defer {
-        const dispatch_depth = self.dispatch_depth;
-        // Only destroy deferred listeners when we exit the outermost dispatch
-        if (dispatch_depth == 1) {
-            for (self.deferred_removals.items) |removal| {
-                removal.list.remove(&removal.listener.node);
-                self.listener_pool.destroy(removal.listener);
-            }
-            self.deferred_removals.clearRetainingCapacity();
-        } else {
-            self.dispatch_depth = dispatch_depth - 1;
-        }
-    }
-
-    // Use the last listener in the list as sentinel - listeners added during dispatch will be after it
-    const last_node = list.last orelse return;
-    const last_listener: *Listener = @alignCast(@fieldParentPtr("node", last_node));
-
-    // Iterate through the list, stopping after we've encountered the last_listener
-    var node = list.first;
-    var is_done = false;
-    while (node) |n| {
-        if (is_done) {
-            break;
-        }
-
-        const listener: *Listener = @alignCast(@fieldParentPtr("node", n));
-        is_done = (listener == last_listener);
-        node = n.next;
-
-        // Skip removed listeners
-        if (listener.removed) {
-            continue;
-        }
-
-        // If the listener has an aborted signal, remove it and skip
-        if (listener.signal) |signal| {
-            if (signal.getAborted()) {
-                self.removeListener(list, listener);
-                continue;
-            }
-        }
-
-        // Remove "once" listeners BEFORE calling them so nested dispatches don't see them
-        if (listener.once) {
-            self.removeListener(list, listener);
-        }
-
-        was_dispatched = true;
-        event._current_target = target;
-
-        switch (listener.function) {
-            .value => |value| try ls.toLocal(value).callWithThis(void, target, .{event}),
-            .string => |string| {
-                const str = try page.call_arena.dupeZ(u8, string.str());
-                try ls.local.eval(str, null);
-            },
-            .object => |obj_global| {
-                const obj = ls.toLocal(obj_global);
-                if (try obj.getFunction("handleEvent")) |handleEvent| {
-                    try handleEvent.callWithThis(void, obj, .{event});
-                }
-            },
-        }
-
-        if (event._stop_immediate_propagation) {
-            return;
-        }
-    }
-}
-
-fn getFunction(handler: anytype, local: *const js.Local) ?js.Function {
-    const T = @TypeOf(handler);
-    const ti = @typeInfo(T);
-
-    if (ti == .null) {
-        return null;
-    }
-    if (ti == .optional) {
-        return getFunction(handler orelse return null, local);
-    }
-    return switch (T) {
-        js.Function => handler,
-        js.Function.Temp => local.toLocal(handler),
-        js.Function.Global => local.toLocal(handler),
-        else => @compileError("handler must be null or \\??js.Function(\\.(Temp|Global))?"),
-    };
-}
-
-fn dispatchNode(self: *EventManager, target: *Node, event: *Event, comptime opts: DispatchOpts) !void {
-    const ShadowRoot = @import("webapi/ShadowRoot.zig");
-
-    {
-        const et = target.asEventTarget();
-        event._target = et;
-        event._dispatch_target = et; // Store original target for composedPath()
-    }
-
-    const page = self.page;
-    var was_handled = false;
-
-    // Create a single scope for all event handlers in this dispatch.
-    // This ensures function handles passed to queueMicrotask remain valid
-    // throughout the entire dispatch, preventing crashes when microtasks run.
-    var ls: js.Local.Scope = undefined;
-    page.js.localScope(&ls);
-    defer {
-        if (was_handled) {
-            ls.local.runMicrotasks();
-        }
-        ls.deinit();
-    }
-
-    const activation_state = ActivationState.create(event, target, page);
-
-    // Defer runs even on early return - ensures event phase is reset
-    // and default actions execute (unless prevented)
-    defer {
-        event._event_phase = .none;
-        event._stop_propagation = false;
-        event._stop_immediate_propagation = false;
-        // Handle checkbox/radio activation rollback or commit
-        if (activation_state) |state| {
-            state.restore(event, page);
-        }
-
-        // Execute default action if not prevented
-        if (event._prevent_default) {
-            // can't return in a defer (╯°□°)╯︵ ┻━┻
-        } else if (event._type_string.eql(comptime .wrap("click"))) {
-            page.handleClick(target) catch |err| {
-                log.warn(.event, "page.click", .{ .err = err });
-            };
-        } else if (event._type_string.eql(comptime .wrap("keydown"))) {
-            page.handleKeydown(target, event) catch |err| {
-                log.warn(.event, "page.keydown", .{ .err = err });
-            };
-        }
-    }
-
-    var path_len: usize = 0;
-    var path_buffer: [128]*EventTarget = undefined;
-
-    var node: ?*Node = target;
-    while (node) |n| {
-        if (path_len >= path_buffer.len) break;
-        path_buffer[path_len] = n.asEventTarget();
-        path_len += 1;
-
-        // Check if this node is a shadow root
-        if (n.is(ShadowRoot)) |shadow| {
-            event._needs_retargeting = true;
-
-            // If event is not composed, stop at shadow boundary
-            if (!event._composed) {
-                break;
-            }
-
-            // Otherwise, jump to the shadow host and continue
-            node = shadow._host.asNode();
-            continue;
-        }
-
-        node = n._parent;
-    }
-
-    // Even though the window isn't part of the DOM, most events propagate
-    // through it in the capture phase (unless we stopped at a shadow boundary)
-    // The only explicit exception is "load"
-    if (event._type_string.eql(comptime .wrap("load")) == false) {
-        if (path_len < path_buffer.len) {
-            path_buffer[path_len] = page.window.asEventTarget();
-            path_len += 1;
-        }
-    }
-
-    const path = path_buffer[0..path_len];
-
-    // Phase 1: Capturing phase (root → target, excluding target)
-    // This happens for all events, regardless of bubbling
-    event._event_phase = .capturing_phase;
-    var i: usize = path_len;
-    while (i > 1) {
-        i -= 1;
-        if (event._stop_propagation) return;
-        const current_target = path[i];
-        if (self.lookup.get(.{
-            .event_target = @intFromPtr(current_target),
-            .type_string = event._type_string,
-        })) |list| {
-            try self.dispatchPhase(list, current_target, event, &was_handled, &ls.local, comptime .init(true, opts));
-        }
-    }
-
-    // Phase 2: At target
-    if (event._stop_propagation) return;
-    event._event_phase = .at_target;
-    const target_et = target.asEventTarget();
-
-    blk: {
-        // Get inline handler (e.g., onclick property) for this target
-        if (self.getInlineHandler(target_et, event)) |inline_handler| {
-            was_handled = true;
-            event._current_target = target_et;
-
-            try ls.toLocal(inline_handler).callWithThis(void, target_et, .{event});
-
-            if (event._stop_propagation) {
-                return;
-            }
-
-            if (event._stop_immediate_propagation) {
-                break :blk;
-            }
-        }
-
-        if (self.lookup.get(.{
-            .type_string = event._type_string,
-            .event_target = @intFromPtr(target_et),
-        })) |list| {
-            try self.dispatchPhase(list, target_et, event, &was_handled, &ls.local, comptime .init(null, opts));
-            if (event._stop_propagation) {
-                return;
-            }
-        }
-    }
-
-    // Phase 3: Bubbling phase (target → root, excluding target)
-    // This only happens if the event bubbles
-    if (event._bubbles) {
-        event._event_phase = .bubbling_phase;
-        for (path[1..]) |current_target| {
-            if (event._stop_propagation) break;
-            if (self.lookup.get(.{
-                .type_string = event._type_string,
-                .event_target = @intFromPtr(current_target),
-            })) |list| {
-                try self.dispatchPhase(list, current_target, event, &was_handled, &ls.local, comptime .init(false, opts));
-            }
-        }
-    }
-}
-
-const DispatchPhaseOpts = struct {
-    capture_only: ?bool = null,
-    apply_ignore: bool = false,
-
-    fn init(capture_only: ?bool, opts: DispatchOpts) DispatchPhaseOpts {
-        return .{
-            .capture_only = capture_only,
-            .apply_ignore = opts.apply_ignore,
-        };
-    }
-};
-
-fn dispatchPhase(self: *EventManager, list: *std.DoublyLinkedList, current_target: *EventTarget, event: *Event, was_handled: *bool, local: *const js.Local, comptime opts: DispatchPhaseOpts) !void {
-    const page = self.page;
-
-    // Track dispatch depth for deferred removal
-    self.dispatch_depth += 1;
-    defer {
-        const dispatch_depth = self.dispatch_depth;
-        // Only destroy deferred listeners when we exit the outermost dispatch
-        if (dispatch_depth == 1) {
-            for (self.deferred_removals.items) |removal| {
-                removal.list.remove(&removal.listener.node);
-                self.listener_pool.destroy(removal.listener);
-            }
-            self.deferred_removals.clearRetainingCapacity();
-        } else {
-            self.dispatch_depth = dispatch_depth - 1;
-        }
-    }
-
-    // Use the last listener in the list as sentinel - listeners added during dispatch will be after it
-    const last_node = list.last orelse return;
-    const last_listener: *Listener = @alignCast(@fieldParentPtr("node", last_node));
-
-    // Iterate through the list, stopping after we've encountered the last_listener
-    var node = list.first;
-    var is_done = false;
-    node_loop: while (node) |n| {
-        if (is_done) {
-            break;
-        }
-
-        const listener: *Listener = @alignCast(@fieldParentPtr("node", n));
-        is_done = (listener == last_listener);
-        node = n.next;
-
-        // Skip non-matching listeners
-        if (comptime opts.capture_only) |capture| {
-            if (listener.capture != capture) {
-                continue;
-            }
-        }
-
-        // Skip removed listeners
-        if (listener.removed) {
-            continue;
-        }
-
-        // If the listener has an aborted signal, remove it and skip
-        if (listener.signal) |signal| {
-            if (signal.getAborted()) {
-                self.removeListener(list, listener);
-                continue;
-            }
-        }
-
-        if (comptime opts.apply_ignore) {
-            for (self.ignore_list.items) |ignored| {
-                if (ignored == listener) {
-                    continue :node_loop;
-                }
-            }
-        }
-
-        // Remove "once" listeners BEFORE calling them so nested dispatches don't see them
-        if (listener.once) {
-            self.removeListener(list, listener);
-        }
-
-        was_handled.* = true;
-        event._current_target = current_target;
-
-        // Compute adjusted target for shadow DOM retargeting (only if needed)
-        const original_target = event._target;
-        if (event._needs_retargeting) {
-            event._target = getAdjustedTarget(original_target, current_target);
-        }
-
-        switch (listener.function) {
-            .value => |value| try local.toLocal(value).callWithThis(void, current_target, .{event}),
-            .string => |string| {
-                const str = try page.call_arena.dupeZ(u8, string.str());
-                try local.eval(str, null);
-            },
-            .object => |obj_global| {
-                const obj = local.toLocal(obj_global);
-                if (try obj.getFunction("handleEvent")) |handleEvent| {
-                    try handleEvent.callWithThis(void, obj, .{event});
-                }
-            },
-        }
-
-        // Restore original target (only if we changed it)
-        if (event._needs_retargeting) {
-            event._target = original_target;
-        }
-
-        if (event._stop_immediate_propagation) {
-            return;
-        }
-    }
-}
-
-fn getInlineHandler(self: *EventManager, target: *EventTarget, event: *Event) ?js.Function.Global {
-    const global_event_handlers = @import("webapi/global_event_handlers.zig");
-    const handler_type = global_event_handlers.fromEventType(event._type_string.str()) orelse return null;
-
-    // Look up the inline handler for this target
-    const html_element = switch (target._type) {
-        .node => |n| n.is(Element.Html) orelse return null,
-        else => return null,
-    };
-
-    return html_element.getAttributeFunction(handler_type, self.page) catch |err| {
-        log.warn(.event, "inline html callback", .{ .type = handler_type, .err = err });
-        return null;
-    };
-}
-
-fn removeListener(self: *EventManager, list: *std.DoublyLinkedList, listener: *Listener) void {
-    // If we're in a dispatch, defer removal to avoid invalidating iteration
-    if (self.dispatch_depth > 0) {
-        listener.removed = true;
-        self.deferred_removals.append(self.arena, .{ .list = list, .listener = listener }) catch unreachable;
-    } else {
-        // Outside dispatch, remove immediately
-        list.remove(&listener.node);
-        self.listener_pool.destroy(listener);
-    }
-}
-
-fn findListener(list: *const std.DoublyLinkedList, callback: Callback, capture: bool) ?*Listener {
-    var node = list.first;
-    while (node) |n| {
-        node = n.next;
-        const listener: *Listener = @alignCast(@fieldParentPtr("node", n));
-        const matches = switch (callback) {
-            .object => |obj| listener.function.eqlObject(obj),
-            .function => |func| listener.function.eqlFunction(func),
-        };
-        if (!matches) {
-            continue;
-        }
-        if (listener.capture != capture) {
-            continue;
-        }
-        return listener;
-    }
-    return null;
-}
-
-const Listener = struct {
-    typ: String,
-    once: bool,
-    capture: bool,
-    passive: bool,
-    function: Function,
-    signal: ?*@import("webapi/AbortSignal.zig") = null,
-    node: std.DoublyLinkedList.Node,
-    removed: bool = false,
-};
-
-const Function = union(enum) {
-    value: js.Function.Global,
-    string: String,
-    object: js.Object.Global,
-
-    fn eqlFunction(self: Function, func: js.Function) bool {
-        return switch (self) {
-            .value => |v| v.isEqual(func),
-            else => false,
-        };
-    }
-
-    fn eqlObject(self: Function, obj: js.Object) bool {
-        return switch (self) {
-            .object => |o| return o.isEqual(obj),
-            else => false,
-        };
-    }
-};
-
-// Computes the adjusted target for shadow DOM event retargeting
-// Returns the lowest shadow-including ancestor of original_target that is
-// also an ancestor-or-self of current_target
-fn getAdjustedTarget(original_target: ?*EventTarget, current_target: *EventTarget) ?*EventTarget {
-    const ShadowRoot = @import("webapi/ShadowRoot.zig");
-
-    const orig_node = switch ((original_target orelse return null)._type) {
-        .node => |n| n,
-        else => return original_target,
-    };
-    const curr_node = switch (current_target._type) {
-        .node => |n| n,
-        else => return original_target,
-    };
-
-    // Walk up from original target, checking if we can reach current target
-    var node: ?*Node = orig_node;
-    while (node) |n| {
-        // Check if current_target is an ancestor of n (or n itself)
-        if (isAncestorOrSelf(curr_node, n)) {
-            return n.asEventTarget();
-        }
-
-        // Cross shadow boundary if needed
-        if (n.is(ShadowRoot)) |shadow| {
-            node = shadow._host.asNode();
-            continue;
-        }
-
-        node = n._parent;
-    }
-
-    return original_target;
-}
-
-// Check if ancestor is an ancestor of (or the same as) node
-// WITHOUT crossing shadow boundaries (just regular DOM tree)
-fn isAncestorOrSelf(ancestor: *Node, node: *Node) bool {
-    if (ancestor == node) {
-        return true;
-    }
-
-    var current: ?*Node = node._parent;
-    while (current) |n| {
-        if (n == ancestor) {
-            return true;
-        }
-        current = n._parent;
-    }
-
-    return false;
-}
-
-// Handles the default action for clicking on input checked/radio. Maybe this
-// could be generalized if needed, but I'm not sure. This wasn't obvious to me
-// but when an input is clicked, it's important to think about both the intent
-// and the actual result. Imagine you have an unchecked checkbox. When clicked,
-// the checkbox immediately becomes checked, and event handlers see this "checked"
-// intent. But a listener can preventDefault() in which case the check we did at
-// the start will be undone.
-// This is a bit more complicated for radio buttons, as the checking/unchecking
-// and the rollback can impact a different radio input. So if you "check" a radio
-// the intent is that it becomes checked and whatever was checked before becomes
-// unchecked, so that if you have to rollback (because of a preventDefault())
-// then both inputs have to revert to their original values.
-const ActivationState = struct {
-    old_checked: bool,
-    input: *Element.Html.Input,
-    previously_checked_radio: ?*Input,
-
-    const Input = Element.Html.Input;
-
-    fn create(event: *const Event, target: *Node, page: *Page) ?ActivationState {
-        if (event._type_string.eql(comptime .wrap("click")) == false) {
-            return null;
-        }
-
-        const input = target.is(Element.Html.Input) orelse return null;
-        if (input._input_type != .checkbox and input._input_type != .radio) {
-            return null;
-        }
-
-        const old_checked = input._checked;
-        var previously_checked_radio: ?*Element.Html.Input = null;
-
-        // For radio buttons, find the currently checked radio in the group
-        if (input._input_type == .radio and !old_checked) {
-            previously_checked_radio = try findCheckedRadioInGroup(input, page);
-        }
-
-        // Toggle checkbox or check radio (which unchecks others in group)
-        const new_checked = if (input._input_type == .checkbox) !old_checked else true;
-        try input.setChecked(new_checked, page);
-
-        return .{
-            .input = input,
-            .old_checked = old_checked,
-            .previously_checked_radio = previously_checked_radio,
-        };
-    }
-
-    fn restore(self: *const ActivationState, event: *const Event, page: *Page) void {
-        const input = self.input;
-        if (event._prevent_default) {
-            // Rollback: restore previous state
-            input._checked = self.old_checked;
-            input._checked_dirty = true;
-            if (self.previously_checked_radio) |prev_radio| {
-                prev_radio._checked = true;
-                prev_radio._checked_dirty = true;
-            }
-            return;
-        }
-
-        // Commit: fire input and change events only if state actually changed
-        // and the element is connected to a document (detached elements don't fire).
-        // For checkboxes, state always changes. For radios, only if was unchecked.
-        const state_changed = (input._input_type == .checkbox) or !self.old_checked;
-        if (state_changed and input.asElement().asNode().isConnected()) {
-            fireEvent(page, input, "input") catch |err| {
-                log.warn(.event, "input event", .{ .err = err });
-            };
-            fireEvent(page, input, "change") catch |err| {
-                log.warn(.event, "change event", .{ .err = err });
-            };
-        }
-    }
-
-    fn findCheckedRadioInGroup(input: *Input, page: *Page) !?*Input {
-        const elem = input.asElement();
-
-        const name = elem.getAttributeSafe(comptime .wrap("name")) orelse return null;
-        if (name.len == 0) {
-            return null;
-        }
-
-        const form = input.getForm(page);
-
-        // Walk from the root of the tree containing this element
-        // This handles both document-attached and orphaned elements
-        const root = elem.asNode().getRootNode(null);
-
-        const TreeWalker = @import("webapi/TreeWalker.zig");
-        var walker = TreeWalker.Full.init(root, .{});
-
-        while (walker.next()) |node| {
-            const other_element = node.is(Element) orelse continue;
-            const other_input = other_element.is(Input) orelse continue;
-
-            if (other_input._input_type != .radio) {
-                continue;
-            }
-
-            // Skip the input we're checking from
-            if (other_input == input) {
-                continue;
-            }
-
-            const other_name = other_element.getAttributeSafe(comptime .wrap("name")) orelse continue;
-            if (!std.mem.eql(u8, name, other_name)) {
-                continue;
-            }
-
-            // Check if same form context
-            const other_form = other_input.getForm(page);
-            if (form) |f| {
-                const of = other_form orelse continue;
-                if (f != of) {
-                    continue; // Different forms
-                }
-            } else if (other_form != null) {
-                continue; // form is null but other has a form
-            }
-
-            if (other_input._checked) {
-                return other_input;
-            }
-        }
-
-        return null;
-    }
-
-    // Fire input or change event
-    fn fireEvent(page: *Page, input: *Input, comptime typ: []const u8) !void {
-        const event = try Event.initTrusted(comptime .wrap(typ), .{
-            .bubbles = true,
-            .cancelable = false,
-        }, page);
-
-        const target = input.asElement().asEventTarget();
-        try page._event_manager.dispatch(target, event);
-    }
-};

src/browser/Factory.zig

@@ -1,461 +0,0 @@
-// Copyright (C) 2023-2026  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-const builtin = @import("builtin");
-const reflect = @import("reflect.zig");
-
-const log = @import("../log.zig");
-const String = @import("../string.zig").String;
-
-const SlabAllocator = @import("../slab.zig").SlabAllocator;
-
-const Page = @import("Page.zig");
-const Node = @import("webapi/Node.zig");
-const Event = @import("webapi/Event.zig");
-const UIEvent = @import("webapi/event/UIEvent.zig");
-const MouseEvent = @import("webapi/event/MouseEvent.zig");
-const Element = @import("webapi/Element.zig");
-const Document = @import("webapi/Document.zig");
-const EventTarget = @import("webapi/EventTarget.zig");
-const XMLHttpRequestEventTarget = @import("webapi/net/XMLHttpRequestEventTarget.zig");
-const Blob = @import("webapi/Blob.zig");
-const AbstractRange = @import("webapi/AbstractRange.zig");
-
-const Allocator = std.mem.Allocator;
-
-const IS_DEBUG = builtin.mode == .Debug;
-const assert = std.debug.assert;
-
-// Shared across all frames of a Page.
-const Factory = @This();
-
-_arena: Allocator,
-_slab: SlabAllocator,
-
-pub fn init(arena: Allocator) !*Factory {
-    const self = try arena.create(Factory);
-    self.* = .{
-        ._arena = arena,
-        ._slab = SlabAllocator.init(arena, 128),
-    };
-    return self;
-}
-
-// this is a root object
-pub fn eventTarget(self: *Factory, child: anytype) !*@TypeOf(child) {
-    return self.eventTargetWithAllocator(self._slab.allocator(), child);
-}
-
-pub fn eventTargetWithAllocator(_: *const Factory, allocator: Allocator, child: anytype) !*@TypeOf(child) {
-    const chain = try PrototypeChain(
-        &.{ EventTarget, @TypeOf(child) },
-    ).allocate(allocator);
-
-    const event_ptr = chain.get(0);
-    event_ptr.* = .{
-        ._type = unionInit(EventTarget.Type, chain.get(1)),
-    };
-    chain.setLeaf(1, child);
-
-    return chain.get(1);
-}
-
-pub fn standaloneEventTarget(self: *Factory, child: anytype) !*EventTarget {
-    const allocator = self._slab.allocator();
-    const et = try allocator.create(EventTarget);
-    et.* = .{ ._type = unionInit(EventTarget.Type, child) };
-    return et;
-}
-
-// this is a root object
-pub fn event(_: *const Factory, arena: Allocator, typ: String, child: anytype) !*@TypeOf(child) {
-    const chain = try PrototypeChain(
-        &.{ Event, @TypeOf(child) },
-    ).allocate(arena);
-
-    // Special case: Event has a _type_string field, so we need manual setup
-    const event_ptr = chain.get(0);
-    event_ptr.* = try eventInit(arena, typ, chain.get(1));
-    chain.setLeaf(1, child);
-
-    return chain.get(1);
-}
-
-pub fn uiEvent(_: *const Factory, arena: Allocator, typ: String, child: anytype) !*@TypeOf(child) {
-    const chain = try PrototypeChain(
-        &.{ Event, UIEvent, @TypeOf(child) },
-    ).allocate(arena);
-
-    // Special case: Event has a _type_string field, so we need manual setup
-    const event_ptr = chain.get(0);
-    event_ptr.* = try eventInit(arena, typ, chain.get(1));
-    chain.setMiddle(1, UIEvent.Type);
-    chain.setLeaf(2, child);
-
-    return chain.get(2);
-}
-
-pub fn mouseEvent(_: *const Factory, arena: Allocator, typ: String, mouse: MouseEvent, child: anytype) !*@TypeOf(child) {
-    const chain = try PrototypeChain(
-        &.{ Event, UIEvent, MouseEvent, @TypeOf(child) },
-    ).allocate(arena);
-
-    // Special case: Event has a _type_string field, so we need manual setup
-    const event_ptr = chain.get(0);
-    event_ptr.* = try eventInit(arena, typ, chain.get(1));
-    chain.setMiddle(1, UIEvent.Type);
-
-    // Set MouseEvent with all its fields
-    const mouse_ptr = chain.get(2);
-    mouse_ptr.* = mouse;
-    mouse_ptr._proto = chain.get(1);
-    mouse_ptr._type = unionInit(MouseEvent.Type, chain.get(3));
-
-    chain.setLeaf(3, child);
-
-    return chain.get(3);
-}
-
-fn PrototypeChain(comptime types: []const type) type {
-    return struct {
-        const Self = @This();
-        memory: []u8,
-
-        fn totalSize() usize {
-            var size: usize = 0;
-            for (types) |T| {
-                size = std.mem.alignForward(usize, size, @alignOf(T));
-                size += @sizeOf(T);
-            }
-            return size;
-        }
-
-        fn maxAlign() std.mem.Alignment {
-            var alignment: std.mem.Alignment = .@"1";
-
-            for (types) |T| {
-                alignment = std.mem.Alignment.max(alignment, std.mem.Alignment.of(T));
-            }
-
-            return alignment;
-        }
-
-        fn getType(comptime index: usize) type {
-            return types[index];
-        }
-
-        fn allocate(allocator: std.mem.Allocator) !Self {
-            const size = comptime Self.totalSize();
-            const alignment = comptime Self.maxAlign();
-
-            const memory = try allocator.alignedAlloc(u8, alignment, size);
-            return .{ .memory = memory };
-        }
-
-        fn get(self: *const Self, comptime index: usize) *getType(index) {
-            var offset: usize = 0;
-            inline for (types, 0..) |T, i| {
-                offset = std.mem.alignForward(usize, offset, @alignOf(T));
-
-                if (i == index) {
-                    return @as(*T, @ptrCast(@alignCast(self.memory.ptr + offset)));
-                }
-                offset += @sizeOf(T);
-            }
-            unreachable;
-        }
-
-        fn set(self: *const Self, comptime index: usize, value: getType(index)) void {
-            const ptr = self.get(index);
-            ptr.* = value;
-        }
-
-        fn setRoot(self: *const Self, comptime T: type) void {
-            const ptr = self.get(0);
-            ptr.* = .{ ._type = unionInit(T, self.get(1)) };
-        }
-
-        fn setMiddle(self: *const Self, comptime index: usize, comptime T: type) void {
-            assert(index >= 1);
-            assert(index < types.len);
-
-            const ptr = self.get(index);
-            ptr.* = .{ ._proto = self.get(index - 1), ._type = unionInit(T, self.get(index + 1)) };
-        }
-
-        fn setMiddleWithValue(self: *const Self, comptime index: usize, comptime T: type, value: anytype) void {
-            assert(index >= 1);
-
-            const ptr = self.get(index);
-            ptr.* = .{ ._proto = self.get(index - 1), ._type = unionInit(T, value) };
-        }
-
-        fn setLeaf(self: *const Self, comptime index: usize, value: anytype) void {
-            assert(index >= 1);
-
-            const ptr = self.get(index);
-            ptr.* = value;
-            ptr._proto = self.get(index - 1);
-        }
-    };
-}
-
-fn AutoPrototypeChain(comptime types: []const type) type {
-    return struct {
-        fn create(allocator: std.mem.Allocator, leaf_value: anytype) !*@TypeOf(leaf_value) {
-            const chain = try PrototypeChain(types).allocate(allocator);
-
-            const RootType = types[0];
-            chain.setRoot(RootType.Type);
-
-            inline for (1..types.len - 1) |i| {
-                const MiddleType = types[i];
-                chain.setMiddle(i, MiddleType.Type);
-            }
-
-            chain.setLeaf(types.len - 1, leaf_value);
-            return chain.get(types.len - 1);
-        }
-    };
-}
-
-fn eventInit(arena: Allocator, typ: String, value: anytype) !Event {
-    // Round to 2ms for privacy (browsers do this)
-    const raw_timestamp = @import("../datetime.zig").milliTimestamp(.monotonic);
-    const time_stamp = (raw_timestamp / 2) * 2;
-
-    return .{
-        ._rc = 0,
-        ._arena = arena,
-        ._type = unionInit(Event.Type, value),
-        ._type_string = typ,
-        ._time_stamp = time_stamp,
-    };
-}
-
-pub fn blob(self: *Factory, child: anytype) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-
-    // Special case: Blob has slice and mime fields, so we need manual setup
-    const chain = try PrototypeChain(
-        &.{ Blob, @TypeOf(child) },
-    ).allocate(allocator);
-
-    const blob_ptr = chain.get(0);
-    blob_ptr.* = .{
-        ._type = unionInit(Blob.Type, chain.get(1)),
-        ._slice = "",
-        ._mime = "",
-    };
-    chain.setLeaf(1, child);
-
-    return chain.get(1);
-}
-
-pub fn abstractRange(self: *Factory, child: anytype, page: *Page) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-    const chain = try PrototypeChain(&.{ AbstractRange, @TypeOf(child) }).allocate(allocator);
-
-    const doc = page.document.asNode();
-    chain.set(0, AbstractRange{
-        ._type = unionInit(AbstractRange.Type, chain.get(1)),
-        ._end_offset = 0,
-        ._start_offset = 0,
-        ._end_container = doc,
-        ._start_container = doc,
-    });
-    chain.setLeaf(1, child);
-    return chain.get(1);
-}
-
-pub fn node(self: *Factory, child: anytype) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-    return try AutoPrototypeChain(
-        &.{ EventTarget, Node, @TypeOf(child) },
-    ).create(allocator, child);
-}
-
-pub fn document(self: *Factory, child: anytype) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-    return try AutoPrototypeChain(
-        &.{ EventTarget, Node, Document, @TypeOf(child) },
-    ).create(allocator, child);
-}
-
-pub fn documentFragment(self: *Factory, child: anytype) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-    return try AutoPrototypeChain(
-        &.{ EventTarget, Node, Node.DocumentFragment, @TypeOf(child) },
-    ).create(allocator, child);
-}
-
-pub fn element(self: *Factory, child: anytype) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-    return try AutoPrototypeChain(
-        &.{ EventTarget, Node, Element, @TypeOf(child) },
-    ).create(allocator, child);
-}
-
-pub fn htmlElement(self: *Factory, child: anytype) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-    return try AutoPrototypeChain(
-        &.{ EventTarget, Node, Element, Element.Html, @TypeOf(child) },
-    ).create(allocator, child);
-}
-
-pub fn htmlMediaElement(self: *Factory, child: anytype) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-    return try AutoPrototypeChain(
-        &.{ EventTarget, Node, Element, Element.Html, Element.Html.Media, @TypeOf(child) },
-    ).create(allocator, child);
-}
-
-pub fn svgElement(self: *Factory, tag_name: []const u8, child: anytype) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-    const ChildT = @TypeOf(child);
-
-    if (ChildT == Element.Svg) {
-        return self.element(child);
-    }
-
-    const chain = try PrototypeChain(
-        &.{ EventTarget, Node, Element, Element.Svg, ChildT },
-    ).allocate(allocator);
-
-    chain.setRoot(EventTarget.Type);
-    chain.setMiddle(1, Node.Type);
-    chain.setMiddle(2, Element.Type);
-
-    // will never allocate, can't fail
-    const tag_name_str = String.init(self._arena, tag_name, .{}) catch unreachable;
-
-    // Manually set Element.Svg with the tag_name
-    chain.set(3, .{
-        ._proto = chain.get(2),
-        ._tag_name = tag_name_str,
-        ._type = unionInit(Element.Svg.Type, chain.get(4)),
-    });
-
-    chain.setLeaf(4, child);
-    return chain.get(4);
-}
-
-pub fn xhrEventTarget(_: *const Factory, allocator: Allocator, child: anytype) !*@TypeOf(child) {
-    return try AutoPrototypeChain(
-        &.{ EventTarget, XMLHttpRequestEventTarget, @TypeOf(child) },
-    ).create(allocator, child);
-}
-
-pub fn textTrackCue(self: *Factory, child: anytype) !*@TypeOf(child) {
-    const allocator = self._slab.allocator();
-    const TextTrackCue = @import("webapi/media/TextTrackCue.zig");
-
-    return try AutoPrototypeChain(
-        &.{ EventTarget, TextTrackCue, @TypeOf(child) },
-    ).create(allocator, child);
-}
-
-pub fn destroy(self: *Factory, value: anytype) void {
-    const S = reflect.Struct(@TypeOf(value));
-
-    if (comptime IS_DEBUG) {
-        // We should always destroy from the leaf down.
-        if (@hasDecl(S, "_prototype_root")) {
-            // A Event{._type == .generic} (or any other similar types)
-            // _should_ be destoyed directly. The _type = .generic is a pseudo
-            // child
-            if (S != Event or value._type != .generic) {
-                log.fatal(.bug, "factory.destroy.event", .{ .type = @typeName(S) });
-                unreachable;
-            }
-        }
-    }
-
-    if (comptime @hasField(S, "_proto")) {
-        self.destroyChain(value, 0, std.mem.Alignment.@"1");
-    } else {
-        self.destroyStandalone(value);
-    }
-}
-
-pub fn destroyStandalone(self: *Factory, value: anytype) void {
-    const allocator = self._slab.allocator();
-    allocator.destroy(value);
-}
-
-fn destroyChain(
-    self: *Factory,
-    value: anytype,
-    old_size: usize,
-    old_align: std.mem.Alignment,
-) void {
-    const S = reflect.Struct(@TypeOf(value));
-    const allocator = self._slab.allocator();
-
-    // aligns the old size to the alignment of this element
-    const current_size = std.mem.alignForward(usize, old_size, @alignOf(S));
-    const new_size = current_size + @sizeOf(S);
-    const new_align = std.mem.Alignment.max(old_align, std.mem.Alignment.of(S));
-
-    if (@hasField(S, "_proto")) {
-        self.destroyChain(value._proto, new_size, new_align);
-    } else {
-        // no proto so this is the head of the chain.
-        // we use this as the ptr to the start of the chain.
-        // and we have summed up the length.
-        assert(@hasDecl(S, "_prototype_root"));
-
-        const memory_ptr: [*]u8 = @ptrCast(@constCast(value));
-        const len = std.mem.alignForward(usize, new_size, new_align.toByteUnits());
-        allocator.rawFree(memory_ptr[0..len], new_align, @returnAddress());
-    }
-}
-
-pub fn createT(self: *Factory, comptime T: type) !*T {
-    const allocator = self._slab.allocator();
-    return try allocator.create(T);
-}
-
-pub fn create(self: *Factory, value: anytype) !*@TypeOf(value) {
-    const ptr = try self.createT(@TypeOf(value));
-    ptr.* = value;
-    return ptr;
-}
-
-fn unionInit(comptime T: type, value: anytype) T {
-    const V = @TypeOf(value);
-    const field_name = comptime unionFieldName(T, V);
-    return @unionInit(T, field_name, value);
-}
-
-// There can be friction between comptime and runtime. Comptime has to
-// account for all possible types, even if some runtime flow makes certain
-// cases impossible. At runtime, we always call `unionFieldName` with the
-// correct struct or pointer type. But at comptime time, `unionFieldName`
-// is called with both variants (S and *S). So we use reflect.Struct().
-// This only works because we never have a union with a field S and another
-// field *S.
-fn unionFieldName(comptime T: type, comptime V: type) []const u8 {
-    inline for (@typeInfo(T).@"union".fields) |field| {
-        if (reflect.Struct(field.type) == reflect.Struct(V)) {
-            return field.name;
-        }
-    }
-    @compileError(@typeName(V) ++ " is not a valid type for " ++ @typeName(T) ++ ".type");
-}

src/browser/Mime.zig

@@ -1,578 +0,0 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-
-const Mime = @This();
-content_type: ContentType,
-params: []const u8 = "",
-// IANA defines max. charset value length as 40.
-// We keep 41 for null-termination since HTML parser expects in this format.
-charset: [41]u8 = default_charset,
-charset_len: usize = default_charset_len,
-
-/// String "UTF-8" continued by null characters.
-const default_charset = .{ 'U', 'T', 'F', '-', '8' } ++ .{0} ** 36;
-const default_charset_len = 5;
-
-/// Mime with unknown Content-Type, empty params and empty charset.
-pub const unknown = Mime{ .content_type = .{ .unknown = {} } };
-
-pub const ContentTypeEnum = enum {
-    text_xml,
-    text_html,
-    text_javascript,
-    text_plain,
-    text_css,
-    image_jpeg,
-    image_gif,
-    image_png,
-    image_webp,
-    application_json,
-    unknown,
-    other,
-};
-
-pub const ContentType = union(ContentTypeEnum) {
-    text_xml: void,
-    text_html: void,
-    text_javascript: void,
-    text_plain: void,
-    text_css: void,
-    image_jpeg: void,
-    image_gif: void,
-    image_png: void,
-    image_webp: void,
-    application_json: void,
-    unknown: void,
-    other: struct { type: []const u8, sub_type: []const u8 },
-};
-
-pub fn contentTypeString(mime: *const Mime) []const u8 {
-    return switch (mime.content_type) {
-        .text_xml => "text/xml",
-        .text_html => "text/html",
-        .text_javascript => "application/javascript",
-        .text_plain => "text/plain",
-        .text_css => "text/css",
-        .image_jpeg => "image/jpeg",
-        .image_png => "image/png",
-        .image_gif => "image/gif",
-        .image_webp => "image/webp",
-        .application_json => "application/json",
-        else => "",
-    };
-}
-
-/// Returns the null-terminated charset value.
-pub fn charsetStringZ(mime: *const Mime) [:0]const u8 {
-    return mime.charset[0..mime.charset_len :0];
-}
-
-pub fn charsetString(mime: *const Mime) []const u8 {
-    return mime.charset[0..mime.charset_len];
-}
-
-/// Removes quotes of value if quotes are given.
-///
-/// Currently we don't validate the charset.
-/// See section 2.3 Naming Requirements:
-/// https://datatracker.ietf.org/doc/rfc2978/
-fn parseCharset(value: []const u8) error{ CharsetTooBig, Invalid }![]const u8 {
-    // Cannot be larger than 40.
-    // https://datatracker.ietf.org/doc/rfc2978/
-    if (value.len > 40) return error.CharsetTooBig;
-
-    // If the first char is a quote, look for a pair.
-    if (value[0] == '"') {
-        if (value.len < 3 or value[value.len - 1] != '"') {
-            return error.Invalid;
-        }
-
-        return value[1 .. value.len - 1];
-    }
-
-    // No quotes.
-    return value;
-}
-
-pub fn parse(input: []u8) !Mime {
-    if (input.len > 255) {
-        return error.TooBig;
-    }
-
-    // Zig's trim API is broken. The return type is always `[]const u8`,
-    // even if the input type is `[]u8`. @constCast is safe here.
-    var normalized = @constCast(std.mem.trim(u8, input, &std.ascii.whitespace));
-    _ = std.ascii.lowerString(normalized, normalized);
-
-    const content_type, const type_len = try parseContentType(normalized);
-    if (type_len >= normalized.len) {
-        return .{ .content_type = content_type };
-    }
-
-    const params = trimLeft(normalized[type_len..]);
-
-    var charset: [41]u8 = default_charset;
-    var charset_len: usize = default_charset_len;
-
-    var it = std.mem.splitScalar(u8, params, ';');
-    while (it.next()) |attr| {
-        const i = std.mem.indexOfScalarPos(u8, attr, 0, '=') orelse continue;
-        const name = trimLeft(attr[0..i]);
-
-        const value = trimRight(attr[i + 1 ..]);
-        if (value.len == 0) {
-            continue;
-        }
-
-        const attribute_name = std.meta.stringToEnum(enum {
-            charset,
-        }, name) orelse continue;
-
-        switch (attribute_name) {
-            .charset => {
-                if (value.len == 0) {
-                    break;
-                }
-
-                const attribute_value = parseCharset(value) catch continue;
-                @memcpy(charset[0..attribute_value.len], attribute_value);
-                // Null-terminate right after attribute value.
-                charset[attribute_value.len] = 0;
-                charset_len = attribute_value.len;
-            },
-        }
-    }
-
-    return .{
-        .params = params,
-        .charset = charset,
-        .charset_len = charset_len,
-        .content_type = content_type,
-    };
-}
-
-pub fn sniff(body: []const u8) ?Mime {
-    // 0x0C is form feed
-    const content = std.mem.trimLeft(u8, body, &.{ ' ', '\t', '\n', '\r', 0x0C });
-    if (content.len == 0) {
-        return null;
-    }
-
-    if (content[0] != '<') {
-        if (std.mem.startsWith(u8, content, &.{ 0xEF, 0xBB, 0xBF })) {
-            // UTF-8 BOM
-            return .{ .content_type = .{ .text_plain = {} } };
-        }
-        if (std.mem.startsWith(u8, content, &.{ 0xFE, 0xFF })) {
-            // UTF-16 big-endian BOM
-            return .{ .content_type = .{ .text_plain = {} } };
-        }
-        if (std.mem.startsWith(u8, content, &.{ 0xFF, 0xFE })) {
-            // UTF-16 little-endian BOM
-            return .{ .content_type = .{ .text_plain = {} } };
-        }
-        return null;
-    }
-
-    // The longest prefix we have is "<!DOCTYPE HTML ", 15 bytes. If we're
-    // here, we already know content[0] == '<', so we can skip that. So 14
-    // bytes.
-
-    // +1 because we don't need the leading '<'
-    var buf: [14]u8 = undefined;
-
-    const stripped = content[1..];
-    const prefix_len = @min(stripped.len, buf.len);
-    const prefix = std.ascii.lowerString(&buf, stripped[0..prefix_len]);
-
-    // we already know it starts with a <
-    const known_prefixes = [_]struct { []const u8, ContentType }{
-        .{ "!doctype html", .{ .text_html = {} } },
-        .{ "html", .{ .text_html = {} } },
-        .{ "script", .{ .text_html = {} } },
-        .{ "iframe", .{ .text_html = {} } },
-        .{ "h1", .{ .text_html = {} } },
-        .{ "div", .{ .text_html = {} } },
-        .{ "font", .{ .text_html = {} } },
-        .{ "table", .{ .text_html = {} } },
-        .{ "a", .{ .text_html = {} } },
-        .{ "style", .{ .text_html = {} } },
-        .{ "title", .{ .text_html = {} } },
-        .{ "b", .{ .text_html = {} } },
-        .{ "body", .{ .text_html = {} } },
-        .{ "br", .{ .text_html = {} } },
-        .{ "p", .{ .text_html = {} } },
-        .{ "!--", .{ .text_html = {} } },
-        .{ "xml", .{ .text_xml = {} } },
-    };
-    inline for (known_prefixes) |kp| {
-        const known_prefix = kp.@"0";
-        if (std.mem.startsWith(u8, prefix, known_prefix) and prefix.len > known_prefix.len) {
-            const next = prefix[known_prefix.len];
-            // a "tag-terminating-byte"
-            if (next == ' ' or next == '>') {
-                return .{ .content_type = kp.@"1" };
-            }
-        }
-    }
-
-    return null;
-}
-
-pub fn isHTML(self: *const Mime) bool {
-    return self.content_type == .text_html;
-}
-
-// we expect value to be lowercase
-fn parseContentType(value: []const u8) !struct { ContentType, usize } {
-    const end = std.mem.indexOfScalarPos(u8, value, 0, ';') orelse value.len;
-    const type_name = trimRight(value[0..end]);
-    const attribute_start = end + 1;
-
-    if (std.meta.stringToEnum(enum {
-        @"text/xml",
-        @"text/html",
-        @"text/css",
-        @"text/plain",
-
-        @"text/javascript",
-        @"application/javascript",
-        @"application/x-javascript",
-
-        @"image/jpeg",
-        @"image/png",
-        @"image/gif",
-        @"image/webp",
-
-        @"application/json",
-    }, type_name)) |known_type| {
-        const ct: ContentType = switch (known_type) {
-            .@"text/xml" => .{ .text_xml = {} },
-            .@"text/html" => .{ .text_html = {} },
-            .@"text/javascript", .@"application/javascript", .@"application/x-javascript" => .{ .text_javascript = {} },
-            .@"text/plain" => .{ .text_plain = {} },
-            .@"text/css" => .{ .text_css = {} },
-            .@"image/jpeg" => .{ .image_jpeg = {} },
-            .@"image/png" => .{ .image_png = {} },
-            .@"image/gif" => .{ .image_gif = {} },
-            .@"image/webp" => .{ .image_webp = {} },
-            .@"application/json" => .{ .application_json = {} },
-        };
-        return .{ ct, attribute_start };
-    }
-
-    const separator = std.mem.indexOfScalarPos(u8, type_name, 0, '/') orelse return error.Invalid;
-
-    const main_type = value[0..separator];
-    const sub_type = trimRight(value[separator + 1 .. end]);
-
-    if (main_type.len == 0 or validType(main_type) == false) {
-        return error.Invalid;
-    }
-    if (sub_type.len == 0 or validType(sub_type) == false) {
-        return error.Invalid;
-    }
-
-    return .{ .{ .other = .{
-        .type = main_type,
-        .sub_type = sub_type,
-    } }, attribute_start };
-}
-
-const VALID_CODEPOINTS = blk: {
-    var v: [256]bool = undefined;
-    for (0..256) |i| {
-        v[i] = std.ascii.isAlphanumeric(i);
-    }
-    for ("!#$%&\\*+-.^'_`|~") |b| {
-        v[b] = true;
-    }
-    break :blk v;
-};
-
-fn validType(value: []const u8) bool {
-    for (value) |b| {
-        if (VALID_CODEPOINTS[b] == false) {
-            return false;
-        }
-    }
-    return true;
-}
-
-fn trimLeft(s: []const u8) []const u8 {
-    return std.mem.trimLeft(u8, s, &std.ascii.whitespace);
-}
-
-fn trimRight(s: []const u8) []const u8 {
-    return std.mem.trimRight(u8, s, &std.ascii.whitespace);
-}
-
-const testing = @import("../testing.zig");
-test "Mime: invalid" {
-    defer testing.reset();
-
-    const invalids = [_][]const u8{
-        "",
-        "text",
-        "text /html",
-        "text/ html",
-        "text / html",
-        "text/html other",
-    };
-
-    for (invalids) |invalid| {
-        const mutable_input = try testing.arena_allocator.dupe(u8, invalid);
-        try testing.expectError(error.Invalid, Mime.parse(mutable_input));
-    }
-}
-
-test "Mime: malformed parameters are ignored" {
-    defer testing.reset();
-
-    // These should all parse successfully as text/html with malformed params ignored
-    const valid_with_malformed_params = [_][]const u8{
-        "text/html; x",
-        "text/html; x=",
-        "text/html; x=  ",
-        "text/html; = ",
-        "text/html;=",
-        "text/html; charset=\"\"",
-        "text/html; charset=\"",
-        "text/html; charset=\"\\",
-        "text/html;\"",
-    };
-
-    for (valid_with_malformed_params) |input| {
-        const mutable_input = try testing.arena_allocator.dupe(u8, input);
-        const mime = try Mime.parse(mutable_input);
-        try testing.expectEqual(.text_html, std.meta.activeTag(mime.content_type));
-    }
-}
-
-test "Mime: parse common" {
-    defer testing.reset();
-
-    try expect(.{ .content_type = .{ .text_xml = {} } }, "text/xml");
-    try expect(.{ .content_type = .{ .text_html = {} } }, "text/html");
-    try expect(.{ .content_type = .{ .text_plain = {} } }, "text/plain");
-
-    try expect(.{ .content_type = .{ .text_xml = {} } }, "text/xml;");
-    try expect(.{ .content_type = .{ .text_html = {} } }, "text/html;");
-    try expect(.{ .content_type = .{ .text_plain = {} } }, "text/plain;");
-
-    try expect(.{ .content_type = .{ .text_xml = {} } }, "  \ttext/xml");
-    try expect(.{ .content_type = .{ .text_html = {} } }, "text/html   ");
-    try expect(.{ .content_type = .{ .text_plain = {} } }, "text/plain \t\t");
-
-    try expect(.{ .content_type = .{ .text_xml = {} } }, "TEXT/xml");
-    try expect(.{ .content_type = .{ .text_html = {} } }, "text/Html");
-    try expect(.{ .content_type = .{ .text_plain = {} } }, "TEXT/PLAIN");
-
-    try expect(.{ .content_type = .{ .text_xml = {} } }, " TeXT/xml");
-    try expect(.{ .content_type = .{ .text_html = {} } }, "teXt/HtML  ;");
-    try expect(.{ .content_type = .{ .text_plain = {} } }, "tExT/PlAiN;");
-
-    try expect(.{ .content_type = .{ .text_javascript = {} } }, "text/javascript");
-    try expect(.{ .content_type = .{ .text_javascript = {} } }, "Application/JavaScript");
-    try expect(.{ .content_type = .{ .text_javascript = {} } }, "application/x-javascript");
-
-    try expect(.{ .content_type = .{ .application_json = {} } }, "application/json");
-    try expect(.{ .content_type = .{ .text_css = {} } }, "text/css");
-
-    try expect(.{ .content_type = .{ .image_jpeg = {} } }, "image/jpeg");
-    try expect(.{ .content_type = .{ .image_png = {} } }, "image/png");
-    try expect(.{ .content_type = .{ .image_gif = {} } }, "image/gif");
-    try expect(.{ .content_type = .{ .image_webp = {} } }, "image/webp");
-}
-
-test "Mime: parse uncommon" {
-    defer testing.reset();
-
-    const text_csv = Expectation{
-        .content_type = .{ .other = .{ .type = "text", .sub_type = "csv" } },
-    };
-    try expect(text_csv, "text/csv");
-    try expect(text_csv, "text/csv;");
-    try expect(text_csv, "  text/csv\t  ");
-    try expect(text_csv, "  text/csv\t  ;");
-
-    try expect(
-        .{ .content_type = .{ .other = .{ .type = "text", .sub_type = "csv" } } },
-        "Text/CSV",
-    );
-}
-
-test "Mime: parse charset" {
-    defer testing.reset();
-
-    try expect(.{
-        .content_type = .{ .text_xml = {} },
-        .charset = "utf-8",
-        .params = "charset=utf-8",
-    }, "text/xml; charset=utf-8");
-
-    try expect(.{
-        .content_type = .{ .text_xml = {} },
-        .charset = "utf-8",
-        .params = "charset=\"utf-8\"",
-    }, "text/xml;charset=\"UTF-8\"");
-
-    try expect(.{
-        .content_type = .{ .text_html = {} },
-        .charset = "iso-8859-1",
-        .params = "charset=\"iso-8859-1\"",
-    }, "text/html; charset=\"iso-8859-1\"");
-
-    try expect(.{
-        .content_type = .{ .text_html = {} },
-        .charset = "iso-8859-1",
-        .params = "charset=\"iso-8859-1\"",
-    }, "text/html; charset=\"ISO-8859-1\"");
-
-    try expect(.{
-        .content_type = .{ .text_xml = {} },
-        .charset = "custom-non-standard-charset-value",
-        .params = "charset=\"custom-non-standard-charset-value\"",
-    }, "text/xml;charset=\"custom-non-standard-charset-value\"");
-
-    try expect(.{
-        .content_type = .{ .text_html = {} },
-        .charset = "UTF-8",
-        .params = "x=\"",
-    }, "text/html;x=\"");
-}
-
-test "Mime: isHTML" {
-    defer testing.reset();
-
-    const assert = struct {
-        fn assert(expected: bool, input: []const u8) !void {
-            const mutable_input = try testing.arena_allocator.dupe(u8, input);
-            var mime = try Mime.parse(mutable_input);
-            try testing.expectEqual(expected, mime.isHTML());
-        }
-    }.assert;
-    try assert(true, "text/html");
-    try assert(true, "text/html;");
-    try assert(true, "text/html; charset=utf-8");
-    try assert(false, "text/htm"); // htm not html
-    try assert(false, "text/plain");
-    try assert(false, "over/9000");
-}
-
-test "Mime: sniff" {
-    try testing.expectEqual(null, Mime.sniff(""));
-    try testing.expectEqual(null, Mime.sniff("<htm"));
-    try testing.expectEqual(null, Mime.sniff("<html!"));
-    try testing.expectEqual(null, Mime.sniff("<a_"));
-    try testing.expectEqual(null, Mime.sniff("<!doctype html"));
-    try testing.expectEqual(null, Mime.sniff("<!doctype  html>"));
-    try testing.expectEqual(null, Mime.sniff("\n  <!doctype  html>"));
-    try testing.expectEqual(null, Mime.sniff("\n \t <font/>"));
-
-    const expectHTML = struct {
-        fn expect(input: []const u8) !void {
-            try testing.expectEqual(.text_html, std.meta.activeTag(Mime.sniff(input).?.content_type));
-        }
-    }.expect;
-
-    try expectHTML("<!doctype html ");
-    try expectHTML("\n  \t    <!DOCTYPE HTML ");
-
-    try expectHTML("<html ");
-    try expectHTML("\n  \t    <HtmL> even more stufff");
-
-    try expectHTML("<script>");
-    try expectHTML("\n  \t    <SCRIpt >alert(document.cookies)</script>");
-
-    try expectHTML("<iframe>");
-    try expectHTML(" \t    <ifRAME >");
-
-    try expectHTML("<h1>");
-    try expectHTML("  <H1>");
-
-    try expectHTML("<div>");
-    try expectHTML("\n\r\r  <DiV>");
-
-    try expectHTML("<font>");
-    try expectHTML("  <fonT>");
-
-    try expectHTML("<table>");
-    try expectHTML("\t\t<TAblE>");
-
-    try expectHTML("<a>");
-    try expectHTML("\n\n<A>");
-
-    try expectHTML("<style>");
-    try expectHTML("    \n\t <STyLE>");
-
-    try expectHTML("<title>");
-    try expectHTML("    \n\t <TITLE>");
-
-    try expectHTML("<b>");
-    try expectHTML("    \n\t <B>");
-
-    try expectHTML("<body>");
-    try expectHTML("    \n\t <BODY>");
-
-    try expectHTML("<br>");
-    try expectHTML("    \n\t <BR>");
-
-    try expectHTML("<p>");
-    try expectHTML("    \n\t <P>");
-
-    try expectHTML("<!-->");
-    try expectHTML("    \n\t <!-->");
-}
-
-const Expectation = struct {
-    content_type: Mime.ContentType,
-    params: []const u8 = "",
-    charset: ?[]const u8 = null,
-};
-
-fn expect(expected: Expectation, input: []const u8) !void {
-    const mutable_input = try testing.arena_allocator.dupe(u8, input);
-
-    const actual = try Mime.parse(mutable_input);
-    try testing.expectEqual(
-        std.meta.activeTag(expected.content_type),
-        std.meta.activeTag(actual.content_type),
-    );
-
-    switch (expected.content_type) {
-        .other => |e| {
-            const a = actual.content_type.other;
-            try testing.expectEqual(e.type, a.type);
-            try testing.expectEqual(e.sub_type, a.sub_type);
-        },
-        else => {}, // already asserted above
-    }
-
-    try testing.expectEqual(expected.params, actual.params);
-
-    if (expected.charset) |ec| {
-        // We remove the null characters for testing purposes here.
-        try testing.expectEqual(ec, actual.charsetString());
-    } else {
-        const m: Mime = .unknown;
-        try testing.expectEqual(m.charsetStringZ(), actual.charsetStringZ());
-    }
-}

src/browser/Scheduler.zig

@@ -0,0 +1,163 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const Allocator = std.mem.Allocator;
+
+const Scheduler = @This();
+
+high_priority: Queue,
+
+// For repeating tasks. We only want to run these if there are other things to
+// do. We don't, for example, want a window.setInterval or the page.runMicrotasks
+// to block the page.wait.
+low_priority: Queue,
+
+// we expect allocator to be the page arena, hence we never call high_priority.deinit
+pub fn init(allocator: Allocator) Scheduler {
+    return .{
+        .high_priority = Queue.init(allocator, {}),
+        .low_priority = Queue.init(allocator, {}),
+    };
+}
+
+pub fn reset(self: *Scheduler) void {
+    // Our allocator is the page arena, it's been reset. We cannot use
+    // clearAndRetainCapacity, since that space is no longer ours
+    self.high_priority.clearAndFree();
+    self.low_priority.clearAndFree();
+}
+
+const AddOpts = struct {
+    name: []const u8 = "",
+    low_priority: bool = false,
+};
+pub fn add(self: *Scheduler, ctx: *anyopaque, func: Task.Func, ms: u32, opts: AddOpts) !void {
+    var low_priority = opts.low_priority;
+    if (ms > 5_000) {
+        // we don't want tasks in the far future to block page.wait from
+        // completing. However, if page.wait is called multiple times (maybe
+        // a CDP driver is wait for something to happen), then we do want
+        // to [eventually] run these when their time is up.
+        low_priority = true;
+    }
+
+    var q = if (low_priority) &self.low_priority else &self.high_priority;
+    return q.add(.{
+        .ms = std.time.milliTimestamp() + ms,
+        .ctx = ctx,
+        .func = func,
+        .name = opts.name,
+    });
+}
+
+pub fn run(self: *Scheduler) !?i32 {
+    _ = try self.runQueue(&self.low_priority);
+    return self.runQueue(&self.high_priority);
+}
+
+fn runQueue(self: *Scheduler, queue: *Queue) !?i32 {
+    // this is O(1)
+    if (queue.count() == 0) {
+        return null;
+    }
+
+    const now = std.time.milliTimestamp();
+
+    var next = queue.peek();
+    while (next) |task| {
+        const time_to_next = task.ms - now;
+        if (time_to_next > 0) {
+            // @intCast is petty safe since we limit tasks to just 5 seconds
+            // in the future
+            return @intCast(time_to_next);
+        }
+
+        if (task.func(task.ctx)) |repeat_delay| {
+            // if we do (now + 0) then our WHILE loop will run endlessly.
+            // no task should ever return 0
+            std.debug.assert(repeat_delay != 0);
+
+            var copy = task;
+            copy.ms = now + repeat_delay;
+            try self.low_priority.add(copy);
+        }
+        _ = queue.remove();
+        next = queue.peek();
+    }
+    return null;
+}
+
+const Task = struct {
+    ms: i64,
+    func: Func,
+    ctx: *anyopaque,
+    name: []const u8,
+
+    const Func = *const fn (ctx: *anyopaque) ?u32;
+};
+
+const Queue = std.PriorityQueue(Task, void, struct {
+    fn compare(_: void, a: Task, b: Task) std.math.Order {
+        return std.math.order(a.ms, b.ms);
+    }
+}.compare);
+
+const testing = @import("../testing.zig");
+test "Scheduler" {
+    defer testing.reset();
+
+    var task = TestTask{ .allocator = testing.arena_allocator };
+
+    var s = Scheduler.init(testing.arena_allocator);
+    try testing.expectEqual(null, s.run());
+    try testing.expectEqual(0, task.calls.items.len);
+
+    try s.add(&task, TestTask.run1, 3, .{});
+
+    try testing.expectDelta(3, try s.run(), 1);
+    try testing.expectEqual(0, task.calls.items.len);
+
+    std.Thread.sleep(std.time.ns_per_ms * 5);
+    try testing.expectEqual(null, s.run());
+    try testing.expectEqualSlices(u32, &.{1}, task.calls.items);
+
+    try s.add(&task, TestTask.run2, 3, .{});
+    try s.add(&task, TestTask.run1, 2, .{});
+
+    std.Thread.sleep(std.time.ns_per_ms * 5);
+    try testing.expectDelta(null, try s.run(), 1);
+    try testing.expectEqualSlices(u32, &.{ 1, 1, 2 }, task.calls.items);
+}
+
+const TestTask = struct {
+    allocator: Allocator,
+    calls: std.ArrayListUnmanaged(u32) = .{},
+
+    fn run1(ctx: *anyopaque) ?u32 {
+        var self: *TestTask = @ptrCast(@alignCast(ctx));
+        self.calls.append(self.allocator, 1) catch unreachable;
+        return null;
+    }
+
+    fn run2(ctx: *anyopaque) ?u32 {
+        var self: *TestTask = @ptrCast(@alignCast(ctx));
+        self.calls.append(self.allocator, 2) catch unreachable;
+        return 2;
+    }
+};

src/browser/Session.zig

@@ -1,384 +0,0 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-const lp = @import("lightpanda");
-const builtin = @import("builtin");
-
-const log = @import("../log.zig");
-
-const js = @import("js/js.zig");
-const storage = @import("webapi/storage/storage.zig");
-const Navigation = @import("webapi/navigation/Navigation.zig");
-const History = @import("webapi/History.zig");
-
-const Page = @import("Page.zig");
-const Browser = @import("Browser.zig");
-const Notification = @import("../Notification.zig");
-
-const Allocator = std.mem.Allocator;
-const IS_DEBUG = builtin.mode == .Debug;
-
-// Session is like a browser's tab.
-// It owns the js env and the loader for all the pages of the session.
-// You can create successively multiple pages for a session, but you must
-// deinit a page before running another one.
-const Session = @This();
-
-browser: *Browser,
-notification: *Notification,
-
-// Used to create our Inspector and in the BrowserContext.
-arena: Allocator,
-
-cookie_jar: storage.Cookie.Jar,
-storage_shed: storage.Shed,
-
-history: History,
-navigation: Navigation,
-
-page: ?Page,
-
-frame_id_gen: u32,
-
-pub fn init(self: *Session, browser: *Browser, notification: *Notification) !void {
-    const allocator = browser.app.allocator;
-    const arena = try browser.arena_pool.acquire();
-    errdefer browser.arena_pool.release(arena);
-
-    self.* = .{
-        .page = null,
-        .arena = arena,
-        .history = .{},
-        .frame_id_gen = 0,
-        // The prototype (EventTarget) for Navigation is created when a Page is created.
-        .navigation = .{ ._proto = undefined },
-        .storage_shed = .{},
-        .browser = browser,
-        .notification = notification,
-        .cookie_jar = storage.Cookie.Jar.init(allocator),
-    };
-}
-
-pub fn deinit(self: *Session) void {
-    if (self.page != null) {
-        self.removePage();
-    }
-    const browser = self.browser;
-
-    self.cookie_jar.deinit();
-    self.storage_shed.deinit(browser.app.allocator);
-    browser.arena_pool.release(self.arena);
-}
-
-// NOTE: the caller is not the owner of the returned value,
-// the pointer on Page is just returned as a convenience
-pub fn createPage(self: *Session) !*Page {
-    lp.assert(self.page == null, "Session.createPage - page not null", .{});
-
-    self.page = @as(Page, undefined);
-    const page = &self.page.?;
-    try Page.init(page, self.nextFrameId(), self, null);
-
-    // Creates a new NavigationEventTarget for this page.
-    try self.navigation.onNewPage(page);
-
-    if (comptime IS_DEBUG) {
-        log.debug(.browser, "create page", .{});
-    }
-    // start JS env
-    // Inform CDP the main page has been created such that additional context for other Worlds can be created as well
-    self.notification.dispatch(.page_created, page);
-
-    return page;
-}
-
-pub fn removePage(self: *Session) void {
-    // Inform CDP the page is going to be removed, allowing other worlds to remove themselves before the main one
-    self.notification.dispatch(.page_remove, .{});
-    lp.assert(self.page != null, "Session.removePage - page is null", .{});
-
-    self.page.?.deinit();
-    self.page = null;
-
-    self.navigation.onRemovePage();
-
-    if (comptime IS_DEBUG) {
-        log.debug(.browser, "remove page", .{});
-    }
-}
-
-pub fn replacePage(self: *Session) !*Page {
-    if (comptime IS_DEBUG) {
-        log.debug(.browser, "replace page", .{});
-    }
-
-    lp.assert(self.page != null, "Session.replacePage null page", .{});
-
-    var current = self.page.?;
-    const frame_id = current._frame_id;
-    const parent = current.parent;
-    current.deinit();
-
-    self.browser.env.memoryPressureNotification(.moderate);
-
-    self.page = @as(Page, undefined);
-    const page = &self.page.?;
-    try Page.init(page, frame_id, self, parent);
-    return page;
-}
-
-pub fn currentPage(self: *Session) ?*Page {
-    return &(self.page orelse return null);
-}
-
-pub const WaitResult = enum {
-    done,
-    no_page,
-    cdp_socket,
-};
-
-pub fn findPage(self: *Session, frame_id: u32) ?*Page {
-    const page = self.currentPage() orelse return null;
-    return if (page._frame_id == frame_id) page else null;
-}
-
-pub fn wait(self: *Session, wait_ms: u32) WaitResult {
-    var page = &(self.page orelse return .no_page);
-    while (true) {
-        const wait_result = self._wait(page, wait_ms) catch |err| {
-            switch (err) {
-                error.JsError => {}, // already logged (with hopefully more context)
-                else => log.err(.browser, "session wait", .{
-                    .err = err,
-                    .url = page.url,
-                }),
-            }
-            return .done;
-        };
-
-        switch (wait_result) {
-            .done => {
-                if (page._queued_navigation == null) {
-                    return .done;
-                }
-                page = self.processScheduledNavigation(page) catch return .done;
-            },
-            else => |result| return result,
-        }
-    }
-}
-
-fn _wait(self: *Session, page: *Page, wait_ms: u32) !WaitResult {
-    var timer = try std.time.Timer.start();
-    var ms_remaining = wait_ms;
-
-    const browser = self.browser;
-    var http_client = browser.http_client;
-
-    // I'd like the page to know NOTHING about cdp_socket / CDP, but the
-    // fact is that the behavior of wait changes depending on whether or
-    // not we're using CDP.
-    // If we aren't using CDP, as soon as we think there's nothing left
-    // to do, we can exit - we'de done.
-    // But if we are using CDP, we should wait for the whole `wait_ms`
-    // because the http_click.tick() also monitors the CDP socket. And while
-    // we could let CDP poll http (like it does for HTTP requests), the fact
-    // is that we know more about the timing of stuff (e.g. how long to
-    // poll/sleep) in the page.
-    const exit_when_done = http_client.cdp_client == null;
-
-    while (true) {
-        switch (page._parse_state) {
-            .pre, .raw, .text, .image => {
-                // The main page hasn't started/finished navigating.
-                // There's no JS to run, and no reason to run the scheduler.
-                if (http_client.active == 0 and exit_when_done) {
-                    // haven't started navigating, I guess.
-                    return .done;
-                }
-                // Either we have active http connections, or we're in CDP
-                // mode with an extra socket. Either way, we're waiting
-                // for http traffic
-                if (try http_client.tick(@intCast(ms_remaining)) == .cdp_socket) {
-                    // exit_when_done is explicitly set when there isn't
-                    // an extra socket, so it should not be possibl to
-                    // get an cdp_socket message when exit_when_done
-                    // is true.
-                    if (IS_DEBUG) {
-                        std.debug.assert(exit_when_done == false);
-                    }
-
-                    // data on a socket we aren't handling, return to caller
-                    return .cdp_socket;
-                }
-            },
-            .html, .complete => {
-                if (page._queued_navigation != null) {
-                    return .done;
-                }
-
-                // The HTML page was parsed. We now either have JS scripts to
-                // download, or scheduled tasks to execute, or both.
-
-                // scheduler.run could trigger new http transfers, so do not
-                // store http_client.active BEFORE this call and then use
-                // it AFTER.
-                const ms_to_next_task = try browser.runMacrotasks();
-
-                // Each call to this runs scheduled load events.
-                try page.dispatchLoad();
-
-                const http_active = http_client.active;
-                const total_network_activity = http_active + http_client.intercepted;
-                if (page._notified_network_almost_idle.check(total_network_activity <= 2)) {
-                    page.notifyNetworkAlmostIdle();
-                }
-                if (page._notified_network_idle.check(total_network_activity == 0)) {
-                    page.notifyNetworkIdle();
-                }
-
-                if (http_active == 0 and exit_when_done) {
-                    // we don't need to consider http_client.intercepted here
-                    // because exit_when_done is true, and that can only be
-                    // the case when interception isn't possible.
-                    if (comptime IS_DEBUG) {
-                        std.debug.assert(http_client.intercepted == 0);
-                    }
-
-                    const ms: u64 = ms_to_next_task orelse blk: {
-                        if (wait_ms - ms_remaining < 100) {
-                            if (comptime builtin.is_test) {
-                                return .done;
-                            }
-                            // Look, we want to exit ASAP, but we don't want
-                            // to exit so fast that we've run none of the
-                            // background jobs.
-                            break :blk 50;
-                        }
-
-                        if (browser.hasBackgroundTasks()) {
-                            // _we_ have nothing to run, but v8 is working on
-                            // background tasks. We'll wait for them.
-                            browser.waitForBackgroundTasks();
-                            break :blk 20;
-                        }
-
-                        // No http transfers, no cdp extra socket, no
-                        // scheduled tasks, we're done.
-                        return .done;
-                    };
-
-                    if (ms > ms_remaining) {
-                        // Same as above, except we have a scheduled task,
-                        // it just happens to be too far into the future
-                        // compared to how long we were told to wait.
-                        return .done;
-                    }
-
-                    // We have a task to run in the not-so-distant future.
-                    // You might think we can just sleep until that task is
-                    // ready, but we should continue to run lowPriority tasks
-                    // in the meantime, and that could unblock things. So
-                    // we'll just sleep for a bit, and then restart our wait
-                    // loop to see if anything new can be processed.
-                    std.Thread.sleep(std.time.ns_per_ms * @as(u64, @intCast(@min(ms, 20))));
-                } else {
-                    // We're here because we either have active HTTP
-                    // connections, or exit_when_done == false (aka, there's
-                    // an cdp_socket registered with the http client).
-                    // We should continue to run lowPriority tasks, so we
-                    // minimize how long we'll poll for network I/O.
-                    var ms_to_wait = @min(200, ms_to_next_task orelse 200);
-                    if (ms_to_wait > 10 and browser.hasBackgroundTasks()) {
-                        // if we have background tasks, we don't want to wait too
-                        // long for a message from the client. We want to go back
-                        // to the top of the loop and run macrotasks.
-                        ms_to_wait = 10;
-                    }
-                    if (try http_client.tick(@min(ms_remaining, ms_to_wait)) == .cdp_socket) {
-                        // data on a socket we aren't handling, return to caller
-                        return .cdp_socket;
-                    }
-                }
-            },
-            .err => |err| {
-                page._parse_state = .{ .raw_done = @errorName(err) };
-                return err;
-            },
-            .raw_done => {
-                if (exit_when_done) {
-                    return .done;
-                }
-                // we _could_ http_client.tick(ms_to_wait), but this has
-                // the same result, and I feel is more correct.
-                return .no_page;
-            },
-        }
-
-        const ms_elapsed = timer.lap() / 1_000_000;
-        if (ms_elapsed >= ms_remaining) {
-            return .done;
-        }
-        ms_remaining -= @intCast(ms_elapsed);
-    }
-}
-
-fn processScheduledNavigation(self: *Session, current_page: *Page) !*Page {
-    const browser = self.browser;
-
-    const qn = current_page._queued_navigation.?;
-    // take ownership of the page's queued navigation
-    current_page._queued_navigation = null;
-    defer browser.arena_pool.release(qn.arena);
-
-    const frame_id, const parent = blk: {
-        const page = &self.page.?;
-        const frame_id = page._frame_id;
-        const parent = page.parent;
-
-        browser.http_client.abort();
-        self.removePage();
-
-        break :blk .{ frame_id, parent };
-    };
-
-    self.page = @as(Page, undefined);
-    const page = &self.page.?;
-    try Page.init(page, frame_id, self, parent);
-
-    // Creates a new NavigationEventTarget for this page.
-    try self.navigation.onNewPage(page);
-
-    // start JS env
-    // Inform CDP the main page has been created such that additional context for other Worlds can be created as well
-    self.notification.dispatch(.page_created, page);
-
-    page.navigate(qn.url, qn.opts) catch |err| {
-        log.err(.browser, "queued navigation error", .{ .err = err, .url = qn.url });
-        return err;
-    };
-
-    return page;
-}
-
-pub fn nextFrameId(self: *Session) u32 {
-    const id = self.frame_id_gen +% 1;
-    self.frame_id_gen = id;
-    return id;
-}

src/browser/SlotChangeMonitor.zig

@@ -0,0 +1,189 @@
+const std = @import("std");
+
+const log = @import("../log.zig");
+const parser = @import("netsurf.zig");
+const collection = @import("dom/html_collection.zig");
+
+const Page = @import("page.zig").Page;
+
+const SlotChangeMonitor = @This();
+
+page: *Page,
+event_node: parser.EventNode,
+slots_changed: std.ArrayList(*parser.Slot),
+
+// Monitors the document in order to trigger slotchange events.
+pub fn init(page: *Page) !*SlotChangeMonitor {
+    // on the heap, we need a stable address for event_node
+    const self = try page.arena.create(SlotChangeMonitor);
+    self.* = .{
+        .page = page,
+        .slots_changed = .empty,
+        .event_node = .{ .func = mutationCallback },
+    };
+    const root = parser.documentToNode(parser.documentHTMLToDocument(page.window.document));
+
+    _ = try parser.eventTargetAddEventListener(
+        parser.toEventTarget(parser.Node, root),
+        "DOMNodeInserted",
+        &self.event_node,
+        false,
+    );
+
+    _ = try parser.eventTargetAddEventListener(
+        parser.toEventTarget(parser.Node, root),
+        "DOMNodeRemoved",
+        &self.event_node,
+        false,
+    );
+
+    _ = try parser.eventTargetAddEventListener(
+        parser.toEventTarget(parser.Node, root),
+        "DOMAttrModified",
+        &self.event_node,
+        false,
+    );
+
+    return self;
+}
+
+// Given a element, finds its slot, if any.
+pub fn findSlot(element: *parser.Element, page: *const Page) !?*parser.Slot {
+    const target_name = (try parser.elementGetAttribute(element, "slot")) orelse return null;
+    return findNamedSlot(element, target_name, page);
+}
+
+// Given an element and a name, find the slo, if any. This is only useful for
+// MutationEvents where findSlot is unreliable because parser.elementGetAttribute(element, "slot")
+// could return the new or old value.
+fn findNamedSlot(element: *parser.Element, target_name: []const u8, page: *const Page) !?*parser.Slot {
+    // I believe elements need to be added as direct descendents of the host,
+    // so we don't need to go find the host, we just grab the parent.
+    const host = parser.nodeParentNode(@ptrCast(element)) orelse return null;
+    const state = page.getNodeState(host) orelse return null;
+    const shadow_root = state.shadow_root orelse return null;
+
+    // if we're here, we found a host, now find the slot
+    var nodes = collection.HTMLCollectionByTagName(
+        @ptrCast(@alignCast(shadow_root.proto)),
+        "slot",
+        .{ .include_root = false },
+    );
+    for (0..1000) |i| {
+        const n = (try nodes.item(@intCast(i))) orelse return null;
+        const slot_name = (try parser.elementGetAttribute(@ptrCast(n), "name")) orelse "";
+        if (std.mem.eql(u8, target_name, slot_name)) {
+            return @ptrCast(n);
+        }
+    }
+    return null;
+}
+
+// Event callback from the mutation event, signaling either the addition of
+// a node, removal of a node, or a change in attribute
+fn mutationCallback(en: *parser.EventNode, event: *parser.Event) void {
+    const mutation_event = parser.eventToMutationEvent(event);
+    const self: *SlotChangeMonitor = @fieldParentPtr("event_node", en);
+    self._mutationCallback(mutation_event) catch |err| {
+        log.err(.web_api, "slot change callback", .{ .err = err });
+    };
+}
+
+fn _mutationCallback(self: *SlotChangeMonitor, event: *parser.MutationEvent) !void {
+    const event_type = parser.eventType(@ptrCast(event));
+    if (std.mem.eql(u8, event_type, "DOMNodeInserted")) {
+        const event_target = parser.eventTarget(@ptrCast(event)) orelse return;
+        return self.nodeAddedOrRemoved(@ptrCast(event_target));
+    }
+
+    if (std.mem.eql(u8, event_type, "DOMNodeRemoved")) {
+        const event_target = parser.eventTarget(@ptrCast(event)) orelse return;
+        return self.nodeAddedOrRemoved(@ptrCast(event_target));
+    }
+
+    if (std.mem.eql(u8, event_type, "DOMAttrModified")) {
+        const attribute_name = try parser.mutationEventAttributeName(event);
+        if (std.mem.eql(u8, attribute_name, "slot") == false) {
+            return;
+        }
+
+        const new_value = parser.mutationEventNewValue(event);
+        const prev_value = parser.mutationEventPrevValue(event);
+        const event_target = parser.eventTarget(@ptrCast(event)) orelse return;
+        return self.nodeAttributeChanged(@ptrCast(event_target), new_value, prev_value);
+    }
+}
+
+// A node was removed or added. If it's an element, and if it has a slot attribute
+// then we'll dispatch a slotchange event.
+fn nodeAddedOrRemoved(self: *SlotChangeMonitor, node: *parser.Node) !void {
+    if (parser.nodeType(node) != .element) {
+        return;
+    }
+    const el: *parser.Element = @ptrCast(node);
+    if (try findSlot(el, self.page)) |slot| {
+        return self.scheduleSlotChange(slot);
+    }
+}
+
+// An attribute was modified. If the attribute is "slot", then we'll trigger 1
+// slotchange for the old slot (if there was one) and 1 slotchange for the new
+// one (if there is one)
+fn nodeAttributeChanged(self: *SlotChangeMonitor, node: *parser.Node, new_value: ?[]const u8, prev_value: ?[]const u8) !void {
+    if (parser.nodeType(node) != .element) {
+        return;
+    }
+
+    const el: *parser.Element = @ptrCast(node);
+    if (try findNamedSlot(el, prev_value orelse "", self.page)) |slot| {
+        try self.scheduleSlotChange(slot);
+    }
+
+    if (try findNamedSlot(el, new_value orelse "", self.page)) |slot| {
+        try self.scheduleSlotChange(slot);
+    }
+}
+
+// OK. Our MutationEvent is not a MutationObserver - it's an older, deprecated
+// API. It gets dispatched in the middle of the change. While I'm sure it has
+// some rules, from our point of view, it fires too early. DOMAttrModified fires
+// before the attribute is actually updated and DOMNodeRemoved before the node
+// is actually removed. This is a problem if the callback will call
+// `slot.assignedNodes`, since that won't return the new state.
+// So, we use the page schedule to schedule the dispatching of the slotchange
+// event.
+fn scheduleSlotChange(self: *SlotChangeMonitor, slot: *parser.Slot) !void {
+    for (self.slots_changed.items) |changed| {
+        if (slot == changed) {
+            return;
+        }
+    }
+
+    try self.slots_changed.append(self.page.arena, slot);
+    if (self.slots_changed.items.len == 1) {
+        // first item added, schedule the callback
+        try self.page.scheduler.add(self, scheduleCallback, 0, .{ .name = "slot change" });
+    }
+}
+
+// Callback from the schedule. Time to dispatch the slotchange event
+fn scheduleCallback(ctx: *anyopaque) ?u32 {
+    var self: *SlotChangeMonitor = @ptrCast(@alignCast(ctx));
+    self._scheduleCallback() catch |err| {
+        log.err(.app, "slot change schedule", .{ .err = err });
+    };
+    return null;
+}
+
+fn _scheduleCallback(self: *SlotChangeMonitor) !void {
+    for (self.slots_changed.items) |slot| {
+        const event = try parser.eventCreate();
+        defer parser.eventDestroy(event);
+        try parser.eventInit(event, "slotchange", .{});
+        _ = try parser.eventTargetDispatchEvent(
+            parser.toEventTarget(parser.Element, @ptrCast(@alignCast(slot))),
+            event,
+        );
+    }
+    self.slots_changed.clearRetainingCapacity();
+}

src/browser/State.zig

@@ -0,0 +1,77 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+// Sometimes we need to extend libdom. For example, its HTMLDocument doesn't
+// have a readyState. We have a couple different options, such as making the
+// correction in libdom directly. Another option stems from the fact that every
+// libdom node has an opaque embedder_data field. This is the struct that we
+// lazily load into that field.
+//
+// It didn't originally start off as a collection of every single extension, but
+// this quickly proved necessary, since different fields are needed on the same
+// data at different levels of the prototype chain. This isn't memory efficient.
+
+const js = @import("js/js.zig");
+const parser = @import("netsurf.zig");
+const DataSet = @import("html/DataSet.zig");
+const ShadowRoot = @import("dom/shadow_root.zig").ShadowRoot;
+const StyleSheet = @import("cssom/StyleSheet.zig");
+const CSSStyleDeclaration = @import("cssom/CSSStyleDeclaration.zig");
+
+// for HTMLScript (but probably needs to be added to more)
+onload: ?js.Function = null,
+onerror: ?js.Function = null,
+
+// for HTMLElement
+style: CSSStyleDeclaration = .empty,
+dataset: ?DataSet = null,
+template_content: ?*parser.DocumentFragment = null,
+
+// For dom/element
+shadow_root: ?*ShadowRoot = null,
+
+// for html/document
+ready_state: ReadyState = .loading,
+
+// for html/HTMLStyleElement
+style_sheet: ?*StyleSheet = null,
+
+// for dom/document
+active_element: ?*parser.Element = null,
+adopted_style_sheets: ?js.Object = null,
+
+// for HTMLSelectElement
+// By default, if no option is explicitly selected, the first option should
+// be selected. However, libdom doesn't do this, and it sets the
+// selectedIndex to -1, which is a valid value for "nothing selected".
+// Therefore, when libdom says the selectedIndex == -1, we don't know if
+// it means that nothing is selected, or if the first option is selected by
+// default.
+// There are cases where this won't work, but when selectedIndex is
+// explicitly set, we set this boolean flag. Then, when we're getting then
+// selectedIndex, if this flag is == false, which is to say that if
+// selectedIndex hasn't been explicitly set AND if we have at least 1 option
+// AND if it isn't a multi select, we can make the 1st item selected by
+// default (by returning selectedIndex == 0).
+explicit_index_set: bool = false,
+
+const ReadyState = enum {
+    loading,
+    interactive,
+    complete,
+};

src/browser/browser.zig

@@ -0,0 +1,119 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const Allocator = std.mem.Allocator;
+const ArenaAllocator = std.heap.ArenaAllocator;
+
+const js = @import("js/js.zig");
+const State = @import("State.zig");
+const App = @import("../app.zig").App;
+const Session = @import("session.zig").Session;
+const Notification = @import("../notification.zig").Notification;
+
+const log = @import("../log.zig");
+const HttpClient = @import("../http/Client.zig");
+
+// Browser is an instance of the browser.
+// You can create multiple browser instances.
+// A browser contains only one session.
+pub const Browser = struct {
+    env: *js.Env,
+    app: *App,
+    session: ?Session,
+    allocator: Allocator,
+    http_client: *HttpClient,
+    call_arena: ArenaAllocator,
+    page_arena: ArenaAllocator,
+    session_arena: ArenaAllocator,
+    transfer_arena: ArenaAllocator,
+    notification: *Notification,
+    state_pool: std.heap.MemoryPool(State),
+
+    pub fn init(app: *App) !Browser {
+        const allocator = app.allocator;
+
+        const env = try js.Env.init(allocator, &app.platform, .{});
+        errdefer env.deinit();
+
+        const notification = try Notification.init(allocator, app.notification);
+        app.http.client.notification = notification;
+        app.http.client.next_request_id = 0; // Should we track ids in CDP only?
+        errdefer notification.deinit();
+
+        return .{
+            .app = app,
+            .env = env,
+            .session = null,
+            .allocator = allocator,
+            .notification = notification,
+            .http_client = app.http.client,
+            .call_arena = ArenaAllocator.init(allocator),
+            .page_arena = ArenaAllocator.init(allocator),
+            .session_arena = ArenaAllocator.init(allocator),
+            .transfer_arena = ArenaAllocator.init(allocator),
+            .state_pool = std.heap.MemoryPool(State).init(allocator),
+        };
+    }
+
+    pub fn deinit(self: *Browser) void {
+        self.closeSession();
+        self.env.deinit();
+        self.call_arena.deinit();
+        self.page_arena.deinit();
+        self.session_arena.deinit();
+        self.transfer_arena.deinit();
+        self.http_client.notification = null;
+        self.notification.deinit();
+        self.state_pool.deinit();
+    }
+
+    pub fn newSession(self: *Browser) !*Session {
+        self.closeSession();
+        self.session = @as(Session, undefined);
+        const session = &self.session.?;
+        try Session.init(session, self);
+        return session;
+    }
+
+    pub fn closeSession(self: *Browser) void {
+        if (self.session) |*session| {
+            session.deinit();
+            self.session = null;
+            _ = self.session_arena.reset(.{ .retain_with_limit = 1 * 1024 * 1024 });
+            self.env.lowMemoryNotification();
+        }
+    }
+
+    pub fn runMicrotasks(self: *const Browser) void {
+        self.env.runMicrotasks();
+    }
+
+    pub fn runMessageLoop(self: *const Browser) void {
+        while (self.env.pumpMessageLoop()) {
+            log.debug(.browser, "pumpMessageLoop", .{});
+        }
+        self.env.runIdleTasks();
+    }
+};
+
+const testing = @import("../testing.zig");
+test "Browser" {
+    try testing.htmlRunner("browser.html");
+}

src/browser/canvas/CanvasRenderingContext2D.zig

@@ -0,0 +1,58 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const color = @import("../cssom/color.zig");
+const Page = @import("../page.zig").Page;
+
+/// This class doesn't implement a `constructor`.
+/// It can be obtained with a call to `HTMLCanvasElement#getContext`.
+/// https://developer.mozilla.org/en-US/docs/Web/API/CanvasRenderingContext2D
+const CanvasRenderingContext2D = @This();
+/// Fill color.
+/// TODO: Add support for `CanvasGradient` and `CanvasPattern`.
+fill_style: color.RGBA = color.RGBA.Named.black,
+
+pub fn _fillRect(
+    self: *const CanvasRenderingContext2D,
+    x: f64,
+    y: f64,
+    width: f64,
+    height: f64,
+) void {
+    _ = self;
+    _ = x;
+    _ = y;
+    _ = width;
+    _ = height;
+}
+
+pub fn get_fillStyle(self: *const CanvasRenderingContext2D, page: *Page) ![]const u8 {
+    var w = std.Io.Writer.Allocating.init(page.call_arena);
+    try self.fill_style.format(&w.writer);
+    return w.written();
+}
+
+pub fn set_fillStyle(
+    self: *CanvasRenderingContext2D,
+    value: []const u8,
+) !void {
+    // Prefer the same fill_style if fails.
+    self.fill_style = color.RGBA.parse(value) catch self.fill_style;
+}

src/browser/canvas/WebGLRenderingContext.zig

@@ -18,20 +18,8 @@
 
 const std = @import("std");
 
-const js = @import("../../js/js.zig");
-const Page = @import("../../Page.zig");
-
-pub fn registerTypes() []const type {
-    return &.{
-        WebGLRenderingContext,
-        // Extension types should be runtime generated. We might want
-        // to revisit this.
-        Extension.Type.WEBGL_debug_renderer_info,
-        Extension.Type.WEBGL_lose_context,
-    };
-}
-
 const WebGLRenderingContext = @This();
+_: u8 = 0,
 
 /// On Chrome and Safari, a call to `getSupportedExtensions` returns total of 39.
 /// The reference for it lists lesser number of extensions:
@@ -69,11 +57,11 @@ pub const Extension = union(enum) {
     WEBGL_compressed_texture_pvrtc: void,
     WEBGL_compressed_texture_s3tc: void,
     WEBGL_compressed_texture_s3tc_srgb: void,
-    WEBGL_debug_renderer_info: *Type.WEBGL_debug_renderer_info,
+    WEBGL_debug_renderer_info: Type.WEBGL_debug_renderer_info,
     WEBGL_debug_shaders: void,
     WEBGL_depth_texture: void,
     WEBGL_draw_buffers: void,
-    WEBGL_lose_context: *Type.WEBGL_lose_context,
+    WEBGL_lose_context: Type.WEBGL_lose_context,
     WEBGL_multi_draw: void,
     WEBGL_polygon_mode: void,
 
@@ -118,93 +106,40 @@ pub const Extension = union(enum) {
     /// Extension types.
     pub const Type = struct {
         pub const WEBGL_debug_renderer_info = struct {
-            _: u8 = 0,
             pub const UNMASKED_VENDOR_WEBGL: u64 = 0x9245;
             pub const UNMASKED_RENDERER_WEBGL: u64 = 0x9246;
 
-            pub const JsApi = struct {
-                pub const bridge = js.Bridge(WEBGL_debug_renderer_info);
+            pub fn get_UNMASKED_VENDOR_WEBGL() u64 {
+                return UNMASKED_VENDOR_WEBGL;
+            }
 
-                pub const Meta = struct {
-                    pub const name = "WEBGL_debug_renderer_info";
-
-                    pub const prototype_chain = bridge.prototypeChain();
-                    pub var class_id: bridge.ClassId = undefined;
-                };
-
-                pub const UNMASKED_VENDOR_WEBGL = bridge.property(WEBGL_debug_renderer_info.UNMASKED_VENDOR_WEBGL, .{ .template = false, .readonly = true });
-                pub const UNMASKED_RENDERER_WEBGL = bridge.property(WEBGL_debug_renderer_info.UNMASKED_RENDERER_WEBGL, .{ .template = false, .readonly = true });
-            };
+            pub fn get_UNMASKED_RENDERER_WEBGL() u64 {
+                return UNMASKED_RENDERER_WEBGL;
+            }
         };
 
         pub const WEBGL_lose_context = struct {
             _: u8 = 0,
-            pub fn loseContext(_: *const WEBGL_lose_context) void {}
-            pub fn restoreContext(_: *const WEBGL_lose_context) void {}
-
-            pub const JsApi = struct {
-                pub const bridge = js.Bridge(WEBGL_lose_context);
-
-                pub const Meta = struct {
-                    pub const name = "WEBGL_lose_context";
-
-                    pub const prototype_chain = bridge.prototypeChain();
-                    pub var class_id: bridge.ClassId = undefined;
-                };
-
-                pub const loseContext = bridge.function(WEBGL_lose_context.loseContext, .{ .noop = true });
-                pub const restoreContext = bridge.function(WEBGL_lose_context.restoreContext, .{ .noop = true });
-            };
+            pub fn _loseContext(_: *const WEBGL_lose_context) void {}
+            pub fn _restoreContext(_: *const WEBGL_lose_context) void {}
         };
     };
 };
 
-/// This actually takes "GLenum" which, in fact, is a fancy way to say number.
-/// Return value also depends on what's being passed as `pname`; we don't really
-/// support any though.
-pub fn getParameter(_: *const WebGLRenderingContext, pname: u32) []const u8 {
-    _ = pname;
-    return "";
-}
-
 /// Enables a WebGL extension.
-pub fn getExtension(_: *const WebGLRenderingContext, name: []const u8, page: *Page) !?Extension {
+pub fn _getExtension(self: *const WebGLRenderingContext, name: []const u8) ?Extension {
+    _ = self;
+
     const tag = Extension.find(name) orelse return null;
 
     return switch (tag) {
-        .WEBGL_debug_renderer_info => {
-            const info = try page._factory.create(Extension.Type.WEBGL_debug_renderer_info{});
-            return .{ .WEBGL_debug_renderer_info = info };
-        },
-        .WEBGL_lose_context => {
-            const ctx = try page._factory.create(Extension.Type.WEBGL_lose_context{});
-            return .{ .WEBGL_lose_context = ctx };
-        },
+        .WEBGL_debug_renderer_info => @unionInit(Extension, "WEBGL_debug_renderer_info", .{}),
+        .WEBGL_lose_context => @unionInit(Extension, "WEBGL_lose_context", .{}),
         inline else => |comptime_enum| @unionInit(Extension, @tagName(comptime_enum), {}),
     };
 }
 
 /// Returns a list of all the supported WebGL extensions.
-pub fn getSupportedExtensions(_: *const WebGLRenderingContext) []const []const u8 {
+pub fn _getSupportedExtensions(_: *const WebGLRenderingContext) []const []const u8 {
     return std.meta.fieldNames(Extension.Kind);
 }
-
-pub const JsApi = struct {
-    pub const bridge = js.Bridge(WebGLRenderingContext);
-
-    pub const Meta = struct {
-        pub const name = "WebGLRenderingContext";
-
-        pub const prototype_chain = bridge.prototypeChain();
-        pub var class_id: bridge.ClassId = undefined;
-    };
-
-    pub const getParameter = bridge.function(WebGLRenderingContext.getParameter, .{});
-    pub const getExtension = bridge.function(WebGLRenderingContext.getExtension, .{});
-    pub const getSupportedExtensions = bridge.function(WebGLRenderingContext.getSupportedExtensions, .{});
-};
-
-const testing = @import("../../../testing.zig");
-test "WebApi: WebGLRenderingContext" {
-    try testing.htmlRunner("canvas/webgl_rendering_context.html", .{});
-}

src/browser/canvas/root.zig

@@ -0,0 +1,13 @@
+//! Canvas API.
+//! https://developer.mozilla.org/en-US/docs/Web/API/Canvas_API
+
+const CanvasRenderingContext2D = @import("CanvasRenderingContext2D.zig");
+const WebGLRenderingContext = @import("WebGLRenderingContext.zig");
+const Extension = WebGLRenderingContext.Extension;
+
+pub const Interfaces = .{
+    CanvasRenderingContext2D,
+    WebGLRenderingContext,
+    Extension.Type.WEBGL_debug_renderer_info,
+    Extension.Type.WEBGL_lose_context,
+};

src/browser/console/console.zig

@@ -0,0 +1,177 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const builtin = @import("builtin");
+const log = @import("../../log.zig");
+
+const js = @import("../js/js.zig");
+const Page = @import("../page.zig").Page;
+
+pub const Console = struct {
+    // TODO: configurable writer
+    timers: std.StringHashMapUnmanaged(u32) = .{},
+    counts: std.StringHashMapUnmanaged(u32) = .{},
+
+    pub fn _lp(values: []js.Object, page: *Page) !void {
+        if (values.len == 0) {
+            return;
+        }
+        log.fatal(.console, "lightpanda", .{ .args = try serializeValues(values, page) });
+    }
+
+    pub fn _log(values: []js.Object, page: *Page) !void {
+        if (values.len == 0) {
+            return;
+        }
+        log.info(.console, "info", .{ .args = try serializeValues(values, page) });
+    }
+
+    pub fn _info(values: []js.Object, page: *Page) !void {
+        return _log(values, page);
+    }
+
+    pub fn _debug(values: []js.Object, page: *Page) !void {
+        if (values.len == 0) {
+            return;
+        }
+        log.debug(.console, "debug", .{ .args = try serializeValues(values, page) });
+    }
+
+    pub fn _warn(values: []js.Object, page: *Page) !void {
+        if (values.len == 0) {
+            return;
+        }
+        log.warn(.console, "warn", .{ .args = try serializeValues(values, page) });
+    }
+
+    pub fn _error(values: []js.Object, page: *Page) !void {
+        if (values.len == 0) {
+            return;
+        }
+
+        log.warn(.console, "error", .{
+            .args = try serializeValues(values, page),
+            .stack = page.stackTrace() catch "???",
+        });
+    }
+
+    pub fn _trace(values: []js.Object, page: *Page) !void {
+        if (values.len == 0) {
+            return;
+        }
+        log.debug(.console, "debug", .{
+            .stack = page.js.stackTrace() catch "???",
+            .args = try serializeValues(values, page),
+        });
+    }
+
+    pub fn _clear() void {}
+
+    pub fn _count(self: *Console, label_: ?[]const u8, page: *Page) !void {
+        const label = label_ orelse "default";
+        const gop = try self.counts.getOrPut(page.arena, label);
+
+        var current: u32 = 0;
+        if (gop.found_existing) {
+            current = gop.value_ptr.*;
+        } else {
+            gop.key_ptr.* = try page.arena.dupe(u8, label);
+        }
+
+        const count = current + 1;
+        gop.value_ptr.* = count;
+
+        log.info(.console, "count", .{ .label = label, .count = count });
+    }
+
+    pub fn _countReset(self: *Console, label_: ?[]const u8) !void {
+        const label = label_ orelse "default";
+        const kv = self.counts.fetchRemove(label) orelse {
+            log.info(.console, "invalid counter", .{ .label = label });
+            return;
+        };
+        log.info(.console, "count reset", .{ .label = label, .count = kv.value });
+    }
+
+    pub fn _time(self: *Console, label_: ?[]const u8, page: *Page) !void {
+        const label = label_ orelse "default";
+        const gop = try self.timers.getOrPut(page.arena, label);
+
+        if (gop.found_existing) {
+            log.info(.console, "duplicate timer", .{ .label = label });
+            return;
+        }
+        gop.key_ptr.* = try page.arena.dupe(u8, label);
+        gop.value_ptr.* = timestamp();
+    }
+
+    pub fn _timeLog(self: *Console, label_: ?[]const u8) void {
+        const elapsed = timestamp();
+        const label = label_ orelse "default";
+        const start = self.timers.get(label) orelse {
+            log.info(.console, "invalid timer", .{ .label = label });
+            return;
+        };
+        log.info(.console, "timer", .{ .label = label, .elapsed = elapsed - start });
+    }
+
+    pub fn _timeStop(self: *Console, label_: ?[]const u8) void {
+        const elapsed = timestamp();
+        const label = label_ orelse "default";
+        const kv = self.timers.fetchRemove(label) orelse {
+            log.info(.console, "invalid timer", .{ .label = label });
+            return;
+        };
+
+        log.warn(.console, "timer stop", .{ .label = label, .elapsed = elapsed - kv.value });
+    }
+
+    pub fn _assert(assertion: js.Object, values: []js.Object, page: *Page) !void {
+        if (assertion.isTruthy()) {
+            return;
+        }
+        var serialized_values: []const u8 = "";
+        if (values.len > 0) {
+            serialized_values = try serializeValues(values, page);
+        }
+        log.info(.console, "assertion failed", .{ .values = serialized_values });
+    }
+
+    fn serializeValues(values: []js.Object, page: *Page) ![]const u8 {
+        if (values.len == 0) {
+            return "";
+        }
+
+        const arena = page.call_arena;
+        const separator = log.separator();
+        var arr: std.ArrayListUnmanaged(u8) = .{};
+
+        for (values, 1..) |value, i| {
+            try arr.appendSlice(arena, separator);
+            try arr.writer(arena).print("{d}: ", .{i});
+            const serialized = if (builtin.mode == .Debug) value.toDetailString() else value.toString();
+            try arr.appendSlice(arena, try serialized);
+        }
+        return arr.items;
+    }
+};
+
+fn timestamp() u32 {
+    return @import("../../datetime.zig").timestamp();
+}

src/browser/crypto/crypto.zig

@@ -1,4 +1,4 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
 //
 // Francis Bouvier <francis@lightpanda.io>
 // Pierre Tachoire <pierre@lightpanda.io>
@@ -18,35 +18,28 @@
 
 const std = @import("std");
 const js = @import("../js/js.zig");
+const uuidv4 = @import("../../id.zig").uuidv4;
 
-const SubtleCrypto = @import("SubtleCrypto.zig");
+// https://w3c.github.io/webcrypto/#crypto-interface
+pub const Crypto = struct {
+    _not_empty: bool = true,
 
-const Crypto = @This();
-_subtle: SubtleCrypto = .{},
-
-pub const init: Crypto = .{};
-
-// We take a js.Value, because we want to return the same instance, not a new
-// TypedArray
-pub fn getRandomValues(_: *const Crypto, js_obj: js.Object) !js.Object {
-    var into = try js_obj.toZig(RandomValues);
-    const buf = into.asBuffer();
-    if (buf.len > 65_536) {
-        return error.QuotaExceeded;
+    pub fn _getRandomValues(_: *const Crypto, js_obj: js.Object) !js.Object {
+        var into = try js_obj.toZig(Crypto, "getRandomValues", RandomValues);
+        const buf = into.asBuffer();
+        if (buf.len > 65_536) {
+            return error.QuotaExceededError;
+        }
+        std.crypto.random.bytes(buf);
+        return js_obj;
     }
-    std.crypto.random.bytes(buf);
-    return js_obj;
-}
 
-pub fn randomUUID(_: *const Crypto) ![36]u8 {
-    var hex: [36]u8 = undefined;
-    @import("../../id.zig").uuidv4(&hex);
-    return hex;
-}
-
-pub fn getSubtle(self: *Crypto) *SubtleCrypto {
-    return &self._subtle;
-}
+    pub fn _randomUUID(_: *const Crypto) [36]u8 {
+        var hex: [36]u8 = undefined;
+        uuidv4(&hex);
+        return hex;
+    }
+};
 
 const RandomValues = union(enum) {
     int8: []i8,
@@ -72,22 +65,7 @@ const RandomValues = union(enum) {
     }
 };
 
-pub const JsApi = struct {
-    pub const bridge = js.Bridge(Crypto);
-
-    pub const Meta = struct {
-        pub const name = "Crypto";
-        pub const prototype_chain = bridge.prototypeChain();
-        pub var class_id: bridge.ClassId = undefined;
-        pub const empty_with_no_proto = true;
-    };
-
-    pub const getRandomValues = bridge.function(Crypto.getRandomValues, .{ .dom_exception = true });
-    pub const randomUUID = bridge.function(Crypto.randomUUID, .{});
-    pub const subtle = bridge.accessor(Crypto.getSubtle, null, .{});
-};
-
 const testing = @import("../../testing.zig");
-test "WebApi: Crypto" {
-    try testing.htmlRunner("crypto.html", .{});
+test "Browser: Crypto" {
+    try testing.htmlRunner("crypto.html");
 }

src/browser/css/Parser.zig

@@ -1,295 +0,0 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-const std = @import("std");
-const Tokenizer = @import("Tokenizer.zig");
-
-pub const Declaration = struct {
-    name: []const u8,
-    value: []const u8,
-    important: bool,
-};
-
-const TokenSpan = struct {
-    token: Tokenizer.Token,
-    start: usize,
-    end: usize,
-};
-
-const TokenStream = struct {
-    tokenizer: Tokenizer,
-    peeked: ?TokenSpan = null,
-
-    fn init(input: []const u8) TokenStream {
-        return .{ .tokenizer = .{ .input = input } };
-    }
-
-    fn nextRaw(self: *TokenStream) ?TokenSpan {
-        const start = self.tokenizer.position;
-        const token = self.tokenizer.next() orelse return null;
-        const end = self.tokenizer.position;
-        return .{ .token = token, .start = start, .end = end };
-    }
-
-    fn next(self: *TokenStream) ?TokenSpan {
-        if (self.peeked) |token| {
-            self.peeked = null;
-            return token;
-        }
-        return self.nextRaw();
-    }
-
-    fn peek(self: *TokenStream) ?TokenSpan {
-        if (self.peeked == null) {
-            self.peeked = self.nextRaw();
-        }
-        return self.peeked;
-    }
-};
-
-pub fn parseDeclarationsList(input: []const u8) DeclarationsIterator {
-    return DeclarationsIterator.init(input);
-}
-
-pub const DeclarationsIterator = struct {
-    input: []const u8,
-    stream: TokenStream,
-
-    pub fn init(input: []const u8) DeclarationsIterator {
-        return .{
-            .input = input,
-            .stream = TokenStream.init(input),
-        };
-    }
-
-    pub fn next(self: *DeclarationsIterator) ?Declaration {
-        while (true) {
-            self.skipTriviaAndSemicolons();
-            const peeked = self.stream.peek() orelse return null;
-
-            switch (peeked.token) {
-                .at_keyword => {
-                    _ = self.stream.next();
-                    self.skipAtRule();
-                },
-                .ident => |name| {
-                    _ = self.stream.next();
-                    if (self.consumeDeclaration(name)) |declaration| {
-                        return declaration;
-                    }
-                },
-                else => {
-                    _ = self.stream.next();
-                    self.skipInvalidDeclaration();
-                },
-            }
-        }
-
-        return null;
-    }
-
-    fn consumeDeclaration(self: *DeclarationsIterator, name: []const u8) ?Declaration {
-        self.skipTrivia();
-
-        const colon = self.stream.next() orelse return null;
-        if (!isColon(colon.token)) {
-            self.skipInvalidDeclaration();
-            return null;
-        }
-
-        const value = self.consumeValue() orelse return null;
-        return .{
-            .name = name,
-            .value = value.value,
-            .important = value.important,
-        };
-    }
-
-    const ValueResult = struct {
-        value: []const u8,
-        important: bool,
-    };
-
-    fn consumeValue(self: *DeclarationsIterator) ?ValueResult {
-        self.skipTrivia();
-
-        var depth: usize = 0;
-        var start: ?usize = null;
-        var last_sig: ?TokenSpan = null;
-        var prev_sig: ?TokenSpan = null;
-
-        while (true) {
-            const peeked = self.stream.peek() orelse break;
-            if (isSemicolon(peeked.token) and depth == 0) {
-                _ = self.stream.next();
-                break;
-            }
-
-            const span = self.stream.next() orelse break;
-            if (isWhitespaceOrComment(span.token)) {
-                continue;
-            }
-
-            if (start == null) start = span.start;
-            prev_sig = last_sig;
-            last_sig = span;
-            updateDepth(span.token, &depth);
-        }
-
-        const value_start = start orelse return null;
-        const last = last_sig orelse return null;
-
-        var important = false;
-        var end_pos = last.end;
-
-        if (isImportantPair(prev_sig, last)) {
-            important = true;
-            const bang = prev_sig orelse return null;
-            if (value_start >= bang.start) return null;
-            end_pos = bang.start;
-        }
-
-        var value_slice = self.input[value_start..end_pos];
-        value_slice = std.mem.trim(u8, value_slice, &std.ascii.whitespace);
-        if (value_slice.len == 0) return null;
-
-        return .{ .value = value_slice, .important = important };
-    }
-
-    fn skipTrivia(self: *DeclarationsIterator) void {
-        while (self.stream.peek()) |peeked| {
-            if (!isWhitespaceOrComment(peeked.token)) break;
-            _ = self.stream.next();
-        }
-    }
-
-    fn skipTriviaAndSemicolons(self: *DeclarationsIterator) void {
-        while (self.stream.peek()) |peeked| {
-            if (isWhitespaceOrComment(peeked.token) or isSemicolon(peeked.token)) {
-                _ = self.stream.next();
-            } else {
-                break;
-            }
-        }
-    }
-
-    fn skipAtRule(self: *DeclarationsIterator) void {
-        var depth: usize = 0;
-        var saw_block = false;
-
-        while (true) {
-            const peeked = self.stream.peek() orelse return;
-            if (!saw_block and isSemicolon(peeked.token) and depth == 0) {
-                _ = self.stream.next();
-                return;
-            }
-
-            const span = self.stream.next() orelse return;
-            if (isWhitespaceOrComment(span.token)) continue;
-
-            if (isBlockStart(span.token)) {
-                depth += 1;
-                saw_block = true;
-            } else if (isBlockEnd(span.token)) {
-                if (depth > 0) depth -= 1;
-                if (saw_block and depth == 0) return;
-            }
-        }
-    }
-
-    fn skipInvalidDeclaration(self: *DeclarationsIterator) void {
-        var depth: usize = 0;
-
-        while (self.stream.peek()) |peeked| {
-            if (isSemicolon(peeked.token) and depth == 0) {
-                _ = self.stream.next();
-                return;
-            }
-
-            const span = self.stream.next() orelse return;
-            if (isWhitespaceOrComment(span.token)) continue;
-            updateDepth(span.token, &depth);
-        }
-    }
-};
-
-fn isWhitespaceOrComment(token: Tokenizer.Token) bool {
-    return switch (token) {
-        .white_space, .comment => true,
-        else => false,
-    };
-}
-
-fn isSemicolon(token: Tokenizer.Token) bool {
-    return switch (token) {
-        .semicolon => true,
-        else => false,
-    };
-}
-
-fn isColon(token: Tokenizer.Token) bool {
-    return switch (token) {
-        .colon => true,
-        else => false,
-    };
-}
-
-fn isBlockStart(token: Tokenizer.Token) bool {
-    return switch (token) {
-        .curly_bracket_block, .square_bracket_block, .parenthesis_block, .function => true,
-        else => false,
-    };
-}
-
-fn isBlockEnd(token: Tokenizer.Token) bool {
-    return switch (token) {
-        .close_curly_bracket, .close_parenthesis, .close_square_bracket => true,
-        else => false,
-    };
-}
-
-fn updateDepth(token: Tokenizer.Token, depth: *usize) void {
-    if (isBlockStart(token)) {
-        depth.* += 1;
-        return;
-    }
-
-    if (isBlockEnd(token)) {
-        if (depth.* > 0) depth.* -= 1;
-    }
-}
-
-fn isImportantPair(prev_sig: ?TokenSpan, last_sig: TokenSpan) bool {
-    if (!isIdentImportant(last_sig.token)) return false;
-    const prev = prev_sig orelse return false;
-    return isBang(prev.token);
-}
-
-fn isIdentImportant(token: Tokenizer.Token) bool {
-    return switch (token) {
-        .ident => |name| std.ascii.eqlIgnoreCase(name, "important"),
-        else => false,
-    };
-}
-
-fn isBang(token: Tokenizer.Token) bool {
-    return switch (token) {
-        .delim => |c| c == '!',
-        else => false,
-    };
-}

src/browser/css/README.md

@@ -0,0 +1,218 @@
+# css
+
+Lightpanda css implements CSS selectors parsing and matching in Zig.
+This package is a port of the Go lib [andybalholm/cascadia](https://github.com/andybalholm/cascadia).
+
+## Usage
+
+### Query parser
+
+```zig
+const css = @import("css.zig");
+
+const selector = try css.parse(alloc, "h1", .{});
+defer selector.deinit(alloc);
+```
+
+### DOM tree match
+
+The lib expects a `Node` interface implementation to match your DOM tree.
+
+```zig
+pub const Node = struct {
+    pub fn firstChild(_: Node) !?Node {
+        return error.TODO;
+    }
+
+    pub fn lastChild(_: Node) !?Node {
+        return error.TODO;
+    }
+
+    pub fn nextSibling(_: Node) !?Node {
+        return error.TODO;
+    }
+
+    pub fn prevSibling(_: Node) !?Node {
+        return error.TODO;
+    }
+
+    pub fn parent(_: Node) !?Node {
+        return error.TODO;
+    }
+
+    pub fn isElement(_: Node) bool {
+        return false;
+    }
+
+    pub fn isDocument(_: Node) bool {
+        return false;
+    }
+
+    pub fn isComment(_: Node) bool {
+        return false;
+    }
+
+    pub fn isText(_: Node) bool {
+        return false;
+    }
+
+    pub fn isEmptyText(_: Node) !bool {
+        return error.TODO;
+    }
+
+    pub fn tag(_: Node) ![]const u8 {
+        return error.TODO;
+    }
+
+    pub fn attr(_: Node, _: []const u8) !?[]const u8 {
+        return error.TODO;
+    }
+
+    pub fn eql(_: Node, _: Node) bool {
+        return false;
+    }
+};
+```
+
+You also need do define a `Matcher` implementing a `match` function to
+accumulate the results.
+
+```zig
+const Matcher = struct {
+    const Nodes = std.ArrayList(Node);
+
+    nodes: Nodes,
+
+    fn init(alloc: std.mem.Allocator) Matcher {
+        return .{ .nodes = Nodes.init(alloc) };
+    }
+
+    fn deinit(m: *Matcher) void {
+        m.nodes.deinit();
+    }
+
+    pub fn match(m: *Matcher, n: Node) !void {
+        try m.nodes.append(n);
+    }
+};
+```
+
+Then you can use the lib itself.
+
+```zig
+var matcher = Matcher.init(alloc);
+defer matcher.deinit();
+
+try css.matchAll(selector, node, &matcher);
+_ = try css.matchFirst(selector, node, &matcher); // returns true if a node matched.
+```
+
+## Features
+
+* [x] parse query selector
+* [x] `matchAll`
+* [x] `matchFirst`
+* [ ] specificity
+
+### Selectors implemented
+
+#### Selectors
+
+* [x] Class selectors
+* [x] Id selectors
+* [x] Type selectors
+* [x] Universal selectors
+* [ ] Nesting selectors
+
+#### Combinators
+
+* [x] Child combinator
+* [ ] Column combinator
+* [x] Descendant combinator
+* [ ] Namespace combinator
+* [x] Next-sibling combinator
+* [x] Selector list combinator
+* [x] Subsequent-sibling combinator
+
+#### Attribute
+
+* [x] `[attr]`
+* [x] `[attr=value]`
+* [x] `[attr|=value]`
+* [x] `[attr^=value]`
+* [x] `[attr$=value]`
+* [ ] `[attr*=value]`
+* [x] `[attr operator value i]`
+* [ ] `[attr operator value s]`
+
+#### Pseudo classes
+
+* [ ] `:active`
+* [ ] `:any-link`
+* [ ] `:autofill`
+* [ ] `:blank Experimental`
+* [x] `:checked`
+* [ ] `:current Experimental`
+* [ ] `:default`
+* [ ] `:defined`
+* [ ] `:dir() Experimental`
+* [x] `:disabled`
+* [x] `:empty`
+* [x] `:enabled`
+* [ ] `:first`
+* [x] `:first-child`
+* [x] `:first-of-type`
+* [ ] `:focus`
+* [ ] `:focus-visible`
+* [ ] `:focus-within`
+* [ ] `:fullscreen`
+* [ ] `:future Experimental`
+* [x] `:has() Experimental`
+* [ ] `:host`
+* [ ] `:host()`
+* [ ] `:host-context() Experimental`
+* [ ] `:hover`
+* [ ] `:indeterminate`
+* [ ] `:in-range`
+* [ ] `:invalid`
+* [ ] `:is()`
+* [x] `:lang()`
+* [x] `:last-child`
+* [x] `:last-of-type`
+* [ ] `:left`
+* [x] `:link`
+* [ ] `:local-link Experimental`
+* [ ] `:modal`
+* [x] `:not()`
+* [x] `:nth-child()`
+* [x] `:nth-last-child()`
+* [x] `:nth-last-of-type()`
+* [x] `:nth-of-type()`
+* [x] `:only-child`
+* [x] `:only-of-type`
+* [ ] `:optional`
+* [ ] `:out-of-range`
+* [ ] `:past Experimental`
+* [ ] `:paused`
+* [ ] `:picture-in-picture`
+* [ ] `:placeholder-shown`
+* [ ] `:playing`
+* [ ] `:read-only`
+* [ ] `:read-write`
+* [ ] `:required`
+* [ ] `:right`
+* [x] `:root`
+* [ ] `:scope`
+* [ ] `:state() Experimental`
+* [ ] `:target`
+* [ ] `:target-within Experimental`
+* [ ] `:user-invalid Experimental`
+* [ ] `:valid`
+* [ ] `:visited`
+* [ ] `:where()`
+* [ ] `:contains()`
+* [ ] `:containsown()`
+* [ ] `:matched()`
+* [ ] `:matchesown()`
+* [x] `:root`
+

src/browser/css/Tokenizer.zig

@@ -1,825 +0,0 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
-//
-// Francis Bouvier <francis@lightpanda.io>
-// Pierre Tachoire <pierre@lightpanda.io>
-//
-// This program is free software: you can redistribute it and/or modify
-// it under the terms of the GNU Affero General Public License as
-// published by the Free Software Foundation, either version 3 of the
-// License, or (at your option) any later version.
-//
-// This program is distributed in the hope that it will be useful,
-// but WITHOUT ANY WARRANTY; without even the implied warranty of
-// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-// GNU Affero General Public License for more details.
-//
-// You should have received a copy of the GNU Affero General Public License
-// along with this program.  If not, see <https://www.gnu.org/licenses/>.
-
-//! This file implements the tokenization step defined in the CSS Syntax Module Level 3 specification.
-//!
-//! The algorithm accepts a valid UTF-8 string and returns a stream of tokens.
-//! The tokenization step never fails, even for complete gibberish.
-//! Validity must then be checked by the parser.
-//!
-//! NOTE: The tokenizer is not thread-safe and does not own any memory, and does not check the validity of utf8.
-//!
-//! See spec for more info: https://drafts.csswg.org/css-syntax/#tokenization
-
-const std = @import("std");
-const builtin = @import("builtin");
-const assert = std.debug.assert;
-
-const Tokenizer = @This();
-
-pub const Token = union(enum) {
-    /// A `<ident-token>`
-    ident: []const u8,
-
-    /// A `<function-token>`
-    ///
-    /// The value (name) does not include the `(` marker.
-    function: []const u8,
-
-    /// A `<at-keyword-token>`
-    ///
-    /// The value does not include the `@` marker.
-    at_keyword: []const u8,
-
-    /// A `<hash-token>` with the type flag set to "id"
-    ///
-    /// The value does not include the `#` marker.
-    id_hash: []const u8, // Hash that is a valid ID selector.
-
-    /// A `<hash-token>` with the type flag set to "unrestricted"
-    ///
-    /// The value does not include the `#` marker.
-    unrestricted_hash: []const u8,
-
-    /// A `<string-token>`
-    ///
-    /// The value does not include the quotes.
-    string: []const u8,
-
-    /// A `<bad-string-token>`
-    ///
-    /// This token always indicates a parse error.
-    bad_string: []const u8,
-
-    /// A `<url-token>`
-    ///
-    /// The value does not include the `url(` `)` markers.  Note that `url( <string-token> )` is represented by a
-    /// `Function` token.
-    url: []const u8,
-
-    /// A `<bad-url-token>`
-    ///
-    /// This token always indicates a parse error.
-    bad_url: []const u8,
-
-    /// A `<delim-token>`
-    delim: u8,
-
-    /// A `<number-token>`
-    number: struct {
-        /// Whether the number had a `+` or `-` sign.
-        ///
-        /// This is used is some cases like the <An+B> micro syntax. (See the `parse_nth` function.)
-        has_sign: bool,
-
-        /// If the origin source did not include a fractional part, the value as an integer.
-        int_value: ?i32,
-
-        /// The value as a float
-        value: f32,
-    },
-
-    /// A `<percentage-token>`
-    percentage: struct {
-        /// Whether the number had a `+` or `-` sign.
-        has_sign: bool,
-
-        /// If the origin source did not include a fractional part, the value as an integer.
-        /// It is **not** divided by 100.
-        int_value: ?i32,
-
-        /// The value as a float, divided by 100 so that the nominal range is 0.0 to 1.0.
-        unit_value: f32,
-    },
-
-    /// A `<dimension-token>`
-    dimension: struct {
-        /// Whether the number had a `+` or `-` sign.
-        ///
-        /// This is used is some cases like the <An+B> micro syntax. (See the `parse_nth` function.)
-        has_sign: bool,
-
-        /// If the origin source did not include a fractional part, the value as an integer.
-        int_value: ?i32,
-
-        /// The value as a float
-        value: f32,
-
-        /// The unit, e.g. "px" in `12px`
-        unit: []const u8,
-    },
-
-    /// A `<unicode-range-token>`
-    unicode_range: struct { bgn: u32, end: i32 },
-
-    /// A `<whitespace-token>`
-    white_space: []const u8,
-
-    /// A `<!--` `<CDO-token>`
-    cdo,
-
-    /// A `-->` `<CDC-token>`
-    cdc,
-
-    /// A `:` `<colon-token>`
-    colon, // :
-
-    /// A `;` `<semicolon-token>`
-    semicolon, // ;
-
-    /// A `,` `<comma-token>`
-    comma, // ,
-
-    /// A `<[-token>`
-    square_bracket_block,
-
-    /// A `<]-token>`
-    ///
-    /// When obtained from one of the `Parser::next*` methods,
-    /// this token is always unmatched and indicates a parse error.
-    close_square_bracket,
-
-    /// A `<(-token>`
-    parenthesis_block,
-
-    /// A `<)-token>`
-    ///
-    /// When obtained from one of the `Parser::next*` methods,
-    /// this token is always unmatched and indicates a parse error.
-    close_parenthesis,
-
-    /// A `<{-token>`
-    curly_bracket_block,
-
-    /// A `<}-token>`
-    ///
-    /// When obtained from one of the `Parser::next*` methods,
-    /// this token is always unmatched and indicates a parse error.
-    close_curly_bracket,
-
-    /// A comment.
-    ///
-    /// The CSS Syntax spec does not generate tokens for comments,
-    /// But we do for simplicity of the interface.
-    ///
-    /// The value does not include the `/*` `*/` markers.
-    comment: []const u8,
-};
-
-input: []const u8,
-
-/// Counted in bytes, not code points. From 0.
-position: usize = 0,
-
-// If true, the input has at least `n` bytes left *after* the current one.
-// That is, `Lexer.byteAt(n)` will not panic.
-fn hasAtLeast(self: *const Tokenizer, n: usize) bool {
-    return self.position + n < self.input.len;
-}
-
-fn isEof(self: *const Tokenizer) bool {
-    return !self.hasAtLeast(0);
-}
-
-fn byteAt(self: *const Tokenizer, offset: usize) u8 {
-    return self.input[self.position + offset];
-}
-
-// Assumes non-EOF
-fn nextByteUnchecked(self: *const Tokenizer) u8 {
-    return self.byteAt(0);
-}
-
-fn nextByte(self: *const Tokenizer) ?u8 {
-    return if (self.isEof())
-        null
-    else
-        self.input[self.position];
-}
-
-fn startsWith(self: *const Tokenizer, needle: []const u8) bool {
-    return std.mem.startsWith(u8, self.input[self.position..], needle);
-}
-
-fn slice(self: *const Tokenizer, start: usize, end: usize) []const u8 {
-    return self.input[start..end];
-}
-
-fn sliceFrom(self: *const Tokenizer, start_pos: usize) []const u8 {
-    return self.slice(start_pos, self.position);
-}
-
-// Advance over N bytes in the input.  This function can advance
-// over ASCII bytes (excluding newlines), or UTF-8 sequence
-// leaders (excluding leaders for 4-byte sequences).
-fn advance(self: *Tokenizer, n: usize) void {
-    if (builtin.mode == .Debug) {
-        // Each byte must either be an ASCII byte or a sequence leader,
-        // but not a 4-byte leader; also newlines are rejected.
-        for (0..n) |i| {
-            const b = self.byteAt(i);
-            assert(b != '\r' and b != '\n' and b != '\x0C');
-            assert(b <= 0x7F or (b & 0xF0 != 0xF0 and b & 0xC0 != 0x80));
-        }
-    }
-    self.position += n;
-}
-
-fn hasNewlineAt(self: *const Tokenizer, offset: usize) bool {
-    if (!self.hasAtLeast(offset)) return false;
-
-    return switch (self.byteAt(offset)) {
-        '\n', '\r', '\x0C' => true,
-        else => false,
-    };
-}
-
-fn hasNonAsciiAt(self: *const Tokenizer, offset: usize) bool {
-    if (!self.hasAtLeast(offset)) return false;
-
-    const byte = self.byteAt(offset);
-    const len_utf8 = std.unicode.utf8ByteSequenceLength(byte) catch return false;
-
-    if (!self.hasAtLeast(offset + len_utf8 - 1)) return false;
-
-    const start = self.position + offset;
-    const bytes = self.slice(start, start + len_utf8);
-
-    const codepoint = std.unicode.utf8Decode(bytes) catch return false;
-
-    // https://drafts.csswg.org/css-syntax/#non-ascii-ident-code-point
-    return switch (codepoint) {
-        '\u{00B7}', '\u{200C}', '\u{200D}', '\u{203F}', '\u{2040}' => true,
-        '\u{00C0}'...'\u{00D6}' => true,
-        '\u{00D8}'...'\u{00F6}' => true,
-        '\u{00F8}'...'\u{037D}' => true,
-        '\u{037F}'...'\u{1FFF}' => true,
-        '\u{2070}'...'\u{218F}' => true,
-        '\u{2C00}'...'\u{2FEF}' => true,
-        '\u{3001}'...'\u{D7FF}' => true,
-        '\u{F900}'...'\u{FDCF}' => true,
-        '\u{FDF0}'...'\u{FFFD}' => true,
-        else => codepoint >= '\u{10000}',
-    };
-}
-
-fn isIdentStart(self: *Tokenizer) bool {
-    if (self.isEof()) return false;
-
-    var b = self.nextByteUnchecked();
-    if (b == '-') {
-        b = if (self.hasAtLeast(1)) self.byteAt(1) else return false;
-    }
-
-    return switch (b) {
-        'a'...'z', 'A'...'Z', '_', 0x0 => true,
-        '\\' => !self.hasNewlineAt(1),
-        else => b > 0x7F, // not is ascii
-    };
-}
-
-fn consumeChar(self: *Tokenizer) void {
-    const byte = self.nextByteUnchecked();
-    const len_utf8 = std.unicode.utf8ByteSequenceLength(byte) catch 1;
-    self.position += len_utf8;
-}
-
-// Given that a newline has been seen, advance over the newline
-// and update the state.
-fn consumeNewline(self: *Tokenizer) void {
-    const byte = self.nextByteUnchecked();
-    assert(byte == '\r' or byte == '\n' or byte == '\x0C');
-
-    self.position += 1;
-    if (byte == '\r' and self.nextByte() == '\n') {
-        self.position += 1;
-    }
-}
-
-fn consumeWhiteSpace(self: *Tokenizer, newline: bool) Token {
-    const start_position = self.position;
-    if (newline) {
-        self.consumeNewline();
-    } else {
-        self.advance(1);
-    }
-    while (!self.isEof()) {
-        const b = self.nextByteUnchecked();
-        switch (b) {
-            ' ', '\t' => {
-                self.advance(1);
-            },
-            '\n', '\x0C', '\r' => {
-                self.consumeNewline();
-            },
-            else => break,
-        }
-    }
-    return .{ .white_space = self.sliceFrom(start_position) };
-}
-
-fn consumeComment(self: *Tokenizer) []const u8 {
-    self.advance(2); // consume "/*"
-    const start_position = self.position;
-    while (!self.isEof()) {
-        switch (self.nextByteUnchecked()) {
-            '*' => {
-                const end_position = self.position;
-                self.advance(1);
-                if (self.nextByte() == '/') {
-                    self.advance(1);
-                    return self.slice(start_position, end_position);
-                }
-            },
-            '\n', '\x0C', '\r' => {
-                self.consumeNewline();
-            },
-            0x0 => self.advance(1),
-            else => self.consumeChar(),
-        }
-    }
-    return self.sliceFrom(start_position);
-}
-
-fn byteToHexDigit(b: u8) ?u32 {
-    return switch (b) {
-        '0'...'9' => b - '0',
-        'a'...'f' => b - 'a' + 10,
-        'A'...'F' => b - 'A' + 10,
-        else => null,
-    };
-}
-
-fn byteToDecimalDigit(b: u8) ?u32 {
-    return if (std.ascii.isDigit(b)) b - '0' else null;
-}
-
-// (value, number of digits up to 6)
-fn consumeHexDigits(self: *Tokenizer) void {
-    var value: u32 = 0;
-    var digits: u32 = 0;
-
-    while (digits < 6 and !self.isEof()) {
-        if (byteToHexDigit(self.nextByteUnchecked())) |digit| {
-            value = value * 16 + digit;
-            digits += 1;
-            self.advance(1);
-        } else {
-            break;
-        }
-    }
-
-    _ = &value;
-}
-
-// Assumes that the U+005C REVERSE SOLIDUS (\) has already been consumed
-// and that the next input character has already been verified
-// to not be a newline.
-fn consumeEscape(self: *Tokenizer) void {
-    if (self.isEof())
-        return; // Escaped EOF
-
-    switch (self.nextByteUnchecked()) {
-        '0'...'9', 'A'...'F', 'a'...'f' => {
-            consumeHexDigits(self);
-
-            if (!self.isEof()) {
-                switch (self.nextByteUnchecked()) {
-                    ' ', '\t' => {
-                        self.advance(1);
-                    },
-                    '\n', '\x0C', '\r' => {
-                        self.consumeNewline();
-                    },
-                    else => {},
-                }
-            }
-        },
-        else => self.consumeChar(),
-    }
-}
-
-/// https://drafts.csswg.org/css-syntax/#consume-string-token
-fn consumeString(self: *Tokenizer, single_quote: bool) Token {
-    self.advance(1); // Skip the initial quote
-
-    // start_pos is at code point boundary, after " or '
-    const start_pos = self.position;
-
-    while (!self.isEof()) {
-        switch (self.nextByteUnchecked()) {
-            '"' => {
-                if (!single_quote) {
-                    const value = self.sliceFrom(start_pos);
-                    self.advance(1);
-                    return .{ .string = value };
-                }
-                self.advance(1);
-            },
-            '\'' => {
-                if (single_quote) {
-                    const value = self.sliceFrom(start_pos);
-                    self.advance(1);
-                    return .{ .string = value };
-                }
-                self.advance(1);
-            },
-            '\n', '\r', '\x0C' => {
-                return .{ .bad_string = self.sliceFrom(start_pos) };
-            },
-            '\\' => {
-                self.advance(1);
-                if (self.isEof())
-                    continue; // escaped EOF, do nothing.
-
-                switch (self.nextByteUnchecked()) {
-                    // Escaped newline
-                    '\n', '\x0C', '\r' => self.consumeNewline(),
-
-                    // Spec calls for replacing escape sequences with characters,
-                    // but this would require allocating a new string.
-                    // Therefore, we leave it as is and let the parser handle the escaping.
-                    else => self.consumeEscape(),
-                }
-            },
-            else => self.consumeChar(),
-        }
-    }
-
-    return .{ .string = self.sliceFrom(start_pos) };
-}
-
-fn consumeName(self: *Tokenizer) []const u8 {
-    // start_pos is the end of the previous token, therefore at a code point boundary
-    const start_pos = self.position;
-
-    while (!self.isEof()) {
-        switch (self.nextByteUnchecked()) {
-            'a'...'z', 'A'...'Z', '0'...'9', '_', '-' => self.advance(1),
-            '\\' => {
-                if (self.hasNewlineAt(1)) {
-                    break;
-                }
-
-                self.advance(1);
-                self.consumeEscape();
-            },
-            0x0 => self.advance(1),
-            '\x80'...'\xFF' => {
-                // Non-ASCII: advance over the complete UTF-8 code point in one step.
-                // Using consumeChar() instead of advance(1) ensures we never land on
-                // a continuation byte, which advance() asserts against.
-                self.consumeChar();
-            },
-            else => {
-                if (self.hasNonAsciiAt(0)) {
-                    self.consumeChar();
-                } else {
-                    break; // ASCII
-                }
-            },
-        }
-    }
-
-    return self.sliceFrom(start_pos);
-}
-
-fn consumeMark(self: *Tokenizer) Token {
-    const byte = self.nextByteUnchecked();
-    self.advance(1);
-    return switch (byte) {
-        ',' => .comma,
-        ':' => .colon,
-        ';' => .semicolon,
-        '(' => .parenthesis_block,
-        ')' => .close_parenthesis,
-        '{' => .curly_bracket_block,
-        '}' => .close_curly_bracket,
-        '[' => .square_bracket_block,
-        ']' => .close_square_bracket,
-        else => unreachable,
-    };
-}
-
-fn consumeNumeric(self: *Tokenizer) Token {
-    // Parse [+-]?\d*(\.\d+)?([eE][+-]?\d+)?
-    // But this is always called so that there is at least one digit in \d*(\.\d+)?
-
-    // Do all the math in f64 so that large numbers overflow to +/-inf
-    // and i32::{MIN, MAX} are within range.
-
-    var sign: f64 = 1.0;
-    var has_sign = false;
-    switch (self.nextByteUnchecked()) {
-        '+' => {
-            has_sign = true;
-        },
-        '-' => {
-            has_sign = true;
-            sign = -1.0;
-        },
-        else => {},
-    }
-    if (has_sign) {
-        self.advance(1);
-    }
-
-    var is_integer = true;
-    var integral_part: f64 = 0.0;
-    var fractional_part: f64 = 0.0;
-
-    while (!self.isEof()) {
-        if (byteToDecimalDigit(self.nextByteUnchecked())) |digit| {
-            integral_part = integral_part * 10.0 + @as(f64, @floatFromInt(digit));
-            self.advance(1);
-        } else {
-            break;
-        }
-    }
-
-    if (self.hasAtLeast(1) and self.nextByteUnchecked() == '.' and std.ascii.isDigit(self.byteAt(1))) {
-        is_integer = false;
-        self.advance(1); // Consume '.'
-
-        var factor: f64 = 0.1;
-        while (!self.isEof()) {
-            if (byteToDecimalDigit(self.nextByteUnchecked())) |digit| {
-                fractional_part += @as(f64, @floatFromInt(digit)) * factor;
-                factor *= 0.1;
-                self.advance(1);
-            } else {
-                break;
-            }
-        }
-    }
-
-    var value = sign * (integral_part + fractional_part);
-
-    blk: {
-        const e = self.nextByte() orelse break :blk;
-        if (e != 'e' and e != 'E') break :blk;
-
-        var mul: f64 = 1.0;
-
-        if (self.hasAtLeast(2) and (self.byteAt(1) == '+' or self.byteAt(1) == '-') and std.ascii.isDigit(self.byteAt(2))) {
-            mul = switch (self.byteAt(1)) {
-                '-' => -1.0,
-                '+' => 1.0,
-                else => unreachable,
-            };
-
-            self.advance(2);
-        } else if (self.hasAtLeast(2) and std.ascii.isDigit(self.byteAt(2))) {
-            self.advance(1);
-        } else {
-            break :blk;
-        }
-
-        is_integer = false;
-
-        var exponent: f64 = 0.0;
-        while (!self.isEof()) {
-            if (byteToDecimalDigit(self.nextByteUnchecked())) |digit| {
-                exponent = exponent * 10.0 + @as(f64, @floatFromInt(digit));
-                self.advance(1);
-            } else {
-                break;
-            }
-        }
-        value *= std.math.pow(f64, 10.0, mul * exponent);
-    }
-
-    const int_value: ?i32 = if (is_integer) blk: {
-        if (value >= std.math.maxInt(i32)) {
-            break :blk std.math.maxInt(i32);
-        }
-
-        if (value <= std.math.minInt(i32)) {
-            break :blk std.math.minInt(i32);
-        }
-
-        break :blk @as(i32, @intFromFloat(value));
-    } else null;
-
-    if (!self.isEof() and self.nextByteUnchecked() == '%') {
-        self.advance(1);
-
-        return .{ .percentage = .{
-            .has_sign = has_sign,
-            .int_value = int_value,
-            .unit_value = @as(f32, @floatCast(value / 100.0)),
-        } };
-    }
-
-    if (isIdentStart(self)) {
-        return .{ .dimension = .{
-            .has_sign = has_sign,
-            .int_value = int_value,
-            .value = @as(f32, @floatCast(value)),
-            .unit = consumeName(self),
-        } };
-    }
-
-    return .{ .number = .{
-        .has_sign = has_sign,
-        .int_value = int_value,
-        .value = @as(f32, @floatCast(value)),
-    } };
-}
-
-fn consumeUnquotedUrl(self: *Tokenizer) ?Token {
-    // TODO: true url parser
-    if (self.nextByte()) |it| {
-        return self.consumeString(it == '\'');
-    }
-
-    return null;
-}
-
-fn consumeIdentLike(self: *Tokenizer) Token {
-    const value = self.consumeName();
-
-    if (!self.isEof() and self.nextByteUnchecked() == '(') {
-        self.advance(1);
-
-        if (std.ascii.eqlIgnoreCase(value, "url")) {
-            if (self.consumeUnquotedUrl()) |result| {
-                return result;
-            }
-        }
-
-        return .{ .function = value };
-    }
-
-    return .{ .ident = value };
-}
-
-pub fn next(self: *Tokenizer) ?Token {
-    if (self.isEof()) {
-        return null;
-    }
-
-    const b = self.nextByteUnchecked();
-    return switch (b) {
-        // Consume comments
-        '/' => {
-            if (self.startsWith("/*")) {
-                return .{ .comment = self.consumeComment() };
-            } else {
-                self.advance(1);
-                return .{ .delim = '/' };
-            }
-        },
-
-        // Consume marks
-        '(', ')', '{', '}', '[', ']', ',', ':', ';' => {
-            return self.consumeMark();
-        },
-
-        // Consume as much whitespace as possible. Return a <whitespace-token>.
-        ' ', '\t' => self.consumeWhiteSpace(false),
-        '\n', '\x0C', '\r' => self.consumeWhiteSpace(true),
-
-        // Consume a string token and return it.
-        '"' => self.consumeString(false),
-        '\'' => self.consumeString(true),
-
-        '0'...'9' => self.consumeNumeric(),
-        'a'...'z', 'A'...'Z', '_', 0x0 => self.consumeIdentLike(),
-
-        '+' => {
-            if ((self.hasAtLeast(1) and std.ascii.isDigit(self.byteAt(1))) or
-                (self.hasAtLeast(2) and self.byteAt(1) == '.' and std.ascii.isDigit(self.byteAt(2))))
-            {
-                return self.consumeNumeric();
-            }
-            self.advance(1);
-            return .{ .delim = '+' };
-        },
-        '-' => {
-            if ((self.hasAtLeast(1) and std.ascii.isDigit(self.byteAt(1))) or
-                (self.hasAtLeast(2) and self.byteAt(1) == '.' and std.ascii.isDigit(self.byteAt(2))))
-            {
-                return self.consumeNumeric();
-            }
-
-            if (self.startsWith("-->")) {
-                self.advance(3);
-                return .cdc;
-            }
-
-            if (isIdentStart(self)) {
-                return self.consumeIdentLike();
-            }
-
-            self.advance(1);
-            return .{ .delim = '-' };
-        },
-        '.' => {
-            if (self.hasAtLeast(1) and std.ascii.isDigit(self.byteAt(1))) {
-                return self.consumeNumeric();
-            }
-            self.advance(1);
-            return .{ .delim = '.' };
-        },
-
-        // Consume hash token
-        '#' => {
-            self.advance(1);
-            if (self.isIdentStart()) {
-                return .{ .id_hash = self.consumeName() };
-            }
-            if (self.nextByte()) |it| {
-                switch (it) {
-                    // Any other valid case here already resulted in IDHash.
-                    '0'...'9', '-' => return .{ .unrestricted_hash = self.consumeName() },
-                    else => {},
-                }
-            }
-            return .{ .delim = '#' };
-        },
-
-        // Consume at-rules
-        '@' => {
-            self.advance(1);
-            return if (isIdentStart(self))
-                .{ .at_keyword = consumeName(self) }
-            else
-                .{ .delim = '@' };
-        },
-
-        '<' => {
-            if (self.startsWith("<!--")) {
-                self.advance(4);
-                return .cdo;
-            } else {
-                self.advance(1);
-                return .{ .delim = '<' };
-            }
-        },
-
-        '\\' => {
-            if (!self.hasNewlineAt(1)) {
-                return self.consumeIdentLike();
-            }
-
-            self.advance(1);
-            return .{ .delim = '\\' };
-        },
-
-        else => {
-            if (b > 0x7F) { // not is ascii
-                return self.consumeIdentLike();
-            }
-
-            self.advance(1);
-            return .{ .delim = b };
-        },
-    };
-}
-
-const testing = std.testing;
-
-fn expectTokensEqual(input: []const u8, tokens: []const Token) !void {
-    var lexer = Tokenizer{ .input = input };
-
-    var i: usize = 0;
-    while (lexer.next()) |token| : (i += 1) {
-        assert(i < tokens.len);
-        try testing.expectEqualDeep(tokens[i], token);
-    }
-
-    try testing.expectEqual(i, tokens.len);
-    try testing.expectEqualDeep(null, lexer.next());
-}
-
-test "smoke" {
-    try expectTokensEqual(
-        \\.lightpanda  {color:red;}
-    , &.{
-        .{ .delim = '.' },
-        .{ .ident = "lightpanda" },
-        .{ .white_space = "  " },
-        .curly_bracket_block,
-        .{ .ident = "color" },
-        .colon,
-        .{ .ident = "red" },
-        .semicolon,
-        .close_curly_bracket,
-    });
-}

src/browser/css/css.zig

@@ -0,0 +1,191 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+// CSS Selector parser and query
+// This package is a rewrite in Zig of Cascadia CSS Selector parser.
+// see https://github.com/andybalholm/cascadia
+const std = @import("std");
+const Selector = @import("selector.zig").Selector;
+const parser = @import("parser.zig");
+
+pub const Interfaces = .{
+    Css,
+};
+
+// https://developer.mozilla.org/en-US/docs/Web/API/CSS
+pub const Css = struct {
+    _not_empty: bool = true,
+
+    pub fn _supports(_: *Css, _: []const u8, _: ?[]const u8) bool {
+        // TODO: Actually respond with which CSS features we support.
+        return true;
+    }
+};
+
+// parse parse a selector string and returns the parsed result or an error.
+pub fn parse(alloc: std.mem.Allocator, s: []const u8, opts: parser.ParseOptions) parser.ParseError!Selector {
+    var p = parser.Parser{ .s = s, .i = 0, .opts = opts };
+    return p.parse(alloc);
+}
+
+// matchFirst call m.match with the first node that matches the selector s, from the
+// descendants of n and returns true. If none matches, it returns false.
+pub fn matchFirst(s: *const Selector, node: anytype, m: anytype) !bool {
+    var child = node.firstChild();
+    while (child) |c| {
+        if (try s.match(c)) {
+            try m.match(c);
+            return true;
+        }
+
+        if (try matchFirst(s, c, m)) return true;
+        child = c.nextSibling();
+    }
+    return false;
+}
+
+// matchAll call m.match with the all the nodes that matches the selector s, from the
+// descendants of n.
+pub fn matchAll(s: *const Selector, node: anytype, m: anytype) !void {
+    var child = node.firstChild();
+    while (child) |c| {
+        if (try s.match(c)) try m.match(c);
+        try matchAll(s, c, m);
+        child = c.nextSibling();
+    }
+}
+
+test "parse" {
+    const alloc = std.testing.allocator;
+
+    const testcases = [_][]const u8{
+        "address",
+        "*",
+        "#foo",
+        "li#t1",
+        "*#t4",
+        ".t1",
+        "p.t1",
+        "div.teST",
+        ".t1.fail",
+        "p.t1.t2",
+        "p.--t1",
+        "p.--t1.--t2",
+        "p[title]",
+        "div[class=\"red\" i]",
+        "address[title=\"foo\"]",
+        "address[title=\"FoOIgnoRECaSe\" i]",
+        "address[title!=\"foo\"]",
+        "address[title!=\"foo\" i]",
+        "p[title!=\"FooBarUFoo\" i]",
+        "[  \t title        ~=       foo    ]",
+        "p[title~=\"FOO\" i]",
+        "p[title~=toofoo i]",
+        "[title~=\"hello world\"]",
+        "[title~=\"hello\" i]",
+        "[title~=\"hello\"          I]",
+        "[lang|=\"en\"]",
+        "[lang|=\"EN\" i]",
+        "[lang|=\"EN\"     i]",
+        "[title^=\"foo\"]",
+        "[title^=\"foo\" i]",
+        "[title$=\"bar\"]",
+        "[title$=\"BAR\" i]",
+        "[title*=\"bar\"]",
+        "[title*=\"BaRu\" i]",
+        "[title*=\"BaRu\" I]",
+        "p[class$=\" \"]",
+        "p[class$=\"\"]",
+        "p[class^=\" \"]",
+        "p[class^=\"\"]",
+        "p[class*=\" \"]",
+        "p[class*=\"\"]",
+        "input[name=Sex][value=F]",
+        "table[border=\"0\"][cellpadding=\"0\"][cellspacing=\"0\"]",
+        ".t1:not(.t2)",
+        "div:not(.t1)",
+        "div:not([class=\"t2\"])",
+        "li:nth-child(odd)",
+        "li:nth-child(even)",
+        "li:nth-child(-n+2)",
+        "li:nth-child(3n+1)",
+        "li:nth-last-child(odd)",
+        "li:nth-last-child(even)",
+        "li:nth-last-child(-n+2)",
+        "li:nth-last-child(3n+1)",
+        "span:first-child",
+        "span:last-child",
+        "p:nth-of-type(2)",
+        "p:nth-last-of-type(2)",
+        "p:last-of-type",
+        "p:first-of-type",
+        "p:only-child",
+        "p:only-of-type",
+        ":empty",
+        "div p",
+        "div table p",
+        "div > p",
+        "p ~ p",
+        "p + p",
+        "li, p",
+        "p +/*This is a comment*/ p",
+        "p:contains(\"that wraps\")",
+        "p:containsOwn(\"that wraps\")",
+        ":containsOwn(\"inner\")",
+        "p:containsOwn(\"block\")",
+        "div:has(#p1)",
+        "div:has(:containsOwn(\"2\"))",
+        "body :has(:containsOwn(\"2\"))",
+        "body :haschild(:containsOwn(\"2\"))",
+        "p:matches([\\d])",
+        "p:matches([a-z])",
+        "p:matches([a-zA-Z])",
+        "p:matches([^\\d])",
+        "p:matches(^(0|a))",
+        "p:matches(^\\d+$)",
+        "p:not(:matches(^\\d+$))",
+        "div :matchesOwn(^\\d+$)",
+        "[href#=(fina)]:not([href#=(\\/\\/[^\\/]+untrusted)])",
+        "[href#=(^https:\\/\\/[^\\/]*\\/?news)]",
+        ":input",
+        ":root",
+        "*:root",
+        "html:nth-child(1)",
+        "*:root:first-child",
+        "*:root:nth-child(1)",
+        "a:not(:root)",
+        "body > *:nth-child(3n+2)",
+        "input:disabled",
+        ":disabled",
+        ":enabled",
+        "div.class1, div.class2",
+    };
+
+    for (testcases) |tc| {
+        const s = parse(alloc, tc, .{}) catch |e| {
+            std.debug.print("query {s}", .{tc});
+            return e;
+        };
+        defer s.deinit(alloc);
+    }
+}
+
+const testing = @import("../../testing.zig");
+test "Browser: CSS" {
+    try testing.htmlRunner("css.html");
+}

src/browser/css/libdom.zig

@@ -0,0 +1,423 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const parser = @import("../netsurf.zig");
+const css = @import("css.zig");
+const Allocator = std.mem.Allocator;
+
+// Node implementation with Netsurf Libdom C lib.
+pub const Node = struct {
+    node: *parser.Node,
+
+    pub fn firstChild(n: Node) ?Node {
+        const c = parser.nodeFirstChild(n.node);
+        if (c) |cc| return .{ .node = cc };
+
+        return null;
+    }
+
+    pub fn lastChild(n: Node) ?Node {
+        const c = parser.nodeLastChild(n.node);
+        if (c) |cc| return .{ .node = cc };
+
+        return null;
+    }
+
+    pub fn nextSibling(n: Node) ?Node {
+        const c = parser.nodeNextSibling(n.node);
+        if (c) |cc| return .{ .node = cc };
+
+        return null;
+    }
+
+    pub fn prevSibling(n: Node) ?Node {
+        const c = parser.nodePreviousSibling(n.node);
+        if (c) |cc| return .{ .node = cc };
+
+        return null;
+    }
+
+    pub fn parent(n: Node) ?Node {
+        const c = parser.nodeParentNode(n.node);
+        if (c) |cc| return .{ .node = cc };
+
+        return null;
+    }
+
+    pub fn isElement(n: Node) bool {
+        return parser.nodeType(n.node) == .element;
+    }
+
+    pub fn isDocument(n: Node) bool {
+        return parser.nodeType(n.node) == .document;
+    }
+
+    pub fn isComment(n: Node) bool {
+        return parser.nodeType(n.node) == .comment;
+    }
+
+    pub fn isText(n: Node) bool {
+        return parser.nodeType(n.node) == .text;
+    }
+
+    pub fn text(n: Node) ?[]const u8 {
+        const data = parser.nodeTextContent(n.node);
+        if (data == null) return null;
+        if (data.?.len == 0) return null;
+
+        return std.mem.trim(u8, data.?, &std.ascii.whitespace);
+    }
+
+    pub fn isEmptyText(n: Node) bool {
+        const data = parser.nodeTextContent(n.node);
+        if (data == null) return true;
+        if (data.?.len == 0) return true;
+
+        return std.mem.trim(u8, data.?, &std.ascii.whitespace).len == 0;
+    }
+
+    pub fn tag(n: Node) ![]const u8 {
+        return parser.nodeName(n.node);
+    }
+
+    pub fn attr(n: Node, key: []const u8) !?[]const u8 {
+        if (!n.isElement()) return null;
+        return try parser.elementGetAttribute(parser.nodeToElement(n.node), key);
+    }
+
+    pub fn eql(a: Node, b: Node) bool {
+        return a.node == b.node;
+    }
+};
+
+const MatcherTest = struct {
+    const Nodes = std.ArrayListUnmanaged(Node);
+
+    nodes: Nodes,
+    allocator: Allocator,
+
+    fn init(allocator: Allocator) MatcherTest {
+        return .{
+            .nodes = .empty,
+            .allocator = allocator,
+        };
+    }
+
+    fn deinit(m: *MatcherTest) void {
+        m.nodes.deinit(m.allocator);
+    }
+
+    fn reset(m: *MatcherTest) void {
+        m.nodes.clearRetainingCapacity();
+    }
+
+    pub fn match(m: *MatcherTest, n: Node) !void {
+        try m.nodes.append(m.allocator, n);
+    }
+};
+
+test "Browser.CSS.Libdom: matchFirst" {
+    const alloc = std.testing.allocator;
+
+    parser.init();
+    defer parser.deinit();
+
+    var matcher = MatcherTest.init(alloc);
+    defer matcher.deinit();
+
+    const testcases = [_]struct {
+        q: []const u8,
+        html: []const u8,
+        exp: usize,
+    }{
+        .{ .q = "address", .html = "<body><address>This address...</address></body>", .exp = 1 },
+        .{ .q = "*", .html = "<!-- comment --><html><head></head><body>text</body></html>", .exp = 1 },
+        .{ .q = "*", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "#foo", .html = "<p id=\"foo\"><p id=\"bar\">", .exp = 1 },
+        .{ .q = "li#t1", .html = "<ul><li id=\"t1\"><p id=\"t1\">", .exp = 1 },
+        .{ .q = ".t3", .html = "<ul><li class=\"t1\"><li class=\"t2 t3\">", .exp = 1 },
+        .{ .q = "*#t4", .html = "<ol><li id=\"t4\"><li id=\"t44\">", .exp = 1 },
+        .{ .q = ".t1", .html = "<ul><li class=\"t1\"><li class=\"t2\">", .exp = 1 },
+        .{ .q = "p.t1", .html = "<p class=\"t1 t2\">", .exp = 1 },
+        .{ .q = "div.teST", .html = "<div class=\"test\">", .exp = 0 },
+        .{ .q = ".t1.fail", .html = "<p class=\"t1 t2\">", .exp = 0 },
+        .{ .q = "p.t1.t2", .html = "<p class=\"t1 t2\">", .exp = 1 },
+        .{ .q = "p.--t1", .html = "<p class=\"--t1 --t2\">", .exp = 1 },
+        .{ .q = "p.--t1.--t2", .html = "<p class=\"--t1 --t2\">", .exp = 1 },
+        .{ .q = "p[title]", .html = "<p><p title=\"title\">", .exp = 1 },
+        .{ .q = "div[class=\"red\" i]", .html = "<div><div class=\"Red\">", .exp = 1 },
+        .{ .q = "address[title=\"foo\"]", .html = "<address><address title=\"foo\"><address title=\"bar\">", .exp = 1 },
+        .{ .q = "address[title=\"FoOIgnoRECaSe\" i]", .html = "<address><address title=\"fooIgnoreCase\"><address title=\"bar\">", .exp = 1 },
+        .{ .q = "address[title!=\"foo\"]", .html = "<address><address title=\"foo\"><address title=\"bar\">", .exp = 1 },
+        .{ .q = "address[title!=\"foo\" i]", .html = "<address><address title=\"FOO\"><address title=\"bar\">", .exp = 1 },
+        .{ .q = "p[title!=\"FooBarUFoo\" i]", .html = "<p title=\"fooBARuFOO\"><p title=\"varfoo\">", .exp = 1 },
+        .{ .q = "[   title        ~=       foo    ]", .html = "<p title=\"tot foo bar\">", .exp = 1 },
+        .{ .q = "p[title~=\"FOO\" i]", .html = "<p title=\"tot foo bar\">", .exp = 1 },
+        .{ .q = "p[title~=toofoo i]", .html = "<p title=\"tot foo bar\">", .exp = 0 },
+        .{ .q = "[title~=\"hello world\"]", .html = "<p title=\"hello world\">", .exp = 0 },
+        .{ .q = "[title~=\"hello\" i]", .html = "<p title=\"HELLO world\">", .exp = 1 },
+        .{ .q = "[title~=\"hello\"          I]", .html = "<p title=\"HELLO world\">", .exp = 1 },
+        .{ .q = "[lang|=\"en\"]", .html = "<p lang=\"en\"><p lang=\"en-gb\"><p lang=\"enough\"><p lang=\"fr-en\">", .exp = 1 },
+        .{ .q = "[lang|=\"EN\" i]", .html = "<p lang=\"en\"><p lang=\"En-gb\"><p lang=\"enough\"><p lang=\"fr-en\">", .exp = 1 },
+        .{ .q = "[lang|=\"EN\"     i]", .html = "<p lang=\"en\"><p lang=\"En-gb\"><p lang=\"enough\"><p lang=\"fr-en\">", .exp = 1 },
+        .{ .q = "[title^=\"foo\"]", .html = "<p title=\"foobar\"><p title=\"barfoo\">", .exp = 1 },
+        .{ .q = "[title^=\"foo\" i]", .html = "<p title=\"FooBAR\"><p title=\"barfoo\">", .exp = 1 },
+        .{ .q = "[title$=\"bar\"]", .html = "<p title=\"foobar\"><p title=\"barfoo\">", .exp = 1 },
+        .{ .q = "[title$=\"BAR\" i]", .html = "<p title=\"foobar\"><p title=\"barfoo\">", .exp = 1 },
+        .{ .q = "[title*=\"bar\"]", .html = "<p title=\"foobarufoo\">", .exp = 1 },
+        .{ .q = "[title*=\"BaRu\" i]", .html = "<p title=\"foobarufoo\">", .exp = 1 },
+        .{ .q = "[title*=\"BaRu\" I]", .html = "<p title=\"foobarufoo\">", .exp = 1 },
+        .{ .q = "p[class$=\" \"]", .html = "<p class=\" \">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class$=\"\"]", .html = "<p class=\"\">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class^=\" \"]", .html = "<p class=\" \">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class^=\"\"]", .html = "<p class=\"\">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class*=\" \"]", .html = "<p class=\" \">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class*=\"\"]", .html = "<p class=\"\">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "input[name=Sex][value=F]", .html = "<input type=\"radio\" name=\"Sex\" value=\"F\"/>", .exp = 1 },
+        .{ .q = "table[border=\"0\"][cellpadding=\"0\"][cellspacing=\"0\"]", .html = "<table border=\"0\" cellpadding=\"0\" cellspacing=\"0\" style=\"table-layout: fixed; width: 100%; border: 0 dashed; border-color: #FFFFFF\"><tr style=\"height:64px\">aaa</tr></table>", .exp = 1 },
+        .{ .q = ".t1:not(.t2)", .html = "<p class=\"t1 t2\">", .exp = 0 },
+        .{ .q = "div:not(.t1)", .html = "<div class=\"t3\">", .exp = 1 },
+        .{ .q = "div:not([class=\"t2\"])", .html = "<div><div class=\"t2\"><div class=\"t3\">", .exp = 1 },
+        .{ .q = "li:nth-child(odd)", .html = "<ol><li id=1><li id=2><li id=3></ol>", .exp = 1 },
+        .{ .q = "li:nth-child(even)", .html = "<ol><li id=1><li id=2><li id=3></ol>", .exp = 1 },
+        .{ .q = "li:nth-child(-n+2)", .html = "<ol><li id=1><li id=2><li id=3></ol>", .exp = 1 },
+        .{ .q = "li:nth-child(3n+1)", .html = "<ol><li id=1><li id=2><li id=3></ol>", .exp = 1 },
+        .{ .q = "li:nth-last-child(odd)", .html = "<ol><li id=1><li id=2><li id=3><li id=4></ol>", .exp = 1 },
+        .{ .q = "li:nth-last-child(even)", .html = "<ol><li id=1><li id=2><li id=3><li id=4></ol>", .exp = 1 },
+        .{ .q = "li:nth-last-child(-n+2)", .html = "<ol><li id=1><li id=2><li id=3><li id=4></ol>", .exp = 1 },
+        .{ .q = "li:nth-last-child(3n+1)", .html = "<ol><li id=1><li id=2><li id=3><li id=4></ol>", .exp = 1 },
+        .{ .q = "span:first-child", .html = "<p>some text <span id=\"1\">and a span</span><span id=\"2\"> and another</span></p>", .exp = 1 },
+        .{ .q = "span:last-child", .html = "<span>a span</span> and some text", .exp = 1 },
+        .{ .q = "p:nth-of-type(2)", .html = "<address></address><p id=1><p id=2>", .exp = 1 },
+        .{ .q = "p:nth-last-of-type(2)", .html = "<address></address><p id=1><p id=2></p><a>", .exp = 1 },
+        .{ .q = "p:last-of-type", .html = "<address></address><p id=1><p id=2></p><a>", .exp = 1 },
+        .{ .q = "p:first-of-type", .html = "<address></address><p id=1><p id=2></p><a>", .exp = 1 },
+        .{ .q = "p:only-child", .html = "<div><p id=\"1\"></p><a></a></div><div><p id=\"2\"></p></div>", .exp = 1 },
+        .{ .q = "p:only-of-type", .html = "<div><p id=\"1\"></p><a></a></div><div><p id=\"2\"></p><p id=\"3\"></p></div>", .exp = 1 },
+        .{ .q = ":empty", .html = "<p id=\"1\"><!-- --><p id=\"2\">Hello<p id=\"3\"><span>", .exp = 1 },
+        .{ .q = "div p", .html = "<div><p id=\"1\"><table><tr><td><p id=\"2\"></table></div><p id=\"3\">", .exp = 1 },
+        .{ .q = "div table p", .html = "<div><p id=\"1\"><table><tr><td><p id=\"2\"></table></div><p id=\"3\">", .exp = 1 },
+        .{ .q = "div > p", .html = "<div><p id=\"1\"><div><p id=\"2\"></div><table><tr><td><p id=\"3\"></table></div>", .exp = 1 },
+        .{ .q = "p ~ p", .html = "<p id=\"1\"><p id=\"2\"></p><address></address><p id=\"3\">", .exp = 1 },
+        .{ .q = "p + p", .html = "<p id=\"1\"></p> <!--comment--> <p id=\"2\"></p><address></address><p id=\"3\">", .exp = 1 },
+        .{ .q = "li, p", .html = "<ul><li></li><li></li></ul><p>", .exp = 1 },
+        .{ .q = "p +/*This is a comment*/ p", .html = "<p id=\"1\"><p id=\"2\"></p><address></address><p id=\"3\">", .exp = 1 },
+        // .{ .q = "p:contains(\"that wraps\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 1 },
+        .{ .q = "p:containsOwn(\"that wraps\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 0 },
+        .{ .q = ":containsOwn(\"inner\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 1 },
+        .{ .q = ":containsOwn(\"Inner\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 0 },
+        .{ .q = "p:containsOwn(\"block\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 1 },
+        // .{ .q = "div:has(#p1)", .html = "<div id=\"d1\"><p id=\"p1\"><span>text content</span></p></div><div id=\"d2\"/>", .exp = 1 },
+        .{ .q = "div:has(:containsOwn(\"2\"))", .html = "<div id=\"d1\"><p id=\"p1\"><span>contents 1</span></p></div> <div id=\"d2\"><p>contents <em>2</em></p></div>", .exp = 1 },
+        .{ .q = "body :has(:containsOwn(\"2\"))", .html = "<body><div id=\"d1\"><p id=\"p1\"><span>contents 1</span></p></div> <div id=\"d2\"><p id=\"p2\">contents <em>2</em></p></div></body>", .exp = 1 },
+        .{ .q = "body :haschild(:containsOwn(\"2\"))", .html = "<body><div id=\"d1\"><p id=\"p1\"><span>contents 1</span></p></div> <div id=\"d2\"><p id=\"p2\">contents <em>2</em></p></div></body>", .exp = 1 },
+        // .{ .q = "p:matches([\\d])", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 1 },
+        // .{ .q = "p:matches([a-z])", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 1 },
+        // .{ .q = "p:matches([a-zA-Z])", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 1 },
+        // .{ .q = "p:matches([^\\d])", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 1 },
+        // .{ .q = "p:matches(^(0|a))", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 1 },
+        // .{ .q = "p:matches(^\\d+$)", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 1 },
+        // .{ .q = "p:not(:matches(^\\d+$))", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 1 },
+        // .{ .q = "div :matchesOwn(^\\d+$)", .html = "<div><p id=\"p1\">01234<em>567</em>89</p><div>", .exp = 1 },
+        // .{ .q = "[href#=(fina)]:not([href#=(\\/\\/[^\\/]+untrusted)])", .html = "<ul> <li><a id=\"a1\" href=\"http://www.google.com/finance\"></a> <li><a id=\"a2\" href=\"http://finance.yahoo.com/\"></a> <li><a id=\"a2\" href=\"http://finance.untrusted.com/\"/> <li><a id=\"a3\" href=\"https://www.google.com/news\"/> <li><a id=\"a4\" href=\"http://news.yahoo.com\"/> </ul>", .exp = 1 },
+        // .{ .q = "[href#=(^https:\\/\\/[^\\/]*\\/?news)]", .html = "<ul> <li><a id=\"a1\" href=\"http://www.google.com/finance\"/> <li><a id=\"a2\" href=\"http://finance.yahoo.com/\"/> <li><a id=\"a3\" href=\"https://www.google.com/news\"></a> <li><a id=\"a4\" href=\"http://news.yahoo.com\"/> </ul>", .exp = 1 },
+        .{ .q = ":input", .html = "<form> <label>Username <input type=\"text\" name=\"username\" /></label> <label>Password <input type=\"password\" name=\"password\" /></label> <label>Country <select name=\"country\"> <option value=\"ca\">Canada</option> <option value=\"us\">United States</option> </select> </label> <label>Bio <textarea name=\"bio\"></textarea></label> <button>Sign up</button> </form>", .exp = 1 },
+        .{ .q = ":root", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "*:root", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "html:nth-child(1)", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "*:root:first-child", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "*:root:nth-child(1)", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "a:not(:root)", .html = "<html><head></head><body><a href=\"http://www.foo.com\"></a></body></html>", .exp = 1 },
+        .{ .q = "body > *:nth-child(3n+2)", .html = "<html><head></head><body><p></p><div></div><span></span><a></a><form></form></body></html>", .exp = 1 },
+        .{ .q = "input:disabled", .html = "<html><head></head><body><fieldset disabled><legend id=\"1\"><input id=\"i1\"/></legend><legend id=\"2\"><input id=\"i2\"/></legend></fieldset></body></html>", .exp = 1 },
+        .{ .q = ":disabled", .html = "<html><head></head><body><fieldset disabled></fieldset></body></html>", .exp = 1 },
+        .{ .q = ":enabled", .html = "<html><head></head><body><fieldset></fieldset></body></html>", .exp = 1 },
+        .{ .q = "div.class1, div.class2", .html = "<div class=class1></div><div class=class2></div><div class=class3></div>", .exp = 1 },
+    };
+
+    for (testcases) |tc| {
+        matcher.reset();
+
+        const doc = try parser.documentHTMLParseFromStr(tc.html);
+        defer parser.documentHTMLClose(doc) catch {};
+
+        const s = css.parse(alloc, tc.q, .{}) catch |e| {
+            std.debug.print("parse, query: {s}\n", .{tc.q});
+            return e;
+        };
+
+        defer s.deinit(alloc);
+
+        const node = Node{ .node = parser.documentHTMLToNode(doc) };
+
+        _ = css.matchFirst(&s, node, &matcher) catch |e| {
+            std.debug.print("match, query: {s}\n", .{tc.q});
+            return e;
+        };
+        std.testing.expectEqual(tc.exp, matcher.nodes.items.len) catch |e| {
+            std.debug.print("expectation, query: {s}\n", .{tc.q});
+            return e;
+        };
+    }
+}
+
+test "Browser.CSS.Libdom: matchAll" {
+    const alloc = std.testing.allocator;
+
+    parser.init();
+    defer parser.deinit();
+
+    var matcher = MatcherTest.init(alloc);
+    defer matcher.deinit();
+
+    const testcases = [_]struct {
+        q: []const u8,
+        html: []const u8,
+        exp: usize,
+    }{
+        .{ .q = "address", .html = "<body><address>This address...</address></body>", .exp = 1 },
+        .{ .q = "*", .html = "<!-- comment --><html><head></head><body>text</body></html>", .exp = 3 },
+        .{ .q = "*", .html = "<html><head></head><body></body></html>", .exp = 3 },
+        .{ .q = "#foo", .html = "<p id=\"foo\"><p id=\"bar\">", .exp = 1 },
+        .{ .q = "li#t1", .html = "<ul><li id=\"t1\"><p id=\"t1\">", .exp = 1 },
+        .{ .q = ".t3", .html = "<ul><li class=\"t1\"><li class=\"t2 t3\">", .exp = 1 },
+        .{ .q = "*#t4", .html = "<ol><li id=\"t4\"><li id=\"t44\">", .exp = 1 },
+        .{ .q = ".t1", .html = "<ul><li class=\"t1\"><li class=\"t2\">", .exp = 1 },
+        .{ .q = "p.t1", .html = "<p class=\"t1 t2\">", .exp = 1 },
+        .{ .q = "div.teST", .html = "<div class=\"test\">", .exp = 0 },
+        .{ .q = ".t1.fail", .html = "<p class=\"t1 t2\">", .exp = 0 },
+        .{ .q = "p.t1.t2", .html = "<p class=\"t1 t2\">", .exp = 1 },
+        .{ .q = "p.--t1", .html = "<p class=\"--t1 --t2\">", .exp = 1 },
+        .{ .q = "p.--t1.--t2", .html = "<p class=\"--t1 --t2\">", .exp = 1 },
+        .{ .q = "p[title]", .html = "<p><p title=\"title\">", .exp = 1 },
+        .{ .q = "div[class=\"red\" i]", .html = "<div><div class=\"Red\">", .exp = 1 },
+        .{ .q = "address[title=\"foo\"]", .html = "<address><address title=\"foo\"><address title=\"bar\">", .exp = 1 },
+        .{ .q = "address[title=\"FoOIgnoRECaSe\" i]", .html = "<address><address title=\"fooIgnoreCase\"><address title=\"bar\">", .exp = 1 },
+        .{ .q = "address[title!=\"foo\"]", .html = "<address><address title=\"foo\"><address title=\"bar\">", .exp = 2 },
+        .{ .q = "address[title!=\"foo\" i]", .html = "<address><address title=\"FOO\"><address title=\"bar\">", .exp = 2 },
+        .{ .q = "p[title!=\"FooBarUFoo\" i]", .html = "<p title=\"fooBARuFOO\"><p title=\"varfoo\">", .exp = 1 },
+        .{ .q = "[   title        ~=       foo    ]", .html = "<p title=\"tot foo bar\">", .exp = 1 },
+        .{ .q = "p[title~=\"FOO\" i]", .html = "<p title=\"tot foo bar\">", .exp = 1 },
+        .{ .q = "p[title~=toofoo i]", .html = "<p title=\"tot foo bar\">", .exp = 0 },
+        .{ .q = "[title~=\"hello world\"]", .html = "<p title=\"hello world\">", .exp = 0 },
+        .{ .q = "[title~=\"hello\" i]", .html = "<p title=\"HELLO world\">", .exp = 1 },
+        .{ .q = "[title~=\"hello\"          I]", .html = "<p title=\"HELLO world\">", .exp = 1 },
+        .{ .q = "[lang|=\"en\"]", .html = "<p lang=\"en\"><p lang=\"en-gb\"><p lang=\"enough\"><p lang=\"fr-en\">", .exp = 2 },
+        .{ .q = "[lang|=\"EN\" i]", .html = "<p lang=\"en\"><p lang=\"En-gb\"><p lang=\"enough\"><p lang=\"fr-en\">", .exp = 2 },
+        .{ .q = "[lang|=\"EN\"     i]", .html = "<p lang=\"en\"><p lang=\"En-gb\"><p lang=\"enough\"><p lang=\"fr-en\">", .exp = 2 },
+        .{ .q = "[title^=\"foo\"]", .html = "<p title=\"foobar\"><p title=\"barfoo\">", .exp = 1 },
+        .{ .q = "[title^=\"foo\" i]", .html = "<p title=\"FooBAR\"><p title=\"barfoo\">", .exp = 1 },
+        .{ .q = "[title$=\"bar\"]", .html = "<p title=\"foobar\"><p title=\"barfoo\">", .exp = 1 },
+        .{ .q = "[title$=\"BAR\" i]", .html = "<p title=\"foobar\"><p title=\"barfoo\">", .exp = 1 },
+        .{ .q = "[title*=\"bar\"]", .html = "<p title=\"foobarufoo\">", .exp = 1 },
+        .{ .q = "[title*=\"BaRu\" i]", .html = "<p title=\"foobarufoo\">", .exp = 1 },
+        .{ .q = "[title*=\"BaRu\" I]", .html = "<p title=\"foobarufoo\">", .exp = 1 },
+        .{ .q = "p[class$=\" \"]", .html = "<p class=\" \">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class$=\"\"]", .html = "<p class=\"\">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class^=\" \"]", .html = "<p class=\" \">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class^=\"\"]", .html = "<p class=\"\">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class*=\" \"]", .html = "<p class=\" \">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "p[class*=\"\"]", .html = "<p class=\"\">This text should be green.</p><p>This text should be green.</p>", .exp = 0 },
+        .{ .q = "input[name=Sex][value=F]", .html = "<input type=\"radio\" name=\"Sex\" value=\"F\"/>", .exp = 1 },
+        .{ .q = "table[border=\"0\"][cellpadding=\"0\"][cellspacing=\"0\"]", .html = "<table border=\"0\" cellpadding=\"0\" cellspacing=\"0\" style=\"table-layout: fixed; width: 100%; border: 0 dashed; border-color: #FFFFFF\"><tr style=\"height:64px\">aaa</tr></table>", .exp = 1 },
+        .{ .q = ".t1:not(.t2)", .html = "<p class=\"t1 t2\">", .exp = 0 },
+        .{ .q = "div:not(.t1)", .html = "<div class=\"t3\">", .exp = 1 },
+        .{ .q = "div:not([class=\"t2\"])", .html = "<div><div class=\"t2\"><div class=\"t3\">", .exp = 2 },
+        .{ .q = "li:nth-child(odd)", .html = "<ol><li id=1><li id=2><li id=3></ol>", .exp = 2 },
+        .{ .q = "li:nth-child(even)", .html = "<ol><li id=1><li id=2><li id=3></ol>", .exp = 1 },
+        .{ .q = "li:nth-child(-n+2)", .html = "<ol><li id=1><li id=2><li id=3></ol>", .exp = 2 },
+        .{ .q = "li:nth-child(3n+1)", .html = "<ol><li id=1><li id=2><li id=3></ol>", .exp = 1 },
+        .{ .q = "li:nth-last-child(odd)", .html = "<ol><li id=1><li id=2><li id=3><li id=4></ol>", .exp = 2 },
+        .{ .q = "li:nth-last-child(even)", .html = "<ol><li id=1><li id=2><li id=3><li id=4></ol>", .exp = 2 },
+        .{ .q = "li:nth-last-child(-n+2)", .html = "<ol><li id=1><li id=2><li id=3><li id=4></ol>", .exp = 2 },
+        .{ .q = "li:nth-last-child(3n+1)", .html = "<ol><li id=1><li id=2><li id=3><li id=4></ol>", .exp = 2 },
+        .{ .q = "span:first-child", .html = "<p>some text <span id=\"1\">and a span</span><span id=\"2\"> and another</span></p>", .exp = 1 },
+        .{ .q = "span:last-child", .html = "<span>a span</span> and some text", .exp = 1 },
+        .{ .q = "p:nth-of-type(2)", .html = "<address></address><p id=1><p id=2>", .exp = 1 },
+        .{ .q = "p:nth-last-of-type(2)", .html = "<address></address><p id=1><p id=2></p><a>", .exp = 1 },
+        .{ .q = "p:last-of-type", .html = "<address></address><p id=1><p id=2></p><a>", .exp = 1 },
+        .{ .q = "p:first-of-type", .html = "<address></address><p id=1><p id=2></p><a>", .exp = 1 },
+        .{ .q = "p:only-child", .html = "<div><p id=\"1\"></p><a></a></div><div><p id=\"2\"></p></div>", .exp = 1 },
+        .{ .q = "p:only-of-type", .html = "<div><p id=\"1\"></p><a></a></div><div><p id=\"2\"></p><p id=\"3\"></p></div>", .exp = 1 },
+        .{ .q = ":empty", .html = "<p id=\"1\"><!-- --><p id=\"2\">Hello<p id=\"3\"><span>", .exp = 3 },
+        .{ .q = "div p", .html = "<div><p id=\"1\"><table><tr><td><p id=\"2\"></table></div><p id=\"3\">", .exp = 2 },
+        .{ .q = "div table p", .html = "<div><p id=\"1\"><table><tr><td><p id=\"2\"></table></div><p id=\"3\">", .exp = 1 },
+        .{ .q = "div > p", .html = "<div><p id=\"1\"><div><p id=\"2\"></div><table><tr><td><p id=\"3\"></table></div>", .exp = 2 },
+        .{ .q = "p ~ p", .html = "<p id=\"1\"><p id=\"2\"></p><address></address><p id=\"3\">", .exp = 2 },
+        .{ .q = "p + p", .html = "<p id=\"1\"></p> <!--comment--> <p id=\"2\"></p><address></address><p id=\"3\">", .exp = 1 },
+        .{ .q = "li, p", .html = "<ul><li></li><li></li></ul><p>", .exp = 3 },
+        .{ .q = "p +/*This is a comment*/ p", .html = "<p id=\"1\"><p id=\"2\"></p><address></address><p id=\"3\">", .exp = 1 },
+        // .{ .q = "p:contains(\"that wraps\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 1 },
+        .{ .q = "p:containsOwn(\"that wraps\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 0 },
+        .{ .q = ":containsOwn(\"inner\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 1 },
+        .{ .q = ":containsOwn(\"Inner\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 0 },
+        .{ .q = "p:containsOwn(\"block\")", .html = "<p>Text block that <span>wraps inner text</span> and continues</p>", .exp = 1 },
+        .{ .q = "div:has(#p1)", .html = "<div id=\"d1\"><p id=\"p1\"><span>text content</span></p></div><div id=\"d2\"/>", .exp = 1 },
+        .{ .q = "div:has(:containsOwn(\"2\"))", .html = "<div id=\"d1\"><p id=\"p1\"><span>contents 1</span></p></div> <div id=\"d2\"><p>contents <em>2</em></p></div>", .exp = 1 },
+        .{ .q = "body :has(:containsOwn(\"2\"))", .html = "<body><div id=\"d1\"><p id=\"p1\"><span>contents 1</span></p></div> <div id=\"d2\"><p id=\"p2\">contents <em>2</em></p></div></body>", .exp = 2 },
+        .{ .q = "body :haschild(:containsOwn(\"2\"))", .html = "<body><div id=\"d1\"><p id=\"p1\"><span>contents 1</span></p></div> <div id=\"d2\"><p id=\"p2\">contents <em>2</em></p></div></body>", .exp = 1 },
+        // .{ .q = "p:matches([\\d])", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 2 },
+        // .{ .q = "p:matches([a-z])", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 1 },
+        // .{ .q = "p:matches([a-zA-Z])", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 2 },
+        // .{ .q = "p:matches([^\\d])", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 2 },
+        // .{ .q = "p:matches(^(0|a))", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 3 },
+        // .{ .q = "p:matches(^\\d+$)", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 1 },
+        // .{ .q = "p:not(:matches(^\\d+$))", .html = "<p id=\"p1\">0123456789</p><p id=\"p2\">abcdef</p><p id=\"p3\">0123ABCD</p>", .exp = 2 },
+        // .{ .q = "div :matchesOwn(^\\d+$)", .html = "<div><p id=\"p1\">01234<em>567</em>89</p><div>", .exp = 2 },
+        // .{ .q = "[href#=(fina)]:not([href#=(\\/\\/[^\\/]+untrusted)])", .html = "<ul> <li><a id=\"a1\" href=\"http://www.google.com/finance\"></a> <li><a id=\"a2\" href=\"http://finance.yahoo.com/\"></a> <li><a id=\"a2\" href=\"http://finance.untrusted.com/\"/> <li><a id=\"a3\" href=\"https://www.google.com/news\"/> <li><a id=\"a4\" href=\"http://news.yahoo.com\"/> </ul>", .exp = 2 },
+        // .{ .q = "[href#=(^https:\\/\\/[^\\/]*\\/?news)]", .html = "<ul> <li><a id=\"a1\" href=\"http://www.google.com/finance\"/> <li><a id=\"a2\" href=\"http://finance.yahoo.com/\"/> <li><a id=\"a3\" href=\"https://www.google.com/news\"></a> <li><a id=\"a4\" href=\"http://news.yahoo.com\"/> </ul>", .exp = 1 },
+        .{ .q = ":input", .html = "<form> <label>Username <input type=\"text\" name=\"username\" /></label> <label>Password <input type=\"password\" name=\"password\" /></label> <label>Country <select name=\"country\"> <option value=\"ca\">Canada</option> <option value=\"us\">United States</option> </select> </label> <label>Bio <textarea name=\"bio\"></textarea></label> <button>Sign up</button> </form>", .exp = 5 },
+        .{ .q = ":root", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "*:root", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "html:nth-child(1)", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "*:root:first-child", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "*:root:nth-child(1)", .html = "<html><head></head><body></body></html>", .exp = 1 },
+        .{ .q = "a:not(:root)", .html = "<html><head></head><body><a href=\"http://www.foo.com\"></a></body></html>", .exp = 1 },
+        .{ .q = "body > *:nth-child(3n+2)", .html = "<html><head></head><body><p></p><div></div><span></span><a></a><form></form></body></html>", .exp = 2 },
+        .{ .q = "input:disabled", .html = "<html><head></head><body><fieldset disabled><legend id=\"1\"><input id=\"i1\"/></legend><legend id=\"2\"><input id=\"i2\"/></legend></fieldset></body></html>", .exp = 1 },
+        .{ .q = ":disabled", .html = "<html><head></head><body><fieldset disabled></fieldset></body></html>", .exp = 1 },
+        .{ .q = ":enabled", .html = "<html><head></head><body><fieldset></fieldset></body></html>", .exp = 1 },
+        .{ .q = "div.class1, div.class2", .html = "<div class=class1></div><div class=class2></div><div class=class3></div>", .exp = 2 },
+    };
+
+    for (testcases) |tc| {
+        matcher.reset();
+
+        const doc = try parser.documentHTMLParseFromStr(tc.html);
+        defer parser.documentHTMLClose(doc) catch {};
+
+        const s = css.parse(alloc, tc.q, .{}) catch |e| {
+            std.debug.print("parse, query: {s}\n", .{tc.q});
+            return e;
+        };
+        defer s.deinit(alloc);
+
+        const node = Node{ .node = parser.documentHTMLToNode(doc) };
+
+        _ = css.matchAll(&s, node, &matcher) catch |e| {
+            std.debug.print("match, query: {s}\n", .{tc.q});
+            return e;
+        };
+        std.testing.expectEqual(tc.exp, matcher.nodes.items.len) catch |e| {
+            std.debug.print("expectation, query: {s}\n", .{tc.q});
+            return e;
+        };
+    }
+}

src/browser/css/parser.zig

@@ -0,0 +1,996 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+// CSS Selector parser
+// This file is a rewrite in Zig of Cascadia CSS Selector parser.
+// see https://github.com/andybalholm/cascadia
+// see https://github.com/andybalholm/cascadia/blob/master/parser.go
+const std = @import("std");
+const ascii = std.ascii;
+const Allocator = std.mem.Allocator;
+
+const selector = @import("selector.zig");
+const Selector = selector.Selector;
+const PseudoClass = selector.PseudoClass;
+const AttributeOP = selector.AttributeOP;
+const Combinator = selector.Combinator;
+
+const REPLACEMENT_CHARACTER = &.{ 239, 191, 189 };
+
+pub const ParseError = error{
+    ExpectedSelector,
+    ExpectedIdentifier,
+    ExpectedName,
+    ExpectedIDSelector,
+    ExpectedClassSelector,
+    ExpectedAttributeSelector,
+    ExpectedString,
+    ExpectedRegexp,
+    ExpectedPseudoClassSelector,
+    ExpectedParenthesis,
+    ExpectedParenthesisClose,
+    ExpectedNthExpression,
+    ExpectedInteger,
+    InvalidEscape,
+    EscapeLineEndingOutsideString,
+    InvalidUnicode,
+    UnicodeIsNotHandled,
+    WriteError,
+    PseudoElementNotAtSelectorEnd,
+    PseudoElementNotUnique,
+    PseudoElementDisabled,
+    InvalidAttributeOperator,
+    InvalidAttributeSelector,
+    InvalidString,
+    InvalidRegexp,
+    InvalidPseudoClassSelector,
+    EmptyPseudoClassSelector,
+    InvalidPseudoClass,
+    InvalidPseudoElement,
+    UnmatchParenthesis,
+    NotHandled,
+    UnknownPseudoSelector,
+    InvalidNthExpression,
+} || PseudoClass.Error || Combinator.Error || std.mem.Allocator.Error;
+
+pub const ParseOptions = struct {
+    accept_pseudo_elts: bool = true,
+};
+
+pub const Parser = struct {
+    s: []const u8, // string to parse
+    i: usize = 0, // current position
+
+    opts: ParseOptions,
+
+    pub fn parse(p: *Parser, allocator: Allocator) ParseError!Selector {
+        return p.parseSelectorGroup(allocator);
+    }
+
+    // skipWhitespace consumes whitespace characters and comments.
+    // It returns true if there was actually anything to skip.
+    fn skipWhitespace(p: *Parser) bool {
+        var i = p.i;
+        while (i < p.s.len) {
+            const c = p.s[i];
+            // Whitespaces.
+            if (ascii.isWhitespace(c)) {
+                i += 1;
+                continue;
+            }
+
+            // Comments.
+            if (c == '/') {
+                if (std.mem.startsWith(u8, p.s[i..], "/*")) {
+                    if (std.mem.indexOf(u8, p.s[i..], "*/")) |end| {
+                        i += end + "*/".len;
+                        continue;
+                    }
+                }
+            }
+            break;
+        }
+
+        if (i > p.i) {
+            p.i = i;
+            return true;
+        }
+
+        return false;
+    }
+
+    // parseSimpleSelectorSequence parses a selector sequence that applies to
+    // a single element.
+    fn parseSimpleSelectorSequence(p: *Parser, allocator: Allocator) ParseError!Selector {
+        if (p.i >= p.s.len) {
+            return ParseError.ExpectedSelector;
+        }
+
+        var buf: std.ArrayListUnmanaged(Selector) = .empty;
+        defer buf.deinit(allocator);
+
+        switch (p.s[p.i]) {
+            '*' => {
+                // It's the universal selector. Just skip over it, since it
+                // doesn't affect the meaning.
+                p.i += 1;
+
+                // other version of universal selector
+                if (p.i + 2 < p.s.len and std.mem.eql(u8, "|*", p.s[p.i .. p.i + 2])) {
+                    p.i += 2;
+                }
+            },
+            '#', '.', '[', ':' => {
+                // There's no type selector. Wait to process the other till the
+                // main loop.
+            },
+            else => try buf.append(allocator, try p.parseTypeSelector(allocator)),
+        }
+
+        var pseudo_elt: ?PseudoClass = null;
+
+        loop: while (p.i < p.s.len) {
+            var ns: Selector = switch (p.s[p.i]) {
+                '#' => try p.parseIDSelector(allocator),
+                '.' => try p.parseClassSelector(allocator),
+                '[' => try p.parseAttributeSelector(allocator),
+                ':' => try p.parsePseudoclassSelector(allocator),
+                else => break :loop,
+            };
+            errdefer ns.deinit(allocator);
+
+            // From https://drafts.csswg.org/selectors-3/#pseudo-elements :
+            // "Only one pseudo-element may appear per selector, and if present
+            // it must appear after the sequence of simple selectors that
+            // represents the subjects of the selector.""
+            switch (ns) {
+                .pseudo_element => |e| {
+                    //  We found a pseudo-element.
+                    //  Only one pseudo-element is accepted per selector.
+                    if (pseudo_elt != null) return ParseError.PseudoElementNotUnique;
+                    if (!p.opts.accept_pseudo_elts) return ParseError.PseudoElementDisabled;
+
+                    pseudo_elt = e;
+                    ns.deinit(allocator);
+                },
+                else => {
+                    if (pseudo_elt != null) return ParseError.PseudoElementNotAtSelectorEnd;
+                    try buf.append(allocator, ns);
+                },
+            }
+        }
+
+        // no need wrap the selectors in compoundSelector
+        if (buf.items.len == 1 and pseudo_elt == null) {
+            return buf.items[0];
+        }
+
+        return .{
+            .compound = .{ .selectors = try buf.toOwnedSlice(allocator), .pseudo_elt = pseudo_elt },
+        };
+    }
+
+    // parseTypeSelector parses a type selector (one that matches by tag name).
+    fn parseTypeSelector(p: *Parser, allocator: Allocator) ParseError!Selector {
+        var buf: std.ArrayListUnmanaged(u8) = .empty;
+        defer buf.deinit(allocator);
+        try p.parseIdentifier(buf.writer(allocator));
+
+        return .{ .tag = try buf.toOwnedSlice(allocator) };
+    }
+
+    // parseIdentifier parses an identifier.
+    fn parseIdentifier(p: *Parser, w: anytype) ParseError!void {
+        const prefix = '-';
+        var numPrefix: usize = 0;
+
+        while (p.s.len > p.i and p.s[p.i] == prefix) {
+            p.i += 1;
+            numPrefix += 1;
+        }
+
+        if (p.s.len <= p.i) {
+            return ParseError.ExpectedSelector;
+        }
+
+        const c = p.s[p.i];
+        if (!(nameStart(c) or c == '\\')) {
+            return ParseError.ExpectedSelector;
+        }
+
+        var ii: usize = 0;
+        while (ii < numPrefix) {
+            w.writeByte(prefix) catch return ParseError.WriteError;
+            ii += 1;
+        }
+        try parseName(p, w);
+    }
+
+    // parseName parses a name (which is like an identifier, but doesn't have
+    // extra restrictions on the first character).
+    fn parseName(p: *Parser, w: anytype) ParseError!void {
+        const sel = p.s;
+        const sel_len = sel.len;
+
+        var i = p.i;
+        var ok = false;
+
+        while (i < sel_len) {
+            const c = sel[i];
+
+            if (nameChar(c)) {
+                const start = i;
+                while (i < sel_len and nameChar(sel[i])) i += 1;
+                w.writeAll(sel[start..i]) catch return ParseError.WriteError;
+                ok = true;
+            } else if (c == '\\') {
+                p.i = i;
+                try p.parseEscape(w);
+                i = p.i;
+                ok = true;
+            } else if (c == 0) {
+                w.writeAll(REPLACEMENT_CHARACTER) catch return ParseError.WriteError;
+                i += 1;
+                if (i == sel_len) {
+                    ok = true;
+                }
+            } else {
+                // default:
+                break;
+            }
+        }
+
+        if (!ok) return ParseError.ExpectedName;
+        p.i = i;
+    }
+
+    // parseEscape parses a backslash escape.
+    // The returned string is owned by the caller.
+    fn parseEscape(p: *Parser, w: anytype) ParseError!void {
+        const sel = p.s;
+        const sel_len = sel.len;
+
+        if (sel_len < p.i + 2 or sel[p.i] != '\\') {
+            p.i += 1;
+            w.writeAll(REPLACEMENT_CHARACTER) catch return ParseError.WriteError;
+            return;
+        }
+
+        const start = p.i + 1;
+        const c = sel[start];
+
+        // unicode escape (hex)
+        if (ascii.isHex(c)) {
+            var i: usize = start;
+            while (i < start + 6 and i < sel_len and ascii.isHex(sel[i])) {
+                i += 1;
+            }
+
+            const v = std.fmt.parseUnsigned(u21, sel[start..i], 16) catch {
+                p.i = i;
+                w.writeAll(REPLACEMENT_CHARACTER) catch return ParseError.WriteError;
+                return;
+            };
+
+            if (sel_len >= i) {
+                if (sel_len > i) {
+                    switch (sel[i]) {
+                        '\r' => {
+                            i += 1;
+                            if (sel_len > i and sel[i] == '\n') i += 1;
+                        },
+                        ' ', '\t', '\n', std.ascii.control_code.ff => i += 1,
+                        else => {},
+                    }
+                }
+                p.i = i;
+                if (v == 0) {
+                    w.writeAll(REPLACEMENT_CHARACTER) catch return ParseError.WriteError;
+                    return;
+                }
+                var buf: [4]u8 = undefined;
+                const ln = std.unicode.utf8Encode(v, &buf) catch {
+                    w.writeAll(REPLACEMENT_CHARACTER) catch return ParseError.WriteError;
+                    return;
+                };
+                w.writeAll(buf[0..ln]) catch return ParseError.WriteError;
+                return;
+            }
+        }
+
+        // Return the literal character after the backslash.
+        p.i += 2;
+        w.writeByte(sel[start]) catch return ParseError.WriteError;
+    }
+
+    // parseIDSelector parses a selector that matches by id attribute.
+    fn parseIDSelector(p: *Parser, allocator: Allocator) ParseError!Selector {
+        if (p.i >= p.s.len) return ParseError.ExpectedIDSelector;
+        if (p.s[p.i] != '#') return ParseError.ExpectedIDSelector;
+
+        p.i += 1;
+
+        var buf: std.ArrayListUnmanaged(u8) = .empty;
+        defer buf.deinit(allocator);
+
+        try p.parseName(buf.writer(allocator));
+        return .{ .id = try buf.toOwnedSlice(allocator) };
+    }
+
+    // parseClassSelector parses a selector that matches by class attribute.
+    fn parseClassSelector(p: *Parser, allocator: Allocator) ParseError!Selector {
+        if (p.i >= p.s.len) return ParseError.ExpectedClassSelector;
+        if (p.s[p.i] != '.') return ParseError.ExpectedClassSelector;
+
+        p.i += 1;
+
+        var buf: std.ArrayListUnmanaged(u8) = .empty;
+        defer buf.deinit(allocator);
+
+        try p.parseIdentifier(buf.writer(allocator));
+        return .{ .class = try buf.toOwnedSlice(allocator) };
+    }
+
+    // parseAttributeSelector parses a selector that matches by attribute value.
+    fn parseAttributeSelector(p: *Parser, allocator: Allocator) ParseError!Selector {
+        if (p.i >= p.s.len) return ParseError.ExpectedAttributeSelector;
+        if (p.s[p.i] != '[') return ParseError.ExpectedAttributeSelector;
+
+        p.i += 1;
+        _ = p.skipWhitespace();
+
+        var buf: std.ArrayListUnmanaged(u8) = .empty;
+        defer buf.deinit(allocator);
+
+        try p.parseIdentifier(buf.writer(allocator));
+        const key = try buf.toOwnedSlice(allocator);
+        errdefer allocator.free(key);
+
+        lowerstr(key);
+
+        _ = p.skipWhitespace();
+        if (p.i >= p.s.len) return ParseError.ExpectedAttributeSelector;
+        if (p.s[p.i] == ']') {
+            p.i += 1;
+            return .{ .attribute = .{ .key = key } };
+        }
+
+        if (p.i + 2 >= p.s.len) return ParseError.ExpectedAttributeSelector;
+
+        const op = try parseAttributeOP(p.s[p.i .. p.i + 2]);
+        p.i += op.len();
+
+        _ = p.skipWhitespace();
+        if (p.i >= p.s.len) return ParseError.ExpectedAttributeSelector;
+
+        buf.clearRetainingCapacity();
+        var is_val: bool = undefined;
+        if (op == .regexp) {
+            is_val = false;
+            try p.parseRegex(buf.writer(allocator));
+        } else {
+            is_val = true;
+            switch (p.s[p.i]) {
+                '\'', '"' => try p.parseString(buf.writer(allocator)),
+                else => try p.parseIdentifier(buf.writer(allocator)),
+            }
+        }
+
+        _ = p.skipWhitespace();
+        if (p.i >= p.s.len) return ParseError.ExpectedAttributeSelector;
+
+        // check if the attribute contains an ignore case flag
+        var ci = false;
+        if (p.s[p.i] == 'i' or p.s[p.i] == 'I') {
+            ci = true;
+            p.i += 1;
+        }
+
+        _ = p.skipWhitespace();
+        if (p.i >= p.s.len) return ParseError.ExpectedAttributeSelector;
+
+        if (p.s[p.i] != ']') return ParseError.InvalidAttributeSelector;
+        p.i += 1;
+
+        return .{ .attribute = .{
+            .key = key,
+            .val = if (is_val) try buf.toOwnedSlice(allocator) else null,
+            .regexp = if (!is_val) try buf.toOwnedSlice(allocator) else null,
+            .op = op,
+            .ci = ci,
+        } };
+    }
+
+    // parseString parses a single- or double-quoted string.
+    fn parseString(p: *Parser, writer: anytype) ParseError!void {
+        const sel = p.s;
+        const sel_len = sel.len;
+
+        var i = p.i;
+        if (sel_len < i + 2) return ParseError.ExpectedString;
+
+        const quote = sel[i];
+        i += 1;
+
+        loop: while (i < sel_len) {
+            switch (sel[i]) {
+                '\\' => {
+                    if (sel_len > i + 1) {
+                        const c = sel[i + 1];
+                        switch (c) {
+                            '\r' => {
+                                if (sel_len > i + 2 and sel[i + 2] == '\n') {
+                                    i += 3;
+                                    continue :loop;
+                                }
+                                i += 2;
+                                continue :loop;
+                            },
+                            '\n', std.ascii.control_code.ff => {
+                                i += 2;
+                                continue :loop;
+                            },
+                            else => {},
+                        }
+                    }
+                    p.i = i;
+                    try p.parseEscape(writer);
+                    i = p.i;
+                },
+                '\r', '\n', std.ascii.control_code.ff => return ParseError.InvalidString,
+                else => |c| {
+                    if (c == quote) break :loop;
+                    const start = i;
+                    while (i < sel_len) {
+                        const cc = sel[i];
+                        if (cc == quote or cc == '\\' or c == '\r' or c == '\n' or c == std.ascii.control_code.ff) break;
+                        i += 1;
+                    }
+                    writer.writeAll(sel[start..i]) catch return ParseError.WriteError;
+                },
+            }
+        }
+
+        if (i >= sel_len) return ParseError.InvalidString;
+
+        // Consume the final quote.
+        i += 1;
+        p.i = i;
+    }
+
+    // parseRegex parses a regular expression; the end is defined by encountering an
+    // unmatched closing ')' or ']' which is not consumed
+    fn parseRegex(p: *Parser, writer: anytype) ParseError!void {
+        var i = p.i;
+        if (p.s.len < i + 2) return ParseError.ExpectedRegexp;
+
+        // number of open parens or brackets;
+        // when it becomes negative, finished parsing regex
+        var open: isize = 0;
+
+        loop: while (i < p.s.len) {
+            switch (p.s[i]) {
+                '(', '[' => open += 1,
+                ')', ']' => {
+                    open -= 1;
+                    if (open < 0) break :loop;
+                },
+                else => {},
+            }
+            i += 1;
+        }
+
+        if (i >= p.s.len) return ParseError.InvalidRegexp;
+        writer.writeAll(p.s[p.i..i]) catch return ParseError.WriteError;
+        p.i = i;
+    }
+
+    // parsePseudoclassSelector parses a pseudoclass selector like :not(p) or a pseudo-element
+    // For backwards compatibility, both ':' and '::' prefix are allowed for pseudo-elements.
+    // https://drafts.csswg.org/selectors-3/#pseudo-elements
+    fn parsePseudoclassSelector(p: *Parser, allocator: Allocator) ParseError!Selector {
+        if (p.i >= p.s.len) return ParseError.ExpectedPseudoClassSelector;
+        if (p.s[p.i] != ':') return ParseError.ExpectedPseudoClassSelector;
+
+        p.i += 1;
+
+        var must_pseudo_elt: bool = false;
+        if (p.i >= p.s.len) return ParseError.EmptyPseudoClassSelector;
+        if (p.s[p.i] == ':') { // we found a pseudo-element
+            must_pseudo_elt = true;
+            p.i += 1;
+        }
+
+        var buf: std.ArrayListUnmanaged(u8) = .empty;
+        defer buf.deinit(allocator);
+
+        try p.parseIdentifier(buf.writer(allocator));
+
+        const pseudo_class = try PseudoClass.parse(buf.items);
+
+        // reset the buffer to reuse it.
+        buf.clearRetainingCapacity();
+
+        if (must_pseudo_elt and !pseudo_class.isPseudoElement()) return ParseError.InvalidPseudoElement;
+
+        switch (pseudo_class) {
+            .not, .has, .haschild => {
+                if (!p.consumeParenthesis()) return ParseError.ExpectedParenthesis;
+
+                const sel = try p.parseSelectorGroup(allocator);
+                if (!p.consumeClosingParenthesis()) return ParseError.ExpectedParenthesisClose;
+
+                const s = try allocator.create(Selector);
+                errdefer allocator.destroy(s);
+                s.* = sel;
+
+                return .{ .pseudo_class_relative = .{ .pseudo_class = pseudo_class, .match = s } };
+            },
+            .contains, .containsown => {
+                if (!p.consumeParenthesis()) return ParseError.ExpectedParenthesis;
+                if (p.i == p.s.len) return ParseError.UnmatchParenthesis;
+
+                switch (p.s[p.i]) {
+                    '\'', '"' => try p.parseString(buf.writer(allocator)),
+                    else => try p.parseString(buf.writer(allocator)),
+                }
+
+                _ = p.skipWhitespace();
+                if (p.i >= p.s.len) return ParseError.InvalidPseudoClass;
+                if (!p.consumeClosingParenthesis()) return ParseError.ExpectedParenthesisClose;
+
+                const val = try buf.toOwnedSlice(allocator);
+                errdefer allocator.free(val);
+
+                return .{ .pseudo_class_contains = .{ .own = pseudo_class == .containsown, .val = val } };
+            },
+            .matches, .matchesown => {
+                if (!p.consumeParenthesis()) return ParseError.ExpectedParenthesis;
+
+                try p.parseRegex(buf.writer(allocator));
+                if (p.i >= p.s.len) return ParseError.InvalidPseudoClassSelector;
+                if (!p.consumeClosingParenthesis()) return ParseError.ExpectedParenthesisClose;
+
+                return .{ .pseudo_class_regexp = .{ .own = pseudo_class == .matchesown, .regexp = try buf.toOwnedSlice(allocator) } };
+            },
+            .nth_child, .nth_last_child, .nth_of_type, .nth_last_of_type => {
+                if (!p.consumeParenthesis()) return ParseError.ExpectedParenthesis;
+                const nth = try p.parseNth(allocator);
+                if (!p.consumeClosingParenthesis()) return ParseError.ExpectedParenthesisClose;
+
+                const last = pseudo_class == .nth_last_child or pseudo_class == .nth_last_of_type;
+                const of_type = pseudo_class == .nth_of_type or pseudo_class == .nth_last_of_type;
+                return .{ .pseudo_class_nth = .{ .a = nth[0], .b = nth[1], .of_type = of_type, .last = last } };
+            },
+            .first_child => return .{ .pseudo_class_nth = .{ .a = 0, .b = 1, .of_type = false, .last = false } },
+            .last_child => return .{ .pseudo_class_nth = .{ .a = 0, .b = 1, .of_type = false, .last = true } },
+            .first_of_type => return .{ .pseudo_class_nth = .{ .a = 0, .b = 1, .of_type = true, .last = false } },
+            .last_of_type => return .{ .pseudo_class_nth = .{ .a = 0, .b = 1, .of_type = true, .last = true } },
+            .only_child => return .{ .pseudo_class_only_child = false },
+            .only_of_type => return .{ .pseudo_class_only_child = true },
+            .input, .empty, .root, .link => return .{ .pseudo_class = pseudo_class },
+            .enabled, .disabled, .checked => return .{ .pseudo_class = pseudo_class },
+            .visible => return .{ .pseudo_class = pseudo_class },
+            .lang => {
+                if (!p.consumeParenthesis()) return ParseError.ExpectedParenthesis;
+                if (p.i == p.s.len) return ParseError.UnmatchParenthesis;
+
+                try p.parseIdentifier(buf.writer(allocator));
+
+                _ = p.skipWhitespace();
+                if (p.i >= p.s.len) return ParseError.InvalidPseudoClass;
+                if (!p.consumeClosingParenthesis()) return ParseError.ExpectedParenthesisClose;
+
+                const val = try buf.toOwnedSlice(allocator);
+                errdefer allocator.free(val);
+                lowerstr(val);
+
+                return .{ .pseudo_class_lang = val };
+            },
+            .visited, .hover, .active, .focus, .target => {
+                // Not applicable in a static context: never match.
+                return .{ .never_match = pseudo_class };
+            },
+            .after, .backdrop, .before, .cue, .first_letter => return .{ .pseudo_element = pseudo_class },
+            .first_line, .grammar_error, .marker, .placeholder => return .{ .pseudo_element = pseudo_class },
+            .selection, .spelling_error => return .{ .pseudo_element = pseudo_class },
+            .modal, .popover_open => return .{ .pseudo_element = pseudo_class },
+        }
+    }
+
+    // consumeParenthesis consumes an opening parenthesis and any following
+    // whitespace. It returns true if there was actually a parenthesis to skip.
+    fn consumeParenthesis(p: *Parser) bool {
+        if (p.i < p.s.len and p.s[p.i] == '(') {
+            p.i += 1;
+            _ = p.skipWhitespace();
+            return true;
+        }
+        return false;
+    }
+
+    // parseSelectorGroup parses a group of selectors, separated by commas.
+    fn parseSelectorGroup(p: *Parser, allocator: Allocator) ParseError!Selector {
+        const s = try p.parseSelector(allocator);
+
+        var buf: std.ArrayListUnmanaged(Selector) = .empty;
+        defer buf.deinit(allocator);
+
+        try buf.append(allocator, s);
+
+        while (p.i < p.s.len) {
+            if (p.s[p.i] != ',') break;
+            p.i += 1;
+            const ss = try p.parseSelector(allocator);
+            try buf.append(allocator, ss);
+        }
+
+        if (buf.items.len == 1) {
+            return buf.items[0];
+        }
+
+        return .{ .group = try buf.toOwnedSlice(allocator) };
+    }
+
+    // parseSelector parses a selector that may include combinators.
+    fn parseSelector(p: *Parser, allocator: Allocator) ParseError!Selector {
+        _ = p.skipWhitespace();
+        var s = try p.parseSimpleSelectorSequence(allocator);
+
+        while (true) {
+            var combinator: Combinator = .empty;
+            if (p.skipWhitespace()) {
+                combinator = .descendant;
+            }
+            if (p.i >= p.s.len) {
+                return s;
+            }
+
+            switch (p.s[p.i]) {
+                '+', '>', '~' => {
+                    combinator = try Combinator.parse(p.s[p.i]);
+                    p.i += 1;
+                    _ = p.skipWhitespace();
+                },
+                // These characters can't begin a selector, but they can legally occur after one.
+                ',', ')' => {
+                    return s;
+                },
+                else => {},
+            }
+
+            if (combinator == .empty) {
+                return s;
+            }
+
+            const c = try p.parseSimpleSelectorSequence(allocator);
+
+            const first = try allocator.create(Selector);
+            errdefer allocator.destroy(first);
+            first.* = s;
+
+            const second = try allocator.create(Selector);
+            errdefer allocator.destroy(second);
+            second.* = c;
+
+            s = Selector{ .combined = .{
+                .first = first,
+                .second = second,
+                .combinator = combinator,
+            } };
+        }
+
+        return s;
+    }
+
+    // consumeClosingParenthesis consumes a closing parenthesis and any preceding
+    // whitespace. It returns true if there was actually a parenthesis to skip.
+    fn consumeClosingParenthesis(p: *Parser) bool {
+        const i = p.i;
+        _ = p.skipWhitespace();
+        if (p.i < p.s.len and p.s[p.i] == ')') {
+            p.i += 1;
+            return true;
+        }
+        p.i = i;
+        return false;
+    }
+
+    // parseInteger parses a  decimal integer.
+    fn parseInteger(p: *Parser) ParseError!isize {
+        var i = p.i;
+        const start = i;
+        while (i < p.s.len and '0' <= p.s[i] and p.s[i] <= '9') i += 1;
+        if (i == start) return ParseError.ExpectedInteger;
+        p.i = i;
+
+        return std.fmt.parseUnsigned(isize, p.s[start..i], 10) catch ParseError.ExpectedInteger;
+    }
+
+    fn parseNthReadN(p: *Parser, a: isize) ParseError![2]isize {
+        _ = p.skipWhitespace();
+        if (p.i >= p.s.len) return ParseError.ExpectedNthExpression;
+
+        return switch (p.s[p.i]) {
+            '+' => {
+                p.i += 1;
+                _ = p.skipWhitespace();
+                const b = try p.parseInteger();
+                return .{ a, b };
+            },
+            '-' => {
+                p.i += 1;
+                _ = p.skipWhitespace();
+                const b = try p.parseInteger();
+                return .{ a, -b };
+            },
+            else => .{ a, 0 },
+        };
+    }
+
+    fn parseNthReadA(p: *Parser, a: isize) ParseError![2]isize {
+        if (p.i >= p.s.len) return ParseError.ExpectedNthExpression;
+        return switch (p.s[p.i]) {
+            'n', 'N' => {
+                p.i += 1;
+                return p.parseNthReadN(a);
+            },
+            else => .{ 0, a },
+        };
+    }
+
+    fn parseNthNegativeA(p: *Parser) ParseError![2]isize {
+        if (p.i >= p.s.len) return ParseError.ExpectedNthExpression;
+        const c = p.s[p.i];
+        if (std.ascii.isDigit(c)) {
+            const a = try p.parseInteger() * -1;
+            return p.parseNthReadA(a);
+        }
+        if (c == 'n' or c == 'N') {
+            p.i += 1;
+            return p.parseNthReadN(-1);
+        }
+
+        return ParseError.InvalidNthExpression;
+    }
+
+    fn parseNthPositiveA(p: *Parser) ParseError![2]isize {
+        if (p.i >= p.s.len) return ParseError.ExpectedNthExpression;
+        const c = p.s[p.i];
+        if (std.ascii.isDigit(c)) {
+            const a = try p.parseInteger();
+            return p.parseNthReadA(a);
+        }
+        if (c == 'n' or c == 'N') {
+            p.i += 1;
+            return p.parseNthReadN(1);
+        }
+
+        return ParseError.InvalidNthExpression;
+    }
+
+    // parseNth parses the argument for :nth-child (normally of the form an+b).
+    fn parseNth(p: *Parser, allocator: Allocator) ParseError![2]isize {
+        // initial state
+        if (p.i >= p.s.len) return ParseError.ExpectedNthExpression;
+        return switch (p.s[p.i]) {
+            '-' => {
+                p.i += 1;
+                return p.parseNthNegativeA();
+            },
+            '+' => {
+                p.i += 1;
+                return p.parseNthPositiveA();
+            },
+            '0', '1', '2', '3', '4', '5', '6', '7', '8', '9' => p.parseNthPositiveA(),
+            'n', 'N' => {
+                p.i += 1;
+                return p.parseNthReadN(1);
+            },
+            'o', 'O', 'e', 'E' => {
+                var buf: std.ArrayListUnmanaged(u8) = .empty;
+                defer buf.deinit(allocator);
+
+                try p.parseName(buf.writer(allocator));
+
+                if (std.ascii.eqlIgnoreCase("odd", buf.items)) return .{ 2, 1 };
+                if (std.ascii.eqlIgnoreCase("even", buf.items)) return .{ 2, 0 };
+
+                return ParseError.InvalidNthExpression;
+            },
+            else => ParseError.InvalidNthExpression,
+        };
+    }
+};
+
+// nameStart returns whether c can be the first character of an identifier
+// (not counting an initial hyphen, or an escape sequence).
+fn nameStart(c: u8) bool {
+    return 'a' <= c and c <= 'z' or 'A' <= c and c <= 'Z' or c == '_' or c > 127 or
+        '0' <= c and c <= '9';
+}
+
+// nameChar returns whether c can be a character within an identifier
+// (not counting an escape sequence).
+fn nameChar(c: u8) bool {
+    return 'a' <= c and c <= 'z' or 'A' <= c and c <= 'Z' or c == '_' or c > 127 or
+        c == '-' or '0' <= c and c <= '9';
+}
+
+fn lowerstr(str: []u8) void {
+    for (str, 0..) |c, i| {
+        str[i] = std.ascii.toLower(c);
+    }
+}
+
+// parseAttributeOP parses an AttributeOP from a string of 1 or 2 bytes.
+fn parseAttributeOP(s: []const u8) ParseError!AttributeOP {
+    if (s.len < 1 or s.len > 2) return ParseError.InvalidAttributeOperator;
+
+    // if the first sign is equal, we don't check anything else.
+    if (s[0] == '=') return .eql;
+
+    if (s.len != 2 or s[1] != '=') return ParseError.InvalidAttributeOperator;
+
+    return switch (s[0]) {
+        '=' => .eql,
+        '!' => .not_eql,
+        '~' => .one_of,
+        '|' => .prefix_hyphen,
+        '^' => .prefix,
+        '$' => .suffix,
+        '*' => .contains,
+        '#' => .regexp,
+        else => ParseError.InvalidAttributeOperator,
+    };
+}
+
+test "parser.skipWhitespace" {
+    const testcases = [_]struct {
+        s: []const u8,
+        i: usize,
+        r: bool,
+    }{
+        .{ .s = "", .i = 0, .r = false },
+        .{ .s = "foo", .i = 0, .r = false },
+        .{ .s = " ", .i = 1, .r = true },
+        .{ .s = " foo", .i = 1, .r = true },
+        .{ .s = "/* foo */ bar", .i = 10, .r = true },
+        .{ .s = "/* foo", .i = 0, .r = false },
+    };
+
+    for (testcases) |tc| {
+        var p = Parser{ .s = tc.s, .opts = .{} };
+        const res = p.skipWhitespace();
+        try std.testing.expectEqual(tc.r, res);
+        try std.testing.expectEqual(tc.i, p.i);
+    }
+}
+
+test "parser.parseIdentifier" {
+    const allocator = std.testing.allocator;
+
+    const testcases = [_]struct {
+        s: []const u8, // given value
+        exp: []const u8, // expected value
+        err: bool = false,
+    }{
+        .{ .s = "x", .exp = "x" },
+        .{ .s = "96", .exp = "96", .err = false },
+        .{ .s = "-x", .exp = "-x" },
+        .{ .s = "r\\e9 sumé", .exp = "résumé" },
+        .{ .s = "r\\0000e9 sumé", .exp = "résumé" },
+        .{ .s = "r\\0000e9sumé", .exp = "résumé" },
+        .{ .s = "a\\\"b", .exp = "a\"b" },
+    };
+
+    var buf: std.ArrayListUnmanaged(u8) = .empty;
+    defer buf.deinit(allocator);
+
+    for (testcases) |tc| {
+        buf.clearRetainingCapacity();
+
+        var p = Parser{ .s = tc.s, .opts = .{} };
+        p.parseIdentifier(buf.writer(allocator)) catch |e| {
+            // if error was expected, continue.
+            if (tc.err) continue;
+
+            std.debug.print("test case {s}\n", .{tc.s});
+            return e;
+        };
+        std.testing.expectEqualDeep(tc.exp, buf.items) catch |e| {
+            std.debug.print("test case {s} : {s}\n", .{ tc.s, buf.items });
+            return e;
+        };
+    }
+}
+
+test "parser.parseString" {
+    const allocator = std.testing.allocator;
+
+    const testcases = [_]struct {
+        s: []const u8, // given value
+        exp: []const u8, // expected value
+        err: bool = false,
+    }{
+        .{ .s = "\"x\"", .exp = "x" },
+        .{ .s = "'x'", .exp = "x" },
+        .{ .s = "'x", .exp = "", .err = true },
+        .{ .s = "'x\\\r\nx'", .exp = "xx" },
+        .{ .s = "\"r\\e9 sumé\"", .exp = "résumé" },
+        .{ .s = "\"r\\0000e9 sumé\"", .exp = "résumé" },
+        .{ .s = "\"r\\0000e9sumé\"", .exp = "résumé" },
+        .{ .s = "\"a\\\"b\"", .exp = "a\"b" },
+        .{ .s = "\"\\\n\"", .exp = "" },
+        .{ .s = "\"hello world\"", .exp = "hello world" },
+    };
+
+    var buf: std.ArrayListUnmanaged(u8) = .empty;
+    defer buf.deinit(allocator);
+
+    for (testcases) |tc| {
+        buf.clearRetainingCapacity();
+
+        var p = Parser{ .s = tc.s, .opts = .{} };
+        p.parseString(buf.writer(allocator)) catch |e| {
+            // if error was expected, continue.
+            if (tc.err) continue;
+
+            std.debug.print("test case {s}\n", .{tc.s});
+            return e;
+        };
+        std.testing.expectEqualDeep(tc.exp, buf.items) catch |e| {
+            std.debug.print("test case {s} : {s}\n", .{ tc.s, buf.items });
+            return e;
+        };
+    }
+}
+
+test "parser.parse" {
+    var arena = std.heap.ArenaAllocator.init(std.testing.allocator);
+    defer arena.deinit();
+    const allocator = arena.allocator();
+
+    const testcases = [_]struct {
+        s: []const u8, // given value
+        exp: Selector, // expected value
+        err: bool = false,
+    }{
+        .{ .s = "root", .exp = .{ .tag = "root" } },
+        .{ .s = ".root", .exp = .{ .class = "root" } },
+        .{ .s = ":root", .exp = .{ .pseudo_class = .root } },
+        .{ .s = ".\\:bar", .exp = .{ .class = ":bar" } },
+        .{ .s = ".foo\\:bar", .exp = .{ .class = "foo:bar" } },
+        .{ .s = "[class=75c0fa18a94b9e3a6b8e14d6cbe688a27f5da10a]", .exp = .{ .attribute = .{ .key = "class", .val = "75c0fa18a94b9e3a6b8e14d6cbe688a27f5da10a", .op = .eql } } },
+    };
+
+    for (testcases) |tc| {
+        var p = Parser{ .s = tc.s, .opts = .{} };
+        const sel = p.parse(allocator) catch |e| {
+            // if error was expected, continue.
+            if (tc.err) continue;
+
+            std.debug.print("test case {s}\n", .{tc.s});
+            return e;
+        };
+        std.testing.expectEqualDeep(tc.exp, sel) catch |e| {
+            std.debug.print("test case {s} : {}\n", .{ tc.s, sel });
+            return e;
+        };
+    }
+}

src/browser/cssom/CSSParser.zig

@@ -0,0 +1,289 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const Allocator = std.mem.Allocator;
+
+const CSSConstants = struct {
+    const IMPORTANT = "!important";
+    const URL_PREFIX = "url(";
+};
+
+const CSSParserState = enum {
+    seek_name,
+    in_name,
+    seek_colon,
+    seek_value,
+    in_value,
+    in_quoted_value,
+    in_single_quoted_value,
+    in_url,
+    in_important,
+};
+
+const CSSDeclaration = struct {
+    name: []const u8,
+    value: []const u8,
+    is_important: bool,
+};
+
+const CSSParser = @This();
+state: CSSParserState,
+name_start: usize,
+name_end: usize,
+value_start: usize,
+position: usize,
+paren_depth: usize,
+escape_next: bool,
+
+pub fn init() CSSParser {
+    return .{
+        .state = .seek_name,
+        .name_start = 0,
+        .name_end = 0,
+        .value_start = 0,
+        .position = 0,
+        .paren_depth = 0,
+        .escape_next = false,
+    };
+}
+
+pub fn parseDeclarations(arena: Allocator, text: []const u8) ![]CSSDeclaration {
+    var parser = init();
+    var declarations: std.ArrayListUnmanaged(CSSDeclaration) = .empty;
+
+    while (parser.position < text.len) {
+        const c = text[parser.position];
+
+        switch (parser.state) {
+            .seek_name => {
+                if (!std.ascii.isWhitespace(c)) {
+                    parser.name_start = parser.position;
+                    parser.state = .in_name;
+                    continue;
+                }
+            },
+            .in_name => {
+                if (c == ':') {
+                    parser.name_end = parser.position;
+                    parser.state = .seek_value;
+                } else if (std.ascii.isWhitespace(c)) {
+                    parser.name_end = parser.position;
+                    parser.state = .seek_colon;
+                }
+            },
+            .seek_colon => {
+                if (c == ':') {
+                    parser.state = .seek_value;
+                } else if (!std.ascii.isWhitespace(c)) {
+                    parser.state = .seek_name;
+                    continue;
+                }
+            },
+            .seek_value => {
+                if (!std.ascii.isWhitespace(c)) {
+                    parser.value_start = parser.position;
+                    if (c == '"') {
+                        parser.state = .in_quoted_value;
+                    } else if (c == '\'') {
+                        parser.state = .in_single_quoted_value;
+                    } else if (c == 'u' and parser.position + CSSConstants.URL_PREFIX.len <= text.len and std.mem.startsWith(u8, text[parser.position..], CSSConstants.URL_PREFIX)) {
+                        parser.state = .in_url;
+                        parser.paren_depth = 1;
+                        parser.position += 3;
+                    } else {
+                        parser.state = .in_value;
+                        continue;
+                    }
+                }
+            },
+            .in_value => {
+                if (parser.escape_next) {
+                    parser.escape_next = false;
+                } else if (c == '\\') {
+                    parser.escape_next = true;
+                } else if (c == '(') {
+                    parser.paren_depth += 1;
+                } else if (c == ')' and parser.paren_depth > 0) {
+                    parser.paren_depth -= 1;
+                } else if (c == ';' and parser.paren_depth == 0) {
+                    try parser.finishDeclaration(arena, &declarations, text);
+                    parser.state = .seek_name;
+                }
+            },
+            .in_quoted_value => {
+                if (parser.escape_next) {
+                    parser.escape_next = false;
+                } else if (c == '\\') {
+                    parser.escape_next = true;
+                } else if (c == '"') {
+                    parser.state = .in_value;
+                }
+            },
+            .in_single_quoted_value => {
+                if (parser.escape_next) {
+                    parser.escape_next = false;
+                } else if (c == '\\') {
+                    parser.escape_next = true;
+                } else if (c == '\'') {
+                    parser.state = .in_value;
+                }
+            },
+            .in_url => {
+                if (parser.escape_next) {
+                    parser.escape_next = false;
+                } else if (c == '\\') {
+                    parser.escape_next = true;
+                } else if (c == '(') {
+                    parser.paren_depth += 1;
+                } else if (c == ')') {
+                    parser.paren_depth -= 1;
+                    if (parser.paren_depth == 0) {
+                        parser.state = .in_value;
+                    }
+                }
+            },
+            .in_important => {},
+        }
+
+        parser.position += 1;
+    }
+
+    try parser.finalize(arena, &declarations, text);
+
+    return declarations.items;
+}
+
+fn finishDeclaration(self: *CSSParser, arena: Allocator, declarations: *std.ArrayListUnmanaged(CSSDeclaration), text: []const u8) !void {
+    const name = std.mem.trim(u8, text[self.name_start..self.name_end], &std.ascii.whitespace);
+    if (name.len == 0) return;
+
+    const raw_value = text[self.value_start..self.position];
+    const value = std.mem.trim(u8, raw_value, &std.ascii.whitespace);
+
+    var final_value = value;
+    var is_important = false;
+
+    if (std.mem.endsWith(u8, value, CSSConstants.IMPORTANT)) {
+        is_important = true;
+        final_value = std.mem.trimRight(u8, value[0 .. value.len - CSSConstants.IMPORTANT.len], &std.ascii.whitespace);
+    }
+
+    try declarations.append(arena, .{
+        .name = name,
+        .value = final_value,
+        .is_important = is_important,
+    });
+}
+
+fn finalize(self: *CSSParser, arena: Allocator, declarations: *std.ArrayListUnmanaged(CSSDeclaration), text: []const u8) !void {
+    if (self.state != .in_value) {
+        return;
+    }
+    return self.finishDeclaration(arena, declarations, text);
+}
+
+const testing = @import("../../testing.zig");
+test "Browser: CSS.Parser - Simple property" {
+    defer testing.reset();
+
+    const text = "color: red;";
+    const allocator = testing.arena_allocator;
+
+    const declarations = try CSSParser.parseDeclarations(allocator, text);
+
+    try testing.expectEqual(1, declarations.len);
+    try testing.expectEqual("color", declarations[0].name);
+    try testing.expectEqual("red", declarations[0].value);
+    try testing.expectEqual(false, declarations[0].is_important);
+}
+
+test "Browser: CSS.Parser - Property with !important" {
+    defer testing.reset();
+    const text = "margin: 10px !important;";
+    const allocator = testing.arena_allocator;
+
+    const declarations = try CSSParser.parseDeclarations(allocator, text);
+
+    try testing.expectEqual(1, declarations.len);
+    try testing.expectEqual("margin", declarations[0].name);
+    try testing.expectEqual("10px", declarations[0].value);
+    try testing.expectEqual(true, declarations[0].is_important);
+}
+
+test "Browser: CSS.Parser - Multiple properties" {
+    defer testing.reset();
+    const text = "color: red; font-size: 12px; margin: 5px !important;";
+    const allocator = testing.arena_allocator;
+
+    const declarations = try CSSParser.parseDeclarations(allocator, text);
+
+    try testing.expect(declarations.len == 3);
+
+    try testing.expectEqual("color", declarations[0].name);
+    try testing.expectEqual("red", declarations[0].value);
+    try testing.expectEqual(false, declarations[0].is_important);
+
+    try testing.expectEqual("font-size", declarations[1].name);
+    try testing.expectEqual("12px", declarations[1].value);
+    try testing.expectEqual(false, declarations[1].is_important);
+
+    try testing.expectEqual("margin", declarations[2].name);
+    try testing.expectEqual("5px", declarations[2].value);
+    try testing.expectEqual(true, declarations[2].is_important);
+}
+
+test "Browser: CSS.Parser - Quoted value with semicolon" {
+    defer testing.reset();
+    const text = "content: \"Hello; world!\";";
+    const allocator = testing.arena_allocator;
+
+    const declarations = try CSSParser.parseDeclarations(allocator, text);
+
+    try testing.expectEqual(1, declarations.len);
+    try testing.expectEqual("content", declarations[0].name);
+    try testing.expectEqual("\"Hello; world!\"", declarations[0].value);
+    try testing.expectEqual(false, declarations[0].is_important);
+}
+
+test "Browser: CSS.Parser - URL value" {
+    defer testing.reset();
+    const text = "background-image: url(\"test.png\");";
+    const allocator = testing.arena_allocator;
+
+    const declarations = try CSSParser.parseDeclarations(allocator, text);
+
+    try testing.expectEqual(1, declarations.len);
+    try testing.expectEqual("background-image", declarations[0].name);
+    try testing.expectEqual("url(\"test.png\")", declarations[0].value);
+    try testing.expectEqual(false, declarations[0].is_important);
+}
+
+test "Browser: CSS.Parser - Whitespace handling" {
+    defer testing.reset();
+    const text = "  color  :  purple  ;  margin  :  10px  ;  ";
+    const allocator = testing.arena_allocator;
+
+    const declarations = try CSSParser.parseDeclarations(allocator, text);
+
+    try testing.expectEqual(2, declarations.len);
+    try testing.expectEqual("color", declarations[0].name);
+    try testing.expectEqual("purple", declarations[0].value);
+    try testing.expectEqual("margin", declarations[1].name);
+    try testing.expectEqual("10px", declarations[1].value);
+}

src/browser/cssom/CSSRule.zig

@@ -0,0 +1,41 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const CSSStyleSheet = @import("CSSStyleSheet.zig");
+
+pub const Interfaces = .{
+    CSSRule,
+    CSSImportRule,
+};
+
+// https://developer.mozilla.org/en-US/docs/Web/API/CSSRule
+const CSSRule = @This();
+css_text: []const u8,
+parent_rule: ?*CSSRule = null,
+parent_stylesheet: ?*CSSStyleSheet = null,
+
+pub const CSSImportRule = struct {
+    pub const prototype = *CSSRule;
+    href: []const u8,
+    layer_name: ?[]const u8,
+    media: void,
+    style_sheet: CSSStyleSheet,
+    supports_text: ?[]const u8,
+};

src/browser/cssom/CSSRuleList.zig

@@ -1,4 +1,4 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
 //
 // Francis Bouvier <francis@lightpanda.io>
 // Pierre Tachoire <pierre@lightpanda.io>
@@ -17,31 +17,35 @@
 // along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
 const std = @import("std");
-const lp = @import("lightpanda");
 
-pub fn main() !void {
-    const allocator = std.heap.c_allocator;
+const CSSRule = @import("CSSRule.zig");
 
-    var platform = try lp.js.Platform.init();
-    defer platform.deinit();
+const CSSImportRule = CSSRule.CSSImportRule;
 
-    const snapshot = try lp.js.Snapshot.create();
-    defer snapshot.deinit();
+const CSSRuleList = @This();
+list: std.ArrayListUnmanaged([]const u8),
 
-    var is_stdout = true;
-    var file = std.fs.File.stdout();
-    var args = try std.process.argsWithAllocator(allocator);
-    _ = args.next(); // executable name
-    if (args.next()) |n| {
-        is_stdout = false;
-        file = try std.fs.cwd().createFile(n, .{});
-    }
-    defer if (!is_stdout) {
-        file.close();
-    };
-
-    var buffer: [4096]u8 = undefined;
-    var writer = file.writer(&buffer);
-    try snapshot.write(&writer.interface);
-    try writer.end();
+pub fn constructor() CSSRuleList {
+    return .{ .list = .empty };
+}
+
+pub fn _item(self: *CSSRuleList, _index: u32) ?CSSRule {
+    const index: usize = @intCast(_index);
+
+    if (index > self.list.items.len) {
+        return null;
+    }
+
+    // todo: for now, just return null.
+    // this depends on properly parsing CSSRule
+    return null;
+}
+
+pub fn get_length(self: *CSSRuleList) u32 {
+    return @intCast(self.list.items.len);
+}
+
+const testing = @import("../../testing.zig");
+test "Browser: CSS.CSSRuleList" {
+    try testing.htmlRunner("cssom/css_rule_list.html");
 }

src/browser/cssom/CSSStyleDeclaration.zig

@@ -0,0 +1,958 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const Page = @import("../page.zig").Page;
+const CSSRule = @import("CSSRule.zig");
+const CSSParser = @import("CSSParser.zig");
+
+const Property = struct {
+    value: []const u8,
+    priority: bool,
+};
+
+const CSSStyleDeclaration = @This();
+
+properties: std.StringArrayHashMapUnmanaged(Property),
+
+pub const empty: CSSStyleDeclaration = .{
+    .properties = .empty,
+};
+
+pub fn get_cssFloat(self: *const CSSStyleDeclaration) []const u8 {
+    return self._getPropertyValue("float");
+}
+
+pub fn set_cssFloat(self: *CSSStyleDeclaration, value: ?[]const u8, page: *Page) !void {
+    const final_value = value orelse "";
+    return self._setProperty("float", final_value, null, page);
+}
+
+pub fn get_cssText(self: *const CSSStyleDeclaration, page: *Page) ![]const u8 {
+    var buffer: std.ArrayListUnmanaged(u8) = .empty;
+    const writer = buffer.writer(page.call_arena);
+    var it = self.properties.iterator();
+    while (it.next()) |entry| {
+        const name = entry.key_ptr.*;
+        const property = entry.value_ptr;
+        const escaped = try escapeCSSValue(page.call_arena, property.value);
+        try writer.print("{s}: {s}", .{ name, escaped });
+        if (property.priority) {
+            try writer.writeAll(" !important; ");
+        } else {
+            try writer.writeAll("; ");
+        }
+    }
+    return buffer.items;
+}
+
+// TODO Propagate also upward to parent node
+pub fn set_cssText(self: *CSSStyleDeclaration, text: []const u8, page: *Page) !void {
+    self.properties.clearRetainingCapacity();
+
+    // call_arena is safe here, because _setProperty will dupe the name
+    // using the page's longer-living arena.
+    const declarations = try CSSParser.parseDeclarations(page.call_arena, text);
+
+    for (declarations) |decl| {
+        if (!isValidPropertyName(decl.name)) {
+            continue;
+        }
+        const priority: ?[]const u8 = if (decl.is_important) "important" else null;
+        try self._setProperty(decl.name, decl.value, priority, page);
+    }
+}
+
+pub fn get_length(self: *const CSSStyleDeclaration) usize {
+    return self.properties.count();
+}
+
+pub fn get_parentRule(_: *const CSSStyleDeclaration) ?CSSRule {
+    return null;
+}
+
+pub fn _getPropertyPriority(self: *const CSSStyleDeclaration, name: []const u8) []const u8 {
+    const property = self.properties.getPtr(name) orelse return "";
+    return if (property.priority) "important" else "";
+}
+
+// TODO should handle properly shorthand properties and canonical forms
+pub fn _getPropertyValue(self: *const CSSStyleDeclaration, name: []const u8) []const u8 {
+    if (self.properties.getPtr(name)) |property| {
+        return property.value;
+    }
+
+    // default to everything being visible (unless it's been explicitly set)
+    if (std.mem.eql(u8, name, "visibility")) {
+        return "visible";
+    }
+
+    return "";
+}
+
+pub fn _item(self: *const CSSStyleDeclaration, index: usize) []const u8 {
+    const values = self.properties.entries.items(.key);
+    if (index >= values.len) {
+        return "";
+    }
+    return values[index];
+}
+
+pub fn _removeProperty(self: *CSSStyleDeclaration, name: []const u8) ![]const u8 {
+    const property = self.properties.fetchOrderedRemove(name) orelse return "";
+    return property.value.value;
+}
+
+pub fn _setProperty(self: *CSSStyleDeclaration, name: []const u8, value: []const u8, priority: ?[]const u8, page: *Page) !void {
+    const gop = try self.properties.getOrPut(page.arena, name);
+    if (!gop.found_existing) {
+        const owned_name = try page.arena.dupe(u8, name);
+        gop.key_ptr.* = owned_name;
+    }
+
+    const owned_value = try page.arena.dupe(u8, value);
+    const is_important = priority != null and std.ascii.eqlIgnoreCase(priority.?, "important");
+    gop.value_ptr.* = .{ .value = owned_value, .priority = is_important };
+}
+
+pub fn named_get(self: *const CSSStyleDeclaration, name: []const u8, _: *bool) []const u8 {
+    return self._getPropertyValue(name);
+}
+
+pub fn named_set(self: *CSSStyleDeclaration, name: []const u8, value: []const u8, _: *bool, page: *Page) !void {
+    return self._setProperty(name, value, null, page);
+}
+
+fn isNumericWithUnit(value: []const u8) bool {
+    if (value.len == 0) {
+        return false;
+    }
+
+    const first = value[0];
+
+    if (!std.ascii.isDigit(first) and first != '+' and first != '-' and first != '.') {
+        return false;
+    }
+
+    var i: usize = 0;
+    var has_digit = false;
+    var decimal_point = false;
+
+    while (i < value.len) : (i += 1) {
+        const c = value[i];
+        if (std.ascii.isDigit(c)) {
+            has_digit = true;
+        } else if (c == '.' and !decimal_point) {
+            decimal_point = true;
+        } else if ((c == 'e' or c == 'E') and has_digit) {
+            if (i + 1 >= value.len) return false;
+            if (value[i + 1] != '+' and value[i + 1] != '-' and !std.ascii.isDigit(value[i + 1])) break;
+            i += 1;
+            if (value[i] == '+' or value[i] == '-') {
+                i += 1;
+            }
+            var has_exp_digits = false;
+            while (i < value.len and std.ascii.isDigit(value[i])) : (i += 1) {
+                has_exp_digits = true;
+            }
+            if (!has_exp_digits) return false;
+            break;
+        } else if (c != '-' and c != '+') {
+            break;
+        }
+    }
+
+    if (!has_digit) {
+        return false;
+    }
+
+    if (i == value.len) {
+        return true;
+    }
+
+    const unit = value[i..];
+    return CSSKeywords.isValidUnit(unit);
+}
+
+pub fn isHexColor(value: []const u8) bool {
+    if (value.len == 0) {
+        return false;
+    }
+    if (value[0] != '#') {
+        return false;
+    }
+
+    const hex_part = value[1..];
+    if (hex_part.len != 3 and hex_part.len != 4 and hex_part.len != 6 and hex_part.len != 8) {
+        return false;
+    }
+
+    for (hex_part) |c| {
+        if (!std.ascii.isHex(c)) {
+            return false;
+        }
+    }
+
+    return true;
+}
+
+fn isMultiValueProperty(value: []const u8) bool {
+    var parts = std.mem.splitAny(u8, value, " ");
+    var multi_value_parts: usize = 0;
+    var all_parts_valid = true;
+
+    while (parts.next()) |part| {
+        if (part.len == 0) continue;
+        multi_value_parts += 1;
+
+        if (isNumericWithUnit(part)) {
+            continue;
+        }
+        if (isHexColor(part)) {
+            continue;
+        }
+        if (CSSKeywords.isKnownKeyword(part)) {
+            continue;
+        }
+        if (CSSKeywords.startsWithFunction(part)) {
+            continue;
+        }
+
+        all_parts_valid = false;
+        break;
+    }
+
+    return multi_value_parts >= 2 and all_parts_valid;
+}
+
+fn isAlreadyQuoted(value: []const u8) bool {
+    return value.len >= 2 and ((value[0] == '"' and value[value.len - 1] == '"') or
+        (value[0] == '\'' and value[value.len - 1] == '\''));
+}
+
+fn isValidPropertyName(name: []const u8) bool {
+    if (name.len == 0) return false;
+
+    if (std.mem.startsWith(u8, name, "--")) {
+        if (name.len == 2) return false;
+        for (name[2..]) |c| {
+            if (!std.ascii.isAlphanumeric(c) and c != '-' and c != '_') {
+                return false;
+            }
+        }
+        return true;
+    }
+
+    const first_char = name[0];
+    if (!std.ascii.isAlphabetic(first_char) and first_char != '-') {
+        return false;
+    }
+
+    if (first_char == '-') {
+        if (name.len < 2) return false;
+
+        if (!std.ascii.isAlphabetic(name[1])) {
+            return false;
+        }
+
+        for (name[2..]) |c| {
+            if (!std.ascii.isAlphanumeric(c) and c != '-') {
+                return false;
+            }
+        }
+    } else {
+        for (name[1..]) |c| {
+            if (!std.ascii.isAlphanumeric(c) and c != '-') {
+                return false;
+            }
+        }
+    }
+
+    return true;
+}
+
+fn extractImportant(value: []const u8) struct { value: []const u8, is_important: bool } {
+    const trimmed = std.mem.trim(u8, value, &std.ascii.whitespace);
+
+    if (std.mem.endsWith(u8, trimmed, "!important")) {
+        const clean_value = std.mem.trimRight(u8, trimmed[0 .. trimmed.len - 10], &std.ascii.whitespace);
+        return .{ .value = clean_value, .is_important = true };
+    }
+
+    return .{ .value = trimmed, .is_important = false };
+}
+
+fn needsQuotes(value: []const u8) bool {
+    if (value.len == 0) return true;
+    if (isAlreadyQuoted(value)) return false;
+
+    if (CSSKeywords.containsSpecialChar(value)) {
+        return true;
+    }
+
+    if (std.mem.indexOfScalar(u8, value, ' ') == null) {
+        return false;
+    }
+
+    const is_url = std.mem.startsWith(u8, value, "url(");
+    const is_function = CSSKeywords.startsWithFunction(value);
+
+    return !isMultiValueProperty(value) and
+        !is_url and
+        !is_function;
+}
+
+fn escapeCSSValue(arena: std.mem.Allocator, value: []const u8) ![]const u8 {
+    if (!needsQuotes(value)) {
+        return value;
+    }
+    var out: std.ArrayListUnmanaged(u8) = .empty;
+
+    // We'll need at least this much space, +2 for the quotes
+    try out.ensureTotalCapacity(arena, value.len + 2);
+    const writer = out.writer(arena);
+
+    try writer.writeByte('"');
+
+    for (value, 0..) |c, i| {
+        switch (c) {
+            '"' => try writer.writeAll("\\\""),
+            '\\' => try writer.writeAll("\\\\"),
+            '\n' => try writer.writeAll("\\A "),
+            '\r' => try writer.writeAll("\\D "),
+            '\t' => try writer.writeAll("\\9 "),
+            0...8, 11, 12, 14...31, 127 => {
+                try writer.print("\\{x}", .{c});
+                if (i + 1 < value.len and std.ascii.isHex(value[i + 1])) {
+                    try writer.writeByte(' ');
+                }
+            },
+            else => try writer.writeByte(c),
+        }
+    }
+
+    try writer.writeByte('"');
+    return out.items;
+}
+
+fn isKnownKeyword(value: []const u8) bool {
+    return CSSKeywords.isKnownKeyword(value);
+}
+
+fn containsSpecialChar(value: []const u8) bool {
+    return CSSKeywords.containsSpecialChar(value);
+}
+
+const CSSKeywords = struct {
+    const BORDER_STYLES = [_][]const u8{
+        "none", "solid", "dotted", "dashed", "double", "groove", "ridge", "inset", "outset",
+    };
+
+    const COLOR_NAMES = [_][]const u8{
+        "black",        "white",   "red", "green", "blue", "yellow", "purple", "gray", "transparent",
+        "currentColor", "inherit",
+    };
+
+    const POSITION_KEYWORDS = [_][]const u8{
+        "auto", "center", "left", "right", "top", "bottom",
+    };
+
+    const BACKGROUND_REPEAT = [_][]const u8{
+        "repeat", "no-repeat", "repeat-x", "repeat-y", "space", "round",
+    };
+
+    const FONT_STYLES = [_][]const u8{
+        "normal", "italic", "oblique", "bold", "bolder", "lighter",
+    };
+
+    const FONT_SIZES = [_][]const u8{
+        "xx-small", "x-small", "small", "medium", "large", "x-large", "xx-large",
+        "smaller",  "larger",
+    };
+
+    const FONT_FAMILIES = [_][]const u8{
+        "serif", "sans-serif", "monospace", "cursive", "fantasy", "system-ui",
+    };
+
+    const CSS_GLOBAL = [_][]const u8{
+        "initial", "inherit", "unset", "revert",
+    };
+
+    const DISPLAY_VALUES = [_][]const u8{
+        "block", "inline", "inline-block", "flex", "grid", "none",
+    };
+
+    const UNITS = [_][]const u8{
+        // LENGTH
+        "px",   "em", "rem", "vw",  "vh",  "vmin", "vmax", "%", "pt", "pc", "in",  "cm",  "mm",
+        "ex",   "ch", "fr",
+
+        // ANGLE
+         "deg", "rad", "grad", "turn",
+
+        // TIME
+        "s", "ms",
+
+        // FREQUENCY
+        "hz", "khz",
+
+        // RESOLUTION
+        "dpi", "dpcm",
+        "dppx",
+    };
+
+    const SPECIAL_CHARS = [_]u8{
+        '"', '\'', ';', '{', '}', '\\', '<', '>', '/', '\n', '\t', '\r', '\x00', '\x7F',
+    };
+
+    const FUNCTIONS = [_][]const u8{
+        "rgb(",             "rgba(",            "hsl(",            "hsla(",      "url(",    "calc(",  "var(",  "attr(",
+        "linear-gradient(", "radial-gradient(", "conic-gradient(", "translate(", "rotate(", "scale(", "skew(", "matrix(",
+    };
+
+    const KEYWORDS = BORDER_STYLES ++ COLOR_NAMES ++ POSITION_KEYWORDS ++
+        BACKGROUND_REPEAT ++ FONT_STYLES ++ FONT_SIZES ++ FONT_FAMILIES ++
+        CSS_GLOBAL ++ DISPLAY_VALUES;
+
+    const MAX_KEYWORD_LEN = lengthOfLongestValue(&KEYWORDS);
+
+    pub fn isKnownKeyword(value: []const u8) bool {
+        if (value.len > MAX_KEYWORD_LEN) {
+            return false;
+        }
+        var buf: [MAX_KEYWORD_LEN]u8 = undefined;
+        const normalized = std.ascii.lowerString(&buf, value);
+
+        for (KEYWORDS) |keyword| {
+            if (std.ascii.eqlIgnoreCase(normalized, keyword)) {
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+    pub fn containsSpecialChar(value: []const u8) bool {
+        return std.mem.indexOfAny(u8, value, &SPECIAL_CHARS) != null;
+    }
+
+    const MAX_UNIT_LEN = lengthOfLongestValue(&UNITS);
+
+    pub fn isValidUnit(unit: []const u8) bool {
+        if (unit.len > MAX_UNIT_LEN) {
+            return false;
+        }
+        var buf: [MAX_UNIT_LEN]u8 = undefined;
+        const normalized = std.ascii.lowerString(&buf, unit);
+
+        for (UNITS) |u| {
+            if (std.mem.eql(u8, normalized, u)) {
+                return true;
+            }
+        }
+        return false;
+    }
+
+    pub fn startsWithFunction(value: []const u8) bool {
+        const pos = std.mem.indexOfScalar(u8, value, '(') orelse return false;
+        if (pos == 0) return false;
+
+        if (std.mem.indexOfScalarPos(u8, value, pos, ')') == null) {
+            return false;
+        }
+        const function_name = value[0..pos];
+        return isValidFunctionName(function_name);
+    }
+
+    fn isValidFunctionName(name: []const u8) bool {
+        if (name.len == 0) return false;
+
+        const first = name[0];
+        if (!std.ascii.isAlphabetic(first) and first != '_' and first != '-') {
+            return false;
+        }
+
+        for (name[1..]) |c| {
+            if (!std.ascii.isAlphanumeric(c) and c != '_' and c != '-') {
+                return false;
+            }
+        }
+
+        return true;
+    }
+};
+
+fn lengthOfLongestValue(values: []const []const u8) usize {
+    var max: usize = 0;
+    for (values) |v| {
+        max = @max(v.len, max);
+    }
+    return max;
+}
+
+const testing = @import("../../testing.zig");
+test "Browser: CSS.StyleDeclaration" {
+    try testing.htmlRunner("cssom/css_style_declaration.html");
+}
+
+test "Browser: CSS.StyleDeclaration: isNumericWithUnit - valid numbers with units" {
+    try testing.expect(isNumericWithUnit("10px"));
+    try testing.expect(isNumericWithUnit("3.14em"));
+    try testing.expect(isNumericWithUnit("-5rem"));
+    try testing.expect(isNumericWithUnit("+12.5%"));
+    try testing.expect(isNumericWithUnit("0vh"));
+    try testing.expect(isNumericWithUnit(".5vw"));
+}
+
+test "Browser: CSS.StyleDeclaration: isNumericWithUnit - scientific notation" {
+    try testing.expect(isNumericWithUnit("1e5px"));
+    try testing.expect(isNumericWithUnit("2.5E-3em"));
+    try testing.expect(isNumericWithUnit("1e+2rem"));
+    try testing.expect(isNumericWithUnit("-3.14e10px"));
+}
+
+test "Browser: CSS.StyleDeclaration: isNumericWithUnit - edge cases and invalid inputs" {
+    try testing.expect(!isNumericWithUnit(""));
+
+    try testing.expect(!isNumericWithUnit("px"));
+    try testing.expect(!isNumericWithUnit("--px"));
+    try testing.expect(!isNumericWithUnit(".px"));
+
+    try testing.expect(!isNumericWithUnit("1e"));
+    try testing.expect(!isNumericWithUnit("1epx"));
+    try testing.expect(!isNumericWithUnit("1e+"));
+    try testing.expect(!isNumericWithUnit("1e+px"));
+
+    try testing.expect(!isNumericWithUnit("1.2.3px"));
+
+    try testing.expect(!isNumericWithUnit("10xyz"));
+    try testing.expect(!isNumericWithUnit("5invalid"));
+
+    try testing.expect(isNumericWithUnit("10"));
+    try testing.expect(isNumericWithUnit("3.14"));
+    try testing.expect(isNumericWithUnit("-5"));
+}
+
+test "Browser: CSS.StyleDeclaration: isHexColor - valid hex colors" {
+    try testing.expect(isHexColor("#000"));
+    try testing.expect(isHexColor("#0000"));
+    try testing.expect(isHexColor("#fff"));
+    try testing.expect(isHexColor("#123456"));
+    try testing.expect(isHexColor("#abcdef"));
+    try testing.expect(isHexColor("#ABCDEF"));
+    try testing.expect(isHexColor("#12345678"));
+}
+
+test "Browser: CSS.StyleDeclaration: isHexColor - invalid hex colors" {
+    try testing.expect(!isHexColor(""));
+    try testing.expect(!isHexColor("#"));
+    try testing.expect(!isHexColor("000"));
+    try testing.expect(!isHexColor("#00"));
+    try testing.expect(!isHexColor("#00000"));
+    try testing.expect(!isHexColor("#0000000"));
+    try testing.expect(!isHexColor("#000000000"));
+    try testing.expect(!isHexColor("#gggggg"));
+    try testing.expect(!isHexColor("#123xyz"));
+}
+
+test "Browser: CSS.StyleDeclaration: isMultiValueProperty - valid multi-value properties" {
+    try testing.expect(isMultiValueProperty("10px 20px"));
+    try testing.expect(isMultiValueProperty("solid red"));
+    try testing.expect(isMultiValueProperty("#fff black"));
+    try testing.expect(isMultiValueProperty("1em 2em 3em 4em"));
+    try testing.expect(isMultiValueProperty("rgb(255,0,0) solid"));
+}
+
+test "Browser: CSS.StyleDeclaration: isMultiValueProperty - invalid multi-value properties" {
+    try testing.expect(!isMultiValueProperty(""));
+    try testing.expect(!isMultiValueProperty("10px"));
+    try testing.expect(!isMultiValueProperty("invalid unknown"));
+    try testing.expect(!isMultiValueProperty("10px invalid"));
+    try testing.expect(!isMultiValueProperty("   "));
+}
+
+test "Browser: CSS.StyleDeclaration: isAlreadyQuoted - various quoting scenarios" {
+    try testing.expect(isAlreadyQuoted("\"hello\""));
+    try testing.expect(isAlreadyQuoted("'world'"));
+    try testing.expect(isAlreadyQuoted("\"\""));
+    try testing.expect(isAlreadyQuoted("''"));
+
+    try testing.expect(!isAlreadyQuoted(""));
+    try testing.expect(!isAlreadyQuoted("hello"));
+    try testing.expect(!isAlreadyQuoted("\""));
+    try testing.expect(!isAlreadyQuoted("'"));
+    try testing.expect(!isAlreadyQuoted("\"hello'"));
+    try testing.expect(!isAlreadyQuoted("'hello\""));
+    try testing.expect(!isAlreadyQuoted("\"hello"));
+    try testing.expect(!isAlreadyQuoted("hello\""));
+}
+
+test "Browser: CSS.StyleDeclaration: isValidPropertyName - valid property names" {
+    try testing.expect(isValidPropertyName("color"));
+    try testing.expect(isValidPropertyName("background-color"));
+    try testing.expect(isValidPropertyName("-webkit-transform"));
+    try testing.expect(isValidPropertyName("font-size"));
+    try testing.expect(isValidPropertyName("margin-top"));
+    try testing.expect(isValidPropertyName("z-index"));
+    try testing.expect(isValidPropertyName("line-height"));
+}
+
+test "Browser: CSS.StyleDeclaration: isValidPropertyName - invalid property names" {
+    try testing.expect(!isValidPropertyName(""));
+    try testing.expect(!isValidPropertyName("123color"));
+    try testing.expect(!isValidPropertyName("color!"));
+    try testing.expect(!isValidPropertyName("color space"));
+    try testing.expect(!isValidPropertyName("@color"));
+    try testing.expect(!isValidPropertyName("color.test"));
+    try testing.expect(!isValidPropertyName("color_test"));
+}
+
+test "Browser: CSS.StyleDeclaration: extractImportant - with and without !important" {
+    var result = extractImportant("red !important");
+    try testing.expect(result.is_important);
+    try testing.expectEqual("red", result.value);
+
+    result = extractImportant("blue");
+    try testing.expect(!result.is_important);
+    try testing.expectEqual("blue", result.value);
+
+    result = extractImportant("  green  !important  ");
+    try testing.expect(result.is_important);
+    try testing.expectEqual("green", result.value);
+
+    result = extractImportant("!important");
+    try testing.expect(result.is_important);
+    try testing.expectEqual("", result.value);
+
+    result = extractImportant("important");
+    try testing.expect(!result.is_important);
+    try testing.expectEqual("important", result.value);
+}
+
+test "Browser: CSS.StyleDeclaration: needsQuotes - various scenarios" {
+    try testing.expect(needsQuotes(""));
+    try testing.expect(needsQuotes("hello world"));
+    try testing.expect(needsQuotes("test;"));
+    try testing.expect(needsQuotes("a{b}"));
+    try testing.expect(needsQuotes("test\"quote"));
+
+    try testing.expect(!needsQuotes("\"already quoted\""));
+    try testing.expect(!needsQuotes("'already quoted'"));
+    try testing.expect(!needsQuotes("url(image.png)"));
+    try testing.expect(!needsQuotes("rgb(255, 0, 0)"));
+    try testing.expect(!needsQuotes("10px 20px"));
+    try testing.expect(!needsQuotes("simple"));
+}
+
+test "Browser: CSS.StyleDeclaration: escapeCSSValue - escaping various characters" {
+    const allocator = testing.arena_allocator;
+
+    var result = try escapeCSSValue(allocator, "simple");
+    try testing.expectEqual("simple", result);
+
+    result = try escapeCSSValue(allocator, "\"already quoted\"");
+    try testing.expectEqual("\"already quoted\"", result);
+
+    result = try escapeCSSValue(allocator, "test\"quote");
+    try testing.expectEqual("\"test\\\"quote\"", result);
+
+    result = try escapeCSSValue(allocator, "test\nline");
+    try testing.expectEqual("\"test\\A line\"", result);
+
+    result = try escapeCSSValue(allocator, "test\\back");
+    try testing.expectEqual("\"test\\\\back\"", result);
+}
+
+test "Browser: CSS.StyleDeclaration: CSSKeywords.isKnownKeyword - case sensitivity" {
+    try testing.expect(CSSKeywords.isKnownKeyword("red"));
+    try testing.expect(CSSKeywords.isKnownKeyword("solid"));
+    try testing.expect(CSSKeywords.isKnownKeyword("center"));
+    try testing.expect(CSSKeywords.isKnownKeyword("inherit"));
+
+    try testing.expect(CSSKeywords.isKnownKeyword("RED"));
+    try testing.expect(CSSKeywords.isKnownKeyword("Red"));
+    try testing.expect(CSSKeywords.isKnownKeyword("SOLID"));
+    try testing.expect(CSSKeywords.isKnownKeyword("Center"));
+
+    try testing.expect(!CSSKeywords.isKnownKeyword("invalid"));
+    try testing.expect(!CSSKeywords.isKnownKeyword("unknown"));
+    try testing.expect(!CSSKeywords.isKnownKeyword(""));
+}
+
+test "Browser: CSS.StyleDeclaration: CSSKeywords.containsSpecialChar - various special characters" {
+    try testing.expect(CSSKeywords.containsSpecialChar("test\"quote"));
+    try testing.expect(CSSKeywords.containsSpecialChar("test'quote"));
+    try testing.expect(CSSKeywords.containsSpecialChar("test;end"));
+    try testing.expect(CSSKeywords.containsSpecialChar("test{brace"));
+    try testing.expect(CSSKeywords.containsSpecialChar("test}brace"));
+    try testing.expect(CSSKeywords.containsSpecialChar("test\\back"));
+    try testing.expect(CSSKeywords.containsSpecialChar("test<angle"));
+    try testing.expect(CSSKeywords.containsSpecialChar("test>angle"));
+    try testing.expect(CSSKeywords.containsSpecialChar("test/slash"));
+
+    try testing.expect(!CSSKeywords.containsSpecialChar("normal-text"));
+    try testing.expect(!CSSKeywords.containsSpecialChar("text123"));
+    try testing.expect(!CSSKeywords.containsSpecialChar(""));
+}
+
+test "Browser: CSS.StyleDeclaration: CSSKeywords.isValidUnit - various units" {
+    try testing.expect(CSSKeywords.isValidUnit("px"));
+    try testing.expect(CSSKeywords.isValidUnit("em"));
+    try testing.expect(CSSKeywords.isValidUnit("rem"));
+    try testing.expect(CSSKeywords.isValidUnit("%"));
+
+    try testing.expect(CSSKeywords.isValidUnit("deg"));
+    try testing.expect(CSSKeywords.isValidUnit("rad"));
+
+    try testing.expect(CSSKeywords.isValidUnit("s"));
+    try testing.expect(CSSKeywords.isValidUnit("ms"));
+
+    try testing.expect(CSSKeywords.isValidUnit("PX"));
+
+    try testing.expect(!CSSKeywords.isValidUnit("invalid"));
+    try testing.expect(!CSSKeywords.isValidUnit(""));
+}
+
+test "Browser: CSS.StyleDeclaration: CSSKeywords.startsWithFunction - function detection" {
+    try testing.expect(CSSKeywords.startsWithFunction("rgb(255, 0, 0)"));
+    try testing.expect(CSSKeywords.startsWithFunction("rgba(255, 0, 0, 0.5)"));
+    try testing.expect(CSSKeywords.startsWithFunction("url(image.png)"));
+    try testing.expect(CSSKeywords.startsWithFunction("calc(100% - 20px)"));
+    try testing.expect(CSSKeywords.startsWithFunction("var(--custom-property)"));
+    try testing.expect(CSSKeywords.startsWithFunction("linear-gradient(to right, red, blue)"));
+
+    try testing.expect(CSSKeywords.startsWithFunction("custom-function(args)"));
+    try testing.expect(CSSKeywords.startsWithFunction("unknown(test)"));
+
+    try testing.expect(!CSSKeywords.startsWithFunction("not-a-function"));
+    try testing.expect(!CSSKeywords.startsWithFunction("missing-paren)"));
+    try testing.expect(!CSSKeywords.startsWithFunction("missing-close("));
+    try testing.expect(!CSSKeywords.startsWithFunction(""));
+    try testing.expect(!CSSKeywords.startsWithFunction("rgb"));
+}
+
+test "Browser: CSS.StyleDeclaration: isNumericWithUnit - whitespace handling" {
+    try testing.expect(!isNumericWithUnit(" 10px"));
+    try testing.expect(!isNumericWithUnit("10 px"));
+    try testing.expect(!isNumericWithUnit("10px "));
+    try testing.expect(!isNumericWithUnit(" 10 px "));
+}
+
+test "Browser: CSS.StyleDeclaration: extractImportant - whitespace edge cases" {
+    var result = extractImportant("   ");
+    try testing.expect(!result.is_important);
+    try testing.expectEqual("", result.value);
+
+    result = extractImportant("\t\n\r !important\t\n");
+    try testing.expect(result.is_important);
+    try testing.expectEqual("", result.value);
+
+    result = extractImportant("red\t!important");
+    try testing.expect(result.is_important);
+    try testing.expectEqual("red", result.value);
+}
+
+test "Browser: CSS.StyleDeclaration: isHexColor - mixed case handling" {
+    try testing.expect(isHexColor("#AbC"));
+    try testing.expect(isHexColor("#123aBc"));
+    try testing.expect(isHexColor("#FFffFF"));
+    try testing.expect(isHexColor("#000FFF"));
+}
+
+test "Browser: CSS.StyleDeclaration: edge case - very long inputs" {
+    const long_valid = "a" ** 1000 ++ "px";
+    try testing.expect(!isNumericWithUnit(long_valid)); // not numeric
+
+    const long_property = "a-" ** 100 ++ "property";
+    try testing.expect(isValidPropertyName(long_property));
+
+    const long_hex = "#" ++ "a" ** 20;
+    try testing.expect(!isHexColor(long_hex));
+}
+
+test "Browser: CSS.StyleDeclaration: boundary conditions - numeric parsing" {
+    try testing.expect(isNumericWithUnit("0px"));
+    try testing.expect(isNumericWithUnit("0.0px"));
+    try testing.expect(isNumericWithUnit(".0px"));
+    try testing.expect(isNumericWithUnit("0.px"));
+
+    try testing.expect(isNumericWithUnit("999999999px"));
+    try testing.expect(isNumericWithUnit("1.7976931348623157e+308px"));
+
+    try testing.expect(isNumericWithUnit("0.000000001px"));
+    try testing.expect(isNumericWithUnit("1e-100px"));
+}
+
+test "Browser: CSS.StyleDeclaration: extractImportant - malformed important declarations" {
+    var result = extractImportant("red ! important");
+    try testing.expect(!result.is_important);
+    try testing.expectEqual("red ! important", result.value);
+
+    result = extractImportant("red !Important");
+    try testing.expect(!result.is_important);
+    try testing.expectEqual("red !Important", result.value);
+
+    result = extractImportant("red !IMPORTANT");
+    try testing.expect(!result.is_important);
+    try testing.expectEqual("red !IMPORTANT", result.value);
+
+    result = extractImportant("!importantred");
+    try testing.expect(!result.is_important);
+    try testing.expectEqual("!importantred", result.value);
+
+    result = extractImportant("red !important !important");
+    try testing.expect(result.is_important);
+    try testing.expectEqual("red !important", result.value);
+}
+
+test "Browser: CSS.StyleDeclaration: isMultiValueProperty - complex spacing scenarios" {
+    try testing.expect(isMultiValueProperty("10px    20px"));
+    try testing.expect(isMultiValueProperty("solid     red"));
+
+    try testing.expect(isMultiValueProperty("  10px 20px  "));
+
+    try testing.expect(!isMultiValueProperty("10px\t20px"));
+    try testing.expect(!isMultiValueProperty("10px\n20px"));
+
+    try testing.expect(isMultiValueProperty("10px   20px   30px"));
+}
+
+test "Browser: CSS.StyleDeclaration: isAlreadyQuoted - edge cases with quotes" {
+    try testing.expect(isAlreadyQuoted("\"'hello'\""));
+    try testing.expect(isAlreadyQuoted("'\"hello\"'"));
+
+    try testing.expect(isAlreadyQuoted("\"hello\\\"world\""));
+    try testing.expect(isAlreadyQuoted("'hello\\'world'"));
+
+    try testing.expect(!isAlreadyQuoted("\"hello"));
+    try testing.expect(!isAlreadyQuoted("hello\""));
+    try testing.expect(!isAlreadyQuoted("'hello"));
+    try testing.expect(!isAlreadyQuoted("hello'"));
+
+    try testing.expect(isAlreadyQuoted("\"a\""));
+    try testing.expect(isAlreadyQuoted("'b'"));
+}
+
+test "Browser: CSS.StyleDeclaration: needsQuotes - function and URL edge cases" {
+    try testing.expect(!needsQuotes("rgb(255, 0, 0)"));
+    try testing.expect(!needsQuotes("calc(100% - 20px)"));
+
+    try testing.expect(!needsQuotes("url(path with spaces.jpg)"));
+
+    try testing.expect(!needsQuotes("linear-gradient(to right, red, blue)"));
+
+    try testing.expect(needsQuotes("rgb(255, 0, 0"));
+}
+
+test "Browser: CSS.StyleDeclaration: escapeCSSValue - control characters and Unicode" {
+    const allocator = testing.arena_allocator;
+
+    var result = try escapeCSSValue(allocator, "test\ttab");
+    try testing.expectEqual("\"test\\9 tab\"", result);
+
+    result = try escapeCSSValue(allocator, "test\rreturn");
+    try testing.expectEqual("\"test\\D return\"", result);
+
+    result = try escapeCSSValue(allocator, "test\x00null");
+    try testing.expectEqual("\"test\\0null\"", result);
+
+    result = try escapeCSSValue(allocator, "test\x7Fdel");
+    try testing.expectEqual("\"test\\7f del\"", result);
+
+    result = try escapeCSSValue(allocator, "test\"quote\nline\\back");
+    try testing.expectEqual("\"test\\\"quote\\A line\\\\back\"", result);
+}
+
+test "Browser: CSS.StyleDeclaration: isValidPropertyName - CSS custom properties and vendor prefixes" {
+    try testing.expect(isValidPropertyName("--custom-color"));
+    try testing.expect(isValidPropertyName("--my-variable"));
+    try testing.expect(isValidPropertyName("--123"));
+
+    try testing.expect(isValidPropertyName("-webkit-transform"));
+    try testing.expect(isValidPropertyName("-moz-border-radius"));
+    try testing.expect(isValidPropertyName("-ms-filter"));
+    try testing.expect(isValidPropertyName("-o-transition"));
+
+    try testing.expect(!isValidPropertyName("-123invalid"));
+    try testing.expect(!isValidPropertyName("--"));
+    try testing.expect(!isValidPropertyName("-"));
+}
+
+test "Browser: CSS.StyleDeclaration: startsWithFunction - case sensitivity and partial matches" {
+    try testing.expect(CSSKeywords.startsWithFunction("RGB(255, 0, 0)"));
+    try testing.expect(CSSKeywords.startsWithFunction("Rgb(255, 0, 0)"));
+    try testing.expect(CSSKeywords.startsWithFunction("URL(image.png)"));
+
+    try testing.expect(CSSKeywords.startsWithFunction("rg(something)"));
+    try testing.expect(CSSKeywords.startsWithFunction("ur(something)"));
+
+    try testing.expect(CSSKeywords.startsWithFunction("rgb(1,2,3)"));
+    try testing.expect(CSSKeywords.startsWithFunction("rgba(1,2,3,4)"));
+
+    try testing.expect(CSSKeywords.startsWithFunction("my-custom-function(args)"));
+    try testing.expect(CSSKeywords.startsWithFunction("function-with-dashes(test)"));
+
+    try testing.expect(!CSSKeywords.startsWithFunction("123function(test)"));
+}
+
+test "Browser: CSS.StyleDeclaration: isHexColor - Unicode and invalid characters" {
+    try testing.expect(!isHexColor("#ghijkl"));
+    try testing.expect(!isHexColor("#12345g"));
+    try testing.expect(!isHexColor("#xyz"));
+
+    try testing.expect(!isHexColor("#АВС"));
+
+    try testing.expect(!isHexColor("#1234567g"));
+    try testing.expect(!isHexColor("#g2345678"));
+}
+
+test "Browser: CSS.StyleDeclaration: complex integration scenarios" {
+    const allocator = testing.arena_allocator;
+
+    try testing.expect(isMultiValueProperty("rgb(255,0,0) url(bg.jpg)"));
+
+    try testing.expect(!needsQuotes("calc(100% - 20px)"));
+
+    const result = try escapeCSSValue(allocator, "fake(function with spaces");
+    try testing.expectEqual("\"fake(function with spaces\"", result);
+
+    const important_result = extractImportant("rgb(255,0,0) !important");
+    try testing.expect(important_result.is_important);
+    try testing.expectEqual("rgb(255,0,0)", important_result.value);
+}
+
+test "Browser: CSS.StyleDeclaration: performance edge cases - empty and minimal inputs" {
+    try testing.expect(!isNumericWithUnit(""));
+    try testing.expect(!isHexColor(""));
+    try testing.expect(!isMultiValueProperty(""));
+    try testing.expect(!isAlreadyQuoted(""));
+    try testing.expect(!isValidPropertyName(""));
+    try testing.expect(needsQuotes(""));
+    try testing.expect(!CSSKeywords.isKnownKeyword(""));
+    try testing.expect(!CSSKeywords.containsSpecialChar(""));
+    try testing.expect(!CSSKeywords.isValidUnit(""));
+    try testing.expect(!CSSKeywords.startsWithFunction(""));
+
+    try testing.expect(!isNumericWithUnit("a"));
+    try testing.expect(!isHexColor("a"));
+    try testing.expect(!isMultiValueProperty("a"));
+    try testing.expect(!isAlreadyQuoted("a"));
+    try testing.expect(isValidPropertyName("a"));
+    try testing.expect(!needsQuotes("a"));
+}

src/browser/cssom/CSSStyleSheet.zig

@@ -0,0 +1,95 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const js = @import("../js/js.zig");
+const Page = @import("../page.zig").Page;
+const StyleSheet = @import("StyleSheet.zig");
+const CSSRuleList = @import("CSSRuleList.zig");
+const CSSImportRule = @import("CSSRule.zig").CSSImportRule;
+
+const CSSStyleSheet = @This();
+pub const prototype = *StyleSheet;
+
+proto: StyleSheet,
+css_rules: CSSRuleList,
+owner_rule: ?*CSSImportRule,
+
+const CSSStyleSheetOpts = struct {
+    base_url: ?[]const u8 = null,
+    // TODO: Suupport media
+    disabled: bool = false,
+};
+
+pub fn constructor(_opts: ?CSSStyleSheetOpts) !CSSStyleSheet {
+    const opts = _opts orelse CSSStyleSheetOpts{};
+    return .{
+        .proto = .{ .disabled = opts.disabled },
+        .css_rules = .constructor(),
+        .owner_rule = null,
+    };
+}
+
+pub fn get_ownerRule(_: *CSSStyleSheet) ?*CSSImportRule {
+    return null;
+}
+
+pub fn get_cssRules(self: *CSSStyleSheet) *CSSRuleList {
+    return &self.css_rules;
+}
+
+pub fn _insertRule(self: *CSSStyleSheet, rule: []const u8, _index: ?usize, page: *Page) !usize {
+    const index = _index orelse 0;
+    if (index > self.css_rules.list.items.len) {
+        return error.IndexSize;
+    }
+
+    const arena = page.arena;
+    try self.css_rules.list.insert(arena, index, try arena.dupe(u8, rule));
+    return index;
+}
+
+pub fn _deleteRule(self: *CSSStyleSheet, index: usize) !void {
+    if (index > self.css_rules.list.items.len) {
+        return error.IndexSize;
+    }
+
+    _ = self.css_rules.list.orderedRemove(index);
+}
+
+pub fn _replace(self: *CSSStyleSheet, text: []const u8, page: *Page) !js.Promise {
+    _ = self;
+    _ = text;
+    // TODO: clear self.css_rules
+    // parse text and re-populate self.css_rules
+
+    return page.js.resolvePromise({});
+}
+
+pub fn _replaceSync(self: *CSSStyleSheet, text: []const u8) !void {
+    _ = self;
+    _ = text;
+    // TODO: clear self.css_rules
+    // parse text and re-populate self.css_rules
+}
+
+const testing = @import("../../testing.zig");
+test "Browser: CSS.StyleSheet" {
+    try testing.htmlRunner("cssom/css_stylesheet.html");
+}

src/browser/cssom/StyleSheet.zig

@@ -0,0 +1,55 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const parser = @import("../netsurf.zig");
+
+// https://developer.mozilla.org/en-US/docs/Web/API/StyleSheet#specifications
+const StyleSheet = @This();
+
+disabled: bool = false,
+href: []const u8 = "",
+owner_node: ?*parser.Node = null,
+parent_stylesheet: ?*StyleSheet = null,
+title: []const u8 = "",
+type: []const u8 = "text/css",
+
+pub fn get_disabled(self: *const StyleSheet) bool {
+    return self.disabled;
+}
+
+pub fn get_href(self: *const StyleSheet) []const u8 {
+    return self.href;
+}
+
+// TODO: media
+
+pub fn get_ownerNode(self: *const StyleSheet) ?*parser.Node {
+    return self.owner_node;
+}
+
+pub fn get_parentStyleSheet(self: *const StyleSheet) ?*StyleSheet {
+    return self.parent_stylesheet;
+}
+
+pub fn get_title(self: *const StyleSheet) []const u8 {
+    return self.title;
+}
+
+pub fn get_type(self: *const StyleSheet) []const u8 {
+    return self.type;
+}

src/browser/cssom/color.zig

@@ -19,23 +19,8 @@
 const std = @import("std");
 const Io = std.Io;
 
-pub fn isHexColor(value: []const u8) bool {
-    if (value.len == 0) {
-        return false;
-    }
-
-    if (value[0] != '#') {
-        return false;
-    }
-
-    const hex_part = value[1..];
-    switch (hex_part.len) {
-        3, 4, 6, 8 => for (hex_part) |c| if (!std.ascii.isHex(c)) return false,
-        else => return false,
-    }
-
-    return true;
-}
+const CSSParser = @import("CSSParser.zig");
+const isHexColor = @import("CSSStyleDeclaration.zig").isHexColor;
 
 pub const RGBA = packed struct(u32) {
     r: u8,

src/browser/cssom/cssom.zig

@@ -1,4 +1,4 @@
-// Copyright (C) 2023-2026  Lightpanda (Selecy SAS)
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
 //
 // Francis Bouvier <francis@lightpanda.io>
 // Pierre Tachoire <pierre@lightpanda.io>
@@ -16,13 +16,10 @@
 // You should have received a copy of the GNU Affero General Public License
 // along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
-// Gets the Parent of child.
-// HtmlElement.of(script) -> *HTMLElement
-pub fn Struct(comptime T: type) type {
-    return switch (@typeInfo(T)) {
-        .pointer => |ptr| ptr.child,
-        .@"struct" => T,
-        .void => T,
-        else => unreachable,
-    };
-}
+pub const Interfaces = .{
+    @import("StyleSheet.zig"),
+    @import("CSSStyleSheet.zig"),
+    @import("CSSStyleDeclaration.zig"),
+    @import("CSSRuleList.zig"),
+    @import("CSSRule.zig").Interfaces,
+};

src/browser/dom/Animation.zig

@@ -0,0 +1,107 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const js = @import("../js/js.zig");
+const Page = @import("../page.zig").Page;
+
+const Animation = @This();
+
+effect: ?js.Object,
+timeline: ?js.Object,
+ready_resolver: ?js.PromiseResolver,
+finished_resolver: ?js.PromiseResolver,
+
+pub fn constructor(effect: ?js.Object, timeline: ?js.Object) !Animation {
+    return .{
+        .effect = if (effect) |eo| try eo.persist() else null,
+        .timeline = if (timeline) |to| try to.persist() else null,
+        .ready_resolver = null,
+        .finished_resolver = null,
+    };
+}
+
+pub fn get_playState(self: *const Animation) []const u8 {
+    _ = self;
+    return "finished";
+}
+
+pub fn get_pending(self: *const Animation) bool {
+    _ = self;
+    return false;
+}
+
+pub fn get_finished(self: *Animation, page: *Page) !js.Promise {
+    if (self.finished_resolver == null) {
+        const resolver = page.js.createPromiseResolver(.none);
+        try resolver.resolve(self);
+        self.finished_resolver = resolver;
+    }
+    return self.finished_resolver.?.promise();
+}
+
+pub fn get_ready(self: *Animation, page: *Page) !js.Promise {
+    // never resolved, because we're always "finished"
+    if (self.ready_resolver == null) {
+        const resolver = page.js.createPromiseResolver(.none);
+        self.ready_resolver = resolver;
+    }
+    return self.ready_resolver.?.promise();
+}
+
+pub fn get_effect(self: *const Animation) ?js.Object {
+    return self.effect;
+}
+
+pub fn set_effect(self: *Animation, effect: js.Object) !void {
+    self.effect = try effect.persist();
+}
+
+pub fn get_timeline(self: *const Animation) ?js.Object {
+    return self.timeline;
+}
+
+pub fn set_timeline(self: *Animation, timeline: js.Object) !void {
+    self.timeline = try timeline.persist();
+}
+
+pub fn _play(self: *const Animation) void {
+    _ = self;
+}
+
+pub fn _pause(self: *const Animation) void {
+    _ = self;
+}
+
+pub fn _cancel(self: *const Animation) void {
+    _ = self;
+}
+
+pub fn _finish(self: *const Animation) void {
+    _ = self;
+}
+
+pub fn _reverse(self: *const Animation) void {
+    _ = self;
+}
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Animation" {
+    try testing.htmlRunner("dom/animation.html");
+}

src/browser/dom/IntersectionObserver.zig

@@ -0,0 +1,329 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const js = @import("../js/js.zig");
+const log = @import("../../log.zig");
+const parser = @import("../netsurf.zig");
+const Page = @import("../page.zig").Page;
+const Node = @import("node.zig").Node;
+const Element = @import("element.zig").Element;
+
+pub const Interfaces = .{
+    IntersectionObserver,
+    Entry,
+};
+
+// This implementation attempts to be as less wrong as possible. Since we don't
+// render, or know how things are positioned, our best guess isn't very good.
+const IntersectionObserver = @This();
+page: *Page,
+root: *parser.Node,
+callback: js.Function,
+event_node: parser.EventNode,
+observed_entries: std.ArrayList(Entry),
+pending_elements: std.ArrayList(*parser.Element),
+ready_elements: std.ArrayList(*parser.Element),
+
+pub fn constructor(callback: js.Function, opts_: ?IntersectionObserverOptions, page: *Page) !*IntersectionObserver {
+    const opts = opts_ orelse IntersectionObserverOptions{};
+
+    const self = try page.arena.create(IntersectionObserver);
+    self.* = .{
+        .page = page,
+        .callback = callback,
+        .ready_elements = .{},
+        .observed_entries = .{},
+        .pending_elements = .{},
+        .event_node = .{ .func = mutationCallback },
+        .root = opts.root orelse parser.documentToNode(parser.documentHTMLToDocument(page.window.document)),
+    };
+
+    _ = try parser.eventTargetAddEventListener(
+        parser.toEventTarget(parser.Node, self.root),
+        "DOMNodeInserted",
+        &self.event_node,
+        false,
+    );
+
+    _ = try parser.eventTargetAddEventListener(
+        parser.toEventTarget(parser.Node, self.root),
+        "DOMNodeRemoved",
+        &self.event_node,
+        false,
+    );
+
+    return self;
+}
+
+pub fn _disconnect(self: *IntersectionObserver) !void {
+    // We don't free as it is on an arena
+    self.ready_elements = .{};
+    self.observed_entries = .{};
+    self.pending_elements = .{};
+}
+
+pub fn _observe(self: *IntersectionObserver, target_element: *parser.Element, page: *Page) !void {
+    for (self.observed_entries.items) |*observer| {
+        if (observer.target == target_element) {
+            return; // Already observed
+        }
+    }
+
+    if (self.isPending(target_element)) {
+        return; // Already pending
+    }
+
+    for (self.ready_elements.items) |element| {
+        if (element == target_element) {
+            return; // Already primed
+        }
+    }
+
+    // We can never fire callbacks synchronously. Code like React expects any
+    // callback to fire in the future (e.g. via microtasks).
+    try self.ready_elements.append(self.page.arena, target_element);
+    if (self.ready_elements.items.len == 1) {
+        // this is our first ready entry, schedule a callback
+        try page.scheduler.add(self, processReady, 0, .{
+            .name = "intersection ready",
+        });
+    }
+}
+
+pub fn _unobserve(self: *IntersectionObserver, target: *parser.Element) !void {
+    if (self.removeObserved(target)) {
+        return;
+    }
+
+    for (self.ready_elements.items, 0..) |el, index| {
+        if (el == target) {
+            _ = self.ready_elements.swapRemove(index);
+            return;
+        }
+    }
+
+    for (self.pending_elements.items, 0..) |el, index| {
+        if (el == target) {
+            _ = self.pending_elements.swapRemove(index);
+            return;
+        }
+    }
+}
+
+pub fn _takeRecords(self: *IntersectionObserver) []Entry {
+    return self.observed_entries.items;
+}
+
+fn processReady(ctx: *anyopaque) ?u32 {
+    const self: *IntersectionObserver = @ptrCast(@alignCast(ctx));
+    self._processReady() catch |err| {
+        log.err(.web_api, "intersection ready", .{ .err = err });
+    };
+    return null;
+}
+
+fn _processReady(self: *IntersectionObserver) !void {
+    defer self.ready_elements.clearRetainingCapacity();
+    for (self.ready_elements.items) |element| {
+        // IntersectionObserver probably doesn't work like what your intuition
+        // thinks. As long as a node has a parent, even if that parent isn't
+        // connected and even if the two nodes don't intersect, it'll fire the
+        // callback once.
+        if (try Node.get_parentNode(@ptrCast(element)) == null) {
+            if (!self.isPending(element)) {
+                try self.pending_elements.append(self.page.arena, element);
+            }
+            continue;
+        }
+        try self.forceObserve(element);
+    }
+}
+
+fn isPending(self: *IntersectionObserver, element: *parser.Element) bool {
+    for (self.pending_elements.items) |el| {
+        if (el == element) {
+            return true;
+        }
+    }
+    return false;
+}
+
+fn mutationCallback(en: *parser.EventNode, event: *parser.Event) void {
+    const mutation_event = parser.eventToMutationEvent(event);
+    const self: *IntersectionObserver = @fieldParentPtr("event_node", en);
+    self._mutationCallback(mutation_event) catch |err| {
+        log.err(.web_api, "mutation callback", .{ .err = err, .source = "intersection observer" });
+    };
+}
+
+fn _mutationCallback(self: *IntersectionObserver, event: *parser.MutationEvent) !void {
+    const event_type = parser.eventType(@ptrCast(event));
+
+    if (std.mem.eql(u8, event_type, "DOMNodeInserted")) {
+        const node = parser.mutationEventRelatedNode(event) catch return orelse return;
+        if (parser.nodeType(node) != .element) {
+            return;
+        }
+        const el: *parser.Element = @ptrCast(node);
+        if (self.removePending(el)) {
+            // It was pending (because it wasn't in the root), but now it is
+            // we should observe it.
+            try self.forceObserve(el);
+        }
+        return;
+    }
+
+    if (std.mem.eql(u8, event_type, "DOMNodeRemoved")) {
+        const node = parser.mutationEventRelatedNode(event) catch return orelse return;
+        if (parser.nodeType(node) != .element) {
+            return;
+        }
+
+        const el: *parser.Element = @ptrCast(node);
+        if (self.removeObserved(el)) {
+            // It _was_ observed, it no longer is in our root, but if it was
+            // to get re-added, it should be observed again (I think), so
+            // we add it to our pending list
+            try self.pending_elements.append(self.page.arena, el);
+        }
+
+        return;
+    }
+
+    // impossible event type
+    unreachable;
+}
+
+// Exists to skip the checks made _observe when called from a DOMNodeInserted
+// event. In such events, the event handler has alread done the necessary
+// checks.
+fn forceObserve(self: *IntersectionObserver, target: *parser.Element) !void {
+    try self.observed_entries.append(self.page.arena, .{
+        .page = self.page,
+        .root = self.root,
+        .target = target,
+    });
+
+    var result: js.Function.Result = undefined;
+    self.callback.tryCall(void, .{self.observed_entries.items}, &result) catch {
+        log.debug(.user_script, "callback error", .{
+            .err = result.exception,
+            .stack = result.stack,
+            .source = "intersection observer",
+        });
+    };
+}
+
+fn removeObserved(self: *IntersectionObserver, target: *parser.Element) bool {
+    for (self.observed_entries.items, 0..) |*observer, index| {
+        if (observer.target == target) {
+            _ = self.observed_entries.swapRemove(index);
+            return true;
+        }
+    }
+    return false;
+}
+
+fn removePending(self: *IntersectionObserver, target: *parser.Element) bool {
+    for (self.pending_elements.items, 0..) |el, index| {
+        if (el == target) {
+            _ = self.pending_elements.swapRemove(index);
+            return true;
+        }
+    }
+    return false;
+}
+
+const IntersectionObserverOptions = struct {
+    root: ?*parser.Node = null, // Element or Document
+    rootMargin: ?[]const u8 = "0px 0px 0px 0px",
+    threshold: ?Threshold = .{ .single = 0.0 },
+
+    const Threshold = union(enum) {
+        single: f32,
+        list: []const f32,
+    };
+};
+
+// https://developer.mozilla.org/en-US/docs/Web/API/Entry
+// https://w3c.github.io/IntersectionObserver/#intersection-observer-entry
+pub const Entry = struct {
+    page: *Page,
+    root: *parser.Node,
+    target: *parser.Element,
+
+    // Returns the bounds rectangle of the target element as a DOMRectReadOnly. The bounds are computed as described in the documentation for Element.getBoundingClientRect().
+    pub fn get_boundingClientRect(self: *const Entry) !Element.DOMRect {
+        return Element._getBoundingClientRect(self.target, self.page);
+    }
+
+    // Returns the ratio of the intersectionRect to the boundingClientRect.
+    pub fn get_intersectionRatio(_: *const Entry) f32 {
+        return 1.0;
+    }
+
+    // Returns a DOMRectReadOnly representing the target's visible area.
+    pub fn get_intersectionRect(self: *const Entry) !Element.DOMRect {
+        return Element._getBoundingClientRect(self.target, self.page);
+    }
+
+    // A Boolean value which is true if the target element intersects with the
+    // intersection observer's root. If this is true, then, the
+    // Entry describes a transition into a state of
+    // intersection; if it's false, then you know the transition is from
+    // intersecting to not-intersecting.
+    pub fn get_isIntersecting(_: *const Entry) bool {
+        return true;
+    }
+
+    // Returns a DOMRectReadOnly for the intersection observer's root.
+    pub fn get_rootBounds(self: *const Entry) !Element.DOMRect {
+        const root = self.root;
+        if (@intFromPtr(root) == @intFromPtr(self.page.window.document)) {
+            return self.page.renderer.boundingRect();
+        }
+
+        const root_type = parser.nodeType(root);
+
+        var element: *parser.Element = undefined;
+        switch (root_type) {
+            .element => element = parser.nodeToElement(root),
+            .document => {
+                const doc = parser.nodeToDocument(root);
+                element = (try parser.documentGetDocumentElement(doc)).?;
+            },
+            else => return error.InvalidState,
+        }
+
+        return Element._getBoundingClientRect(element, self.page);
+    }
+
+    // The Element whose intersection with the root changed.
+    pub fn get_target(self: *const Entry) *parser.Element {
+        return self.target;
+    }
+
+    // TODO: pub fn get_time(self: *const Entry)
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.IntersectionObserver" {
+    try testing.htmlRunner("dom/intersection_observer.html");
+}

src/browser/dom/MessageChannel.zig

@@ -0,0 +1,288 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const log = @import("../../log.zig");
+const parser = @import("../netsurf.zig");
+
+const js = @import("../js/js.zig");
+const Page = @import("../page.zig").Page;
+const EventTarget = @import("../dom/event_target.zig").EventTarget;
+const EventHandler = @import("../events/event.zig").EventHandler;
+
+const Allocator = std.mem.Allocator;
+
+const MAX_QUEUE_SIZE = 10;
+
+pub const Interfaces = .{ MessageChannel, MessagePort };
+
+const MessageChannel = @This();
+
+port1: *MessagePort,
+port2: *MessagePort,
+
+pub fn constructor(page: *Page) !MessageChannel {
+    // Why do we allocate this rather than storing directly in the struct?
+    // https://github.com/lightpanda-io/project/discussions/165
+    const port1 = try page.arena.create(MessagePort);
+    const port2 = try page.arena.create(MessagePort);
+    port1.* = .{
+        .pair = port2,
+    };
+    port2.* = .{
+        .pair = port1,
+    };
+
+    return .{
+        .port1 = port1,
+        .port2 = port2,
+    };
+}
+
+pub fn get_port1(self: *const MessageChannel) *MessagePort {
+    return self.port1;
+}
+
+pub fn get_port2(self: *const MessageChannel) *MessagePort {
+    return self.port2;
+}
+
+pub const MessagePort = struct {
+    pub const prototype = *EventTarget;
+
+    proto: parser.EventTargetTBase = .{ .internal_target_type = .message_port },
+
+    pair: *MessagePort,
+    closed: bool = false,
+    started: bool = false,
+    onmessage_cbk: ?js.Function = null,
+    onmessageerror_cbk: ?js.Function = null,
+    // This is the queue of messages to dispatch to THIS MessagePort when the
+    // MessagePort is started.
+    queue: std.ArrayListUnmanaged(js.Object) = .empty,
+
+    pub const PostMessageOption = union(enum) {
+        transfer: js.Object,
+        options: Opts,
+
+        pub const Opts = struct {
+            transfer: js.Object,
+        };
+    };
+
+    pub fn _postMessage(self: *MessagePort, obj: js.Object, opts_: ?PostMessageOption, page: *Page) !void {
+        if (self.closed) {
+            return;
+        }
+
+        if (opts_ != null) {
+            log.warn(.web_api, "not implemented", .{ .feature = "MessagePort postMessage options" });
+        }
+
+        try self.pair.dispatchOrQueue(obj, page.arena);
+    }
+
+    // Start impacts the ability to receive a message.
+    // Given pair1 (started) and pair2 (not started), then:
+    //    pair2.postMessage('x'); //will be dispatched to pair1.onmessage
+    //    pair1.postMessage('x'); // will be queued until pair2 is started
+    pub fn _start(self: *MessagePort) !void {
+        if (self.started) {
+            return;
+        }
+        self.started = true;
+        for (self.queue.items) |data| {
+            try self.dispatch(data);
+        }
+        // we'll never use this queue again, but it's allocated with an arena
+        // we don't even need to clear it, but it seems a bit safer to do at
+        // least that
+        self.queue.clearRetainingCapacity();
+    }
+
+    // Closing seems to stop both the publishing and receiving of messages,
+    // effectively rendering the channel useless. It cannot be reversed.
+    pub fn _close(self: *MessagePort) void {
+        self.closed = true;
+        self.pair.closed = true;
+    }
+
+    pub fn get_onmessage(self: *MessagePort) ?js.Function {
+        return self.onmessage_cbk;
+    }
+    pub fn get_onmessageerror(self: *MessagePort) ?js.Function {
+        return self.onmessageerror_cbk;
+    }
+
+    pub fn set_onmessage(self: *MessagePort, listener: EventHandler.Listener, page: *Page) !void {
+        if (self.onmessage_cbk) |cbk| {
+            try self.unregister("message", cbk.id);
+        }
+        self.onmessage_cbk = try self.register(page.arena, "message", listener);
+
+        // When onmessage is set directly, then it's like start() was called.
+        // If addEventListener('message') is used, the app has to call start()
+        // explicitly.
+        try self._start();
+    }
+
+    pub fn set_onmessageerror(self: *MessagePort, listener: EventHandler.Listener, page: *Page) !void {
+        if (self.onmessageerror_cbk) |cbk| {
+            try self.unregister("messageerror", cbk.id);
+        }
+        self.onmessageerror_cbk = try self.register(page.arena, "messageerror", listener);
+    }
+
+    // called from our pair. If port1.postMessage("x") is called, then this
+    // will be called on port2.
+    fn dispatchOrQueue(self: *MessagePort, obj: js.Object, arena: Allocator) !void {
+        // our pair should have checked this already
+        std.debug.assert(self.closed == false);
+
+        if (self.started) {
+            return self.dispatch(try obj.persist());
+        }
+
+        if (self.queue.items.len > MAX_QUEUE_SIZE) {
+            // This isn't part of the spec, but not putting a limit is reckless
+            return error.MessageQueueLimit;
+        }
+        return self.queue.append(arena, try obj.persist());
+    }
+
+    fn dispatch(self: *MessagePort, obj: js.Object) !void {
+        // obj is already persisted, don't use `MessageEvent.constructor`, but
+        // go directly to `init`, which assumes persisted objects.
+        var evt = try MessageEvent.init(.{ .data = obj });
+        _ = try parser.eventTargetDispatchEvent(
+            parser.toEventTarget(MessagePort, self),
+            @as(*parser.Event, @ptrCast(&evt)),
+        );
+    }
+
+    fn register(
+        self: *MessagePort,
+        alloc: Allocator,
+        typ: []const u8,
+        listener: EventHandler.Listener,
+    ) !?js.Function {
+        const target = @as(*parser.EventTarget, @ptrCast(self));
+        const eh = (try EventHandler.register(alloc, target, typ, listener, null)) orelse unreachable;
+        return eh.callback;
+    }
+
+    fn unregister(self: *MessagePort, typ: []const u8, cbk_id: usize) !void {
+        const et = @as(*parser.EventTarget, @ptrCast(self));
+        const lst = try parser.eventTargetHasListener(et, typ, false, cbk_id);
+        if (lst == null) {
+            return;
+        }
+        try parser.eventTargetRemoveEventListener(et, typ, lst.?, false);
+    }
+};
+
+pub const MessageEvent = struct {
+    const Event = @import("../events/event.zig").Event;
+    const DOMException = @import("exceptions.zig").DOMException;
+
+    pub const prototype = *Event;
+    pub const Exception = DOMException;
+    pub const union_make_copy = true;
+
+    proto: parser.Event,
+    data: ?js.Object,
+
+    // You would think if port1 sends to port2, the source would be port2
+    // (which is how I read the documentation), but it appears to always be
+    // null. It can always be set explicitly via the constructor;
+    source: ?js.Object,
+
+    origin: []const u8,
+
+    // This is used for Server-Sent events. Appears to always be an empty
+    // string for MessagePort messages.
+    last_event_id: []const u8,
+
+    // This might be related to the "transfer" option of postMessage which
+    // we don't yet support. For "normal" message, it's always an empty array.
+    // Though it could be set explicitly via the constructor
+    ports: []*MessagePort,
+
+    const Options = struct {
+        data: ?js.Object = null,
+        source: ?js.Object = null,
+        origin: []const u8 = "",
+        lastEventId: []const u8 = "",
+        ports: []*MessagePort = &.{},
+    };
+
+    pub fn constructor(opts: Options) !MessageEvent {
+        return init(.{
+            .data = if (opts.data) |obj| try obj.persist() else null,
+            .source = if (opts.source) |obj| try obj.persist() else null,
+            .ports = opts.ports,
+            .origin = opts.origin,
+            .lastEventId = opts.lastEventId,
+        });
+    }
+
+    // This is like "constructor", but it assumes js.Objects have already been
+    // persisted. Necessary because this `new MessageEvent()` can be called
+    // directly from JS OR from a port.postMessage. In the latter case, data
+    // may have already been persisted (as it might need to be queued);
+    fn init(opts: Options) !MessageEvent {
+        const event = try parser.eventCreate();
+        defer parser.eventDestroy(event);
+        try parser.eventInit(event, "message", .{});
+        parser.eventSetInternalType(event, .message_event);
+
+        return .{
+            .proto = event.*,
+            .data = opts.data,
+            .source = opts.source,
+            .ports = opts.ports,
+            .origin = opts.origin,
+            .last_event_id = opts.lastEventId,
+        };
+    }
+
+    pub fn get_data(self: *const MessageEvent) !?js.Object {
+        return self.data;
+    }
+
+    pub fn get_origin(self: *const MessageEvent) []const u8 {
+        return self.origin;
+    }
+
+    pub fn get_source(self: *const MessageEvent) ?js.Object {
+        return self.source;
+    }
+
+    pub fn get_ports(self: *const MessageEvent) []*MessagePort {
+        return self.ports;
+    }
+
+    pub fn get_lastEventId(self: *const MessageEvent) []const u8 {
+        return self.last_event_id;
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.MessageChannel" {
+    try testing.htmlRunner("dom/message_channel.html");
+}

src/browser/dom/attribute.zig

@@ -0,0 +1,75 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+const parser = @import("../netsurf.zig");
+
+const Node = @import("node.zig").Node;
+
+// WEB IDL https://dom.spec.whatwg.org/#attr
+pub const Attr = struct {
+    pub const Self = parser.Attribute;
+    pub const prototype = *Node;
+    pub const subtype = .node;
+
+    pub fn get_namespaceURI(self: *parser.Attribute) ?[]const u8 {
+        return parser.nodeGetNamespace(parser.attributeToNode(self));
+    }
+
+    pub fn get_prefix(self: *parser.Attribute) ?[]const u8 {
+        return parser.nodeGetPrefix(parser.attributeToNode(self));
+    }
+
+    pub fn get_localName(self: *parser.Attribute) ![]const u8 {
+        return parser.nodeLocalName(parser.attributeToNode(self));
+    }
+
+    pub fn get_name(self: *parser.Attribute) ![]const u8 {
+        return parser.attributeGetName(self);
+    }
+
+    pub fn get_value(self: *parser.Attribute) !?[]const u8 {
+        return parser.attributeGetValue(self);
+    }
+
+    pub fn set_value(self: *parser.Attribute, v: []const u8) !?[]const u8 {
+        if (try parser.attributeGetOwnerElement(self)) |el| {
+            // if possible, go through the element, as that triggers a
+            // DOMAttrModified event (which MutationObserver cares about)
+            const name = try parser.attributeGetName(self);
+            try parser.elementSetAttribute(el, name, v);
+        } else {
+            try parser.attributeSetValue(self, v);
+        }
+        return v;
+    }
+
+    pub fn get_ownerElement(self: *parser.Attribute) !?*parser.Element {
+        return try parser.attributeGetOwnerElement(self);
+    }
+
+    pub fn get_specified(_: *parser.Attribute) bool {
+        return true;
+    }
+};
+
+// Tests
+// -----
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Attribute" {
+    try testing.htmlRunner("dom/attribute.html");
+}

src/browser/dom/cdata_section.zig

@@ -0,0 +1,28 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const parser = @import("../netsurf.zig");
+
+const Text = @import("text.zig").Text;
+
+// https://dom.spec.whatwg.org/#cdatasection
+pub const CDATASection = struct {
+    pub const Self = parser.CDATASection;
+    pub const prototype = *Text;
+    pub const subtype = .node;
+};

src/browser/dom/character_data.zig

@@ -0,0 +1,134 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const parser = @import("../netsurf.zig");
+
+const Node = @import("node.zig").Node;
+const Comment = @import("comment.zig").Comment;
+const Text = @import("text.zig");
+const ProcessingInstruction = @import("processing_instruction.zig").ProcessingInstruction;
+const Element = @import("element.zig").Element;
+const ElementUnion = @import("element.zig").Union;
+
+// CharacterData interfaces
+pub const Interfaces = .{
+    Comment,
+    Text.Text,
+    Text.Interfaces,
+    ProcessingInstruction,
+};
+
+// CharacterData implementation
+pub const CharacterData = struct {
+    pub const Self = parser.CharacterData;
+    pub const prototype = *Node;
+    pub const subtype = .node;
+
+    // JS funcs
+    // --------
+
+    // Read attributes
+
+    pub fn get_length(self: *parser.CharacterData) !u32 {
+        return try parser.characterDataLength(self);
+    }
+
+    pub fn get_nextElementSibling(self: *parser.CharacterData) !?ElementUnion {
+        const res = parser.nodeNextElementSibling(parser.characterDataToNode(self));
+        if (res == null) {
+            return null;
+        }
+        return try Element.toInterface(res.?);
+    }
+
+    pub fn get_previousElementSibling(self: *parser.CharacterData) !?ElementUnion {
+        const res = parser.nodePreviousElementSibling(parser.characterDataToNode(self));
+        if (res == null) {
+            return null;
+        }
+        return try Element.toInterface(res.?);
+    }
+
+    // Read/Write attributes
+
+    pub fn get_data(self: *parser.CharacterData) []const u8 {
+        return parser.characterDataData(self);
+    }
+
+    pub fn set_data(self: *parser.CharacterData, data: []const u8) !void {
+        return try parser.characterDataSetData(self, data);
+    }
+
+    // JS methods
+    // ----------
+
+    pub fn _appendData(self: *parser.CharacterData, data: []const u8) !void {
+        return try parser.characterDataAppendData(self, data);
+    }
+
+    pub fn _deleteData(self: *parser.CharacterData, offset: u32, count: u32) !void {
+        return try parser.characterDataDeleteData(self, offset, count);
+    }
+
+    pub fn _insertData(self: *parser.CharacterData, offset: u32, data: []const u8) !void {
+        return try parser.characterDataInsertData(self, offset, data);
+    }
+
+    pub fn _replaceData(self: *parser.CharacterData, offset: u32, count: u32, data: []const u8) !void {
+        return try parser.characterDataReplaceData(self, offset, count, data);
+    }
+
+    pub fn _substringData(self: *parser.CharacterData, offset: u32, count: u32) ![]const u8 {
+        return parser.characterDataSubstringData(self, offset, count);
+    }
+
+    // netsurf's CharacterData (text, comment) doesn't implement the
+    // dom_node_get_attributes and thus will crash if we try to call nodeIsEqualNode.
+    pub fn _isEqualNode(self: *parser.CharacterData, other_node: *parser.Node) bool {
+        if (parser.nodeType(@ptrCast(@alignCast(self))) != parser.nodeType(other_node)) {
+            return false;
+        }
+
+        const other: *parser.CharacterData = @ptrCast(other_node);
+        if (std.mem.eql(u8, get_data(self), get_data(other)) == false) {
+            return false;
+        }
+
+        return true;
+    }
+
+    pub fn _before(self: *parser.CharacterData, nodes: []const Node.NodeOrText) !void {
+        const ref_node = parser.characterDataToNode(self);
+        return Node.before(ref_node, nodes);
+    }
+
+    pub fn _after(self: *parser.CharacterData, nodes: []const Node.NodeOrText) !void {
+        const ref_node = parser.characterDataToNode(self);
+        return Node.after(ref_node, nodes);
+    }
+};
+
+// Tests
+// -----
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.CharacterData" {
+    try testing.htmlRunner("dom/character_data.html");
+}

src/browser/dom/comment.zig

@@ -1,4 +1,4 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
 //
 // Francis Bouvier <francis@lightpanda.io>
 // Pierre Tachoire <pierre@lightpanda.io>
@@ -16,31 +16,30 @@
 // You should have received a copy of the GNU Affero General Public License
 // along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
-const String = @import("../../../../string.zig").String;
+const parser = @import("../netsurf.zig");
 
-const js = @import("../../../js/js.zig");
+const CharacterData = @import("character_data.zig").CharacterData;
 
-const Node = @import("../../Node.zig");
-const Element = @import("../../Element.zig");
-const HtmlElement = @import("../Html.zig");
+const Page = @import("../page.zig").Page;
 
-const Generic = @This();
-_tag_name: String,
-_tag: Element.Tag,
-_proto: *HtmlElement,
+// https://dom.spec.whatwg.org/#interface-comment
+pub const Comment = struct {
+    pub const Self = parser.Comment;
+    pub const prototype = *CharacterData;
+    pub const subtype = .node;
 
-pub fn asElement(self: *Generic) *Element {
-    return self._proto._proto;
-}
-pub fn asNode(self: *Generic) *Node {
-    return self.asElement().asNode();
-}
-
-pub const JsApi = struct {
-    pub const bridge = js.Bridge(Generic);
-
-    pub const Meta = struct {
-        pub const prototype_chain = bridge.prototypeChain();
-        pub var class_id: bridge.ClassId = undefined;
-    };
+    pub fn constructor(data: ?[]const u8, page: *const Page) !*parser.Comment {
+        return parser.documentCreateComment(
+            parser.documentHTMLToDocument(page.window.document),
+            data orelse "",
+        );
+    }
 };
+
+// Tests
+// -----
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Comment" {
+    try testing.htmlRunner("dom/comment.html");
+}

src/browser/dom/css.zig

@@ -0,0 +1,80 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const parser = @import("../netsurf.zig");
+
+const css = @import("../css/css.zig");
+const Node = @import("../css/libdom.zig").Node;
+const NodeList = @import("nodelist.zig").NodeList;
+
+const MatchFirst = struct {
+    n: ?*parser.Node = null,
+
+    pub fn match(m: *MatchFirst, n: Node) !void {
+        m.n = n.node;
+    }
+};
+
+pub fn querySelector(alloc: std.mem.Allocator, n: *parser.Node, selector: []const u8) !?*parser.Node {
+    const ps = try css.parse(alloc, selector, .{ .accept_pseudo_elts = true });
+    defer ps.deinit(alloc);
+
+    var m = MatchFirst{};
+
+    _ = try css.matchFirst(&ps, Node{ .node = n }, &m);
+    return m.n;
+}
+
+const MatchAll = struct {
+    alloc: std.mem.Allocator,
+    nl: NodeList,
+
+    fn init(alloc: std.mem.Allocator) MatchAll {
+        return .{
+            .alloc = alloc,
+            .nl = .{},
+        };
+    }
+
+    fn deinit(m: *MatchAll) void {
+        m.nl.deinit(m.alloc);
+    }
+
+    pub fn match(m: *MatchAll, n: Node) !void {
+        try m.nl.append(m.alloc, n.node);
+    }
+
+    fn toOwnedList(m: *MatchAll) NodeList {
+        // reset it.
+        defer m.nl = .{};
+        return m.nl;
+    }
+};
+
+pub fn querySelectorAll(alloc: std.mem.Allocator, n: *parser.Node, selector: []const u8) !NodeList {
+    const ps = try css.parse(alloc, selector, .{ .accept_pseudo_elts = true });
+    defer ps.deinit(alloc);
+
+    var m = MatchAll.init(alloc);
+    defer m.deinit();
+
+    try css.matchAll(&ps, Node{ .node = n }, &m);
+    return m.toOwnedList();
+}

src/browser/dom/document.zig

@@ -0,0 +1,385 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const log = @import("../../log.zig");
+
+const js = @import("../js/js.zig");
+const parser = @import("../netsurf.zig");
+const Page = @import("../page.zig").Page;
+
+const Node = @import("node.zig").Node;
+const NodeList = @import("nodelist.zig").NodeList;
+const NodeUnion = @import("node.zig").Union;
+
+const collection = @import("html_collection.zig");
+const css = @import("css.zig");
+
+const Element = @import("element.zig").Element;
+const ElementUnion = @import("element.zig").Union;
+const TreeWalker = @import("tree_walker.zig").TreeWalker;
+const CSSStyleSheet = @import("../cssom/CSSStyleSheet.zig");
+const NodeIterator = @import("node_iterator.zig").NodeIterator;
+const Range = @import("range.zig").Range;
+
+const CustomEvent = @import("../events/custom_event.zig").CustomEvent;
+
+const DOMImplementation = @import("implementation.zig").DOMImplementation;
+
+// WEB IDL https://dom.spec.whatwg.org/#document
+pub const Document = struct {
+    pub const Self = parser.Document;
+    pub const prototype = *Node;
+    pub const subtype = .node;
+
+    pub fn constructor(page: *const Page) !*parser.DocumentHTML {
+        const doc = try parser.documentCreateDocument(
+            try parser.documentHTMLGetTitle(page.window.document),
+        );
+
+        // we have to work w/ document instead of html document.
+        const ddoc = parser.documentHTMLToDocument(doc);
+        const ccur = parser.documentHTMLToDocument(page.window.document);
+        try parser.documentSetDocumentURI(ddoc, try parser.documentGetDocumentURI(ccur));
+        try parser.documentSetInputEncoding(ddoc, try parser.documentGetInputEncoding(ccur));
+
+        return doc;
+    }
+
+    // JS funcs
+    // --------
+    pub fn get_implementation(_: *parser.Document) DOMImplementation {
+        return DOMImplementation{};
+    }
+
+    pub fn get_documentElement(self: *parser.Document) !?ElementUnion {
+        const e = try parser.documentGetDocumentElement(self);
+        if (e == null) return null;
+        return try Element.toInterface(e.?);
+    }
+
+    pub fn get_documentURI(self: *parser.Document) ![]const u8 {
+        return try parser.documentGetDocumentURI(self);
+    }
+
+    pub fn get_URL(self: *parser.Document) ![]const u8 {
+        return try get_documentURI(self);
+    }
+
+    // TODO implement contentType
+    pub fn get_contentType(self: *parser.Document) []const u8 {
+        _ = self;
+        return "text/html";
+    }
+
+    // TODO implement compactMode
+    pub fn get_compatMode(self: *parser.Document) []const u8 {
+        _ = self;
+        return "CSS1Compat";
+    }
+
+    pub fn get_characterSet(self: *parser.Document) ![]const u8 {
+        return try parser.documentGetInputEncoding(self);
+    }
+
+    // alias of get_characterSet
+    pub fn get_charset(self: *parser.Document) ![]const u8 {
+        return try get_characterSet(self);
+    }
+
+    // alias of get_characterSet
+    pub fn get_inputEncoding(self: *parser.Document) ![]const u8 {
+        return try get_characterSet(self);
+    }
+
+    pub fn get_doctype(self: *parser.Document) !?*parser.DocumentType {
+        return try parser.documentGetDoctype(self);
+    }
+
+    pub fn _createEvent(_: *parser.Document, eventCstr: []const u8) !union(enum) {
+        base: *parser.Event,
+        custom: CustomEvent,
+    } {
+        const eqlIgnoreCase = std.ascii.eqlIgnoreCase;
+
+        if (eqlIgnoreCase(eventCstr, "Event") or eqlIgnoreCase(eventCstr, "Events") or eqlIgnoreCase(eventCstr, "HTMLEvents")) {
+            return .{ .base = try parser.eventCreate() };
+        }
+
+        // Not documented in MDN but supported in Chrome.
+        // This is actually both instance of `Event` and `CustomEvent`.
+        if (std.ascii.eqlIgnoreCase(eventCstr, "CustomEvent")) {
+            return .{ .custom = try CustomEvent.constructor(eventCstr, null) };
+        }
+
+        return error.NotSupported;
+    }
+
+    pub fn _getElementById(self: *parser.Document, id: []const u8) !?ElementUnion {
+        const e = try parser.documentGetElementById(self, id) orelse return null;
+        return try Element.toInterface(e);
+    }
+
+    pub fn _createElement(self: *parser.Document, tag_name: []const u8) !ElementUnion {
+        // The element’s namespace is the HTML namespace when document is an HTML document
+        // https://dom.spec.whatwg.org/#ref-for-dom-document-createelement%E2%91%A0
+        const e = try parser.documentCreateElementNS(self, "http://www.w3.org/1999/xhtml", tag_name);
+        return Element.toInterface(e);
+    }
+
+    pub fn _createElementNS(self: *parser.Document, ns: []const u8, tag_name: []const u8) !ElementUnion {
+        const e = try parser.documentCreateElementNS(self, ns, tag_name);
+        return try Element.toInterface(e);
+    }
+
+    // We can't simply use libdom dom_document_get_elements_by_tag_name here.
+    // Indeed, netsurf implemented a previous dom spec when
+    // getElementsByTagName returned a NodeList.
+    // But since
+    // https://github.com/whatwg/dom/commit/190700b7c12ecfd3b5ebdb359ab1d6ea9cbf7749
+    // the spec changed to return an HTMLCollection instead.
+    // That's why we reimplemented getElementsByTagName by using an
+    // HTMLCollection in zig here.
+    pub fn _getElementsByTagName(self: *parser.Document, tag_name: js.String) !collection.HTMLCollection {
+        return collection.HTMLCollectionByTagName(parser.documentToNode(self), tag_name.string, .{
+            .include_root = true,
+        });
+    }
+
+    pub fn _getElementsByClassName(self: *parser.Document, class_names: js.String) !collection.HTMLCollection {
+        return collection.HTMLCollectionByClassName(parser.documentToNode(self), class_names.string, .{
+            .include_root = true,
+        });
+    }
+
+    pub fn _createDocumentFragment(self: *parser.Document) !*parser.DocumentFragment {
+        return try parser.documentCreateDocumentFragment(self);
+    }
+
+    pub fn _createTextNode(self: *parser.Document, data: []const u8) !*parser.Text {
+        return try parser.documentCreateTextNode(self, data);
+    }
+
+    pub fn _createCDATASection(self: *parser.Document, data: []const u8) !*parser.CDATASection {
+        return try parser.documentCreateCDATASection(self, data);
+    }
+
+    pub fn _createComment(self: *parser.Document, data: []const u8) !*parser.Comment {
+        return try parser.documentCreateComment(self, data);
+    }
+
+    pub fn _createProcessingInstruction(self: *parser.Document, target: []const u8, data: []const u8) !*parser.ProcessingInstruction {
+        return try parser.documentCreateProcessingInstruction(self, target, data);
+    }
+
+    pub fn _importNode(self: *parser.Document, node: *parser.Node, deep: ?bool) !NodeUnion {
+        const n = try parser.documentImportNode(self, node, deep orelse false);
+        return try Node.toInterface(n);
+    }
+
+    pub fn _adoptNode(self: *parser.Document, node: *parser.Node) !NodeUnion {
+        const n = try parser.documentAdoptNode(self, node);
+        return try Node.toInterface(n);
+    }
+
+    pub fn _createAttribute(self: *parser.Document, name: []const u8) !*parser.Attribute {
+        return try parser.documentCreateAttribute(self, name);
+    }
+
+    pub fn _createAttributeNS(self: *parser.Document, ns: []const u8, qname: []const u8) !*parser.Attribute {
+        return try parser.documentCreateAttributeNS(self, ns, qname);
+    }
+
+    // ParentNode
+    // https://dom.spec.whatwg.org/#parentnode
+    pub fn get_children(self: *parser.Document) !collection.HTMLCollection {
+        return collection.HTMLCollectionChildren(parser.documentToNode(self), .{
+            .include_root = false,
+        });
+    }
+
+    pub fn get_firstElementChild(self: *parser.Document) !?ElementUnion {
+        const elt = try parser.documentGetDocumentElement(self) orelse return null;
+        return try Element.toInterface(elt);
+    }
+
+    pub fn get_lastElementChild(self: *parser.Document) !?ElementUnion {
+        const elt = try parser.documentGetDocumentElement(self) orelse return null;
+        return try Element.toInterface(elt);
+    }
+
+    pub fn get_childElementCount(self: *parser.Document) !u32 {
+        _ = try parser.documentGetDocumentElement(self) orelse return 0;
+        return 1;
+    }
+
+    pub fn _querySelector(self: *parser.Document, selector: []const u8, page: *Page) !?ElementUnion {
+        if (selector.len == 0) return null;
+
+        const n = try css.querySelector(page.call_arena, parser.documentToNode(self), selector);
+
+        if (n == null) return null;
+
+        return try Element.toInterface(parser.nodeToElement(n.?));
+    }
+
+    pub fn _querySelectorAll(self: *parser.Document, selector: []const u8, page: *Page) !NodeList {
+        return css.querySelectorAll(page.arena, parser.documentToNode(self), selector);
+    }
+
+    pub fn _prepend(self: *parser.Document, nodes: []const Node.NodeOrText) !void {
+        return Node.prepend(parser.documentToNode(self), nodes);
+    }
+
+    pub fn _append(self: *parser.Document, nodes: []const Node.NodeOrText) !void {
+        return Node.append(parser.documentToNode(self), nodes);
+    }
+
+    pub fn _replaceChildren(self: *parser.Document, nodes: []const Node.NodeOrText) !void {
+        return Node.replaceChildren(parser.documentToNode(self), nodes);
+    }
+
+    pub fn _createTreeWalker(_: *parser.Document, root: *parser.Node, what_to_show: ?TreeWalker.WhatToShow, filter: ?TreeWalker.TreeWalkerOpts) !TreeWalker {
+        return TreeWalker.init(root, what_to_show, filter);
+    }
+
+    pub fn _createNodeIterator(_: *parser.Document, root: *parser.Node, what_to_show: ?NodeIterator.WhatToShow, filter: ?NodeIterator.NodeIteratorOpts) !NodeIterator {
+        return NodeIterator.init(root, what_to_show, filter);
+    }
+
+    pub fn getActiveElement(self: *parser.Document, page: *Page) !?*parser.Element {
+        if (page.getNodeState(@ptrCast(@alignCast(self)))) |state| {
+            if (state.active_element) |ae| {
+                return ae;
+            }
+        }
+
+        if (try parser.documentHTMLBody(page.window.document)) |body| {
+            return @ptrCast(@alignCast(body));
+        }
+
+        return try parser.documentGetDocumentElement(self);
+    }
+
+    pub fn get_activeElement(self: *parser.Document, page: *Page) !?ElementUnion {
+        const ae = (try getActiveElement(self, page)) orelse return null;
+        return try Element.toInterface(ae);
+    }
+
+    // TODO: some elements can't be focused, like if they're disabled
+    // but there doesn't seem to be a generic way to check this. For example
+    // we could look for the "disabled" attribute, but that's only meaningful
+    // on certain types, and libdom's vtable doesn't seem to expose this.
+    pub fn setFocus(self: *parser.Document, e: *parser.ElementHTML, page: *Page) !void {
+        const state = try page.getOrCreateNodeState(@ptrCast(@alignCast(self)));
+        state.active_element = @ptrCast(e);
+    }
+
+    pub fn _createRange(_: *parser.Document, page: *Page) Range {
+        return Range.constructor(page);
+    }
+
+    // TODO: dummy implementation
+    pub fn get_styleSheets(_: *parser.Document) []CSSStyleSheet {
+        return &.{};
+    }
+
+    pub fn get_adoptedStyleSheets(self: *parser.Document, page: *Page) !js.Object {
+        const state = try page.getOrCreateNodeState(@ptrCast(@alignCast(self)));
+        if (state.adopted_style_sheets) |obj| {
+            return obj;
+        }
+
+        const obj = try page.js.createArray(0).persist();
+        state.adopted_style_sheets = obj;
+        return obj;
+    }
+
+    pub fn set_adoptedStyleSheets(self: *parser.Document, sheets: js.Object, page: *Page) !void {
+        const state = try page.getOrCreateNodeState(@ptrCast(@alignCast(self)));
+        state.adopted_style_sheets = try sheets.persist();
+    }
+
+    pub fn _hasFocus(_: *parser.Document) bool {
+        log.debug(.web_api, "not implemented", .{ .feature = "Document hasFocus" });
+        return true;
+    }
+
+    pub fn _open(_: *parser.Document, page: *Page) !*parser.DocumentHTML {
+        if (page.open) {
+            return page.window.document;
+        }
+
+        // This implementation is invalid.
+        // According to MDN, we should cleanup registered listeners.
+        // So we sould cleanup previous DOM memory.
+        // But this implementation is more simple for now.
+        const html_doc = try parser.documentHTMLParseFromStr("");
+        try page.setDocument(html_doc);
+        page.open = true;
+
+        return page.window.document;
+    }
+
+    pub fn _close(_: *parser.Document, page: *Page) !void {
+        page.open = false;
+    }
+
+    pub fn _write(self: *parser.Document, str: []const u8, page: *Page) !void {
+        _ = try _open(self, page);
+
+        const document = parser.documentHTMLToDocument(page.window.document);
+        const fragment = try parser.documentParseFragmentFromStr(document, str);
+        const fragment_node = parser.documentFragmentToNode(fragment);
+
+        const fragment_html = parser.nodeFirstChild(fragment_node) orelse return;
+        const fragment_head = parser.nodeFirstChild(fragment_html) orelse return;
+        const fragment_body = parser.nodeNextSibling(fragment_head) orelse return;
+
+        const document_node = parser.documentToNode(document);
+        const document_html = parser.nodeFirstChild(document_node) orelse return;
+        const document_head = parser.nodeFirstChild(document_html) orelse return;
+        const document_body = parser.nodeNextSibling(document_head) orelse return;
+
+        {
+            const children = try parser.nodeGetChildNodes(fragment_head);
+            // always index 0, because nodeAppendChild moves the node out of
+            // the nodeList and into the new tree
+            while (parser.nodeListItem(children, 0)) |child| {
+                _ = try parser.nodeAppendChild(document_head, child);
+            }
+        }
+
+        {
+            const children = try parser.nodeGetChildNodes(fragment_body);
+            // always index 0, because nodeAppendChild moves the node out of
+            // the nodeList and into the new tree
+            while (parser.nodeListItem(children, 0)) |child| {
+                _ = try parser.nodeAppendChild(document_body, child);
+            }
+        }
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Document" {
+    try testing.htmlRunner("dom/document.html");
+}
+test "Browser: DOM.Document.write" {
+    try testing.htmlRunner("dom/document_write.html");
+}

src/browser/dom/document_fragment.zig

@@ -0,0 +1,96 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const css = @import("css.zig");
+const parser = @import("../netsurf.zig");
+const Page = @import("../page.zig").Page;
+const NodeList = @import("nodelist.zig").NodeList;
+const Element = @import("element.zig").Element;
+const ElementUnion = @import("element.zig").Union;
+const collection = @import("html_collection.zig");
+
+const Node = @import("node.zig").Node;
+
+// WEB IDL https://dom.spec.whatwg.org/#documentfragment
+pub const DocumentFragment = struct {
+    pub const Self = parser.DocumentFragment;
+    pub const prototype = *Node;
+    pub const subtype = .node;
+
+    pub fn constructor(page: *const Page) !*parser.DocumentFragment {
+        return parser.documentCreateDocumentFragment(
+            parser.documentHTMLToDocument(page.window.document),
+        );
+    }
+
+    pub fn _isEqualNode(self: *parser.DocumentFragment, other_node: *parser.Node) bool {
+        const other_type = parser.nodeType(other_node);
+        if (other_type != .document_fragment) {
+            return false;
+        }
+        _ = self;
+        return true;
+    }
+
+    pub fn _prepend(self: *parser.DocumentFragment, nodes: []const Node.NodeOrText) !void {
+        return Node.prepend(parser.documentFragmentToNode(self), nodes);
+    }
+
+    pub fn _append(self: *parser.DocumentFragment, nodes: []const Node.NodeOrText) !void {
+        return Node.append(parser.documentFragmentToNode(self), nodes);
+    }
+
+    pub fn _replaceChildren(self: *parser.DocumentFragment, nodes: []const Node.NodeOrText) !void {
+        return Node.replaceChildren(parser.documentFragmentToNode(self), nodes);
+    }
+
+    pub fn _querySelector(self: *parser.DocumentFragment, selector: []const u8, page: *Page) !?ElementUnion {
+        if (selector.len == 0) return null;
+
+        const n = try css.querySelector(page.call_arena, parser.documentFragmentToNode(self), selector);
+
+        if (n == null) return null;
+
+        return try Element.toInterface(parser.nodeToElement(n.?));
+    }
+
+    pub fn _querySelectorAll(self: *parser.DocumentFragment, selector: []const u8, page: *Page) !NodeList {
+        return css.querySelectorAll(page.arena, parser.documentFragmentToNode(self), selector);
+    }
+
+    pub fn get_childElementCount(self: *parser.DocumentFragment) !u32 {
+        var children = try get_children(self);
+        return children.get_length();
+    }
+
+    pub fn get_children(self: *parser.DocumentFragment) !collection.HTMLCollection {
+        return collection.HTMLCollectionChildren(parser.documentFragmentToNode(self), .{
+            .include_root = false,
+        });
+    }
+
+    pub fn _getElementById(self: *parser.DocumentFragment, id: []const u8) !?ElementUnion {
+        const e = try parser.nodeGetElementById(@ptrCast(@alignCast(self)), id) orelse return null;
+        return try Element.toInterface(e);
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.DocumentFragment" {
+    try testing.htmlRunner("dom/document_fragment.html");
+}

src/browser/dom/document_type.zig

@@ -0,0 +1,67 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const parser = @import("../netsurf.zig");
+
+const Node = @import("node.zig").Node;
+
+// WEB IDL https://dom.spec.whatwg.org/#documenttype
+pub const DocumentType = struct {
+    pub const Self = parser.DocumentType;
+    pub const prototype = *Node;
+    pub const subtype = .node;
+
+    pub fn get_name(self: *parser.DocumentType) ![]const u8 {
+        return parser.documentTypeGetName(self);
+    }
+
+    pub fn get_publicId(self: *parser.DocumentType) []const u8 {
+        return parser.documentTypeGetPublicId(self);
+    }
+
+    pub fn get_systemId(self: *parser.DocumentType) []const u8 {
+        return parser.documentTypeGetSystemId(self);
+    }
+
+    // netsurf's DocumentType doesn't implement the dom_node_get_attributes
+    // and thus will crash if we try to call nodeIsEqualNode.
+    pub fn _isEqualNode(self: *parser.DocumentType, other_node: *parser.Node) !bool {
+        if (parser.nodeType(other_node) != .document_type) {
+            return false;
+        }
+
+        const other: *parser.DocumentType = @ptrCast(other_node);
+        if (std.mem.eql(u8, try get_name(self), try get_name(other)) == false) {
+            return false;
+        }
+        if (std.mem.eql(u8, get_publicId(self), get_publicId(other)) == false) {
+            return false;
+        }
+        if (std.mem.eql(u8, get_systemId(self), get_systemId(other)) == false) {
+            return false;
+        }
+        return true;
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.DocumentType" {
+    try testing.htmlRunner("dom/document_type.html");
+}

src/browser/dom/dom.zig

@@ -0,0 +1,56 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const DOMException = @import("exceptions.zig").DOMException;
+const EventTarget = @import("event_target.zig").EventTarget;
+const DOMImplementation = @import("implementation.zig").DOMImplementation;
+const NamedNodeMap = @import("namednodemap.zig").NamedNodeMap;
+const DOMTokenList = @import("token_list.zig");
+const NodeList = @import("nodelist.zig");
+const Node = @import("node.zig");
+const ResizeObserver = @import("resize_observer.zig");
+const MutationObserver = @import("mutation_observer.zig");
+const DOMParser = @import("dom_parser.zig").DOMParser;
+const TreeWalker = @import("tree_walker.zig").TreeWalker;
+const NodeIterator = @import("node_iterator.zig").NodeIterator;
+const NodeFilter = @import("node_filter.zig").NodeFilter;
+const PerformanceObserver = @import("performance_observer.zig").PerformanceObserver;
+
+pub const Interfaces = .{
+    DOMException,
+    EventTarget,
+    DOMImplementation,
+    NamedNodeMap,
+    NamedNodeMap.Iterator,
+    DOMTokenList.Interfaces,
+    NodeList.Interfaces,
+    Node.Node,
+    Node.Interfaces,
+    ResizeObserver.Interfaces,
+    MutationObserver.Interfaces,
+    DOMParser,
+    TreeWalker,
+    NodeIterator,
+    NodeFilter,
+    @import("performance.zig").Interfaces,
+    PerformanceObserver,
+    @import("range.zig").Interfaces,
+    @import("Animation.zig"),
+    @import("MessageChannel.zig").Interfaces,
+    @import("IntersectionObserver.zig").Interfaces,
+};

src/browser/dom/dom_parser.zig

@@ -1,4 +1,4 @@
-// Copyright (C) 2023-2026  Lightpanda (Selecy SAS)
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
 //
 // Francis Bouvier <francis@lightpanda.io>
 // Pierre Tachoire <pierre@lightpanda.io>
@@ -16,27 +16,26 @@
 // You should have received a copy of the GNU Affero General Public License
 // along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
-const js = @import("../../../js/js.zig");
-const Node = @import("../../Node.zig");
-const Element = @import("../../Element.zig");
-const HtmlElement = @import("../Html.zig");
+const std = @import("std");
+const parser = @import("../netsurf.zig");
 
-const DList = @This();
-_proto: *HtmlElement,
+// https://developer.mozilla.org/en-US/docs/Web/API/DOMParser
+pub const DOMParser = struct {
+    pub fn constructor() !DOMParser {
+        return .{};
+    }
 
-pub fn asElement(self: *DList) *Element {
-    return self._proto._proto;
-}
-pub fn asNode(self: *DList) *Node {
-    return self.asElement().asNode();
-}
+    pub fn _parseFromString(_: *DOMParser, string: []const u8, mime_type: []const u8) !*parser.DocumentHTML {
+        if (!std.mem.eql(u8, mime_type, "text/html")) {
+            // TODO: Support XML
+            return error.TypeError;
+        }
 
-pub const JsApi = struct {
-    pub const bridge = js.Bridge(DList);
-
-    pub const Meta = struct {
-        pub const name = "HTMLDListElement";
-        pub const prototype_chain = bridge.prototypeChain();
-        pub var class_id: bridge.ClassId = undefined;
-    };
+        return try parser.documentHTMLParseFromStr(string);
+    }
 };
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Parser" {
+    try testing.htmlRunner("dom/dom_parser.html");
+}

src/browser/dom/element.zig

@@ -0,0 +1,686 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const js = @import("../js/js.zig");
+const parser = @import("../netsurf.zig");
+const Page = @import("../page.zig").Page;
+
+const css = @import("css.zig");
+const log = @import("../../log.zig");
+const dump = @import("../dump.zig");
+const collection = @import("html_collection.zig");
+
+const Node = @import("node.zig").Node;
+const Walker = @import("walker.zig").WalkerDepthFirst;
+const NodeList = @import("nodelist.zig").NodeList;
+const HTMLElem = @import("../html/elements.zig");
+const ShadowRoot = @import("../dom/shadow_root.zig").ShadowRoot;
+
+const Animation = @import("Animation.zig");
+
+pub const Union = @import("../html/elements.zig").Union;
+
+// WEB IDL https://dom.spec.whatwg.org/#element
+pub const Element = struct {
+    pub const Self = parser.Element;
+    pub const prototype = *Node;
+    pub const subtype = .node;
+
+    pub const DOMRect = struct {
+        x: f64,
+        y: f64,
+        width: f64,
+        height: f64,
+        bottom: f64,
+        right: f64,
+        top: f64,
+        left: f64,
+    };
+
+    pub fn toInterface(e: *parser.Element) !Union {
+        return toInterfaceT(Union, e);
+    }
+
+    pub fn toInterfaceT(comptime T: type, e: *parser.Element) !T {
+        const tagname = try parser.elementGetTagName(e) orelse {
+            // If the owner's document is HTML, assume we have an HTMLElement.
+            const doc = parser.nodeOwnerDocument(parser.elementToNode(e));
+            if (doc != null and !doc.?.is_html) {
+                return .{ .HTMLElement = @as(*parser.ElementHTML, @ptrCast(e)) };
+            }
+
+            return .{ .Element = e };
+        };
+
+        // TODO SVGElement and MathML are not supported yet.
+
+        const tag = parser.Tag.fromString(tagname) catch {
+            // If the owner's document is HTML, assume we have an HTMLElement.
+            const doc = parser.nodeOwnerDocument(parser.elementToNode(e));
+            if (doc != null and doc.?.is_html) {
+                return .{ .HTMLElement = @as(*parser.ElementHTML, @ptrCast(e)) };
+            }
+
+            return .{ .Element = e };
+        };
+
+        return HTMLElem.toInterfaceFromTag(T, e, tag);
+    }
+
+    // JS funcs
+    // --------
+
+    pub fn get_namespaceURI(self: *parser.Element) ?[]const u8 {
+        return parser.nodeGetNamespace(parser.elementToNode(self));
+    }
+
+    pub fn get_prefix(self: *parser.Element) ?[]const u8 {
+        return parser.nodeGetPrefix(parser.elementToNode(self));
+    }
+
+    pub fn get_localName(self: *parser.Element) ![]const u8 {
+        return try parser.nodeLocalName(parser.elementToNode(self));
+    }
+
+    pub fn get_tagName(self: *parser.Element) ![]const u8 {
+        return try parser.nodeName(parser.elementToNode(self));
+    }
+
+    pub fn get_dir(self: *parser.Element) ![]const u8 {
+        return try parser.elementGetAttribute(self, "dir") orelse "";
+    }
+
+    pub fn set_dir(self: *parser.Element, dir: []const u8) !void {
+        return parser.elementSetAttribute(self, "dir", dir);
+    }
+
+    pub fn get_id(self: *parser.Element) ![]const u8 {
+        return try parser.elementGetAttribute(self, "id") orelse "";
+    }
+
+    pub fn set_id(self: *parser.Element, id: []const u8) !void {
+        return try parser.elementSetAttribute(self, "id", id);
+    }
+
+    pub fn get_className(self: *parser.Element) ![]const u8 {
+        return try parser.elementGetAttribute(self, "class") orelse "";
+    }
+
+    pub fn set_className(self: *parser.Element, class: []const u8) !void {
+        return try parser.elementSetAttribute(self, "class", class);
+    }
+
+    pub fn get_slot(self: *parser.Element) ![]const u8 {
+        return try parser.elementGetAttribute(self, "slot") orelse "";
+    }
+
+    pub fn set_slot(self: *parser.Element, slot: []const u8) !void {
+        return try parser.elementSetAttribute(self, "slot", slot);
+    }
+
+    pub fn get_assignedSlot(self: *parser.Element, page: *const Page) !?*parser.Slot {
+        return @import("../SlotChangeMonitor.zig").findSlot(self, page);
+    }
+
+    pub fn get_classList(self: *parser.Element) !*parser.TokenList {
+        return try parser.tokenListCreate(self, "class");
+    }
+
+    pub fn get_attributes(self: *parser.Element) !*parser.NamedNodeMap {
+        // An element must have non-nil attributes.
+        return try parser.nodeGetAttributes(parser.elementToNode(self)) orelse unreachable;
+    }
+
+    pub fn get_innerHTML(self: *parser.Element, page: *Page) ![]const u8 {
+        var aw = std.Io.Writer.Allocating.init(page.call_arena);
+        try dump.writeChildren(parser.elementToNode(self), .{}, &aw.writer);
+        return aw.written();
+    }
+
+    pub fn get_outerHTML(self: *parser.Element, page: *Page) ![]const u8 {
+        var aw = std.Io.Writer.Allocating.init(page.call_arena);
+        try dump.writeNode(parser.elementToNode(self), .{}, &aw.writer);
+        return aw.written();
+    }
+
+    pub fn set_innerHTML(self: *parser.Element, str: []const u8, page: *Page) !void {
+        const node = parser.elementToNode(self);
+        const doc = parser.nodeOwnerDocument(node) orelse return parser.DOMError.WrongDocument;
+        // parse the fragment
+        const fragment = try parser.documentParseFragmentFromStr(doc, str);
+
+        // remove existing children
+        try Node.removeChildren(node);
+
+        const fragment_node = parser.documentFragmentToNode(fragment);
+
+        // I'm not sure what the exact behavior is supposed to be. Initially,
+        // we were only copying the body of the document fragment. But it seems
+        // like head elements should be copied too. Specifically, some sites
+        // create script tags via innerHTML, which we need to capture.
+        // If you play with this in a browser, you should notice that the
+        // behavior is different depending on whether you're in a blank page
+        // or an actual document. In a blank page, something like:
+        //    x.innerHTML = '<script></script>';
+        // does _not_ create an empty script, but in a real page, it does. Weird.
+        const html = parser.nodeFirstChild(fragment_node) orelse return;
+        const head = parser.nodeFirstChild(html) orelse return;
+        const body = parser.nodeNextSibling(head) orelse return;
+
+        if (try parser.elementTag(self) == .template) {
+            // HTMLElementTemplate is special. We don't append these as children
+            // of the template, but instead set its content as the body of the
+            // fragment. Simpler to do this by copying the body children into
+            // a new fragment
+            const clean = try parser.documentCreateDocumentFragment(doc);
+            const children = try parser.nodeGetChildNodes(body);
+            // always index 0, because nodeAppendChild moves the node out of
+            // the nodeList and into the new tree
+            while (parser.nodeListItem(children, 0)) |child| {
+                _ = try parser.nodeAppendChild(@ptrCast(@alignCast(clean)), child);
+            }
+
+            const state = try page.getOrCreateNodeState(node);
+            state.template_content = clean;
+            return;
+        }
+
+        // For any node other than a template, we copy the head and body elements
+        // as child nodes of the element
+        {
+            // First, copy some of the head element
+            const children = try parser.nodeGetChildNodes(head);
+            // always index 0, because nodeAppendChild moves the node out of
+            // the nodeList and into the new tree
+            while (parser.nodeListItem(children, 0)) |child| {
+                _ = try parser.nodeAppendChild(node, child);
+            }
+        }
+
+        {
+            const children = try parser.nodeGetChildNodes(body);
+            // always index 0, because nodeAppendChild moves the node out of
+            // the nodeList and into the new tree
+            while (parser.nodeListItem(children, 0)) |child| {
+                _ = try parser.nodeAppendChild(node, child);
+            }
+        }
+    }
+
+    /// Parses the given `input` string and inserts its children to an element at given `position`.
+    /// https://developer.mozilla.org/en-US/docs/Web/API/Element/insertAdjacentHTML
+    ///
+    /// TODO: Support for XML parsing and `TrustedHTML` instances.
+    pub fn _insertAdjacentHTML(self: *parser.Element, position: []const u8, input: []const u8) !void {
+        const self_node = parser.elementToNode(self);
+        const doc = parser.nodeOwnerDocument(self_node) orelse {
+            return parser.DOMError.WrongDocument;
+        };
+
+        // Parse the fragment.
+        // Should return error.Syntax on fail?
+        const fragment = try parser.documentParseFragmentFromStr(doc, input);
+        const fragment_node = parser.documentFragmentToNode(fragment);
+
+        // We always get it wrapped like so:
+        // <html><head></head><body>{ ... }</body></html>
+        // None of the following can be null.
+        const maybe_html = parser.nodeFirstChild(fragment_node);
+        std.debug.assert(maybe_html != null);
+        const html = maybe_html orelse return;
+
+        const maybe_body = parser.nodeLastChild(html);
+        std.debug.assert(maybe_body != null);
+        const body = maybe_body orelse return;
+
+        const children = try parser.nodeGetChildNodes(body);
+
+        // * `target_node` is `*Node` (where we actually insert),
+        // * `prev_node` is `?*Node`.
+        const target_node, const prev_node = blk: {
+            // Prefer case-sensitive match.
+            // "beforeend" was the most common case in my tests; we might adjust the order
+            // depending on which ones websites prefer most.
+            if (std.mem.eql(u8, position, "beforeend")) {
+                break :blk .{ self_node, null };
+            }
+
+            if (std.mem.eql(u8, position, "afterbegin")) {
+                // Get the first child; null indicates there are no children.
+                const first_child = parser.nodeFirstChild(self_node);
+                break :blk .{ self_node, first_child };
+            }
+
+            if (std.mem.eql(u8, position, "beforebegin")) {
+                // The node must have a parent node in order to use this variant.
+                const parent = parser.nodeParentNode(self_node) orelse return error.NoModificationAllowed;
+                // Parent cannot be Document.
+                // Should have checks for document_fragment and document_type?
+                if (parser.nodeType(parent) == .document) {
+                    return error.NoModificationAllowed;
+                }
+
+                break :blk .{ parent, self_node };
+            }
+
+            if (std.mem.eql(u8, position, "afterend")) {
+                // The node must have a parent node in order to use this variant.
+                const parent = parser.nodeParentNode(self_node) orelse return error.NoModificationAllowed;
+                // Parent cannot be Document.
+                if (parser.nodeType(parent) == .document) {
+                    return error.NoModificationAllowed;
+                }
+                // Get the next sibling or null; null indicates our node is the only one.
+                const sibling = parser.nodeNextSibling(self_node);
+                break :blk .{ parent, sibling };
+            }
+
+            // Thrown if:
+            // * position is not one of the four listed values.
+            // * The input is XML that is not well-formed.
+            return error.Syntax;
+        };
+
+        while (parser.nodeListItem(children, 0)) |child| {
+            _ = try parser.nodeInsertBefore(target_node, child, prev_node);
+        }
+    }
+
+    // The closest() method of the Element interface traverses the element and its parents (heading toward the document root) until it finds a node that matches the specified CSS selector.
+    // Returns the closest ancestor Element or itself, which matches the selectors. If there are no such element, null.
+    pub fn _closest(self: *parser.Element, selector: []const u8, page: *Page) !?*parser.Element {
+        const cssParse = @import("../css/css.zig").parse;
+        const CssNodeWrap = @import("../css/libdom.zig").Node;
+        const select = try cssParse(page.call_arena, selector, .{});
+
+        var current: CssNodeWrap = .{ .node = parser.elementToNode(self) };
+        while (true) {
+            if (try select.match(current)) {
+                if (!current.isElement()) {
+                    log.err(.browser, "closest invalid type", .{ .type = try current.tag() });
+                    return null;
+                }
+                return parser.nodeToElement(current.node);
+            }
+            current = current.parent() orelse return null;
+        }
+    }
+
+    // don't use parser.nodeHasAttributes(...) because that returns true/false
+    // based on the type, e.g. a node never as attributes, an element always has
+    // attributes. But, Element.hasAttributes is supposed to return true only
+    // if the element has at least 1 attribute.
+    pub fn _hasAttributes(self: *parser.Element) !bool {
+        // an element _must_ have at least an empty attribute
+        const node_map = try parser.nodeGetAttributes(parser.elementToNode(self)) orelse unreachable;
+        return try parser.namedNodeMapGetLength(node_map) > 0;
+    }
+
+    pub fn _getAttribute(self: *parser.Element, qname: []const u8) !?[]const u8 {
+        return try parser.elementGetAttribute(self, qname);
+    }
+
+    pub fn _getAttributeNS(self: *parser.Element, ns: []const u8, qname: []const u8) !?[]const u8 {
+        return try parser.elementGetAttributeNS(self, ns, qname);
+    }
+
+    pub fn _setAttribute(self: *parser.Element, qname: []const u8, value: []const u8) !void {
+        return try parser.elementSetAttribute(self, qname, value);
+    }
+
+    pub fn _setAttributeNS(self: *parser.Element, ns: []const u8, qname: []const u8, value: []const u8) !void {
+        return try parser.elementSetAttributeNS(self, ns, qname, value);
+    }
+
+    pub fn _removeAttribute(self: *parser.Element, qname: []const u8) !void {
+        return try parser.elementRemoveAttribute(self, qname);
+    }
+
+    pub fn _removeAttributeNS(self: *parser.Element, ns: []const u8, qname: []const u8) !void {
+        return try parser.elementRemoveAttributeNS(self, ns, qname);
+    }
+
+    pub fn _hasAttribute(self: *parser.Element, qname: []const u8) !bool {
+        return try parser.elementHasAttribute(self, qname);
+    }
+
+    pub fn _hasAttributeNS(self: *parser.Element, ns: []const u8, qname: []const u8) !bool {
+        return try parser.elementHasAttributeNS(self, ns, qname);
+    }
+
+    // https://dom.spec.whatwg.org/#dom-element-toggleattribute
+    pub fn _toggleAttribute(self: *parser.Element, qname: []u8, force: ?bool) !bool {
+        _ = std.ascii.lowerString(qname, qname);
+        const exists = try parser.elementHasAttribute(self, qname);
+
+        // If attribute is null, then:
+        if (!exists) {
+            // If force is not given or is true, create an attribute whose
+            // local name is qualifiedName, value is the empty string and node
+            // document is this’s node document, then append this attribute to
+            // this, and then return true.
+            if (force == null or force.?) {
+                try parser.elementSetAttribute(self, qname, "");
+                return true;
+            }
+            if (try parser.validateName(qname) == false) {
+                return parser.DOMError.InvalidCharacter;
+            }
+
+            // Return false.
+            return false;
+        }
+
+        // Otherwise, if force is not given or is false, remove an attribute
+        // given qualifiedName and this, and then return false.
+        if (force == null or !force.?) {
+            try parser.elementRemoveAttribute(self, qname);
+            return false;
+        }
+
+        // Return true.
+        return true;
+    }
+
+    pub fn _getAttributeNames(self: *parser.Element, page: *Page) ![]const []const u8 {
+        const attributes = try parser.nodeGetAttributes(@ptrCast(self)) orelse return &.{};
+        const ln = try parser.namedNodeMapGetLength(attributes);
+
+        const names = try page.call_arena.alloc([]const u8, ln);
+        var at: usize = 0;
+
+        for (0..ln) |i| {
+            const attribute = try parser.namedNodeMapItem(attributes, @intCast(i)) orelse break;
+            names[at] = try parser.attributeGetName(attribute);
+            at += 1;
+        }
+
+        return names[0..at];
+    }
+
+    pub fn _getAttributeNode(self: *parser.Element, name: []const u8) !?*parser.Attribute {
+        return try parser.elementGetAttributeNode(self, name);
+    }
+
+    pub fn _getAttributeNodeNS(self: *parser.Element, ns: []const u8, name: []const u8) !?*parser.Attribute {
+        return try parser.elementGetAttributeNodeNS(self, ns, name);
+    }
+
+    pub fn _setAttributeNode(self: *parser.Element, attr: *parser.Attribute) !?*parser.Attribute {
+        return try parser.elementSetAttributeNode(self, attr);
+    }
+
+    pub fn _setAttributeNodeNS(self: *parser.Element, attr: *parser.Attribute) !?*parser.Attribute {
+        return try parser.elementSetAttributeNodeNS(self, attr);
+    }
+
+    pub fn _removeAttributeNode(self: *parser.Element, attr: *parser.Attribute) !*parser.Attribute {
+        return try parser.elementRemoveAttributeNode(self, attr);
+    }
+
+    pub fn _getElementsByTagName(self: *parser.Element, tag_name: js.String) !collection.HTMLCollection {
+        return collection.HTMLCollectionByTagName(
+            parser.elementToNode(self),
+            tag_name.string,
+            .{ .include_root = false },
+        );
+    }
+
+    pub fn _getElementsByClassName(self: *parser.Element, class_names: js.String) !collection.HTMLCollection {
+        return try collection.HTMLCollectionByClassName(
+            parser.elementToNode(self),
+            class_names.string,
+            .{ .include_root = false },
+        );
+    }
+
+    // ParentNode
+    // https://dom.spec.whatwg.org/#parentnode
+    pub fn get_children(self: *parser.Element) !collection.HTMLCollection {
+        return collection.HTMLCollectionChildren(parser.elementToNode(self), .{
+            .include_root = false,
+        });
+    }
+
+    pub fn get_firstElementChild(self: *parser.Element) !?Union {
+        var children = try get_children(self);
+        return try children._item(0);
+    }
+
+    pub fn get_lastElementChild(self: *parser.Element) !?Union {
+        // TODO we could check the last child node first, if it's an element,
+        // we can return it directly instead of looping twice over the
+        // children.
+        var children = try get_children(self);
+        const ln = try children.get_length();
+        if (ln == 0) return null;
+        return try children._item(ln - 1);
+    }
+
+    pub fn get_childElementCount(self: *parser.Element) !u32 {
+        var children = try get_children(self);
+        return try children.get_length();
+    }
+
+    // NonDocumentTypeChildNode
+    // https://dom.spec.whatwg.org/#interface-nondocumenttypechildnode
+    pub fn get_previousElementSibling(self: *parser.Element) !?Union {
+        const res = parser.nodePreviousElementSibling(parser.elementToNode(self));
+        if (res == null) return null;
+        return try toInterface(res.?);
+    }
+
+    pub fn get_nextElementSibling(self: *parser.Element) !?Union {
+        const res = parser.nodeNextElementSibling(parser.elementToNode(self));
+        if (res == null) return null;
+        return try toInterface(res.?);
+    }
+
+    fn getElementById(self: *parser.Element, id: []const u8) !?*parser.Node {
+        // walk over the node tree fo find the node by id.
+        const root = parser.elementToNode(self);
+        const walker = Walker{};
+        var next: ?*parser.Node = null;
+        while (true) {
+            next = try walker.get_next(root, next) orelse return null;
+            // ignore non-element nodes.
+            if (parser.nodeType(next.?) != .element) {
+                continue;
+            }
+            const e = parser.nodeToElement(next.?);
+            if (std.mem.eql(u8, id, try get_id(e))) return next;
+        }
+    }
+
+    pub fn _querySelector(self: *parser.Element, selector: []const u8, page: *Page) !?Union {
+        if (selector.len == 0) return null;
+
+        const n = try css.querySelector(page.call_arena, parser.elementToNode(self), selector);
+
+        if (n == null) return null;
+
+        return try toInterface(parser.nodeToElement(n.?));
+    }
+
+    pub fn _querySelectorAll(self: *parser.Element, selector: []const u8, page: *Page) !NodeList {
+        return css.querySelectorAll(page.arena, parser.elementToNode(self), selector);
+    }
+
+    pub fn _prepend(self: *parser.Element, nodes: []const Node.NodeOrText) !void {
+        return Node.prepend(parser.elementToNode(self), nodes);
+    }
+
+    pub fn _append(self: *parser.Element, nodes: []const Node.NodeOrText) !void {
+        return Node.append(parser.elementToNode(self), nodes);
+    }
+
+    pub fn _before(self: *parser.Element, nodes: []const Node.NodeOrText) !void {
+        const ref_node = parser.elementToNode(self);
+        return Node.before(ref_node, nodes);
+    }
+
+    pub fn _after(self: *parser.Element, nodes: []const Node.NodeOrText) !void {
+        const ref_node = parser.elementToNode(self);
+        return Node.after(ref_node, nodes);
+    }
+
+    pub fn _replaceChildren(self: *parser.Element, nodes: []const Node.NodeOrText) !void {
+        return Node.replaceChildren(parser.elementToNode(self), nodes);
+    }
+
+    // A DOMRect object providing information about the size of an element and its position relative to the viewport.
+    // Returns a 0 DOMRect object if the element is eventually detached from the main window
+    pub fn _getBoundingClientRect(self: *parser.Element, page: *Page) !DOMRect {
+        // Since we are lazy rendering we need to do this check. We could store the renderer in a viewport such that it could cache these, but it would require tracking changes.
+        if (!page.isNodeAttached(parser.elementToNode(self))) {
+            return DOMRect{
+                .x = 0,
+                .y = 0,
+                .width = 0,
+                .height = 0,
+                .bottom = 0,
+                .right = 0,
+                .top = 0,
+                .left = 0,
+            };
+        }
+        return page.renderer.getRect(self);
+    }
+
+    // Returns a collection of DOMRect objects that indicate the bounding rectangles for each CSS border box in a client.
+    // We do not render so it only always return the element's bounding rect.
+    // Returns an empty array if the element is eventually detached from the main window
+    pub fn _getClientRects(self: *parser.Element, page: *Page) ![]DOMRect {
+        if (!page.isNodeAttached(parser.elementToNode(self))) {
+            return &.{};
+        }
+        const heap_ptr = try page.call_arena.create(DOMRect);
+        heap_ptr.* = try page.renderer.getRect(self);
+        return heap_ptr[0..1];
+    }
+
+    pub fn get_clientWidth(_: *parser.Element, page: *Page) u32 {
+        return page.renderer.width();
+    }
+
+    pub fn get_clientHeight(_: *parser.Element, page: *Page) u32 {
+        return page.renderer.height();
+    }
+
+    pub fn _matches(self: *parser.Element, selectors: []const u8, page: *Page) !bool {
+        const cssParse = @import("../css/css.zig").parse;
+        const CssNodeWrap = @import("../css/libdom.zig").Node;
+        const s = try cssParse(page.call_arena, selectors, .{});
+        return s.match(CssNodeWrap{ .node = parser.elementToNode(self) });
+    }
+
+    pub fn _scrollIntoViewIfNeeded(_: *parser.Element, center_if_needed: ?bool) void {
+        _ = center_if_needed;
+    }
+
+    const CheckVisibilityOpts = struct {
+        contentVisibilityAuto: bool,
+        opacityProperty: bool,
+        visibilityProperty: bool,
+        checkVisibilityCSS: bool,
+        checkOpacity: bool,
+    };
+
+    pub fn _checkVisibility(self: *parser.Element, opts: ?CheckVisibilityOpts) bool {
+        _ = self;
+        _ = opts;
+        return true;
+    }
+
+    const AttachShadowOpts = struct {
+        mode: []const u8, // must be specified
+    };
+    pub fn _attachShadow(self: *parser.Element, opts: AttachShadowOpts, page: *Page) !*ShadowRoot {
+        const mode = std.meta.stringToEnum(ShadowRoot.Mode, opts.mode) orelse return error.InvalidArgument;
+        const state = try page.getOrCreateNodeState(@ptrCast(@alignCast(self)));
+        if (state.shadow_root) |sr| {
+            if (mode != sr.mode) {
+                // this is the behavior per the spec
+                return error.NotSupportedError;
+            }
+
+            try Node.removeChildren(@ptrCast(@alignCast(sr.proto)));
+            return sr;
+        }
+
+        // Not sure what to do if there is no owner document
+        const doc = parser.nodeOwnerDocument(@ptrCast(self)) orelse return error.InvalidArgument;
+        const fragment = try parser.documentCreateDocumentFragment(doc);
+        const sr = try page.arena.create(ShadowRoot);
+        sr.* = .{
+            .host = self,
+            .mode = mode,
+            .proto = fragment,
+        };
+        state.shadow_root = sr;
+        parser.documentFragmentSetHost(sr.proto, @ptrCast(@alignCast(self)));
+
+        // Storing the ShadowRoot on the element makes sense, it's the ShadowRoot's
+        // parent. When we render, we go top-down, so we'll have the element, get
+        // its shadowroot, and go on. that's what the above code does.
+        // But we sometimes need to go bottom-up, e.g when we have a slot element
+        // and want to find the containing parent. Unforatunately , we don't have
+        // that link, so we need to create it. In the DOM, the ShadowRoot is
+        // represented by this DocumentFragment (it's the ShadowRoot's base prototype)
+        // So we can also store the ShadowRoot in the DocumentFragment's state.
+        const fragment_state = try page.getOrCreateNodeState(@ptrCast(@alignCast(fragment)));
+        fragment_state.shadow_root = sr;
+
+        return sr;
+    }
+
+    pub fn get_shadowRoot(self: *parser.Element, page: *Page) ?*ShadowRoot {
+        const state = page.getNodeState(@ptrCast(@alignCast(self))) orelse return null;
+        const sr = state.shadow_root orelse return null;
+        if (sr.mode == .closed) {
+            return null;
+        }
+        return sr;
+    }
+
+    pub fn _animate(self: *parser.Element, effect: js.Object, opts: js.Object) !Animation {
+        _ = self;
+        _ = opts;
+        return Animation.constructor(effect, null);
+    }
+
+    pub fn _remove(self: *parser.Element) !void {
+        // TODO: This hasn't been tested to make sure all references to this
+        // node are properly updated. A lot of libdom is lazy and will look
+        // for related elements JIT by walking the tree, but there could be
+        // cases in libdom or the Zig WebAPI where this reference is kept
+        const as_node: *parser.Node = @ptrCast(self);
+        const parent = parser.nodeParentNode(as_node) orelse return;
+        _ = try Node._removeChild(parent, as_node);
+    }
+};
+
+// Tests
+// -----
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Element" {
+    try testing.htmlRunner("dom/element.html");
+}

src/browser/dom/event_target.zig

@@ -0,0 +1,175 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const parser = @import("../netsurf.zig");
+const Page = @import("../page.zig").Page;
+
+const EventHandler = @import("../events/event.zig").EventHandler;
+
+const DOMException = @import("exceptions.zig").DOMException;
+const nod = @import("node.zig");
+
+pub const Union = union(enum) {
+    node: nod.Union,
+    xhr: *@import("../xhr/xhr.zig").XMLHttpRequest,
+    plain: *parser.EventTarget,
+    message_port: *@import("MessageChannel.zig").MessagePort,
+    screen: *@import("../html/screen.zig").Screen,
+    screen_orientation: *@import("../html/screen.zig").ScreenOrientation,
+    performance: *@import("performance.zig").Performance,
+    media_query_list: *@import("../html/media_query_list.zig").MediaQueryList,
+    navigation: *@import("../navigation/Navigation.zig"),
+};
+
+// EventTarget implementation
+pub const EventTarget = struct {
+    pub const Self = parser.EventTarget;
+    pub const Exception = DOMException;
+
+    // Extend libdom event target for pure zig struct.
+    base: parser.EventTargetTBase = parser.EventTargetTBase{ .internal_target_type = .plain },
+
+    pub fn toInterface(et: *parser.EventTarget, page: *Page) !Union {
+        // libdom assumes that all event targets are libdom nodes. They are not.
+
+        switch (parser.eventTargetInternalType(et)) {
+            .libdom_node => {
+                return .{ .node = try nod.Node.toInterface(@as(*parser.Node, @ptrCast(et))) };
+            },
+            .plain => return .{ .plain = et },
+            .abort_signal => {
+                // AbortSignal is a special case, it has its own internal type.
+                // We return it as a node, but we need to handle it differently.
+                return .{ .node = .{ .AbortSignal = @fieldParentPtr("proto", @as(*parser.EventTargetTBase, @ptrCast(et))) } };
+            },
+            .window => {
+                // The window is a common non-node target, but it's easy to handle as its a singleton.
+                std.debug.assert(@intFromPtr(et) == @intFromPtr(&page.window.base));
+                return .{ .node = .{ .Window = &page.window } };
+            },
+            .xhr => {
+                const XMLHttpRequestEventTarget = @import("../xhr/event_target.zig").XMLHttpRequestEventTarget;
+                const base: *XMLHttpRequestEventTarget = @fieldParentPtr("base", @as(*parser.EventTargetTBase, @ptrCast(et)));
+                return .{ .xhr = @fieldParentPtr("proto", base) };
+            },
+            .message_port => {
+                return .{ .message_port = @fieldParentPtr("proto", @as(*parser.EventTargetTBase, @ptrCast(et))) };
+            },
+            .screen => {
+                return .{ .screen = @fieldParentPtr("proto", @as(*parser.EventTargetTBase, @ptrCast(et))) };
+            },
+            .screen_orientation => {
+                return .{ .screen_orientation = @fieldParentPtr("proto", @as(*parser.EventTargetTBase, @ptrCast(et))) };
+            },
+            .performance => {
+                return .{ .performance = @fieldParentPtr("base", @as(*parser.EventTargetTBase, @ptrCast(et))) };
+            },
+            .media_query_list => {
+                return .{ .media_query_list = @fieldParentPtr("base", @as(*parser.EventTargetTBase, @ptrCast(et))) };
+            },
+            .navigation => {
+                const NavigationEventTarget = @import("../navigation/NavigationEventTarget.zig");
+                const base: *NavigationEventTarget = @fieldParentPtr("base", @as(*parser.EventTargetTBase, @ptrCast(et)));
+                return .{ .navigation = @fieldParentPtr("proto", base) };
+            },
+        }
+    }
+
+    // JS funcs
+    // --------
+    pub fn constructor(page: *Page) !*parser.EventTarget {
+        const et = try page.arena.create(EventTarget);
+        et.* = .{};
+        return @ptrCast(&et.base);
+    }
+
+    pub fn _addEventListener(
+        self: *parser.EventTarget,
+        typ: []const u8,
+        listener: EventHandler.Listener,
+        opts: ?EventHandler.Opts,
+        page: *Page,
+    ) !void {
+        _ = try EventHandler.register(page.arena, self, typ, listener, opts);
+        if (std.mem.eql(u8, typ, "slotchange")) {
+            try page.registerSlotChangeMonitor();
+        }
+    }
+
+    const RemoveEventListenerOpts = union(enum) {
+        opts: Opts,
+        capture: bool,
+
+        const Opts = struct {
+            capture: ?bool,
+        };
+    };
+
+    pub fn _removeEventListener(
+        self: *parser.EventTarget,
+        typ: []const u8,
+        listener: EventHandler.Listener,
+        opts_: ?RemoveEventListenerOpts,
+    ) !void {
+        var capture = false;
+        if (opts_) |opts| {
+            capture = switch (opts) {
+                .capture => |c| c,
+                .opts => |o| o.capture orelse false,
+            };
+        }
+
+        const cbk = (try listener.callback(self)) orelse return;
+
+        // check if event target has already this listener
+        const lst = try parser.eventTargetHasListener(
+            self,
+            typ,
+            capture,
+            cbk.id,
+        );
+        if (lst == null) {
+            return;
+        }
+
+        // remove listener
+        try parser.eventTargetRemoveEventListener(
+            self,
+            typ,
+            lst.?,
+            capture,
+        );
+    }
+
+    pub fn _dispatchEvent(self: *parser.EventTarget, event: *parser.Event, page: *Page) !bool {
+        const res = try parser.eventTargetDispatchEvent(self, event);
+
+        if (!parser.eventBubbles(event) or parser.eventIsStopped(event)) {
+            return res;
+        }
+
+        try page.window.dispatchForDocumentTarget(event);
+        return true;
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.EventTarget" {
+    try testing.htmlRunner("dom/event_target.html");
+}

src/browser/dom/exceptions.zig

@@ -0,0 +1,224 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const allocPrint = std.fmt.allocPrint;
+
+const parser = @import("../netsurf.zig");
+const Page = @import("../page.zig").Page;
+
+// https://webidl.spec.whatwg.org/#idl-DOMException
+pub const DOMException = struct {
+    err: ?parser.DOMError,
+    str: []const u8,
+
+    pub const ErrorSet = parser.DOMError;
+
+    // static attributes
+    pub const _INDEX_SIZE_ERR = 1;
+    pub const _DOMSTRING_SIZE_ERR = 2;
+    pub const _HIERARCHY_REQUEST_ERR = 3;
+    pub const _WRONG_DOCUMENT_ERR = 4;
+    pub const _INVALID_CHARACTER_ERR = 5;
+    pub const _NO_DATA_ALLOWED_ERR = 6;
+    pub const _NO_MODIFICATION_ALLOWED_ERR = 7;
+    pub const _NOT_FOUND_ERR = 8;
+    pub const _NOT_SUPPORTED_ERR = 9;
+    pub const _INUSE_ATTRIBUTE_ERR = 10;
+    pub const _INVALID_STATE_ERR = 11;
+    pub const _SYNTAX_ERR = 12;
+    pub const _INVALID_MODIFICATION_ERR = 13;
+    pub const _NAMESPACE_ERR = 14;
+    pub const _INVALID_ACCESS_ERR = 15;
+    pub const _VALIDATION_ERR = 16;
+    pub const _TYPE_MISMATCH_ERR = 17;
+    pub const _SECURITY_ERR = 18;
+    pub const _NETWORK_ERR = 19;
+    pub const _ABORT_ERR = 20;
+    pub const _URL_MISMATCH_ERR = 21;
+    pub const _QUOTA_EXCEEDED_ERR = 22;
+    pub const _TIMEOUT_ERR = 23;
+    pub const _INVALID_NODE_TYPE_ERR = 24;
+    pub const _DATA_CLONE_ERR = 25;
+
+    pub fn constructor(message_: ?[]const u8, name_: ?[]const u8, page: *const Page) !DOMException {
+        const message = message_ orelse "";
+        const err = if (name_) |n| error_from_str(n) else null;
+        const fixed_name = name(err);
+
+        if (message.len == 0) return .{ .err = err, .str = fixed_name };
+
+        const str = try allocPrint(page.arena, "{s}: {s}", .{ fixed_name, message });
+        return .{ .err = err, .str = str };
+    }
+
+    // TODO: deinit
+    pub fn init(alloc: std.mem.Allocator, err: anyerror, caller_name: []const u8) !DOMException {
+        const dom_error = @as(parser.DOMError, @errorCast(err));
+        const error_name = DOMException.name(dom_error);
+        const str = switch (dom_error) {
+            error.HierarchyRequest => try allocPrint(
+                alloc,
+                "{s}: Failed to execute '{s}' on 'Node': The new child element contains the parent.",
+                .{ error_name, caller_name },
+            ),
+            // todo add more custom error messages
+            else => try allocPrint(
+                alloc,
+                "{s}: Failed to execute '{s}' : {s}",
+                .{ error_name, caller_name, error_name },
+            ),
+            error.NoError => unreachable,
+        };
+        return .{ .err = dom_error, .str = str };
+    }
+
+    fn error_from_str(name_: []const u8) ?parser.DOMError {
+        // @speed: Consider length first, left as is for maintainability, awaiting switch on string support
+        if (std.mem.eql(u8, name_, "IndexSizeError")) return error.IndexSize;
+        if (std.mem.eql(u8, name_, "StringSizeError")) return error.StringSize;
+        if (std.mem.eql(u8, name_, "HierarchyRequestError")) return error.HierarchyRequest;
+        if (std.mem.eql(u8, name_, "WrongDocumentError")) return error.WrongDocument;
+        if (std.mem.eql(u8, name_, "InvalidCharacterError")) return error.InvalidCharacter;
+        if (std.mem.eql(u8, name_, "NoDataAllowedError")) return error.NoDataAllowed;
+        if (std.mem.eql(u8, name_, "NoModificationAllowedError")) return error.NoModificationAllowed;
+        if (std.mem.eql(u8, name_, "NotFoundError")) return error.NotFound;
+        if (std.mem.eql(u8, name_, "NotSupportedError")) return error.NotSupported;
+        if (std.mem.eql(u8, name_, "InuseAttributeError")) return error.InuseAttribute;
+        if (std.mem.eql(u8, name_, "InvalidStateError")) return error.InvalidState;
+        if (std.mem.eql(u8, name_, "SyntaxError")) return error.Syntax;
+        if (std.mem.eql(u8, name_, "InvalidModificationError")) return error.InvalidModification;
+        if (std.mem.eql(u8, name_, "NamespaceError")) return error.Namespace;
+        if (std.mem.eql(u8, name_, "InvalidAccessError")) return error.InvalidAccess;
+        if (std.mem.eql(u8, name_, "ValidationError")) return error.Validation;
+        if (std.mem.eql(u8, name_, "TypeMismatchError")) return error.TypeMismatch;
+        if (std.mem.eql(u8, name_, "SecurityError")) return error.Security;
+        if (std.mem.eql(u8, name_, "NetworkError")) return error.Network;
+        if (std.mem.eql(u8, name_, "AbortError")) return error.Abort;
+        if (std.mem.eql(u8, name_, "URLismatchError")) return error.URLismatch;
+        if (std.mem.eql(u8, name_, "QuotaExceededError")) return error.QuotaExceeded;
+        if (std.mem.eql(u8, name_, "TimeoutError")) return error.Timeout;
+        if (std.mem.eql(u8, name_, "InvalidNodeTypeError")) return error.InvalidNodeType;
+        if (std.mem.eql(u8, name_, "DataCloneError")) return error.DataClone;
+
+        // custom netsurf error
+        if (std.mem.eql(u8, name_, "UnspecifiedEventTypeError")) return error.UnspecifiedEventType;
+        if (std.mem.eql(u8, name_, "DispatchRequestError")) return error.DispatchRequest;
+        if (std.mem.eql(u8, name_, "NoMemoryError")) return error.NoMemory;
+        if (std.mem.eql(u8, name_, "AttributeWrongTypeError")) return error.AttributeWrongType;
+        return null;
+    }
+
+    fn name(err_: ?parser.DOMError) []const u8 {
+        const err = err_ orelse return "Error";
+
+        return switch (err) {
+            error.IndexSize => "IndexSizeError",
+            error.StringSize => "StringSizeError", // Legacy: DOMSTRING_SIZE_ERR
+            error.HierarchyRequest => "HierarchyRequestError",
+            error.WrongDocument => "WrongDocumentError",
+            error.InvalidCharacter => "InvalidCharacterError",
+            error.NoDataAllowed => "NoDataAllowedError", // Legacy: NO_DATA_ALLOWED_ERR
+            error.NoModificationAllowed => "NoModificationAllowedError",
+            error.NotFound => "NotFoundError",
+            error.NotSupported => "NotSupportedError",
+            error.InuseAttribute => "InuseAttributeError",
+            error.InvalidState => "InvalidStateError",
+            error.Syntax => "SyntaxError",
+            error.InvalidModification => "InvalidModificationError",
+            error.Namespace => "NamespaceError",
+            error.InvalidAccess => "InvalidAccessError",
+            error.Validation => "ValidationError", // Legacy: VALIDATION_ERR
+            error.TypeMismatch => "TypeMismatchError",
+            error.Security => "SecurityError",
+            error.Network => "NetworkError",
+            error.Abort => "AbortError",
+            error.URLismatch => "URLismatchError",
+            error.QuotaExceeded => "QuotaExceededError",
+            error.Timeout => "TimeoutError",
+            error.InvalidNodeType => "InvalidNodeTypeError",
+            error.DataClone => "DataCloneError",
+            error.NoError => unreachable,
+
+            // custom netsurf error
+            error.UnspecifiedEventType => "UnspecifiedEventTypeError",
+            error.DispatchRequest => "DispatchRequestError",
+            error.NoMemory => "NoMemoryError",
+            error.AttributeWrongType => "AttributeWrongTypeError",
+        };
+    }
+
+    // JS properties and methods
+
+    pub fn get_code(self: *const DOMException) u8 {
+        const err = self.err orelse return 0;
+        return switch (err) {
+            error.IndexSize => 1,
+            error.StringSize => 2,
+            error.HierarchyRequest => 3,
+            error.WrongDocument => 4,
+            error.InvalidCharacter => 5,
+            error.NoDataAllowed => 6,
+            error.NoModificationAllowed => 7,
+            error.NotFound => 8,
+            error.NotSupported => 9,
+            error.InuseAttribute => 10,
+            error.InvalidState => 11,
+            error.Syntax => 12,
+            error.InvalidModification => 13,
+            error.Namespace => 14,
+            error.InvalidAccess => 15,
+            error.Validation => 16,
+            error.TypeMismatch => 17,
+            error.Security => 18,
+            error.Network => 19,
+            error.Abort => 20,
+            error.URLismatch => 21,
+            error.QuotaExceeded => 22,
+            error.Timeout => 23,
+            error.InvalidNodeType => 24,
+            error.DataClone => 25,
+            error.NoError => unreachable,
+
+            // custom netsurf error
+            error.UnspecifiedEventType => 128,
+            error.DispatchRequest => 129,
+            error.NoMemory => 130,
+            error.AttributeWrongType => 131,
+        };
+    }
+
+    pub fn get_name(self: *const DOMException) []const u8 {
+        return DOMException.name(self.err);
+    }
+
+    pub fn get_message(self: *const DOMException) []const u8 {
+        const errName = DOMException.name(self.err);
+        if (self.str.len <= errName.len + 2) return "";
+        return self.str[errName.len + 2 ..]; // ! Requires str is formatted as "{name}: {message}"
+    }
+
+    pub fn _toString(self: *const DOMException) []const u8 {
+        return self.str;
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Exceptions" {
+    try testing.htmlRunner("dom/exceptions.html");
+}

src/browser/dom/html_collection.zig

@@ -0,0 +1,454 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const Allocator = std.mem.Allocator;
+
+const parser = @import("../netsurf.zig");
+
+const Element = @import("element.zig").Element;
+const Union = @import("element.zig").Union;
+const Walker = @import("walker.zig").Walker;
+
+const Matcher = union(enum) {
+    matchByName: MatchByName,
+    matchByTagName: MatchByTagName,
+    matchByClassName: MatchByClassName,
+    matchByLinks: MatchByLinks,
+    matchByAnchors: MatchByAnchors,
+    matchTrue: struct {},
+    matchFalse: struct {},
+
+    pub fn match(self: Matcher, node: *parser.Node) !bool {
+        switch (self) {
+            .matchTrue => return true,
+            .matchFalse => return false,
+            .matchByLinks => return MatchByLinks.match(node),
+            .matchByAnchors => return MatchByAnchors.match(node),
+            inline else => |m| return m.match(node),
+        }
+    }
+};
+
+pub const MatchByTagName = struct {
+    // tag is used to select node against their name.
+    // tag comparison is case insensitive.
+    tag: []const u8,
+    is_wildcard: bool,
+
+    fn init(tag_name: []const u8) MatchByTagName {
+        if (std.mem.eql(u8, tag_name, "*")) {
+            return .{ .tag = "*", .is_wildcard = true };
+        }
+
+        return .{
+            .tag = tag_name,
+            .is_wildcard = false,
+        };
+    }
+
+    pub fn match(self: MatchByTagName, node: *parser.Node) !bool {
+        return self.is_wildcard or std.ascii.eqlIgnoreCase(self.tag, try parser.nodeName(node));
+    }
+};
+
+pub fn HTMLCollectionByTagName(
+    root: ?*parser.Node,
+    tag_name: []const u8,
+    opts: Opts,
+) HTMLCollection {
+    return .{
+        .root = root,
+        .walker = .{ .walkerDepthFirst = .{} },
+        .matcher = .{ .matchByTagName = MatchByTagName.init(tag_name) },
+        .mutable = opts.mutable,
+        .include_root = opts.include_root,
+    };
+}
+
+pub const MatchByClassName = struct {
+    class_names: []const u8,
+
+    fn init(class_names: []const u8) !MatchByClassName {
+        return .{
+            .class_names = class_names,
+        };
+    }
+
+    pub fn match(self: MatchByClassName, node: *parser.Node) !bool {
+        const e = parser.nodeToElement(node);
+
+        var it = std.mem.splitScalar(u8, self.class_names, ' ');
+        while (it.next()) |c| {
+            if (!try parser.elementHasClass(e, c)) {
+                return false;
+            }
+        }
+
+        return true;
+    }
+};
+
+pub fn HTMLCollectionByClassName(
+    root: ?*parser.Node,
+    class_names: []const u8,
+    opts: Opts,
+) !HTMLCollection {
+    return HTMLCollection{
+        .root = root,
+        .walker = .{ .walkerDepthFirst = .{} },
+        .matcher = .{ .matchByClassName = try MatchByClassName.init(class_names) },
+        .mutable = opts.mutable,
+        .include_root = opts.include_root,
+    };
+}
+
+pub const MatchByName = struct {
+    name: []const u8,
+
+    fn init(name: []const u8) !MatchByName {
+        return .{ .name = name };
+    }
+
+    pub fn match(self: MatchByName, node: *parser.Node) !bool {
+        const e = parser.nodeToElement(node);
+        const nname = try parser.elementGetAttribute(e, "name") orelse return false;
+        return std.mem.eql(u8, self.name, nname);
+    }
+};
+
+pub fn HTMLCollectionByName(
+    root: ?*parser.Node,
+    name: []const u8,
+    opts: Opts,
+) !HTMLCollection {
+    return HTMLCollection{
+        .root = root,
+        .walker = .{ .walkerDepthFirst = .{} },
+        .matcher = .{ .matchByName = try MatchByName.init(name) },
+        .mutable = opts.mutable,
+        .include_root = opts.include_root,
+    };
+}
+
+// HTMLAllCollection is a special type: instances of it are falsy. It's the only
+// object in the WebAPI that behaves like this - in fact, it's even a special
+// case in the JavaScript spec.
+// This is important, because a lot of browser detection rely on this behavior
+// to determine what browser is running.
+
+// It's also possible to use an instance like a function:
+//   document.all(3)
+//   document.all('some_id')
+pub const HTMLAllCollection = struct {
+    pub const prototype = *HTMLCollection;
+
+    proto: HTMLCollection,
+
+    pub const mark_as_undetectable = true;
+
+    pub fn init(root: ?*parser.Node) HTMLAllCollection {
+        return .{ .proto = .{
+            .root = root,
+            .walker = .{ .walkerDepthFirst = .{} },
+            .matcher = .{ .matchTrue = .{} },
+            .include_root = true,
+        } };
+    }
+
+    const CAllAsFunctionArg = union(enum) {
+        index: u32,
+        id: []const u8,
+    };
+
+    pub fn jsCallAsFunction(self: *HTMLAllCollection, arg: CAllAsFunctionArg) !?Union {
+        return switch (arg) {
+            .index => |i| self.proto._item(i),
+            .id => |id| self.proto._namedItem(id),
+        };
+    }
+};
+
+pub fn HTMLCollectionChildren(
+    root: ?*parser.Node,
+    opts: Opts,
+) HTMLCollection {
+    return HTMLCollection{
+        .root = root,
+        .walker = .{ .walkerChildren = .{} },
+        .matcher = .{ .matchTrue = .{} },
+        .mutable = opts.mutable,
+        .include_root = opts.include_root,
+    };
+}
+
+pub fn HTMLCollectionEmpty() HTMLCollection {
+    return .{
+        .root = null,
+        .walker = .{ .walkerNone = .{} },
+        .matcher = .{ .matchFalse = .{} },
+        .include_root = false,
+    };
+}
+
+// MatchByLinks matches the a and area elements in the Document that have href
+// attributes.
+// https://html.spec.whatwg.org/#dom-document-links
+pub const MatchByLinks = struct {
+    pub fn match(node: *parser.Node) !bool {
+        const tag = try parser.nodeName(node);
+        if (!std.ascii.eqlIgnoreCase(tag, "a") and !std.ascii.eqlIgnoreCase(tag, "area")) {
+            return false;
+        }
+        const elem = @as(*parser.Element, @ptrCast(node));
+        return parser.elementHasAttribute(elem, "href");
+    }
+};
+
+pub fn HTMLCollectionByLinks(root: ?*parser.Node, opts: Opts) HTMLCollection {
+    return .{
+        .root = root,
+        .walker = .{ .walkerDepthFirst = .{} },
+        .matcher = .{ .matchByLinks = .{} },
+        .mutable = opts.mutable,
+        .include_root = opts.include_root,
+    };
+}
+
+// MatchByAnchors matches the a elements in the Document that have name
+// attributes.
+// https://html.spec.whatwg.org/#dom-document-anchors
+pub const MatchByAnchors = struct {
+    pub fn match(node: *parser.Node) !bool {
+        const tag = try parser.nodeName(node);
+        if (!std.ascii.eqlIgnoreCase(tag, "a")) return false;
+
+        const elem = @as(*parser.Element, @ptrCast(node));
+        return parser.elementHasAttribute(elem, "name");
+    }
+};
+
+pub fn HTMLCollectionByAnchors(root: ?*parser.Node, opts: Opts) HTMLCollection {
+    return .{
+        .root = root,
+        .walker = .{ .walkerDepthFirst = .{} },
+        .matcher = .{ .matchByAnchors = .{} },
+        .mutable = opts.mutable,
+        .include_root = opts.include_root,
+    };
+}
+
+pub const HTMLCollectionIterator = struct {
+    coll: *HTMLCollection,
+    index: u32 = 0,
+
+    pub const Return = struct {
+        value: ?Union,
+        done: bool,
+    };
+
+    pub fn _next(self: *HTMLCollectionIterator) !Return {
+        const e = try self.coll._item(self.index);
+        if (e == null) {
+            return Return{
+                .value = null,
+                .done = true,
+            };
+        }
+
+        self.index += 1;
+        return Return{
+            .value = e,
+            .done = false,
+        };
+    }
+};
+
+const Opts = struct {
+    include_root: bool,
+    mutable: bool = false,
+};
+
+// WEB IDL https://dom.spec.whatwg.org/#htmlcollection
+// HTMLCollection is re implemented in zig here because libdom
+// dom_html_collection expects a comparison function callback as argument.
+// But we wanted a dynamically comparison here, according to the match tagname.
+pub const HTMLCollection = struct {
+    matcher: Matcher,
+    walker: Walker,
+
+    root: ?*parser.Node,
+
+    // By default the HTMLCollection walk on the root's descendant only.
+    // But on somes cases, like for dom document, we want to walk over the root
+    // itself.
+    include_root: bool = false,
+
+    mutable: bool = false,
+
+    // save a state for the collection to improve the _item speed.
+    cur_idx: ?u32 = null,
+    cur_node: ?*parser.Node = null,
+
+    // start returns the first node to walk on.
+    fn start(self: *const HTMLCollection) !?*parser.Node {
+        if (self.root == null) return null;
+
+        if (self.include_root) {
+            return self.root.?;
+        }
+
+        return try self.walker.get_next(self.root.?, null);
+    }
+
+    pub fn _symbol_iterator(self: *HTMLCollection) HTMLCollectionIterator {
+        return HTMLCollectionIterator{
+            .coll = self,
+        };
+    }
+
+    /// get_length computes the collection's length dynamically according to
+    /// the current root structure.
+    // TODO: nodes retrieved must be de-referenced.
+    pub fn get_length(self: *HTMLCollection) !u32 {
+        if (self.root == null) return 0;
+
+        var len: u32 = 0;
+        var node = try self.start() orelse return 0;
+
+        while (true) {
+            if (parser.nodeType(node) == .element) {
+                if (try self.matcher.match(node)) {
+                    len += 1;
+                }
+            }
+
+            node = try self.walker.get_next(self.root.?, node) orelse break;
+        }
+
+        return len;
+    }
+
+    pub fn item(self: *HTMLCollection, index: u32) !?*parser.Node {
+        if (self.root == null) return null;
+
+        var i: u32 = 0;
+        var node: *parser.Node = undefined;
+
+        // Use the current state to improve speed if possible.
+        if (self.mutable == false and self.cur_idx != null and index >= self.cur_idx.?) {
+            i = self.cur_idx.?;
+            node = self.cur_node.?;
+        } else {
+            node = try self.start() orelse return null;
+        }
+
+        while (true) {
+            if (parser.nodeType(node) == .element) {
+                if (try self.matcher.match(node)) {
+                    // check if we found the searched element.
+                    if (i == index) {
+                        // save the current state
+                        self.cur_node = node;
+                        self.cur_idx = i;
+
+                        return node;
+                    }
+
+                    i += 1;
+                }
+            }
+
+            node = try self.walker.get_next(self.root.?, node) orelse break;
+        }
+
+        return null;
+    }
+
+    pub fn _item(self: *HTMLCollection, index: u32) !?Union {
+        const node = try self.item(index) orelse return null;
+        const e = @as(*parser.Element, @ptrCast(node));
+        return try Element.toInterface(e);
+    }
+
+    pub fn _namedItem(self: *const HTMLCollection, name: []const u8) !?Union {
+        if (self.root == null) return null;
+        if (name.len == 0) return null;
+
+        var node = try self.start() orelse return null;
+
+        while (true) {
+            if (parser.nodeType(node) == .element) {
+                if (try self.matcher.match(node)) {
+                    const elem = @as(*parser.Element, @ptrCast(node));
+
+                    var attr = try parser.elementGetAttribute(elem, "id");
+                    // check if the node id corresponds to the name argument.
+                    if (attr != null and std.mem.eql(u8, name, attr.?)) {
+                        return try Element.toInterface(elem);
+                    }
+
+                    attr = try parser.elementGetAttribute(elem, "name");
+                    // check if the node id corresponds to the name argument.
+                    if (attr != null and std.mem.eql(u8, name, attr.?)) {
+                        return try Element.toInterface(elem);
+                    }
+                }
+            }
+
+            node = try self.walker.get_next(self.root.?, node) orelse break;
+        }
+
+        return null;
+    }
+
+    fn item_name(elt: *parser.Element) !?[]const u8 {
+        if (try parser.elementGetAttribute(elt, "id")) |v| {
+            return v;
+        }
+        if (try parser.elementGetAttribute(elt, "name")) |v| {
+            return v;
+        }
+
+        return null;
+    }
+
+    pub fn indexed_get(self: *HTMLCollection, index: u32, has_value: *bool) !?Union {
+        return (try _item(self, index)) orelse {
+            has_value.* = false;
+            return undefined;
+        };
+    }
+
+    pub fn named_get(self: *const HTMLCollection, name: []const u8, has_value: *bool) !?Union {
+        // Even though an entry might have an empty id, the spec says
+        // that namedItem("") should always return null
+        if (name.len == 0) {
+            return null;
+        }
+        return (try _namedItem(self, name)) orelse {
+            has_value.* = false;
+            return undefined;
+        };
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.HTMLCollection" {
+    try testing.htmlRunner("dom/html_collection.html");
+}

src/browser/dom/implementation.zig

@@ -0,0 +1,56 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const parser = @import("../netsurf.zig");
+const DOMException = @import("exceptions.zig").DOMException;
+
+// WEB IDL https://dom.spec.whatwg.org/#domimplementation
+pub const DOMImplementation = struct {
+    pub const Exception = DOMException;
+
+    pub fn _createDocumentType(
+        _: *DOMImplementation,
+        qname: [:0]const u8,
+        publicId: [:0]const u8,
+        systemId: [:0]const u8,
+    ) !*parser.DocumentType {
+        return try parser.domImplementationCreateDocumentType(qname, publicId, systemId);
+    }
+
+    pub fn _createDocument(
+        _: *DOMImplementation,
+        namespace: ?[:0]const u8,
+        qname: ?[:0]const u8,
+        doctype: ?*parser.DocumentType,
+    ) !*parser.Document {
+        return try parser.domImplementationCreateDocument(namespace, qname, doctype);
+    }
+
+    pub fn _createHTMLDocument(_: *DOMImplementation, title: ?[]const u8) !*parser.DocumentHTML {
+        return try parser.domImplementationCreateHTMLDocument(title);
+    }
+
+    pub fn _hasFeature(_: *DOMImplementation) bool {
+        return true;
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Implementation" {
+    try testing.htmlRunner("dom/implementation.html");
+}

src/browser/dom/mutation_observer.zig

@@ -0,0 +1,407 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const js = @import("../js/js.zig");
+const log = @import("../../log.zig");
+const parser = @import("../netsurf.zig");
+const Page = @import("../page.zig").Page;
+
+const NodeList = @import("nodelist.zig").NodeList;
+
+pub const Interfaces = .{
+    MutationObserver,
+    MutationRecord,
+};
+
+const Walker = @import("../dom/walker.zig").WalkerChildren;
+
+// WEB IDL https://dom.spec.whatwg.org/#interface-mutationobserver
+pub const MutationObserver = struct {
+    page: *Page,
+    cbk: js.Function,
+    scheduled: bool,
+    observers: std.ArrayListUnmanaged(*Observer),
+
+    // List of records which were observed. When the call scope ends, we need to
+    // execute our callback with it.
+    observed: std.ArrayListUnmanaged(MutationRecord),
+
+    pub fn constructor(cbk: js.Function, page: *Page) !MutationObserver {
+        return .{
+            .cbk = cbk,
+            .page = page,
+            .observed = .{},
+            .scheduled = false,
+            .observers = .empty,
+        };
+    }
+
+    pub fn _observe(self: *MutationObserver, node: *parser.Node, options_: ?Options) !void {
+        const arena = self.page.arena;
+        var options = options_ orelse Options{};
+        if (options.attributeFilter.len > 0) {
+            options.attributeFilter = try arena.dupe([]const u8, options.attributeFilter);
+        }
+
+        const observer = try arena.create(Observer);
+        observer.* = .{
+            .node = node,
+            .options = options,
+            .mutation_observer = self,
+            .event_node = .{ .id = self.cbk.id, .func = Observer.handle },
+        };
+
+        try self.observers.append(arena, observer);
+
+        // register node's events
+        if (options.childList or options.subtree) {
+            observer.dom_node_inserted_listener = try parser.eventTargetAddEventListener(
+                parser.toEventTarget(parser.Node, node),
+                "DOMNodeInserted",
+                &observer.event_node,
+                false,
+            );
+            observer.dom_node_removed_listener = try parser.eventTargetAddEventListener(
+                parser.toEventTarget(parser.Node, node),
+                "DOMNodeRemoved",
+                &observer.event_node,
+                false,
+            );
+        }
+        if (options.attr()) {
+            observer.dom_node_attribute_modified_listener = try parser.eventTargetAddEventListener(
+                parser.toEventTarget(parser.Node, node),
+                "DOMAttrModified",
+                &observer.event_node,
+                false,
+            );
+        }
+        if (options.cdata()) {
+            observer.dom_cdata_modified_listener = try parser.eventTargetAddEventListener(
+                parser.toEventTarget(parser.Node, node),
+                "DOMCharacterDataModified",
+                &observer.event_node,
+                false,
+            );
+        }
+        if (options.subtree) {
+            observer.dom_subtree_modified_listener = try parser.eventTargetAddEventListener(
+                parser.toEventTarget(parser.Node, node),
+                "DOMSubtreeModified",
+                &observer.event_node,
+                false,
+            );
+        }
+    }
+
+    fn callback(ctx: *anyopaque) ?u32 {
+        const self: *MutationObserver = @ptrCast(@alignCast(ctx));
+        self.scheduled = false;
+
+        const records = self.observed.items;
+        if (records.len == 0) {
+            return null;
+        }
+
+        defer self.observed.clearRetainingCapacity();
+
+        var result: js.Function.Result = undefined;
+        self.cbk.tryCallWithThis(void, self, .{records}, &result) catch {
+            log.debug(.user_script, "callback error", .{
+                .err = result.exception,
+                .stack = result.stack,
+                .source = "mutation observer",
+            });
+        };
+        return null;
+    }
+
+    pub fn _disconnect(self: *MutationObserver) !void {
+        for (self.observers.items) |observer| {
+            const event_target = parser.toEventTarget(parser.Node, observer.node);
+            if (observer.dom_node_inserted_listener) |listener| {
+                try parser.eventTargetRemoveEventListener(
+                    event_target,
+                    "DOMNodeInserted",
+                    listener,
+                    false,
+                );
+            }
+
+            if (observer.dom_node_removed_listener) |listener| {
+                try parser.eventTargetRemoveEventListener(
+                    event_target,
+                    "DOMNodeRemoved",
+                    listener,
+                    false,
+                );
+            }
+
+            if (observer.dom_node_attribute_modified_listener) |listener| {
+                try parser.eventTargetRemoveEventListener(
+                    event_target,
+                    "DOMAttrModified",
+                    listener,
+                    false,
+                );
+            }
+
+            if (observer.dom_cdata_modified_listener) |listener| {
+                try parser.eventTargetRemoveEventListener(
+                    event_target,
+                    "DOMCharacterDataModified",
+                    listener,
+                    false,
+                );
+            }
+
+            if (observer.dom_subtree_modified_listener) |listener| {
+                try parser.eventTargetRemoveEventListener(
+                    event_target,
+                    "DOMSubtreeModified",
+                    listener,
+                    false,
+                );
+            }
+        }
+        self.observers.clearRetainingCapacity();
+    }
+
+    // TODO
+    pub fn _takeRecords(_: *const MutationObserver) ?[]const u8 {
+        return &[_]u8{};
+    }
+};
+
+pub const MutationRecord = struct {
+    type: []const u8,
+    target: *parser.Node,
+    added_nodes: NodeList = .{},
+    removed_nodes: NodeList = .{},
+    previous_sibling: ?*parser.Node = null,
+    next_sibling: ?*parser.Node = null,
+    attribute_name: ?[]const u8 = null,
+    attribute_namespace: ?[]const u8 = null,
+    old_value: ?[]const u8 = null,
+
+    pub fn get_type(self: *const MutationRecord) []const u8 {
+        return self.type;
+    }
+
+    pub fn get_addedNodes(self: *MutationRecord) *NodeList {
+        return &self.added_nodes;
+    }
+
+    pub fn get_removedNodes(self: *MutationRecord) *NodeList {
+        return &self.removed_nodes;
+    }
+
+    pub fn get_target(self: *const MutationRecord) *parser.Node {
+        return self.target;
+    }
+
+    pub fn get_attributeName(self: *const MutationRecord) ?[]const u8 {
+        return self.attribute_name;
+    }
+
+    pub fn get_attributeNamespace(self: *const MutationRecord) ?[]const u8 {
+        return self.attribute_namespace;
+    }
+
+    pub fn get_previousSibling(self: *const MutationRecord) ?*parser.Node {
+        return self.previous_sibling;
+    }
+
+    pub fn get_nextSibling(self: *const MutationRecord) ?*parser.Node {
+        return self.next_sibling;
+    }
+
+    pub fn get_oldValue(self: *const MutationRecord) ?[]const u8 {
+        return self.old_value;
+    }
+};
+
+const Options = struct {
+    childList: bool = false,
+    attributes: bool = false,
+    characterData: bool = false,
+    subtree: bool = false,
+    attributeOldValue: bool = false,
+    characterDataOldValue: bool = false,
+    attributeFilter: [][]const u8 = &.{},
+
+    fn attr(self: Options) bool {
+        return self.attributes or self.attributeOldValue or self.attributeFilter.len > 0;
+    }
+
+    fn cdata(self: Options) bool {
+        return self.characterData or self.characterDataOldValue;
+    }
+};
+
+const Observer = struct {
+    node: *parser.Node,
+    options: Options,
+
+    // reference back to the MutationObserver so that we can access the arena
+    // and batch the mutation records.
+    mutation_observer: *MutationObserver,
+
+    event_node: parser.EventNode,
+
+    dom_node_inserted_listener: ?*parser.EventListener = null,
+    dom_node_removed_listener: ?*parser.EventListener = null,
+    dom_node_attribute_modified_listener: ?*parser.EventListener = null,
+    dom_cdata_modified_listener: ?*parser.EventListener = null,
+    dom_subtree_modified_listener: ?*parser.EventListener = null,
+
+    fn appliesTo(
+        self: *const Observer,
+        target: *parser.Node,
+        event_type: MutationEventType,
+        event: *parser.MutationEvent,
+    ) !bool {
+        if (event_type == .DOMAttrModified and self.options.attributeFilter.len > 0) {
+            const attribute_name = try parser.mutationEventAttributeName(event);
+            for (self.options.attributeFilter) |needle| blk: {
+                if (std.mem.eql(u8, attribute_name, needle)) {
+                    break :blk;
+                }
+            }
+            return false;
+        }
+
+        // mutation on any target is always ok.
+        if (self.options.subtree) {
+            return true;
+        }
+
+        // if target equals node, alway ok.
+        if (target == self.node) {
+            return true;
+        }
+
+        // no subtree, no same target and no childlist, always noky.
+        if (!self.options.childList) {
+            return false;
+        }
+
+        // target must be a child of o.node
+        const walker = Walker{};
+        var next: ?*parser.Node = null;
+        while (true) {
+            next = walker.get_next(self.node, next) catch break orelse break;
+            if (next.? == target) {
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+    fn handle(en: *parser.EventNode, event: *parser.Event) void {
+        const self: *Observer = @fieldParentPtr("event_node", en);
+        self._handle(event) catch |err| {
+            log.err(.web_api, "handle error", .{ .err = err, .source = "mutation observer" });
+        };
+    }
+
+    fn _handle(self: *Observer, event: *parser.Event) !void {
+        var mutation_observer = self.mutation_observer;
+
+        const node = blk: {
+            const event_target = parser.eventTarget(event) orelse return;
+            break :blk parser.eventTargetToNode(event_target);
+        };
+
+        const mutation_event = parser.eventToMutationEvent(event);
+        const event_type = blk: {
+            const t = parser.eventType(event);
+            break :blk std.meta.stringToEnum(MutationEventType, t) orelse return;
+        };
+
+        if (try self.appliesTo(node, event_type, mutation_event) == false) {
+            return;
+        }
+
+        var record = MutationRecord{
+            .target = self.node,
+            .type = event_type.recordType(),
+        };
+
+        const arena = mutation_observer.page.arena;
+        switch (event_type) {
+            .DOMAttrModified => {
+                record.attribute_name = parser.mutationEventAttributeName(mutation_event) catch null;
+                if (self.options.attributeOldValue) {
+                    record.old_value = parser.mutationEventPrevValue(mutation_event);
+                }
+            },
+            .DOMCharacterDataModified => {
+                if (self.options.characterDataOldValue) {
+                    record.old_value = parser.mutationEventPrevValue(mutation_event);
+                }
+            },
+            .DOMNodeInserted => {
+                if (parser.mutationEventRelatedNode(mutation_event) catch null) |related_node| {
+                    try record.added_nodes.append(arena, related_node);
+                }
+            },
+            .DOMNodeRemoved => {
+                if (parser.mutationEventRelatedNode(mutation_event) catch null) |related_node| {
+                    try record.removed_nodes.append(arena, related_node);
+                }
+            },
+        }
+
+        try mutation_observer.observed.append(arena, record);
+
+        if (mutation_observer.scheduled == false) {
+            mutation_observer.scheduled = true;
+            try mutation_observer.page.scheduler.add(
+                mutation_observer,
+                MutationObserver.callback,
+                0,
+                .{ .name = "mutation_observer" },
+            );
+        }
+    }
+};
+
+const MutationEventType = enum {
+    DOMAttrModified,
+    DOMCharacterDataModified,
+    DOMNodeInserted,
+    DOMNodeRemoved,
+
+    fn recordType(self: MutationEventType) []const u8 {
+        return switch (self) {
+            .DOMAttrModified => "attributes",
+            .DOMCharacterDataModified => "characterData",
+            .DOMNodeInserted => "childList",
+            .DOMNodeRemoved => "childList",
+        };
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.MutationObserver" {
+    try testing.htmlRunner("dom/mutation_observer.html");
+}

src/browser/dom/namednodemap.zig

@@ -0,0 +1,121 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const parser = @import("../netsurf.zig");
+
+const DOMException = @import("exceptions.zig").DOMException;
+
+// WEB IDL https://dom.spec.whatwg.org/#namednodemap
+pub const NamedNodeMap = struct {
+    pub const Self = parser.NamedNodeMap;
+
+    pub const Exception = DOMException;
+    pub const Iterator = NamedNodeMapIterator;
+
+    // TODO implement LegacyUnenumerableNamedProperties.
+    // https://webidl.spec.whatwg.org/#LegacyUnenumerableNamedProperties
+
+    pub fn get_length(self: *parser.NamedNodeMap) !u32 {
+        return try parser.namedNodeMapGetLength(self);
+    }
+
+    pub fn _item(self: *parser.NamedNodeMap, index: u32) !?*parser.Attribute {
+        return try parser.namedNodeMapItem(self, index);
+    }
+
+    pub fn _getNamedItem(self: *parser.NamedNodeMap, qname: []const u8) !?*parser.Attribute {
+        return try parser.namedNodeMapGetNamedItem(self, qname);
+    }
+
+    pub fn _getNamedItemNS(
+        self: *parser.NamedNodeMap,
+        namespace: []const u8,
+        localname: []const u8,
+    ) !?*parser.Attribute {
+        return try parser.namedNodeMapGetNamedItemNS(self, namespace, localname);
+    }
+
+    pub fn _setNamedItem(self: *parser.NamedNodeMap, attr: *parser.Attribute) !?*parser.Attribute {
+        return try parser.namedNodeMapSetNamedItem(self, attr);
+    }
+
+    pub fn _setNamedItemNS(self: *parser.NamedNodeMap, attr: *parser.Attribute) !?*parser.Attribute {
+        return try parser.namedNodeMapSetNamedItemNS(self, attr);
+    }
+
+    pub fn _removeNamedItem(self: *parser.NamedNodeMap, qname: []const u8) !*parser.Attribute {
+        return try parser.namedNodeMapRemoveNamedItem(self, qname);
+    }
+
+    pub fn _removeNamedItemNS(
+        self: *parser.NamedNodeMap,
+        namespace: []const u8,
+        localname: []const u8,
+    ) !*parser.Attribute {
+        return try parser.namedNodeMapRemoveNamedItemNS(self, namespace, localname);
+    }
+
+    pub fn indexed_get(self: *parser.NamedNodeMap, index: u32, has_value: *bool) !*parser.Attribute {
+        return (try _item(self, index)) orelse {
+            has_value.* = false;
+            return undefined;
+        };
+    }
+
+    pub fn named_get(self: *parser.NamedNodeMap, name: []const u8, has_value: *bool) !*parser.Attribute {
+        return (try _getNamedItem(self, name)) orelse {
+            has_value.* = false;
+            return undefined;
+        };
+    }
+
+    pub fn _symbol_iterator(self: *parser.NamedNodeMap) NamedNodeMapIterator {
+        return .{ .map = self };
+    }
+};
+
+pub const NamedNodeMapIterator = struct {
+    index: u32 = 0,
+    map: *parser.NamedNodeMap,
+
+    pub const Return = struct {
+        done: bool,
+        value: ?*parser.Attribute,
+    };
+
+    pub fn _next(self: *NamedNodeMapIterator) !Return {
+        const e = try NamedNodeMap._item(self.map, self.index);
+        if (e == null) {
+            return .{
+                .value = null,
+                .done = true,
+            };
+        }
+
+        self.index += 1;
+        return .{
+            .value = e,
+            .done = false,
+        };
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.NamedNodeMap" {
+    try testing.htmlRunner("dom/named_node_map.html");
+}

src/browser/dom/node.zig

@@ -0,0 +1,665 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const log = @import("../../log.zig");
+const parser = @import("../netsurf.zig");
+const generate = @import("../js/generate.zig");
+
+const Page = @import("../page.zig").Page;
+const EventTarget = @import("event_target.zig").EventTarget;
+
+// DOM
+const Attr = @import("attribute.zig").Attr;
+const CData = @import("character_data.zig");
+const Element = @import("element.zig").Element;
+const ElementUnion = @import("element.zig").Union;
+const NodeList = @import("nodelist.zig").NodeList;
+const Document = @import("document.zig").Document;
+const DocumentType = @import("document_type.zig").DocumentType;
+const DocumentFragment = @import("document_fragment.zig").DocumentFragment;
+const HTMLCollection = @import("html_collection.zig").HTMLCollection;
+const HTMLAllCollection = @import("html_collection.zig").HTMLAllCollection;
+const HTMLCollectionIterator = @import("html_collection.zig").HTMLCollectionIterator;
+const ShadowRoot = @import("shadow_root.zig").ShadowRoot;
+const Walker = @import("walker.zig").WalkerDepthFirst;
+
+// HTML
+const HTML = @import("../html/html.zig");
+
+// Node interfaces
+pub const Interfaces = .{
+    Attr,
+    CData.CharacterData,
+    CData.Interfaces,
+    Element,
+    Document,
+    DocumentType,
+    DocumentFragment,
+    HTMLCollection,
+    HTMLAllCollection,
+    HTMLCollectionIterator,
+    HTML.Interfaces,
+};
+
+pub const Union = generate.Union(Interfaces);
+
+// Node implementation
+pub const Node = struct {
+    pub const Self = parser.Node;
+    pub const prototype = *EventTarget;
+    pub const subtype = .node;
+
+    pub fn toInterface(node: *parser.Node) !Union {
+        return switch (parser.nodeType(node)) {
+            .element => try Element.toInterfaceT(
+                Union,
+                @as(*parser.Element, @ptrCast(node)),
+            ),
+            .comment => .{ .Comment = @as(*parser.Comment, @ptrCast(node)) },
+            .text => .{ .Text = @as(*parser.Text, @ptrCast(node)) },
+            .cdata_section => .{ .CDATASection = @as(*parser.CDATASection, @ptrCast(node)) },
+            .processing_instruction => .{ .ProcessingInstruction = @as(*parser.ProcessingInstruction, @ptrCast(node)) },
+            .document => blk: {
+                const doc: *parser.Document = @ptrCast(node);
+                if (doc.is_html) {
+                    break :blk .{ .HTMLDocument = @as(*parser.DocumentHTML, @ptrCast(node)) };
+                }
+
+                break :blk .{ .Document = doc };
+            },
+            .document_type => .{ .DocumentType = @as(*parser.DocumentType, @ptrCast(node)) },
+            .attribute => .{ .Attr = @as(*parser.Attribute, @ptrCast(node)) },
+            .document_fragment => .{ .DocumentFragment = @as(*parser.DocumentFragment, @ptrCast(node)) },
+            else => @panic("node type not handled"), // TODO
+        };
+    }
+
+    // class attributes
+
+    pub const _ELEMENT_NODE = @intFromEnum(parser.NodeType.element);
+    pub const _ATTRIBUTE_NODE = @intFromEnum(parser.NodeType.attribute);
+    pub const _TEXT_NODE = @intFromEnum(parser.NodeType.text);
+    pub const _CDATA_SECTION_NODE = @intFromEnum(parser.NodeType.cdata_section);
+    pub const _PROCESSING_INSTRUCTION_NODE = @intFromEnum(parser.NodeType.processing_instruction);
+    pub const _COMMENT_NODE = @intFromEnum(parser.NodeType.comment);
+    pub const _DOCUMENT_NODE = @intFromEnum(parser.NodeType.document);
+    pub const _DOCUMENT_TYPE_NODE = @intFromEnum(parser.NodeType.document_type);
+    pub const _DOCUMENT_FRAGMENT_NODE = @intFromEnum(parser.NodeType.document_fragment);
+
+    // These 3 are deprecated, but both Chrome and Firefox still expose them
+    pub const _ENTITY_REFERENCE_NODE = @intFromEnum(parser.NodeType.entity_reference);
+    pub const _ENTITY_NODE = @intFromEnum(parser.NodeType.entity);
+    pub const _NOTATION_NODE = @intFromEnum(parser.NodeType.notation);
+
+    pub const _DOCUMENT_POSITION_DISCONNECTED = @intFromEnum(parser.DocumentPosition.disconnected);
+    pub const _DOCUMENT_POSITION_PRECEDING = @intFromEnum(parser.DocumentPosition.preceding);
+    pub const _DOCUMENT_POSITION_FOLLOWING = @intFromEnum(parser.DocumentPosition.following);
+    pub const _DOCUMENT_POSITION_CONTAINS = @intFromEnum(parser.DocumentPosition.contains);
+    pub const _DOCUMENT_POSITION_CONTAINED_BY = @intFromEnum(parser.DocumentPosition.contained_by);
+    pub const _DOCUMENT_POSITION_IMPLEMENTATION_SPECIFIC = @intFromEnum(parser.DocumentPosition.implementation_specific);
+
+    // JS funcs
+    // --------
+
+    // Read-only attributes
+    pub fn get_baseURI(_: *parser.Node, page: *Page) ![]const u8 {
+        return page.url.raw;
+    }
+
+    pub fn get_firstChild(self: *parser.Node) !?Union {
+        const res = parser.nodeFirstChild(self);
+        if (res == null) {
+            return null;
+        }
+        return try Node.toInterface(res.?);
+    }
+
+    pub fn get_lastChild(self: *parser.Node) !?Union {
+        const res = parser.nodeLastChild(self);
+        if (res == null) {
+            return null;
+        }
+        return try Node.toInterface(res.?);
+    }
+
+    pub fn get_nextSibling(self: *parser.Node) !?Union {
+        const res = parser.nodeNextSibling(self);
+        if (res == null) {
+            return null;
+        }
+        return try Node.toInterface(res.?);
+    }
+
+    pub fn get_previousSibling(self: *parser.Node) !?Union {
+        const res = parser.nodePreviousSibling(self);
+        if (res == null) {
+            return null;
+        }
+        return try Node.toInterface(res.?);
+    }
+
+    pub fn get_parentNode(self: *parser.Node) !?Union {
+        const res = parser.nodeParentNode(self);
+        if (res == null) {
+            return null;
+        }
+        return try Node.toInterface(res.?);
+    }
+
+    pub fn get_parentElement(self: *parser.Node) !?ElementUnion {
+        const res = parser.nodeParentElement(self);
+        if (res == null) {
+            return null;
+        }
+        return try Element.toInterface(res.?);
+    }
+
+    pub fn get_nodeName(self: *parser.Node) ![]const u8 {
+        return try parser.nodeName(self);
+    }
+
+    pub fn get_nodeType(self: *parser.Node) !u8 {
+        return @intFromEnum(parser.nodeType(self));
+    }
+
+    pub fn get_ownerDocument(self: *parser.Node) !?*parser.DocumentHTML {
+        const res = parser.nodeOwnerDocument(self);
+        if (res == null) {
+            return null;
+        }
+        return @as(*parser.DocumentHTML, @ptrCast(res.?));
+    }
+
+    pub fn get_isConnected(self: *parser.Node) !bool {
+        var node = self;
+        while (true) {
+            const node_type = parser.nodeType(node);
+            if (node_type == .document) {
+                return true;
+            }
+
+            if (parser.nodeParentNode(node)) |parent| {
+                // didn't find a document, but node has a parent, let's see
+                // if it's connected;
+                node = parent;
+                continue;
+            }
+
+            if (node_type != .document_fragment) {
+                // doesn't have a parent and isn't a document_fragment
+                // can't be connected
+                return false;
+            }
+
+            if (parser.documentFragmentGetHost(@ptrCast(node))) |host| {
+                // node doesn't have a parent, but it's a document fragment
+                // with a host. The host is like the parent, but we only want to
+                // traverse up (or down) to it in specific cases, like isConnected.
+                node = host;
+                continue;
+            }
+            return false;
+        }
+    }
+
+    // Read/Write attributes
+
+    pub fn get_nodeValue(self: *parser.Node) !?[]const u8 {
+        return parser.nodeValue(self);
+    }
+
+    pub fn set_nodeValue(self: *parser.Node, data: []u8) !void {
+        try parser.nodeSetValue(self, data);
+    }
+
+    pub fn get_textContent(self: *parser.Node) ?[]const u8 {
+        return parser.nodeTextContent(self);
+    }
+
+    pub fn set_textContent(self: *parser.Node, data: []u8) !void {
+        return try parser.nodeSetTextContent(self, data);
+    }
+
+    // Methods
+
+    pub fn _appendChild(self: *parser.Node, child: *parser.Node) !Union {
+        const self_owner = parser.nodeOwnerDocument(self);
+        const child_owner = parser.nodeOwnerDocument(child);
+
+        // If the node to be inserted has a different ownerDocument than the parent node,
+        // modern browsers automatically adopt the node and its descendants into
+        // the parent's ownerDocument.
+        // This process is known as adoption.
+        // (7.1) https://dom.spec.whatwg.org/#concept-node-insert
+        if (child_owner == null or (self_owner != null and child_owner.? != self_owner.?)) {
+            const w = Walker{};
+            var current = child;
+            while (true) {
+                current.owner = self_owner;
+                current = try w.get_next(child, current) orelse break;
+            }
+        }
+
+        // TODO: DocumentFragment special case
+        const res = try parser.nodeAppendChild(self, child);
+        return try Node.toInterface(res);
+    }
+
+    pub fn _cloneNode(self: *parser.Node, deep: ?bool) !Union {
+        const clone = try parser.nodeCloneNode(self, deep orelse false);
+        return try Node.toInterface(clone);
+    }
+
+    pub fn _compareDocumentPosition(self: *parser.Node, other: *parser.Node) !u32 {
+        if (self == other) {
+            return 0;
+        }
+
+        const docself = parser.nodeOwnerDocument(self) orelse blk: {
+            if (parser.nodeType(self) == .document) {
+                break :blk @as(*parser.Document, @ptrCast(self));
+            }
+            break :blk null;
+        };
+        const docother = parser.nodeOwnerDocument(other) orelse blk: {
+            if (parser.nodeType(other) == .document) {
+                break :blk @as(*parser.Document, @ptrCast(other));
+            }
+            break :blk null;
+        };
+
+        // Both are in different document.
+        if (docself == null or docother == null or docself.? != docother.?) {
+            return @intFromEnum(parser.DocumentPosition.disconnected) +
+                @intFromEnum(parser.DocumentPosition.implementation_specific) +
+                @intFromEnum(parser.DocumentPosition.preceding);
+        }
+
+        if (@intFromPtr(self) == @intFromPtr(docself.?)) {
+            // if self is the document, and we already know other is in the
+            // document, then other is contained by and following self.
+            return @intFromEnum(parser.DocumentPosition.following) +
+                @intFromEnum(parser.DocumentPosition.contained_by);
+        }
+
+        const rootself = parser.nodeGetRootNode(self);
+        const rootother = parser.nodeGetRootNode(other);
+        if (rootself != rootother) {
+            return @intFromEnum(parser.DocumentPosition.disconnected) +
+                @intFromEnum(parser.DocumentPosition.implementation_specific) +
+                @intFromEnum(parser.DocumentPosition.preceding);
+        }
+
+        // TODO Both are in a different trees in the same document.
+
+        const w = Walker{};
+        var next: ?*parser.Node = null;
+
+        // Is other a descendant of self?
+        while (true) {
+            next = try w.get_next(self, next) orelse break;
+            if (other == next) {
+                return @intFromEnum(parser.DocumentPosition.following) +
+                    @intFromEnum(parser.DocumentPosition.contained_by);
+            }
+        }
+
+        // Is self a descendant of other?
+        next = null;
+        while (true) {
+            next = try w.get_next(other, next) orelse break;
+            if (self == next) {
+                return @intFromEnum(parser.DocumentPosition.contains) +
+                    @intFromEnum(parser.DocumentPosition.preceding);
+            }
+        }
+
+        next = null;
+        while (true) {
+            next = try w.get_next(parser.documentToNode(docself.?), next) orelse break;
+            if (other == next) {
+                // other precedes self.
+                return @intFromEnum(parser.DocumentPosition.preceding);
+            }
+            if (self == next) {
+                // other follows self.
+                return @intFromEnum(parser.DocumentPosition.following);
+            }
+        }
+
+        return 0;
+    }
+
+    pub fn _contains(self: *parser.Node, other: *parser.Node) bool {
+        return parser.nodeContains(self, other);
+    }
+
+    // Returns itself or ancestor object inheriting from Node.
+    // - An Element inside a standard web page will return an HTMLDocument object representing the entire page (or <iframe>).
+    // - An Element inside a shadow DOM will return the associated ShadowRoot.
+    // - An Element that is not attached to a document or a shadow tree will return the root of the DOM tree it belongs to
+    const GetRootNodeResult = union(enum) {
+        shadow_root: *ShadowRoot,
+        node: Union,
+    };
+    pub fn _getRootNode(self: *parser.Node, options: ?struct { composed: bool = false }, page: *Page) !GetRootNodeResult {
+        const composed = if (options) |opts| opts.composed else false;
+
+        var current_root = parser.nodeGetRootNode(self);
+
+        while (true) {
+            const node_type = parser.nodeType(current_root);
+
+            if (node_type == .document_fragment) {
+                if (parser.documentFragmentGetHost(@ptrCast(current_root))) |host| {
+                    if (page.getNodeState(host)) |state| {
+                        if (state.shadow_root) |sr| {
+                            if (!composed) {
+                                return .{ .shadow_root = sr };
+                            }
+                            current_root = parser.nodeGetRootNode(@ptrCast(sr.host));
+                            continue;
+                        }
+                    }
+                }
+            }
+            break;
+        }
+
+        return .{ .node = try Node.toInterface(current_root) };
+    }
+
+    pub fn _hasChildNodes(self: *parser.Node) bool {
+        return parser.nodeHasChildNodes(self);
+    }
+
+    fn is_template(self: *parser.Node) !bool {
+        if (parser.nodeType(self) != .element) {
+            return false;
+        }
+
+        const e = parser.nodeToElement(self);
+        return try parser.elementTag(e) == .template;
+    }
+
+    pub fn get_childNodes(self: *parser.Node, page: *Page) !NodeList {
+        // special case for template:
+        // > The Node.childNodes property of the <template> element is always empty
+        // https://developer.mozilla.org/en-US/docs/Web/HTML/Reference/Elements/template#usage_notes
+        if (try is_template(self)) {
+            return .{};
+        }
+
+        const allocator = page.arena;
+        var list: NodeList = .{};
+
+        var n = parser.nodeFirstChild(self) orelse return list;
+        while (true) {
+            try list.append(allocator, n);
+            n = parser.nodeNextSibling(n) orelse return list;
+        }
+    }
+
+    pub fn _insertBefore(self: *parser.Node, new_node: *parser.Node, ref_node_: ?*parser.Node) !Union {
+        if (ref_node_ == null) {
+            return _appendChild(self, new_node);
+        }
+
+        const self_owner = parser.nodeOwnerDocument(self);
+        const new_node_owner = parser.nodeOwnerDocument(new_node);
+
+        // If the node to be inserted has a different ownerDocument than the parent node,
+        // modern browsers automatically adopt the node and its descendants into
+        // the parent's ownerDocument.
+        // This process is known as adoption.
+        // (7.1) https://dom.spec.whatwg.org/#concept-node-insert
+        if (new_node_owner == null or (self_owner != null and new_node_owner.? != self_owner.?)) {
+            const w = Walker{};
+            var current = new_node;
+            while (true) {
+                current.owner = self_owner;
+                current = try w.get_next(new_node, current) orelse break;
+            }
+        }
+
+        return Node.toInterface(try parser.nodeInsertBefore(self, new_node, ref_node_.?));
+    }
+
+    pub fn _isDefaultNamespace(self: *parser.Node, namespace: ?[]const u8) !bool {
+        return parser.nodeIsDefaultNamespace(self, namespace);
+    }
+
+    pub fn _isEqualNode(self: *parser.Node, other: *parser.Node) !bool {
+        // TODO: other is not an optional parameter, but can be null.
+        return try parser.nodeIsEqualNode(self, other);
+    }
+
+    pub fn _isSameNode(self: *parser.Node, other: *parser.Node) bool {
+        // TODO: other is not an optional parameter, but can be null.
+        // NOTE: there is no need to use isSameNode(); instead use the === strict equality operator
+        return parser.nodeIsSameNode(self, other);
+    }
+
+    pub fn _lookupPrefix(self: *parser.Node, namespace: ?[]const u8) !?[]const u8 {
+        // TODO: other is not an optional parameter, but can be null.
+        if (namespace == null) {
+            return null;
+        }
+        if (std.mem.eql(u8, namespace.?, "")) {
+            return null;
+        }
+        return try parser.nodeLookupPrefix(self, namespace.?);
+    }
+
+    pub fn _lookupNamespaceURI(self: *parser.Node, prefix: ?[]const u8) !?[]const u8 {
+        // TODO: other is not an optional parameter, but can be null.
+        return try parser.nodeLookupNamespaceURI(self, prefix);
+    }
+
+    pub fn _normalize(self: *parser.Node) !void {
+        return try parser.nodeNormalize(self);
+    }
+
+    pub fn _removeChild(self: *parser.Node, child: *parser.Node) !Union {
+        const res = try parser.nodeRemoveChild(self, child);
+        return try Node.toInterface(res);
+    }
+
+    pub fn _replaceChild(self: *parser.Node, new_child: *parser.Node, old_child: *parser.Node) !Union {
+        const res = try parser.nodeReplaceChild(self, new_child, old_child);
+        return try Node.toInterface(res);
+    }
+
+    // Check if the hierarchy node tree constraints are respected.
+    // For now, it checks only if new nodes are not self.
+    // TODO implements the others constraints.
+    // see https://dom.spec.whatwg.org/#concept-node-tree
+    pub fn hierarchy(self: *parser.Node, nodes: []const NodeOrText) bool {
+        for (nodes) |n| {
+            if (n.is(self)) {
+                return false;
+            }
+        }
+        return true;
+    }
+
+    pub fn prepend(self: *parser.Node, nodes: []const NodeOrText) !void {
+        if (nodes.len == 0) {
+            return;
+        }
+
+        // check hierarchy
+        if (!hierarchy(self, nodes)) {
+            return parser.DOMError.HierarchyRequest;
+        }
+
+        const doc = (parser.nodeOwnerDocument(self)) orelse return;
+
+        if (parser.nodeFirstChild(self)) |first| {
+            for (nodes) |node| {
+                _ = try parser.nodeInsertBefore(self, try node.toNode(doc), first);
+            }
+            return;
+        }
+
+        for (nodes) |node| {
+            _ = try parser.nodeAppendChild(self, try node.toNode(doc));
+        }
+    }
+
+    pub fn append(self: *parser.Node, nodes: []const NodeOrText) !void {
+        if (nodes.len == 0) {
+            return;
+        }
+
+        // check hierarchy
+        if (!hierarchy(self, nodes)) {
+            return parser.DOMError.HierarchyRequest;
+        }
+
+        const doc = (parser.nodeOwnerDocument(self)) orelse return;
+        for (nodes) |node| {
+            _ = try parser.nodeAppendChild(self, try node.toNode(doc));
+        }
+    }
+
+    pub fn replaceChildren(self: *parser.Node, nodes: []const NodeOrText) !void {
+        if (nodes.len == 0) {
+            return;
+        }
+
+        // check hierarchy
+        if (!hierarchy(self, nodes)) {
+            return parser.DOMError.HierarchyRequest;
+        }
+
+        // remove existing children
+        try removeChildren(self);
+
+        const doc = (parser.nodeOwnerDocument(self)) orelse return;
+        // add new children
+        for (nodes) |node| {
+            _ = try parser.nodeAppendChild(self, try node.toNode(doc));
+        }
+    }
+
+    pub fn removeChildren(self: *parser.Node) !void {
+        if (!parser.nodeHasChildNodes(self)) return;
+
+        const children = try parser.nodeGetChildNodes(self);
+        const ln = parser.nodeListLength(children);
+        var i: u32 = 0;
+        while (i < ln) {
+            defer i += 1;
+            // we always retrieve the 0 index child on purpose: libdom nodelist
+            // are dynamic. So the next child to remove is always as pos 0.
+            const child = parser.nodeListItem(children, 0) orelse continue;
+            _ = try parser.nodeRemoveChild(self, child);
+        }
+    }
+
+    pub fn before(self: *parser.Node, nodes: []const NodeOrText) !void {
+        const parent = parser.nodeParentNode(self) orelse return;
+        const doc = (parser.nodeOwnerDocument(parent)) orelse return;
+
+        var sibling: ?*parser.Node = self;
+        // have to find the first sibling that isn't in nodes
+        CHECK: while (sibling) |s| {
+            for (nodes) |n| {
+                if (n.is(s)) {
+                    sibling = parser.nodePreviousSibling(s);
+                    continue :CHECK;
+                }
+            }
+            break;
+        }
+
+        if (sibling == null) {
+            sibling = parser.nodeFirstChild(parent);
+        }
+
+        if (sibling) |ref_node| {
+            for (nodes) |node| {
+                _ = try parser.nodeInsertBefore(parent, try node.toNode(doc), ref_node);
+            }
+            return;
+        }
+
+        return Node.prepend(self, nodes);
+    }
+
+    pub fn after(self: *parser.Node, nodes: []const NodeOrText) !void {
+        const parent = parser.nodeParentNode(self) orelse return;
+        const doc = (parser.nodeOwnerDocument(parent)) orelse return;
+
+        // have to find the first sibling that isn't in nodes
+        var sibling = parser.nodeNextSibling(self);
+        CHECK: while (sibling) |s| {
+            for (nodes) |n| {
+                if (n.is(s)) {
+                    sibling = parser.nodeNextSibling(s);
+                    continue :CHECK;
+                }
+            }
+            break;
+        }
+
+        if (sibling) |ref_node| {
+            for (nodes) |node| {
+                _ = try parser.nodeInsertBefore(parent, try node.toNode(doc), ref_node);
+            }
+            return;
+        }
+
+        for (nodes) |node| {
+            _ = try parser.nodeAppendChild(parent, try node.toNode(doc));
+        }
+    }
+
+    // A lot of functions take either a node or text input.
+    // The text input is to be converted into a Text node.
+    pub const NodeOrText = union(enum) {
+        text: []const u8,
+        node: *parser.Node,
+
+        fn toNode(self: NodeOrText, doc: *parser.Document) !*parser.Node {
+            return switch (self) {
+                .node => |n| n,
+                .text => |txt| @ptrCast(@alignCast(try parser.documentCreateTextNode(doc, txt))),
+            };
+        }
+
+        // Whether the node represented by the NodeOrText is the same as the
+        // given Node. Always false for text values as these represent as-of-yet
+        // created Text nodes.
+        fn is(self: NodeOrText, other: *parser.Node) bool {
+            return switch (self) {
+                .text => false,
+                .node => |n| n == other,
+            };
+        }
+    };
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Node" {
+    try testing.htmlRunner("dom/node.html");
+    try testing.htmlRunner("dom/node_owner.html");
+}

src/browser/dom/node_filter.zig

@@ -0,0 +1,83 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const js = @import("../js/js.zig");
+const parser = @import("../netsurf.zig");
+const Node = @import("node.zig").Node;
+
+pub const NodeFilter = struct {
+    pub const _FILTER_ACCEPT: u16 = 1;
+    pub const _FILTER_REJECT: u16 = 2;
+    pub const _FILTER_SKIP: u16 = 3;
+
+    pub const _SHOW_ALL: u32 = std.math.maxInt(u32);
+    pub const _SHOW_ELEMENT: u32 = 0b1;
+    pub const _SHOW_ATTRIBUTE: u32 = 0b10;
+    pub const _SHOW_TEXT: u32 = 0b100;
+    pub const _SHOW_CDATA_SECTION: u32 = 0b1000;
+    pub const _SHOW_ENTITY_REFERENCE: u32 = 0b10000;
+    pub const _SHOW_ENTITY: u32 = 0b100000;
+    pub const _SHOW_PROCESSING_INSTRUCTION: u32 = 0b1000000;
+    pub const _SHOW_COMMENT: u32 = 0b10000000;
+    pub const _SHOW_DOCUMENT: u32 = 0b100000000;
+    pub const _SHOW_DOCUMENT_TYPE: u32 = 0b1000000000;
+    pub const _SHOW_DOCUMENT_FRAGMENT: u32 = 0b10000000000;
+    pub const _SHOW_NOTATION: u32 = 0b100000000000;
+};
+
+const VerifyResult = enum { accept, skip, reject };
+
+pub fn verify(what_to_show: u32, filter: ?js.Function, node: *parser.Node) !VerifyResult {
+    const node_type = parser.nodeType(node);
+
+    // Verify that we can show this node type.
+    // Per the DOM spec, what_to_show filters which nodes to return, but should
+    // still traverse children. So we return .skip (not .reject) when the node
+    // type doesn't match.
+    if (!switch (node_type) {
+        .attribute => what_to_show & NodeFilter._SHOW_ATTRIBUTE != 0,
+        .cdata_section => what_to_show & NodeFilter._SHOW_CDATA_SECTION != 0,
+        .comment => what_to_show & NodeFilter._SHOW_COMMENT != 0,
+        .document => what_to_show & NodeFilter._SHOW_DOCUMENT != 0,
+        .document_fragment => what_to_show & NodeFilter._SHOW_DOCUMENT_FRAGMENT != 0,
+        .document_type => what_to_show & NodeFilter._SHOW_DOCUMENT_TYPE != 0,
+        .element => what_to_show & NodeFilter._SHOW_ELEMENT != 0,
+        .entity => what_to_show & NodeFilter._SHOW_ENTITY != 0,
+        .entity_reference => what_to_show & NodeFilter._SHOW_ENTITY_REFERENCE != 0,
+        .notation => what_to_show & NodeFilter._SHOW_NOTATION != 0,
+        .processing_instruction => what_to_show & NodeFilter._SHOW_PROCESSING_INSTRUCTION != 0,
+        .text => what_to_show & NodeFilter._SHOW_TEXT != 0,
+    }) return .skip;
+
+    // Verify that we aren't filtering it out.
+    if (filter) |f| {
+        const acceptance = try f.call(u16, .{try Node.toInterface(node)});
+        return switch (acceptance) {
+            NodeFilter._FILTER_ACCEPT => .accept,
+            NodeFilter._FILTER_REJECT => .reject,
+            NodeFilter._FILTER_SKIP => .skip,
+            else => .reject,
+        };
+    } else return .accept;
+}
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.NodeFilter" {
+    try testing.htmlRunner("dom/node_filter.html");
+}

src/browser/dom/node_iterator.zig

@@ -0,0 +1,302 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const js = @import("../js/js.zig");
+const parser = @import("../netsurf.zig");
+const NodeFilter = @import("node_filter.zig");
+const Node = @import("node.zig").Node;
+const NodeUnion = @import("node.zig").Union;
+const DOMException = @import("exceptions.zig").DOMException;
+
+// https://developer.mozilla.org/en-US/docs/Web/API/NodeIterator
+// While this is similar to TreeWalker it has its own implementation as there are several subtle differences
+// For example:
+// - nextNode returns the reference node, whereas TreeWalker returns the next node
+// - Skip and reject are equivalent for NodeIterator, for TreeWalker they are different
+pub const NodeIterator = struct {
+    pub const Exception = DOMException;
+
+    root: *parser.Node,
+    reference_node: *parser.Node,
+    what_to_show: u32,
+    filter: ?NodeIteratorOpts,
+    filter_func: ?js.Function,
+    pointer_before_current: bool = true,
+    // used to track / block recursive filters
+    is_in_callback: bool = false,
+
+    // One of the few cases where null and undefined resolve to different default.
+    // We need the raw JsObject so that we can probe the tri state:
+    // null, undefined or i32.
+    pub const WhatToShow = js.Object;
+
+    pub const NodeIteratorOpts = union(enum) {
+        function: js.Function,
+        object: struct { acceptNode: js.Function },
+    };
+
+    pub fn init(node: *parser.Node, what_to_show_: ?WhatToShow, filter: ?NodeIteratorOpts) !NodeIterator {
+        var filter_func: ?js.Function = null;
+        if (filter) |f| {
+            filter_func = switch (f) {
+                .function => |func| func,
+                .object => |o| o.acceptNode,
+            };
+        }
+
+        var what_to_show: u32 = undefined;
+        if (what_to_show_) |wts| {
+            switch (try wts.triState(NodeIterator, "what_to_show", u32)) {
+                .null => what_to_show = 0,
+                .undefined => what_to_show = NodeFilter.NodeFilter._SHOW_ALL,
+                .value => |v| what_to_show = v,
+            }
+        } else {
+            what_to_show = NodeFilter.NodeFilter._SHOW_ALL;
+        }
+
+        return .{
+            .root = node,
+            .filter = filter,
+            .reference_node = node,
+            .filter_func = filter_func,
+            .what_to_show = what_to_show,
+        };
+    }
+
+    pub fn get_filter(self: *const NodeIterator) ?NodeIteratorOpts {
+        return self.filter;
+    }
+
+    pub fn get_pointerBeforeReferenceNode(self: *const NodeIterator) bool {
+        return self.pointer_before_current;
+    }
+
+    pub fn get_referenceNode(self: *const NodeIterator) !NodeUnion {
+        return try Node.toInterface(self.reference_node);
+    }
+
+    pub fn get_root(self: *const NodeIterator) !NodeUnion {
+        return try Node.toInterface(self.root);
+    }
+
+    pub fn get_whatToShow(self: *const NodeIterator) u32 {
+        return self.what_to_show;
+    }
+
+    pub fn _nextNode(self: *NodeIterator) !?NodeUnion {
+        try self.callbackStart();
+        defer self.callbackEnd();
+
+        if (self.pointer_before_current) {
+            // Unlike TreeWalker, NodeIterator starts at the first node
+            if (.accept == try NodeFilter.verify(self.what_to_show, self.filter_func, self.reference_node)) {
+                self.pointer_before_current = false;
+                return try Node.toInterface(self.reference_node);
+            }
+        }
+
+        if (try self.firstChild(self.reference_node)) |child| {
+            self.reference_node = child;
+            self.pointer_before_current = false;
+            return try Node.toInterface(child);
+        }
+
+        var current = self.reference_node;
+        while (current != self.root) {
+            // Try to get next sibling (including .skip/.reject nodes we need to descend into)
+            if (try self.nextSiblingOrSkipReject(current)) |result| {
+                if (result.should_descend) {
+                    // This is a .skip/.reject node - try to find acceptable children within it
+                    if (try self.firstChild(result.node)) |child| {
+                        self.reference_node = child;
+                        return try Node.toInterface(child);
+                    }
+                    // No acceptable children, continue looking at this node's siblings
+                    current = result.node;
+                    continue;
+                }
+                // This is an .accept node - return it
+                self.reference_node = result.node;
+                return try Node.toInterface(result.node);
+            }
+
+            current = (parser.nodeParentNode(current)) orelse break;
+        }
+
+        return null;
+    }
+
+    pub fn _previousNode(self: *NodeIterator) !?NodeUnion {
+        try self.callbackStart();
+        defer self.callbackEnd();
+
+        if (!self.pointer_before_current) {
+            if (.accept == try NodeFilter.verify(self.what_to_show, self.filter_func, self.reference_node)) {
+                self.pointer_before_current = true;
+                // Still need to verify as last may be first as well
+                return try Node.toInterface(self.reference_node);
+            }
+        }
+        if (self.reference_node == self.root) {
+            return null;
+        }
+
+        var current = self.reference_node;
+        while (parser.nodePreviousSibling(current)) |previous| {
+            current = previous;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, current)) {
+                .accept => {
+                    // Get last child if it has one.
+                    if (try self.lastChild(current)) |child| {
+                        self.reference_node = child;
+                        return try Node.toInterface(child);
+                    }
+
+                    // Otherwise, this node is our previous one.
+                    self.reference_node = current;
+                    return try Node.toInterface(current);
+                },
+                .reject, .skip => {
+                    // Get last child if it has one.
+                    if (try self.lastChild(current)) |child| {
+                        self.reference_node = child;
+                        return try Node.toInterface(child);
+                    }
+                },
+            }
+        }
+
+        if (current != self.root) {
+            if (try self.parentNode(current)) |parent| {
+                self.reference_node = parent;
+                return try Node.toInterface(parent);
+            }
+        }
+
+        return null;
+    }
+
+    pub fn _detach(self: *const NodeIterator) void {
+        // no-op as per spec
+        _ = self;
+    }
+
+    fn firstChild(self: *const NodeIterator, node: *parser.Node) !?*parser.Node {
+        const children = try parser.nodeGetChildNodes(node);
+        const child_count = parser.nodeListLength(children);
+
+        for (0..child_count) |i| {
+            const index: u32 = @intCast(i);
+            const child = (parser.nodeListItem(children, index)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, child)) {
+                .accept => return child, // NOTE: Skip and reject are equivalent for NodeIterator, this is different from TreeWalker
+                .reject, .skip => if (try self.firstChild(child)) |gchild| return gchild,
+            }
+        }
+
+        return null;
+    }
+
+    fn lastChild(self: *const NodeIterator, node: *parser.Node) !?*parser.Node {
+        const children = try parser.nodeGetChildNodes(node);
+        const child_count = parser.nodeListLength(children);
+
+        var index: u32 = child_count;
+        while (index > 0) {
+            index -= 1;
+            const child = (parser.nodeListItem(children, index)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, child)) {
+                .accept => return child, // NOTE: Skip and reject are equivalent for NodeIterator, this is different from TreeWalker
+                .reject, .skip => if (try self.lastChild(child)) |gchild| return gchild,
+            }
+        }
+
+        return null;
+    }
+
+    // This implementation is actually the same as :TreeWalker
+    fn parentNode(self: *const NodeIterator, node: *parser.Node) !?*parser.Node {
+        if (self.root == node) return null;
+
+        var current = node;
+        while (true) {
+            if (current == self.root) return null;
+            current = (parser.nodeParentNode(current)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, current)) {
+                .accept => return current,
+                .reject, .skip => continue,
+            }
+        }
+    }
+
+    // This implementation is actually the same as :TreeWalker
+    fn nextSibling(self: *const NodeIterator, node: *parser.Node) !?*parser.Node {
+        var current = node;
+
+        while (true) {
+            current = (parser.nodeNextSibling(current)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, current)) {
+                .accept => return current,
+                .skip, .reject => continue,
+            }
+        }
+
+        return null;
+    }
+
+    // Get the next sibling that is either acceptable or should be descended into (skip/reject)
+    fn nextSiblingOrSkipReject(self: *const NodeIterator, node: *parser.Node) !?struct { node: *parser.Node, should_descend: bool } {
+        var current = node;
+
+        while (true) {
+            current = (parser.nodeNextSibling(current)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, current)) {
+                .accept => return .{ .node = current, .should_descend = false },
+                .skip, .reject => return .{ .node = current, .should_descend = true },
+            }
+        }
+
+        return null;
+    }
+
+    fn callbackStart(self: *NodeIterator) !void {
+        if (self.is_in_callback) {
+            // this is the correct DOMExeption
+            return error.InvalidState;
+        }
+        self.is_in_callback = true;
+    }
+
+    fn callbackEnd(self: *NodeIterator) void {
+        self.is_in_callback = false;
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.NodeIterator" {
+    try testing.htmlRunner("dom/node_iterator.html");
+}

src/browser/dom/nodelist.zig

@@ -0,0 +1,188 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const Allocator = std.mem.Allocator;
+
+const js = @import("../js/js.zig");
+const log = @import("../../log.zig");
+const parser = @import("../netsurf.zig");
+
+const NodeUnion = @import("node.zig").Union;
+const Node = @import("node.zig").Node;
+
+const U32Iterator = @import("../iterator/iterator.zig").U32Iterator;
+
+const DOMException = @import("exceptions.zig").DOMException;
+
+pub const Interfaces = .{
+    NodeListIterator,
+    NodeList,
+};
+
+pub const NodeListIterator = struct {
+    coll: *NodeList,
+    index: u32 = 0,
+
+    pub const Return = struct {
+        value: ?NodeUnion,
+        done: bool,
+    };
+
+    pub fn _next(self: *NodeListIterator) !Return {
+        const e = try self.coll._item(self.index);
+        if (e == null) {
+            return Return{
+                .value = null,
+                .done = true,
+            };
+        }
+
+        self.index += 1;
+        return Return{
+            .value = e,
+            .done = false,
+        };
+    }
+};
+
+pub const NodeListEntriesIterator = struct {
+    coll: *NodeList,
+    index: u32 = 0,
+
+    pub const Return = struct {
+        value: ?NodeUnion,
+        done: bool,
+    };
+
+    pub fn _next(self: *NodeListEntriesIterator) !Return {
+        const e = try self.coll._item(self.index);
+        if (e == null) {
+            return Return{
+                .value = null,
+                .done = true,
+            };
+        }
+
+        self.index += 1;
+        return Return{
+            .value = e,
+            .done = false,
+        };
+    }
+};
+
+// Nodelist is implemented in pure Zig b/c libdom's NodeList doesn't allow to
+// append nodes.
+// WEB IDL https://dom.spec.whatwg.org/#nodelist
+//
+// TODO: a Nodelist can be either static or live. But the current
+// implementation allows only static nodelist.
+// see https://dom.spec.whatwg.org/#old-style-collections
+pub const NodeList = struct {
+    pub const Exception = DOMException;
+    const NodesArrayList = std.ArrayListUnmanaged(*parser.Node);
+
+    nodes: NodesArrayList = .{},
+
+    pub fn deinit(self: *NodeList, allocator: Allocator) void {
+        self.nodes.deinit(allocator);
+    }
+
+    pub fn ensureTotalCapacity(self: *NodeList, allocator: Allocator, n: usize) !void {
+        return self.nodes.ensureTotalCapacity(allocator, n);
+    }
+
+    pub fn append(self: *NodeList, allocator: Allocator, node: *parser.Node) !void {
+        try self.nodes.append(allocator, node);
+    }
+
+    pub fn appendAssumeCapacity(self: *NodeList, node: *parser.Node) void {
+        self.nodes.appendAssumeCapacity(node);
+    }
+
+    pub fn get_length(self: *const NodeList) u32 {
+        return @intCast(self.nodes.items.len);
+    }
+
+    pub fn _item(self: *const NodeList, index: u32) !?NodeUnion {
+        if (index >= self.nodes.items.len) {
+            return null;
+        }
+
+        const n = self.nodes.items[index];
+        return try Node.toInterface(n);
+    }
+
+    // This code works, but it's _MUCH_ slower than using postAttach. The benefit
+    // of this version, is that it's "live"..but we're talking many orders of
+    // magnitude slower.
+    //
+    // You can test it by commenting out `postAttach`, uncommenting this and
+    // running:
+    //    zig build wpt --  tests/wpt/dom/nodes/NodeList-static-length-getter-tampered-indexOf-1.html
+    //
+    // I think this _is_ the right way to do it, but I must be doing something
+    // wrong to make it so slow.
+    // pub fn indexed_get(self: *const NodeList, index: u32, has_value: *bool) !?NodeUnion {
+    //     return (try self._item(index)) orelse {
+    //         has_value.* = false;
+    //         return null;
+    //     };
+    // }
+
+    pub fn _forEach(self: *NodeList, cbk: js.Function) !void { // TODO handle thisArg
+        for (self.nodes.items, 0..) |n, i| {
+            const ii: u32 = @intCast(i);
+            var result: js.Function.Result = undefined;
+            cbk.tryCall(void, .{ n, ii, self }, &result) catch {
+                log.debug(.user_script, "forEach callback", .{ .err = result.exception, .stack = result.stack });
+            };
+        }
+    }
+
+    pub fn _keys(self: *NodeList) U32Iterator {
+        return .{
+            .length = self.get_length(),
+        };
+    }
+
+    pub fn _values(self: *NodeList) NodeListIterator {
+        return .{
+            .coll = self,
+        };
+    }
+
+    pub fn _symbol_iterator(self: *NodeList) NodeListIterator {
+        return self._values();
+    }
+
+    // TODO entries() https://developer.mozilla.org/en-US/docs/Web/API/NodeList/entries
+    pub fn postAttach(self: *NodeList, js_this: js.This) !void {
+        const len = self.get_length();
+        for (0..len) |i| {
+            const node = try self._item(@intCast(i)) orelse unreachable;
+            try js_this.setIndex(@intCast(i), node, .{});
+        }
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.NodeList" {
+    try testing.htmlRunner("dom/node_list.html");
+}

src/browser/dom/performance.zig

@@ -0,0 +1,206 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const js = @import("../js/js.zig");
+const parser = @import("../netsurf.zig");
+const EventTarget = @import("../dom/event_target.zig").EventTarget;
+const Page = @import("../page.zig").Page;
+
+const milliTimestamp = @import("../../datetime.zig").milliTimestamp;
+
+pub const Interfaces = .{
+    Performance,
+    PerformanceEntry,
+    PerformanceMark,
+};
+
+// https://developer.mozilla.org/en-US/docs/Web/API/Performance
+pub const Performance = struct {
+    pub const prototype = *EventTarget;
+
+    // Extend libdom event target for pure zig struct.
+    base: parser.EventTargetTBase = parser.EventTargetTBase{ .internal_target_type = .performance },
+
+    time_origin: u64,
+    // if (Window.crossOriginIsolated) -> Resolution in isolated contexts:       5 microseconds
+    // else                            -> Resolution in non-isolated contexts: 100 microseconds
+    const ms_resolution = 100;
+
+    pub fn init() Performance {
+        return .{
+            .time_origin = milliTimestamp(),
+        };
+    }
+
+    pub fn get_timeOrigin(self: *const Performance) u64 {
+        return self.time_origin;
+    }
+
+    pub fn reset(self: *Performance) void {
+        self.time_origin = milliTimestamp();
+    }
+
+    pub fn _now(self: *const Performance) u64 {
+        return milliTimestamp() - self.time_origin;
+    }
+
+    pub fn _mark(_: *Performance, name: js.String, _options: ?PerformanceMark.Options, page: *Page) !PerformanceMark {
+        const mark: PerformanceMark = try .constructor(name, _options, page);
+        // TODO: Should store this in an entries list
+        return mark;
+    }
+
+    // TODO: fn _mark should record the marks in a lookup
+    pub fn _clearMarks(_: *Performance, name: ?[]const u8) void {
+        _ = name;
+    }
+
+    // TODO: fn _measures should record the marks in a lookup
+    pub fn _clearMeasures(_: *Performance, name: ?[]const u8) void {
+        _ = name;
+    }
+
+    // TODO: fn _measures should record the marks in a lookup
+    pub fn _getEntriesByName(_: *Performance, name: []const u8, typ: ?[]const u8) []PerformanceEntry {
+        _ = name;
+        _ = typ;
+        return &.{};
+    }
+
+    // TODO: fn _measures should record the marks in a lookup
+    pub fn _getEntriesByType(_: *Performance, typ: []const u8) []PerformanceEntry {
+        _ = typ;
+        return &.{};
+    }
+};
+
+// https://developer.mozilla.org/en-US/docs/Web/API/PerformanceEntry
+pub const PerformanceEntry = struct {
+    const PerformanceEntryType = enum {
+        element,
+        event,
+        first_input,
+        largest_contentful_paint,
+        layout_shift,
+        long_animation_frame,
+        longtask,
+        mark,
+        measure,
+        navigation,
+        paint,
+        resource,
+        taskattribution,
+        visibility_state,
+
+        pub fn toString(self: PerformanceEntryType) []const u8 {
+            return switch (self) {
+                .first_input => "first-input",
+                .largest_contentful_paint => "largest-contentful-paint",
+                .layout_shift => "layout-shift",
+                .long_animation_frame => "long-animation-frame",
+                .visibility_state => "visibility-state",
+                else => @tagName(self),
+            };
+        }
+    };
+
+    duration: f64 = 0.0,
+    entry_type: PerformanceEntryType,
+    name: []const u8,
+    start_time: f64 = 0.0,
+
+    pub fn get_duration(self: *const PerformanceEntry) f64 {
+        return self.duration;
+    }
+
+    pub fn get_entryType(self: *const PerformanceEntry) PerformanceEntryType {
+        return self.entry_type;
+    }
+
+    pub fn get_name(self: *const PerformanceEntry) []const u8 {
+        return self.name;
+    }
+
+    pub fn get_startTime(self: *const PerformanceEntry) f64 {
+        return self.start_time;
+    }
+};
+
+// https://developer.mozilla.org/en-US/docs/Web/API/PerformanceMark
+pub const PerformanceMark = struct {
+    pub const prototype = *PerformanceEntry;
+
+    proto: PerformanceEntry,
+    detail: ?js.Object,
+
+    const Options = struct {
+        detail: ?js.Object = null,
+        startTime: ?f64 = null,
+    };
+
+    pub fn constructor(name: js.String, _options: ?Options, page: *Page) !PerformanceMark {
+        const perf = &page.window.performance;
+
+        const options = _options orelse Options{};
+        const start_time = options.startTime orelse @as(f64, @floatFromInt(perf._now()));
+
+        if (start_time < 0.0) {
+            return error.TypeError;
+        }
+
+        const detail = if (options.detail) |d| try d.persist() else null;
+        const proto = PerformanceEntry{ .name = name.string, .entry_type = .mark, .start_time = start_time };
+
+        return .{ .proto = proto, .detail = detail };
+    }
+
+    pub fn get_detail(self: *const PerformanceMark) ?js.Object {
+        return self.detail;
+    }
+};
+
+const testing = @import("./../../testing.zig");
+
+test "Performance: get_timeOrigin" {
+    var perf = Performance.init();
+    const time_origin = perf.get_timeOrigin();
+    try testing.expect(time_origin >= 0);
+}
+
+test "Performance: now" {
+    var perf = Performance.init();
+
+    // Monotonically increasing
+    var now = perf._now();
+    while (now <= 0) { // Loop for now to not be 0
+        try testing.expectEqual(now, 0);
+        now = perf._now();
+    }
+
+    var after = perf._now();
+    while (after <= now) { // Loop until after > now
+        try testing.expectEqual(after, now);
+        after = perf._now();
+    }
+}
+
+test "Browser: Performance.Mark" {
+    try testing.htmlRunner("dom/performance.html");
+}

src/browser/dom/performance_observer.zig

@@ -0,0 +1,58 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const js = @import("../js/js.zig");
+
+const PerformanceEntry = @import("performance.zig").PerformanceEntry;
+
+// https://developer.mozilla.org/en-US/docs/Web/API/PerformanceObserver
+pub const PerformanceObserver = struct {
+    pub const _supportedEntryTypes = [0][]const u8{};
+
+    pub fn constructor(cbk: js.Function) PerformanceObserver {
+        _ = cbk;
+        return .{};
+    }
+
+    pub fn _observe(self: *const PerformanceObserver, options_: ?Options) void {
+        _ = self;
+        _ = options_;
+        return;
+    }
+
+    pub fn _disconnect(self: *PerformanceObserver) void {
+        _ = self;
+    }
+
+    pub fn _takeRecords(_: *const PerformanceObserver) []PerformanceEntry {
+        return &[_]PerformanceEntry{};
+    }
+};
+
+const Options = struct {
+    buffered: ?bool = null,
+    durationThreshold: ?f64 = null,
+    entryTypes: ?[]const []const u8 = null,
+    type: ?[]const u8 = null,
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.PerformanceObserver" {
+    try testing.htmlRunner("dom/performance_observer.html");
+}

src/browser/dom/processing_instruction.zig

@@ -0,0 +1,92 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const parser = @import("../netsurf.zig");
+const Node = @import("node.zig").Node;
+const Page = @import("../page.zig").Page;
+
+// https://dom.spec.whatwg.org/#processinginstruction
+pub const ProcessingInstruction = struct {
+    pub const Self = parser.ProcessingInstruction;
+
+    // TODO for libdom processing instruction inherit from node.
+    // But the spec says it must inherit from CDATA.
+    pub const prototype = *Node;
+    pub const subtype = .node;
+
+    pub fn get_target(self: *parser.ProcessingInstruction) ![]const u8 {
+        // libdom stores the ProcessingInstruction target in the node's name.
+        return try parser.nodeName(parser.processingInstructionToNode(self));
+    }
+
+    // There's something wrong when we try to clone a ProcessInstruction normally.
+    // The resulting object can't be cast back into a node (it crashes). This is
+    // a simple workaround.
+    pub fn _cloneNode(self: *parser.ProcessingInstruction, _: ?bool, page: *Page) !*parser.ProcessingInstruction {
+        return try parser.documentCreateProcessingInstruction(
+            @ptrCast(page.window.document),
+            try get_target(self),
+            (try get_data(self)) orelse "",
+        );
+    }
+
+    pub fn get_data(self: *parser.ProcessingInstruction) !?[]const u8 {
+        return parser.nodeValue(parser.processingInstructionToNode(self));
+    }
+
+    pub fn set_data(self: *parser.ProcessingInstruction, data: []u8) !void {
+        try parser.nodeSetValue(parser.processingInstructionToNode(self), data);
+    }
+
+    // netsurf's ProcessInstruction doesn't implement the dom_node_get_attributes
+    // and thus will crash if we try to call nodeIsEqualNode.
+    pub fn _isEqualNode(self: *parser.ProcessingInstruction, other_node: *parser.Node) !bool {
+        if (parser.nodeType(other_node) != .processing_instruction) {
+            return false;
+        }
+
+        const other: *parser.ProcessingInstruction = @ptrCast(other_node);
+
+        if (std.mem.eql(u8, try get_target(self), try get_target(other)) == false) {
+            return false;
+        }
+
+        {
+            const self_data = try get_data(self);
+            const other_data = try get_data(other);
+            if (self_data == null and other_data != null) {
+                return false;
+            }
+            if (self_data != null and other_data == null) {
+                return false;
+            }
+            if (std.mem.eql(u8, self_data.?, other_data.?) == false) {
+                return false;
+            }
+        }
+
+        return true;
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.ProcessingInstruction" {
+    try testing.htmlRunner("dom/processing_instruction.html");
+}

src/browser/dom/range.zig

@@ -0,0 +1,390 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const parser = @import("../netsurf.zig");
+const Page = @import("../page.zig").Page;
+
+const Node = @import("node.zig").Node;
+const NodeUnion = @import("node.zig").Union;
+const DOMException = @import("exceptions.zig").DOMException;
+
+pub const Interfaces = .{
+    AbstractRange,
+    Range,
+};
+
+pub const AbstractRange = struct {
+    collapsed: bool,
+    end_node: *parser.Node,
+    end_offset: u32,
+    start_node: *parser.Node,
+    start_offset: u32,
+
+    pub fn updateCollapsed(self: *AbstractRange) void {
+        // TODO: Eventually, compare properly.
+        self.collapsed = false;
+    }
+
+    pub fn get_collapsed(self: *const AbstractRange) bool {
+        return self.collapsed;
+    }
+
+    pub fn get_endContainer(self: *const AbstractRange) !NodeUnion {
+        return Node.toInterface(self.end_node);
+    }
+
+    pub fn get_endOffset(self: *const AbstractRange) u32 {
+        return self.end_offset;
+    }
+
+    pub fn get_startContainer(self: *const AbstractRange) !NodeUnion {
+        return Node.toInterface(self.start_node);
+    }
+
+    pub fn get_startOffset(self: *const AbstractRange) u32 {
+        return self.start_offset;
+    }
+};
+
+pub const Range = struct {
+    pub const Exception = DOMException;
+    pub const prototype = *AbstractRange;
+
+    proto: AbstractRange,
+
+    pub const _START_TO_START = 0;
+    pub const _START_TO_END = 1;
+    pub const _END_TO_END = 2;
+    pub const _END_TO_START = 3;
+
+    // The Range() constructor returns a newly created Range object whose start
+    // and end is the global Document object.
+    // https://developer.mozilla.org/en-US/docs/Web/API/Range/Range
+    pub fn constructor(page: *Page) Range {
+        const proto: AbstractRange = .{
+            .collapsed = true,
+            .end_node = parser.documentHTMLToNode(page.window.document),
+            .end_offset = 0,
+            .start_node = parser.documentHTMLToNode(page.window.document),
+            .start_offset = 0,
+        };
+
+        return .{ .proto = proto };
+    }
+
+    pub fn _setStart(self: *Range, node: *parser.Node, offset_: i32) !void {
+        try ensureValidOffset(node, offset_);
+        const offset: u32 = @intCast(offset_);
+        const position = compare(node, offset, self.proto.end_node, self.proto.end_offset) catch |err| switch (err) {
+            error.WrongDocument => blk: {
+                // allow a node with a different root than the current, or
+                // a disconnected one. Treat it as if it's "after", so that
+                // we also update the end_offset and end_node.
+                break :blk 1;
+            },
+            else => return err,
+        };
+
+        if (position == 1) {
+            // if we're setting the node after the current end, the end must
+            // be set too.
+            self.proto.end_offset = offset;
+            self.proto.end_node = node;
+        }
+        self.proto.start_node = node;
+        self.proto.start_offset = offset;
+        self.proto.updateCollapsed();
+    }
+
+    pub fn _setStartBefore(self: *Range, node: *parser.Node) !void {
+        const parent, const index = try getParentAndIndex(node);
+        self.proto.start_node = parent;
+        self.proto.start_offset = index;
+    }
+
+    pub fn _setStartAfter(self: *Range, node: *parser.Node) !void {
+        const parent, const index = try getParentAndIndex(node);
+        self.proto.start_node = parent;
+        self.proto.start_offset = index + 1;
+    }
+
+    pub fn _setEnd(self: *Range, node: *parser.Node, offset_: i32) !void {
+        try ensureValidOffset(node, offset_);
+        const offset: u32 = @intCast(offset_);
+
+        const position = compare(node, offset, self.proto.start_node, self.proto.start_offset) catch |err| switch (err) {
+            error.WrongDocument => blk: {
+                // allow a node with a different root than the current, or
+                // a disconnected one. Treat it as if it's "before", so that
+                // we also update the end_offset and end_node.
+                break :blk -1;
+            },
+            else => return err,
+        };
+
+        if (position == -1) {
+            // if we're setting the node before the current start, the start
+            // must be set too.
+            self.proto.start_offset = offset;
+            self.proto.start_node = node;
+        }
+
+        self.proto.end_node = node;
+        self.proto.end_offset = offset;
+        self.proto.updateCollapsed();
+    }
+
+    pub fn _setEndBefore(self: *Range, node: *parser.Node) !void {
+        const parent, const index = try getParentAndIndex(node);
+        self.proto.end_node = parent;
+        self.proto.end_offset = index;
+    }
+
+    pub fn _setEndAfter(self: *Range, node: *parser.Node) !void {
+        const parent, const index = try getParentAndIndex(node);
+        self.proto.end_node = parent;
+        self.proto.end_offset = index + 1;
+    }
+
+    pub fn _createContextualFragment(_: *Range, fragment: []const u8, page: *Page) !*parser.DocumentFragment {
+        const document_html = page.window.document;
+        const document = parser.documentHTMLToDocument(document_html);
+        const doc_frag = try parser.documentParseFragmentFromStr(document, fragment);
+        return doc_frag;
+    }
+
+    pub fn _selectNodeContents(self: *Range, node: *parser.Node) !void {
+        self.proto.start_node = node;
+        self.proto.start_offset = 0;
+        self.proto.end_node = node;
+
+        // Set end_offset
+        switch (parser.nodeType(node)) {
+            .text, .cdata_section, .comment, .processing_instruction => {
+                // For text-like nodes, end_offset should be the length of the text data
+                if (parser.nodeValue(node)) |text_data| {
+                    self.proto.end_offset = @intCast(text_data.len);
+                } else {
+                    self.proto.end_offset = 0;
+                }
+            },
+            else => {
+                // For element and other nodes, end_offset is the number of children
+                const child_nodes = try parser.nodeGetChildNodes(node);
+                const child_count = parser.nodeListLength(child_nodes);
+                self.proto.end_offset = @intCast(child_count);
+            },
+        }
+
+        self.proto.updateCollapsed();
+    }
+
+    // creates a copy
+    pub fn _cloneRange(self: *const Range) Range {
+        return .{
+            .proto = .{
+                .collapsed = self.proto.collapsed,
+                .end_node = self.proto.end_node,
+                .end_offset = self.proto.end_offset,
+                .start_node = self.proto.start_node,
+                .start_offset = self.proto.start_offset,
+            },
+        };
+    }
+
+    pub fn _comparePoint(self: *const Range, node: *parser.Node, offset_: i32) !i32 {
+        const start = self.proto.start_node;
+        if (parser.nodeGetRootNode(start) != parser.nodeGetRootNode(node)) {
+            // WPT really wants this error to be first. Later, when we check
+            // if the relative position is 'disconnected', it'll also catch this
+            // case, but WPT will complain because it sometimes also sends
+            // invalid offsets, and it wants WrongDocument to be raised.
+            return error.WrongDocument;
+        }
+
+        if (parser.nodeType(node) == .document_type) {
+            return error.InvalidNodeType;
+        }
+
+        try ensureValidOffset(node, offset_);
+
+        const offset: u32 = @intCast(offset_);
+        if (try compare(node, offset, start, self.proto.start_offset) == -1) {
+            return -1;
+        }
+
+        if (try compare(node, offset, self.proto.end_node, self.proto.end_offset) == 1) {
+            return 1;
+        }
+
+        return 0;
+    }
+
+    pub fn _isPointInRange(self: *const Range, node: *parser.Node, offset_: i32) !bool {
+        return self._comparePoint(node, offset_) catch |err| switch (err) {
+            error.WrongDocument => return false,
+            else => return err,
+        } == 0;
+    }
+
+    pub fn _intersectsNode(self: *const Range, node: *parser.Node) !bool {
+        const start_root = parser.nodeGetRootNode(self.proto.start_node);
+        const node_root = parser.nodeGetRootNode(node);
+        if (start_root != node_root) {
+            return false;
+        }
+
+        const parent, const index = getParentAndIndex(node) catch |err| switch (err) {
+            error.InvalidNodeType => return true, // if node has no parent, we return true.
+            else => return err,
+        };
+
+        if (try compare(parent, index + 1, self.proto.start_node, self.proto.start_offset) != 1) {
+            // node isn't after start, can't intersect
+            return false;
+        }
+
+        if (try compare(parent, index, self.proto.end_node, self.proto.end_offset) != -1) {
+            // node isn't before end, can't intersect
+            return false;
+        }
+
+        return true;
+    }
+
+    pub fn _compareBoundaryPoints(self: *const Range, how: i32, other: *const Range) !i32 {
+        return switch (how) {
+            _START_TO_START => compare(self.proto.start_node, self.proto.start_offset, other.proto.start_node, other.proto.start_offset),
+            _START_TO_END => compare(self.proto.start_node, self.proto.start_offset, other.proto.end_node, other.proto.end_offset),
+            _END_TO_END => compare(self.proto.end_node, self.proto.end_offset, other.proto.end_node, other.proto.end_offset),
+            _END_TO_START => compare(self.proto.end_node, self.proto.end_offset, other.proto.start_node, other.proto.start_offset),
+            else => error.NotSupported, // this is the correct DOM Exception to return
+        };
+    }
+
+    // The Range.detach() method does nothing. It used to disable the Range
+    // object and enable the browser to release associated resources. The
+    // method has been kept for compatibility.
+    // https://developer.mozilla.org/en-US/docs/Web/API/Range/detach
+    pub fn _detach(_: *Range) void {}
+};
+
+fn ensureValidOffset(node: *parser.Node, offset: i32) !void {
+    if (offset < 0) {
+        return error.IndexSize;
+    }
+
+    // not >= because 0 seems to represent the node itself.
+    if (offset > try nodeLength(node)) {
+        return error.IndexSize;
+    }
+}
+
+fn nodeLength(node: *parser.Node) !usize {
+    switch (try isTextual(node)) {
+        true => return ((parser.nodeTextContent(node)) orelse "").len,
+        false => {
+            const children = try parser.nodeGetChildNodes(node);
+            return @intCast(parser.nodeListLength(children));
+        },
+    }
+}
+
+fn isTextual(node: *parser.Node) !bool {
+    return switch (parser.nodeType(node)) {
+        .text, .comment, .cdata_section => true,
+        else => false,
+    };
+}
+
+fn getParentAndIndex(child: *parser.Node) !struct { *parser.Node, u32 } {
+    const parent = (parser.nodeParentNode(child)) orelse return error.InvalidNodeType;
+    const children = try parser.nodeGetChildNodes(parent);
+    const ln = parser.nodeListLength(children);
+    var i: u32 = 0;
+    while (i < ln) {
+        defer i += 1;
+        const c = parser.nodeListItem(children, i) orelse continue;
+        if (c == child) {
+            return .{ parent, i };
+        }
+    }
+
+    // should not be possible to reach this point
+    return error.InvalidNodeType;
+}
+
+// implementation is largely copied from the WPT helper called getPosition in
+// the common.js of the dom folder.
+fn compare(node_a: *parser.Node, offset_a: u32, node_b: *parser.Node, offset_b: u32) !i32 {
+    if (node_a == node_b) {
+        // This is a simple and common case, where the two nodes are the same
+        // We just need to compare their offsets
+        if (offset_a == offset_b) {
+            return 0;
+        }
+        return if (offset_a < offset_b) -1 else 1;
+    }
+
+    // We're probably comparing two different nodes. "Probably", because the
+    // above case on considered the offset if the two nodes were the same
+    // as-is. They could still be the same here, if we first consider the
+    // offset.
+    const position = try Node._compareDocumentPosition(node_b, node_a);
+    if (position & @intFromEnum(parser.DocumentPosition.disconnected) == @intFromEnum(parser.DocumentPosition.disconnected)) {
+        return error.WrongDocument;
+    }
+
+    if (position & @intFromEnum(parser.DocumentPosition.following) == @intFromEnum(parser.DocumentPosition.following)) {
+        return switch (try compare(node_b, offset_b, node_a, offset_a)) {
+            -1 => 1,
+            1 => -1,
+            else => unreachable,
+        };
+    }
+
+    if (position & @intFromEnum(parser.DocumentPosition.contains) == @intFromEnum(parser.DocumentPosition.contains)) {
+        // node_a contains node_b
+        var child = node_b;
+        while (parser.nodeParentNode(child)) |parent| {
+            if (parent == node_a) {
+                // child.parentNode == node_a
+                break;
+            }
+            child = parent;
+        } else {
+            // this should not happen, because  Node._compareDocumentPosition
+            // has told us that node_a contains node_b, so one of node_b's
+            // parent's MUST be node_a. But somehow we do end up here sometimes.
+            return -1;
+        }
+
+        const child_parent, const child_index = try getParentAndIndex(child);
+        std.debug.assert(node_a == child_parent);
+        return if (offset_a <= child_index) -1 else 1;
+    }
+
+    return -1;
+}
+
+const testing = @import("../../testing.zig");
+test "Browser: Range" {
+    try testing.htmlRunner("dom/range.html");
+}

src/browser/dom/resize_observer.zig

@@ -17,32 +17,38 @@
 // along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
 const js = @import("../js/js.zig");
+const parser = @import("../netsurf.zig");
 
-const Document = @import("Document.zig");
-const Node = @import("Node.zig");
-
-const XMLDocument = @This();
-
-_proto: *Document,
-
-pub fn asDocument(self: *XMLDocument) *Document {
-    return self._proto;
-}
-
-pub fn asNode(self: *XMLDocument) *Node {
-    return self._proto.asNode();
-}
-
-pub fn asEventTarget(self: *XMLDocument) *@import("EventTarget.zig") {
-    return self._proto.asEventTarget();
-}
-
-pub const JsApi = struct {
-    pub const bridge = js.Bridge(XMLDocument);
-
-    pub const Meta = struct {
-        pub const name = "XMLDocument";
-        pub const prototype_chain = bridge.prototypeChain();
-        pub var class_id: bridge.ClassId = undefined;
-    };
+pub const Interfaces = .{
+    ResizeObserver,
+};
+
+// WEB IDL https://drafts.csswg.org/resize-observer/#resize-observer-interface
+pub const ResizeObserver = struct {
+    pub fn constructor(cbk: js.Function) ResizeObserver {
+        _ = cbk;
+        return .{};
+    }
+
+    pub fn _observe(self: *const ResizeObserver, element: *parser.Element, options_: ?Options) void {
+        _ = self;
+        _ = element;
+        _ = options_;
+        return;
+    }
+
+    pub fn _unobserve(self: *const ResizeObserver, element: *parser.Element) void {
+        _ = self;
+        _ = element;
+        return;
+    }
+
+    // TODO
+    pub fn _disconnect(self: *ResizeObserver) void {
+        _ = self;
+    }
+};
+
+const Options = struct {
+    box: []const u8,
 };

src/browser/dom/shadow_root.zig

@@ -0,0 +1,101 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const dump = @import("../dump.zig");
+const parser = @import("../netsurf.zig");
+
+const js = @import(".././js/js.zig");
+const Page = @import("../page.zig").Page;
+const Node = @import("node.zig").Node;
+const Element = @import("element.zig").Element;
+const ElementUnion = @import("element.zig").Union;
+
+// WEB IDL https://dom.spec.whatwg.org/#interface-shadowroot
+pub const ShadowRoot = struct {
+    pub const prototype = *parser.DocumentFragment;
+    pub const subtype = .node;
+
+    mode: Mode,
+    host: *parser.Element,
+    proto: *parser.DocumentFragment,
+    adopted_style_sheets: ?js.Object = null,
+
+    pub const Mode = enum {
+        open,
+        closed,
+    };
+
+    pub fn get_host(self: *const ShadowRoot) !ElementUnion {
+        return Element.toInterface(self.host);
+    }
+
+    pub fn get_adoptedStyleSheets(self: *ShadowRoot, page: *Page) !js.Object {
+        if (self.adopted_style_sheets) |obj| {
+            return obj;
+        }
+
+        const obj = try page.js.createArray(0).persist();
+        self.adopted_style_sheets = obj;
+        return obj;
+    }
+
+    pub fn set_adoptedStyleSheets(self: *ShadowRoot, sheets: js.Object) !void {
+        self.adopted_style_sheets = try sheets.persist();
+    }
+
+    pub fn get_innerHTML(self: *ShadowRoot, page: *Page) ![]const u8 {
+        var aw = std.Io.Writer.Allocating.init(page.call_arena);
+        try dump.writeChildren(parser.documentFragmentToNode(self.proto), .{}, &aw.writer);
+        return aw.written();
+    }
+
+    pub fn set_innerHTML(self: *ShadowRoot, str_: ?[]const u8) !void {
+        const sr_doc = parser.documentFragmentToNode(self.proto);
+        const doc = parser.nodeOwnerDocument(sr_doc) orelse return parser.DOMError.WrongDocument;
+        try Node.removeChildren(sr_doc);
+        const str = str_ orelse return;
+
+        const fragment = try parser.documentParseFragmentFromStr(doc, str);
+        const fragment_node = parser.documentFragmentToNode(fragment);
+
+        // Element.set_innerHTML also has some weirdness here. It isn't clear
+        // what should and shouldn't be set. Whatever string you pass to libdom,
+        // it always creates a full HTML document, with an html, head and body
+        // element.
+        // For ShadowRoot, it appears the only the children within the body should
+        // be set.
+        const html = parser.nodeFirstChild(fragment_node) orelse return;
+        const head = parser.nodeFirstChild(html) orelse return;
+        const body = parser.nodeNextSibling(head) orelse return;
+
+        const children = try parser.nodeGetChildNodes(body);
+        const ln = parser.nodeListLength(children);
+        for (0..ln) |_| {
+            // always index 0, because nodeAppendChild moves the node out of
+            // the nodeList and into the new tree
+            const child = parser.nodeListItem(children, 0) orelse continue;
+            _ = try parser.nodeAppendChild(sr_doc, child);
+        }
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.ShadowRoot" {
+    try testing.htmlRunner("dom/shadow_root.html");
+}

src/browser/dom/text.zig

@@ -0,0 +1,62 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const parser = @import("../netsurf.zig");
+const Page = @import("../page.zig").Page;
+
+const CharacterData = @import("character_data.zig").CharacterData;
+const CDATASection = @import("cdata_section.zig").CDATASection;
+
+// Text interfaces
+pub const Interfaces = .{
+    CDATASection,
+};
+
+pub const Text = struct {
+    pub const Self = parser.Text;
+    pub const prototype = *CharacterData;
+    pub const subtype = .node;
+
+    pub fn constructor(data: ?[]const u8, page: *const Page) !*parser.Text {
+        return parser.documentCreateTextNode(
+            parser.documentHTMLToDocument(page.window.document),
+            data orelse "",
+        );
+    }
+
+    // JS funcs
+    // --------
+
+    // Read attributes
+
+    pub fn get_wholeText(self: *parser.Text) ![]const u8 {
+        return try parser.textWholdeText(self);
+    }
+
+    // JS methods
+    // ----------
+
+    pub fn _splitText(self: *parser.Text, offset: u32) !*parser.Text {
+        return try parser.textSplitText(self, offset);
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.Text" {
+    try testing.htmlRunner("dom/text.html");
+}

src/browser/dom/token_list.zig

@@ -0,0 +1,174 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const js = @import("../js/js.zig");
+const log = @import("../../log.zig");
+const parser = @import("../netsurf.zig");
+const iterator = @import("../iterator/iterator.zig");
+
+const DOMException = @import("exceptions.zig").DOMException;
+
+pub const Interfaces = .{
+    DOMTokenList,
+    DOMTokenListIterable,
+    TokenListEntriesIterator,
+    TokenListEntriesIterator.Iterable,
+};
+
+// https://dom.spec.whatwg.org/#domtokenlist
+pub const DOMTokenList = struct {
+    pub const Self = parser.TokenList;
+    pub const Exception = DOMException;
+
+    pub fn get_length(self: *parser.TokenList) !u32 {
+        return parser.tokenListGetLength(self);
+    }
+
+    pub fn _item(self: *parser.TokenList, index: u32) !?[]const u8 {
+        return parser.tokenListItem(self, index);
+    }
+
+    pub fn _contains(self: *parser.TokenList, token: []const u8) !bool {
+        return parser.tokenListContains(self, token);
+    }
+
+    pub fn _add(self: *parser.TokenList, tokens: []const []const u8) !void {
+        for (tokens) |token| {
+            try parser.tokenListAdd(self, token);
+        }
+    }
+
+    pub fn _remove(self: *parser.TokenList, tokens: []const []const u8) !void {
+        for (tokens) |token| {
+            try parser.tokenListRemove(self, token);
+        }
+    }
+
+    /// If token is the empty string, then throw a "SyntaxError" DOMException.
+    /// If token contains any ASCII whitespace, then throw an
+    /// "InvalidCharacterError" DOMException.
+    fn validateToken(token: []const u8) !void {
+        if (token.len == 0) {
+            return parser.DOMError.Syntax;
+        }
+        for (token) |c| {
+            if (std.ascii.isWhitespace(c)) return parser.DOMError.InvalidCharacter;
+        }
+    }
+
+    pub fn _toggle(self: *parser.TokenList, token: []const u8, force: ?bool) !bool {
+        try validateToken(token);
+        const exists = try parser.tokenListContains(self, token);
+        if (exists) {
+            if (force == null or force.? == false) {
+                try parser.tokenListRemove(self, token);
+                return false;
+            }
+            return true;
+        }
+
+        if (force == null or force.? == true) {
+            try parser.tokenListAdd(self, token);
+            return true;
+        }
+        return false;
+    }
+
+    pub fn _replace(self: *parser.TokenList, token: []const u8, new: []const u8) !bool {
+        try validateToken(token);
+        try validateToken(new);
+        const exists = try parser.tokenListContains(self, token);
+        if (!exists) return false;
+        try parser.tokenListRemove(self, token);
+        try parser.tokenListAdd(self, new);
+        return true;
+    }
+
+    // TODO to implement.
+    pub fn _supports(_: *parser.TokenList, token: []const u8) !bool {
+        try validateToken(token);
+        return error.TypeError;
+    }
+
+    pub fn get_value(self: *parser.TokenList) !?[]const u8 {
+        return (try parser.tokenListGetValue(self)) orelse "";
+    }
+
+    pub fn set_value(self: *parser.TokenList, value: []const u8) !void {
+        return parser.tokenListSetValue(self, value);
+    }
+
+    pub fn _toString(self: *parser.TokenList) ![]const u8 {
+        return (try get_value(self)) orelse "";
+    }
+
+    pub fn _keys(self: *parser.TokenList) !iterator.U32Iterator {
+        return .{ .length = try get_length(self) };
+    }
+
+    pub fn _values(self: *parser.TokenList) DOMTokenListIterable {
+        return DOMTokenListIterable.init(.{ .token_list = self });
+    }
+
+    pub fn _entries(self: *parser.TokenList) TokenListEntriesIterator {
+        return TokenListEntriesIterator.init(.{ .token_list = self });
+    }
+
+    pub fn _symbol_iterator(self: *parser.TokenList) DOMTokenListIterable {
+        return _values(self);
+    }
+
+    // TODO handle thisArg
+    pub fn _forEach(self: *parser.TokenList, cbk: js.Function, this_arg: js.Object) !void {
+        var entries = _entries(self);
+        while (try entries._next()) |entry| {
+            var result: js.Function.Result = undefined;
+            cbk.tryCallWithThis(void, this_arg, .{ entry.@"1", entry.@"0", self }, &result) catch {
+                log.debug(.user_script, "callback error", .{
+                    .err = result.exception,
+                    .stack = result.stack,
+                    .soure = "tokenList foreach",
+                });
+            };
+        }
+    }
+};
+
+const DOMTokenListIterable = iterator.Iterable(Iterator, "DOMTokenListIterable");
+const TokenListEntriesIterator = iterator.NumericEntries(Iterator, "TokenListEntriesIterator");
+
+pub const Iterator = struct {
+    index: u32 = 0,
+    token_list: *parser.TokenList,
+
+    // used when wrapped in an iterator.NumericEntries
+    pub const Error = parser.DOMError;
+
+    pub fn _next(self: *Iterator) !?[]const u8 {
+        const index = self.index;
+        self.index = index + 1;
+        return DOMTokenList._item(self.token_list, index);
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: DOM.TokenList" {
+    try testing.htmlRunner("dom/token_list.html");
+}

src/browser/dom/tree_walker.zig

@@ -0,0 +1,314 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const js = @import("../js/js.zig");
+const parser = @import("../netsurf.zig");
+
+const NodeFilter = @import("node_filter.zig");
+const Node = @import("node.zig").Node;
+const NodeUnion = @import("node.zig").Union;
+
+// https://developer.mozilla.org/en-US/docs/Web/API/TreeWalker
+pub const TreeWalker = struct {
+    root: *parser.Node,
+    current_node: *parser.Node,
+    what_to_show: u32,
+    filter: ?TreeWalkerOpts,
+    filter_func: ?js.Function,
+
+    // One of the few cases where null and undefined resolve to different default.
+    // We need the raw JsObject so that we can probe the tri state:
+    // null, undefined or i32.
+    pub const WhatToShow = js.Object;
+
+    pub const TreeWalkerOpts = union(enum) {
+        function: js.Function,
+        object: struct { acceptNode: js.Function },
+    };
+
+    pub fn init(node: *parser.Node, what_to_show_: ?WhatToShow, filter: ?TreeWalkerOpts) !TreeWalker {
+        var filter_func: ?js.Function = null;
+
+        if (filter) |f| {
+            filter_func = switch (f) {
+                .function => |func| func,
+                .object => |o| o.acceptNode,
+            };
+        }
+
+        var what_to_show: u32 = undefined;
+        if (what_to_show_) |wts| {
+            switch (try wts.triState(TreeWalker, "what_to_show", u32)) {
+                .null => what_to_show = 0,
+                .undefined => what_to_show = NodeFilter.NodeFilter._SHOW_ALL,
+                .value => |v| what_to_show = v,
+            }
+        } else {
+            what_to_show = NodeFilter.NodeFilter._SHOW_ALL;
+        }
+
+        return .{
+            .root = node,
+            .current_node = node,
+            .what_to_show = what_to_show,
+            .filter = filter,
+            .filter_func = filter_func,
+        };
+    }
+
+    pub fn get_root(self: *TreeWalker) !NodeUnion {
+        return try Node.toInterface(self.root);
+    }
+
+    pub fn get_currentNode(self: *TreeWalker) !NodeUnion {
+        return try Node.toInterface(self.current_node);
+    }
+
+    pub fn get_whatToShow(self: *TreeWalker) u32 {
+        return self.what_to_show;
+    }
+
+    pub fn get_filter(self: *TreeWalker) ?TreeWalkerOpts {
+        return self.filter;
+    }
+
+    pub fn set_currentNode(self: *TreeWalker, node: *parser.Node) !void {
+        self.current_node = node;
+    }
+
+    fn firstChild(self: *const TreeWalker, node: *parser.Node) !?*parser.Node {
+        const children = try parser.nodeGetChildNodes(node);
+        const child_count = parser.nodeListLength(children);
+
+        for (0..child_count) |i| {
+            const index: u32 = @intCast(i);
+            const child = (parser.nodeListItem(children, index)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, child)) {
+                .accept => return child,
+                .reject => continue,
+                .skip => if (try self.firstChild(child)) |gchild| return gchild,
+            }
+        }
+
+        return null;
+    }
+
+    fn lastChild(self: *const TreeWalker, node: *parser.Node) !?*parser.Node {
+        const children = try parser.nodeGetChildNodes(node);
+        const child_count = parser.nodeListLength(children);
+
+        var index: u32 = child_count;
+        while (index > 0) {
+            index -= 1;
+            const child = (parser.nodeListItem(children, index)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, child)) {
+                .accept => return child,
+                .reject => continue,
+                .skip => if (try self.lastChild(child)) |gchild| return gchild,
+            }
+        }
+
+        return null;
+    }
+
+    fn nextSibling(self: *const TreeWalker, node: *parser.Node) !?*parser.Node {
+        var current = node;
+
+        while (true) {
+            current = (parser.nodeNextSibling(current)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, current)) {
+                .accept => return current,
+                .skip, .reject => continue,
+            }
+        }
+
+        return null;
+    }
+
+    // Get the next sibling that is either acceptable or should be descended into (skip)
+    fn nextSiblingOrSkip(self: *const TreeWalker, node: *parser.Node) !?struct { node: *parser.Node, should_descend: bool } {
+        var current = node;
+
+        while (true) {
+            current = (parser.nodeNextSibling(current)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, current)) {
+                .accept => return .{ .node = current, .should_descend = false },
+                .skip => return .{ .node = current, .should_descend = true },
+                .reject => continue,
+            }
+        }
+
+        return null;
+    }
+
+    fn previousSibling(self: *const TreeWalker, node: *parser.Node) !?*parser.Node {
+        var current = node;
+
+        while (true) {
+            current = (parser.nodePreviousSibling(current)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, current)) {
+                .accept => return current,
+                .skip, .reject => continue,
+            }
+        }
+
+        return null;
+    }
+
+    fn parentNode(self: *const TreeWalker, node: *parser.Node) !?*parser.Node {
+        if (self.root == node) return null;
+
+        var current = node;
+        while (true) {
+            if (current == self.root) return null;
+            current = (parser.nodeParentNode(current)) orelse return null;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, current)) {
+                .accept => return current,
+                .reject, .skip => continue,
+            }
+        }
+    }
+
+    pub fn _firstChild(self: *TreeWalker) !?NodeUnion {
+        if (try self.firstChild(self.current_node)) |child| {
+            self.current_node = child;
+            return try Node.toInterface(child);
+        }
+
+        return null;
+    }
+
+    pub fn _lastChild(self: *TreeWalker) !?NodeUnion {
+        if (try self.lastChild(self.current_node)) |child| {
+            self.current_node = child;
+            return try Node.toInterface(child);
+        }
+
+        return null;
+    }
+
+    pub fn _nextNode(self: *TreeWalker) !?NodeUnion {
+        var current = self.current_node;
+
+        // First, try to go to first child of current node
+        if (try self.firstChild(current)) |child| {
+            self.current_node = child;
+            return try Node.toInterface(child);
+        }
+
+        // No acceptable children, move to next node in tree
+        while (current != self.root) {
+            const result = try self.nextSiblingOrSkip(current) orelse {
+                // No next sibling, go up to parent and continue
+                // or, if there is no parent, we're done
+                current = (parser.nodeParentNode(current)) orelse break;
+                continue;
+            };
+
+            if (!result.should_descend) {
+                // This is an .accept node - return it
+                self.current_node = result.node;
+                return try Node.toInterface(result.node);
+            }
+
+            // This is a .skip node - try to find acceptable children within it
+            if (try self.firstChild(result.node)) |child| {
+                self.current_node = child;
+                return try Node.toInterface(child);
+            }
+            // No acceptable children, continue looking at this node's siblings
+            current = result.node;
+        }
+
+        return null;
+    }
+
+    pub fn _nextSibling(self: *TreeWalker) !?NodeUnion {
+        if (try self.nextSibling(self.current_node)) |sibling| {
+            self.current_node = sibling;
+            return try Node.toInterface(sibling);
+        }
+
+        return null;
+    }
+
+    pub fn _parentNode(self: *TreeWalker) !?NodeUnion {
+        if (try self.parentNode(self.current_node)) |parent| {
+            self.current_node = parent;
+            return try Node.toInterface(parent);
+        }
+
+        return null;
+    }
+
+    pub fn _previousNode(self: *TreeWalker) !?NodeUnion {
+        if (self.current_node == self.root) return null;
+
+        var current = self.current_node;
+        while (parser.nodePreviousSibling(current)) |previous| {
+            current = previous;
+
+            switch (try NodeFilter.verify(self.what_to_show, self.filter_func, current)) {
+                .accept => {
+                    // Get last child if it has one.
+                    if (try self.lastChild(current)) |child| {
+                        self.current_node = child;
+                        return try Node.toInterface(child);
+                    }
+
+                    // Otherwise, this node is our previous one.
+                    self.current_node = current;
+                    return try Node.toInterface(current);
+                },
+                .reject => continue,
+                .skip => {
+                    // Get last child if it has one.
+                    if (try self.lastChild(current)) |child| {
+                        self.current_node = child;
+                        return try Node.toInterface(child);
+                    }
+                },
+            }
+        }
+
+        if (current != self.root) {
+            if (try self.parentNode(current)) |parent| {
+                self.current_node = parent;
+                return try Node.toInterface(parent);
+            }
+        }
+
+        return null;
+    }
+
+    pub fn _previousSibling(self: *TreeWalker) !?NodeUnion {
+        if (try self.previousSibling(self.current_node)) |sibling| {
+            self.current_node = sibling;
+            return try Node.toInterface(sibling);
+        }
+
+        return null;
+    }
+};

src/browser/dom/walker.zig

@@ -0,0 +1,102 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const parser = @import("../netsurf.zig");
+
+pub const Walker = union(enum) {
+    walkerDepthFirst: WalkerDepthFirst,
+    walkerChildren: WalkerChildren,
+    walkerNone: WalkerNone,
+
+    pub fn get_next(self: Walker, root: *parser.Node, cur: ?*parser.Node) !?*parser.Node {
+        switch (self) {
+            inline else => |case| return case.get_next(root, cur),
+        }
+    }
+};
+
+// WalkerDepthFirst iterates over the DOM tree to return the next following
+// node or null at the end.
+//
+// This implementation is a zig version of Netsurf code.
+// http://source.netsurf-browser.org/libdom.git/tree/src/html/html_collection.c#n177
+//
+// The iteration is a depth first as required by the specification.
+// https://dom.spec.whatwg.org/#htmlcollection
+// https://dom.spec.whatwg.org/#concept-tree-order
+pub const WalkerDepthFirst = struct {
+    pub fn get_next(_: WalkerDepthFirst, root: *parser.Node, cur: ?*parser.Node) !?*parser.Node {
+        var n = cur orelse root;
+
+        // TODO deinit next
+        if (parser.nodeFirstChild(n)) |next| {
+            return next;
+        }
+
+        // TODO deinit next
+        if (parser.nodeNextSibling(n)) |next| {
+            return next;
+        }
+
+        // TODO deinit parent
+        // Back to the parent of cur.
+        // If cur has no parent, then the iteration is over.
+        var parent = parser.nodeParentNode(n) orelse return null;
+
+        // TODO deinit lastchild
+        var lastchild = parser.nodeLastChild(parent);
+        while (n != root and n == lastchild) {
+            n = parent;
+
+            // TODO deinit parent
+            // Back to the prev's parent.
+            // If prev has no parent, then the loop must stop.
+            parent = parser.nodeParentNode(n) orelse break;
+
+            // TODO deinit lastchild
+            lastchild = parser.nodeLastChild(parent);
+        }
+
+        if (n == root) {
+            return null;
+        }
+
+        return parser.nodeNextSibling(n);
+    }
+};
+
+// WalkerChildren iterates over the root's children only.
+pub const WalkerChildren = struct {
+    pub fn get_next(_: WalkerChildren, root: *parser.Node, cur: ?*parser.Node) !?*parser.Node {
+        // On walk start, we return the first root's child.
+        if (cur == null) return parser.nodeFirstChild(root);
+
+        // If cur is root, then return null.
+        // This is a special case, if the root is included in the walk, we
+        // don't want to go further to find children.
+        if (root == cur.?) return null;
+
+        return parser.nodeNextSibling(cur.?);
+    }
+};
+
+pub const WalkerNone = struct {
+    pub fn get_next(_: WalkerNone, _: *parser.Node, _: ?*parser.Node) !?*parser.Node {
+        return null;
+    }
+};

src/browser/dump.zig

@@ -1,4 +1,4 @@
-// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
 //
 // Francis Bouvier <francis@lightpanda.io>
 // Pierre Tachoire <pierre@lightpanda.io>
@@ -17,344 +17,323 @@
 // along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
 const std = @import("std");
-const Page = @import("Page.zig");
-const Node = @import("webapi/Node.zig");
-const Slot = @import("webapi/element/html/Slot.zig");
-const IFrame = @import("webapi/element/html/IFrame.zig");
 
-const IS_DEBUG = @import("builtin").mode == .Debug;
+const parser = @import("netsurf.zig");
+const Page = @import("page.zig").Page;
+const Walker = @import("dom/walker.zig").WalkerChildren;
 
 pub const Opts = struct {
-    with_base: bool = false,
-    with_frames: bool = false,
-    strip: Opts.Strip = .{},
-    shadow: Opts.Shadow = .rendered,
+    // set to include element shadowroots in the dump
+    page: ?*const Page = null,
 
-    pub const Strip = struct {
+    strip_mode: StripMode = .{},
+
+    pub const StripMode = struct {
         js: bool = false,
         ui: bool = false,
         css: bool = false,
     };
-
-    pub const Shadow = enum {
-        // Skip shadow DOM entirely (innerHTML/outerHTML)
-        skip,
-
-        // Dump everyhting (like "view source")
-        complete,
-
-        // Resolve slot elements (like what actually gets rendered)
-        rendered,
-    };
 };
 
-pub fn root(doc: *Node.Document, opts: Opts, writer: *std.Io.Writer, page: *Page) !void {
-    if (doc.is(Node.Document.HTMLDocument)) |html_doc| {
-        blk: {
-            // Ideally we just render the doctype which is part of the document
-            if (doc.asNode().firstChild()) |first| {
-                if (first._type == .document_type) {
-                    break :blk;
-                }
-            }
-            // But if the doc has no child, or the first child isn't a doctype
-            // well force it.
-            try writer.writeAll("<!DOCTYPE html>");
-        }
+// writer must be a std.io.Writer
+pub fn writeHTML(doc: *parser.Document, opts: Opts, writer: *std.Io.Writer) !void {
+    try writer.writeAll("<!DOCTYPE html>\n");
+    try writeChildren(parser.documentToNode(doc), opts, writer);
+    try writer.writeAll("\n");
+}
 
-        if (opts.with_base) {
-            const parent = if (html_doc.getHead()) |head| head.asNode() else doc.asNode();
-            const base = try doc.createElement("base", null, page);
-            try base.setAttributeSafe(comptime .wrap("base"), .wrap(page.base()), page);
-            _ = try parent.insertBefore(base.asNode(), parent.firstChild(), page);
-        }
+// Spec: https://www.w3.org/TR/xml/#sec-prolog-dtd
+pub fn writeDocType(doc_type: *parser.DocumentType, writer: *std.Io.Writer) !void {
+    try writer.writeAll("<!DOCTYPE ");
+    try writer.writeAll(try parser.documentTypeGetName(doc_type));
+
+    const public_id = parser.documentTypeGetPublicId(doc_type);
+    const system_id = parser.documentTypeGetSystemId(doc_type);
+    if (public_id.len != 0 and system_id.len != 0) {
+        try writer.writeAll(" PUBLIC \"");
+        try writeEscapedAttributeValue(writer, public_id);
+        try writer.writeAll("\" \"");
+        try writeEscapedAttributeValue(writer, system_id);
+        try writer.writeAll("\"");
+    } else if (public_id.len != 0) {
+        try writer.writeAll(" PUBLIC \"");
+        try writeEscapedAttributeValue(writer, public_id);
+        try writer.writeAll("\"");
+    } else if (system_id.len != 0) {
+        try writer.writeAll(" SYSTEM \"");
+        try writeEscapedAttributeValue(writer, system_id);
+        try writer.writeAll("\"");
     }
-
-    return deep(doc.asNode(), opts, writer, page);
+    // Internal subset is not implemented
+    try writer.writeAll(">");
 }
 
-pub fn deep(node: *Node, opts: Opts, writer: *std.Io.Writer, page: *Page) error{WriteFailed}!void {
-    return _deep(node, opts, false, writer, page);
-}
-
-fn _deep(node: *Node, opts: Opts, comptime force_slot: bool, writer: *std.Io.Writer, page: *Page) error{WriteFailed}!void {
-    switch (node._type) {
-        .cdata => |cd| {
-            if (node.is(Node.CData.Comment)) |_| {
-                try writer.writeAll("<!--");
-                try writer.writeAll(cd.getData().str());
-                try writer.writeAll("-->");
-            } else if (node.is(Node.CData.ProcessingInstruction)) |pi| {
-                try writer.writeAll("<?");
-                try writer.writeAll(pi._target);
-                try writer.writeAll(" ");
-                try writer.writeAll(cd.getData().str());
-                try writer.writeAll("?>");
-            } else {
-                if (shouldEscapeText(node._parent)) {
-                    try writeEscapedText(cd.getData().str(), writer);
-                } else {
-                    try writer.writeAll(cd.getData().str());
-                }
-            }
-        },
-        .element => |el| {
-            if (shouldStripElement(el, opts)) {
+pub fn writeNode(node: *parser.Node, opts: Opts, writer: *std.Io.Writer) anyerror!void {
+    switch (parser.nodeType(node)) {
+        .element => {
+            // open the tag
+            const tag_type = try parser.nodeHTMLGetTagType(node) orelse .undef;
+            if (try isStripped(tag_type, node, opts.strip_mode)) {
                 return;
             }
 
-            // When opts.shadow == .rendered, we normally skip any element with
-            // a slot attribute. Only the "active" element will get rendered into
-            // the <slot name="X">. However, the `deep` function is itself used
-            // to render that "active" content, so when we're trying to render
-            // it, we don't want to skip it.
-            if ((comptime force_slot == false) and opts.shadow == .rendered) {
-                if (el.getAttributeSafe(comptime .wrap("slot"))) |_| {
-                    // Skip - will be rendered by the Slot if it's the active container
-                    return;
+            const tag = try parser.nodeLocalName(node);
+            try writer.writeAll("<");
+            try writer.writeAll(tag);
+
+            // write the attributes
+            const _map = try parser.nodeGetAttributes(node);
+            if (_map) |map| {
+                const ln = try parser.namedNodeMapGetLength(map);
+                for (0..ln) |i| {
+                    const attr = try parser.namedNodeMapItem(map, @intCast(i)) orelse break;
+                    try writer.writeAll(" ");
+                    try writer.writeAll(try parser.attributeGetName(attr));
+                    try writer.writeAll("=\"");
+                    const attribute_value = try parser.attributeGetValue(attr) orelse "";
+                    try writeEscapedAttributeValue(writer, attribute_value);
+                    try writer.writeAll("\"");
                 }
             }
 
-            try el.format(writer);
+            try writer.writeAll(">");
 
-            if (opts.shadow == .rendered) {
-                if (el.is(Slot)) |slot| {
-                    try dumpSlotContent(slot, opts, writer, page);
-                    return writer.writeAll("</slot>");
-                }
-            }
-            if (opts.shadow != .skip) {
-                if (page._element_shadow_roots.get(el)) |shadow| {
-                    try children(shadow.asNode(), opts, writer, page);
-                    // In rendered mode, light DOM is only shown through slots, not directly
-                    if (opts.shadow == .rendered) {
-                        // Skip rendering light DOM children
-                        if (!isVoidElement(el)) {
-                            try writer.writeAll("</");
-                            try writer.writeAll(el.getTagNameDump());
-                            try writer.writeByte('>');
-                        }
-                        return;
+            if (opts.page) |page| {
+                if (page.getNodeState(node)) |state| {
+                    if (state.shadow_root) |sr| {
+                        try writeChildren(@ptrCast(@alignCast(sr.proto)), opts, writer);
                     }
                 }
             }
 
-            if (opts.with_frames and el.is(IFrame) != null) {
-                const frame = el.as(IFrame);
-                if (frame.getContentDocument()) |doc| {
-                    // A frame's document should always ahave a page, but
-                    // I'm not willing to crash a release build on that assertion.
-                    if (comptime IS_DEBUG) {
-                        std.debug.assert(doc._page != null);
-                    }
-                    if (doc._page) |frame_page| {
-                        try writer.writeByte('\n');
-                        root(doc, opts, writer, frame_page) catch return error.WriteFailed;
-                        try writer.writeByte('\n');
-                    }
-                }
+            // void elements can't have any content.
+            if (try isVoid(parser.nodeToElement(node))) return;
+
+            if (tag_type == .script) {
+                try writer.writeAll(parser.nodeTextContent(node) orelse "");
             } else {
-                try children(node, opts, writer, page);
+                // write the children
+                // TODO avoid recursion
+                try writeChildren(node, opts, writer);
             }
 
-            if (!isVoidElement(el)) {
-                try writer.writeAll("</");
-                try writer.writeAll(el.getTagNameDump());
-                try writer.writeByte('>');
-            }
+            // close the tag
+            try writer.writeAll("</");
+            try writer.writeAll(tag);
+            try writer.writeAll(">");
         },
-        .document => try children(node, opts, writer, page),
-        .document_type => |dt| {
-            try writer.writeAll("<!DOCTYPE ");
-            try writer.writeAll(dt.getName());
-
-            const public_id = dt.getPublicId();
-            const system_id = dt.getSystemId();
-            if (public_id.len != 0 and system_id.len != 0) {
-                try writer.writeAll(" PUBLIC \"");
-                try writeEscapedText(public_id, writer);
-                try writer.writeAll("\" \"");
-                try writeEscapedText(system_id, writer);
-                try writer.writeByte('"');
-            } else if (public_id.len != 0) {
-                try writer.writeAll(" PUBLIC \"");
-                try writeEscapedText(public_id, writer);
-                try writer.writeByte('"');
-            } else if (system_id.len != 0) {
-                try writer.writeAll(" SYSTEM \"");
-                try writeEscapedText(system_id, writer);
-                try writer.writeByte('"');
-            }
-            try writer.writeAll(">\n");
+        .text => {
+            const v = parser.nodeValue(node) orelse return;
+            try writeEscapedTextNode(writer, v);
         },
-        .document_fragment => try children(node, opts, writer, page),
-        .attribute => {
-            // Not called normally, but can be called via XMLSerializer.serializeToString
-            // in which case it should return an empty string
-            try writer.writeAll("");
+        .cdata_section => {
+            const v = parser.nodeValue(node) orelse return;
+            try writer.writeAll("<![CDATA[");
+            try writer.writeAll(v);
+            try writer.writeAll("]]>");
         },
+        .comment => {
+            const v = parser.nodeValue(node) orelse return;
+            try writer.writeAll("<!--");
+            try writer.writeAll(v);
+            try writer.writeAll("-->");
+        },
+        // TODO handle processing instruction dump
+        .processing_instruction => return,
+        // document fragment is outside of the main document DOM, so we
+        // don't output it.
+        .document_fragment => return,
+        // document will never be called, but required for completeness.
+        .document => return,
+        // done globally instead, but required for completeness. Only the outer DOCTYPE should be written
+        .document_type => return,
+        // deprecated
+        .attribute => return,
+        .entity_reference => return,
+        .entity => return,
+        .notation => return,
     }
 }
 
-pub fn children(parent: *Node, opts: Opts, writer: *std.Io.Writer, page: *Page) !void {
-    var it = parent.childrenIterator();
-    while (it.next()) |child| {
-        try deep(child, opts, writer, page);
+// writer must be a std.io.Writer
+pub fn writeChildren(root: *parser.Node, opts: Opts, writer: *std.Io.Writer) !void {
+    const walker = Walker{};
+    var next: ?*parser.Node = null;
+    while (true) {
+        next = try walker.get_next(root, next) orelse break;
+        try writeNode(next.?, opts, writer);
     }
 }
 
-pub fn toJSON(node: *Node, writer: *std.json.Stringify) !void {
-    try writer.beginObject();
-
-    try writer.objectField("type");
-    switch (node.type) {
-        .cdata => {
-            try writer.write("cdata");
-        },
-        .document => {
-            try writer.write("document");
-        },
-        .document_type => {
-            try writer.write("document_type");
-        },
-        .element => |*el| {
-            try writer.write("element");
-            try writer.objectField("tag");
-            try writer.write(el.tagName());
-
-            try writer.objectField("attributes");
-            try writer.beginObject();
-            var it = el.attributeIterator();
-            while (it.next()) |attr| {
-                try writer.objectField(attr.name);
-                try writer.write(attr.value);
-            }
-            try writer.endObject();
-        },
+fn isStripped(tag_type: parser.Tag, node: *parser.Node, strip_mode: Opts.StripMode) !bool {
+    if (strip_mode.js and try isJsRelated(tag_type, node)) {
+        return true;
     }
 
-    try writer.objectField("children");
-    try writer.beginArray();
-    var it = node.childrenIterator();
-    while (it.next()) |child| {
-        try toJSON(child, writer);
-    }
-    try writer.endArray();
-    try writer.endObject();
-}
-
-fn dumpSlotContent(slot: *Slot, opts: Opts, writer: *std.Io.Writer, page: *Page) !void {
-    const assigned = slot.assignedNodes(null, page) catch return;
-
-    if (assigned.len > 0) {
-        for (assigned) |assigned_node| {
-            try _deep(assigned_node, opts, true, writer, page);
-        }
-    } else {
-        try children(slot.asNode(), opts, writer, page);
-    }
-}
-
-fn isVoidElement(el: *const Node.Element) bool {
-    return switch (el._type) {
-        .html => |html| switch (html._type) {
-            .br, .hr, .img, .input, .link, .meta => true,
-            else => false,
-        },
-        .svg => false,
-    };
-}
-
-fn shouldStripElement(el: *const Node.Element, opts: Opts) bool {
-    const tag_name = el.getTagNameDump();
-
-    if (opts.strip.js) {
-        if (std.mem.eql(u8, tag_name, "script")) return true;
-        if (std.mem.eql(u8, tag_name, "noscript")) return true;
-
-        if (std.mem.eql(u8, tag_name, "link")) {
-            if (el.getAttributeSafe(comptime .wrap("as"))) |as| {
-                if (std.mem.eql(u8, as, "script")) return true;
-            }
-            if (el.getAttributeSafe(comptime .wrap("rel"))) |rel| {
-                if (std.mem.eql(u8, rel, "modulepreload") or std.mem.eql(u8, rel, "preload")) {
-                    if (el.getAttributeSafe(comptime .wrap("as"))) |as| {
-                        if (std.mem.eql(u8, as, "script")) return true;
-                    }
-                }
-            }
-        }
+    if (strip_mode.css and try isCssRelated(tag_type, node)) {
+        return true;
     }
 
-    if (opts.strip.css or opts.strip.ui) {
-        if (std.mem.eql(u8, tag_name, "style")) return true;
-
-        if (std.mem.eql(u8, tag_name, "link")) {
-            if (el.getAttributeSafe(comptime .wrap("rel"))) |rel| {
-                if (std.mem.eql(u8, rel, "stylesheet")) return true;
-            }
-        }
+    if (strip_mode.ui and try isUIRelated(tag_type, node)) {
+        return true;
     }
-
-    if (opts.strip.ui) {
-        if (std.mem.eql(u8, tag_name, "img")) return true;
-        if (std.mem.eql(u8, tag_name, "picture")) return true;
-        if (std.mem.eql(u8, tag_name, "video")) return true;
-        if (std.mem.eql(u8, tag_name, "audio")) return true;
-        if (std.mem.eql(u8, tag_name, "svg")) return true;
-        if (std.mem.eql(u8, tag_name, "canvas")) return true;
-        if (std.mem.eql(u8, tag_name, "iframe")) return true;
-    }
-
     return false;
 }
 
-fn shouldEscapeText(node_: ?*Node) bool {
-    const node = node_ orelse return true;
-    if (node.is(Node.Element.Html.Script) != null) {
-        return false;
+fn isJsRelated(tag_type: parser.Tag, node: *parser.Node) !bool {
+    if (tag_type == .script) {
+        return true;
     }
-    // When scripting is enabled, <noscript> is a raw text element per the HTML spec
-    // (https://html.spec.whatwg.org/multipage/parsing.html#serialising-html-fragments).
-    // Its text content must not be HTML-escaped during serialization.
-    if (node.is(Node.Element.Html.Generic)) |generic| {
-        if (generic._tag == .noscript) return false;
+    if (tag_type == .link) {
+        const el = parser.nodeToElement(node);
+        const as = try parser.elementGetAttribute(el, "as") orelse return false;
+        if (!std.ascii.eqlIgnoreCase(as, "script")) {
+            return false;
+        }
+
+        const rel = try parser.elementGetAttribute(el, "rel") orelse return false;
+        return std.ascii.eqlIgnoreCase(rel, "preload");
     }
-    return true;
+    return false;
 }
-fn writeEscapedText(text: []const u8, writer: *std.Io.Writer) !void {
-    // Fast path: if no special characters, write directly
-    const first_special = std.mem.indexOfAnyPos(u8, text, 0, &.{ '&', '<', '>', 194 }) orelse {
-        return writer.writeAll(text);
+
+fn isCssRelated(tag_type: parser.Tag, node: *parser.Node) !bool {
+    if (tag_type == .style) {
+        return true;
+    }
+    if (tag_type == .link) {
+        const el = parser.nodeToElement(node);
+        const rel = try parser.elementGetAttribute(el, "rel") orelse return false;
+        return std.ascii.eqlIgnoreCase(rel, "stylesheet");
+    }
+    return false;
+}
+
+fn isUIRelated(tag_type: parser.Tag, node: *parser.Node) !bool {
+    if (try isCssRelated(tag_type, node)) {
+        return true;
+    }
+    if (tag_type == .img or tag_type == .picture or tag_type == .video) {
+        return true;
+    }
+    if (tag_type == .undef) {
+        const name = try parser.nodeLocalName(node);
+        if (std.mem.eql(u8, name, "svg")) {
+            return true;
+        }
+    }
+    return false;
+}
+
+// area, base, br, col, embed, hr, img, input, link, meta, source, track, wbr
+// https://html.spec.whatwg.org/#void-elements
+fn isVoid(elem: *parser.Element) !bool {
+    const tag = try parser.elementTag(elem);
+    return switch (tag) {
+        .area, .base, .br, .col, .embed, .hr, .img, .input, .link => true,
+        .meta, .source, .track, .wbr => true,
+        else => false,
     };
+}
 
-    try writer.writeAll(text[0..first_special]);
-    var remaining = try writeEscapedByte(text, first_special, writer);
-
-    while (std.mem.indexOfAnyPos(u8, remaining, 0, &.{ '&', '<', '>', 194 })) |offset| {
-        try writer.writeAll(remaining[0..offset]);
-        remaining = try writeEscapedByte(remaining, offset, writer);
-    }
-
-    if (remaining.len > 0) {
-        try writer.writeAll(remaining);
+fn writeEscapedTextNode(writer: *std.Io.Writer, value: []const u8) !void {
+    var v = value;
+    while (v.len > 0) {
+        const index = std.mem.indexOfAnyPos(u8, v, 0, &.{ '&', '<', '>', 194 }) orelse {
+            return writer.writeAll(v);
+        };
+        try writer.writeAll(v[0..index]);
+        switch (v[index]) {
+            '&' => try writer.writeAll("&amp;"),
+            '<' => try writer.writeAll("&lt;"),
+            '>' => try writer.writeAll("&gt;"),
+            194 => {
+                // non breaking space
+                if (v.len > index + 1 and v[index + 1] == 160) {
+                    try writer.writeAll("&nbsp;");
+                    v = v[index + 2 ..];
+                    continue;
+                }
+                try writer.writeByte(194);
+            },
+            else => unreachable,
+        }
+        v = v[index + 1 ..];
     }
 }
 
-fn writeEscapedByte(input: []const u8, index: usize, writer: *std.Io.Writer) ![]const u8 {
-    switch (input[index]) {
-        '&' => try writer.writeAll("&amp;"),
-        '<' => try writer.writeAll("&lt;"),
-        '>' => try writer.writeAll("&gt;"),
-        194 => {
-            // non breaking space
-            if (input.len > index + 1 and input[index + 1] == 160) {
-                try writer.writeAll("&nbsp;");
-                return input[index + 2 ..];
-            }
-            try writer.writeByte(194);
-        },
-        else => unreachable,
+fn writeEscapedAttributeValue(writer: *std.Io.Writer, value: []const u8) !void {
+    var v = value;
+    while (v.len > 0) {
+        const index = std.mem.indexOfAnyPos(u8, v, 0, &.{ '&', '<', '>', '"' }) orelse {
+            return writer.writeAll(v);
+        };
+        try writer.writeAll(v[0..index]);
+        switch (v[index]) {
+            '&' => try writer.writeAll("&amp;"),
+            '<' => try writer.writeAll("&lt;"),
+            '>' => try writer.writeAll("&gt;"),
+            '"' => try writer.writeAll("&quot;"),
+            else => unreachable,
+        }
+        v = v[index + 1 ..];
     }
-    return input[index + 1 ..];
+}
+
+const testing = std.testing;
+test "dump.writeHTML" {
+    parser.init();
+    defer parser.deinit();
+
+    try testWriteHTML(
+        "<div id=\"content\">Over 9000!</div>",
+        "<div id=\"content\">Over 9000!</div>",
+    );
+
+    try testWriteHTML(
+        "<root><!-- a comment --></root>",
+        "<root><!-- a comment --></root>",
+    );
+
+    try testWriteHTML(
+        "<p>&lt; &gt; &amp;</p>",
+        "<p>&lt; &gt; &amp;</p>",
+    );
+
+    try testWriteHTML(
+        "<p id=\"&quot;&gt;&lt;&amp;&quot;''\">wat?</p>",
+        "<p id='\">&lt;&amp;&quot;&#39;&apos;'>wat?</p>",
+    );
+
+    try testWriteFullHTML(
+        \\<!DOCTYPE html>
+        \\<html><head><title>It's over what?</title><meta name="a" value="b">
+        \\</head><body>9000</body></html>
+        \\
+    , "<html><title>It's over what?</title><meta name=a value=\"b\">\n<body>9000");
+
+    try testWriteHTML(
+        "<p>hi</p><script>alert(power > 9000)</script>",
+        "<p>hi</p><script>alert(power > 9000)</script>",
+    );
+}
+
+fn testWriteHTML(comptime expected_body: []const u8, src: []const u8) !void {
+    const expected =
+        "<!DOCTYPE html>\n<html><head></head><body>" ++
+        expected_body ++
+        "</body></html>\n";
+    return testWriteFullHTML(expected, src);
+}
+
+fn testWriteFullHTML(comptime expected: []const u8, src: []const u8) !void {
+    var aw = std.Io.Writer.Allocating.init(testing.allocator);
+    defer aw.deinit();
+
+    const doc_html = try parser.documentHTMLParseFromStr(src);
+    defer parser.documentHTMLClose(doc_html) catch {};
+
+    const doc = parser.documentHTMLToDocument(doc_html);
+    try writeHTML(doc, .{}, &aw.writer);
+    try testing.expectEqualStrings(expected, aw.written());
 }

src/browser/encoding/TextDecoder.zig

@@ -0,0 +1,102 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+const log = @import("../../log.zig");
+
+const Page = @import("../page.zig").Page;
+
+// https://encoding.spec.whatwg.org/#interface-textdecoder
+const TextDecoder = @This();
+
+const SupportedLabels = enum {
+    utf8,
+    @"utf-8",
+    @"unicode-1-1-utf-8",
+};
+
+const Options = struct {
+    fatal: bool = false,
+    ignoreBOM: bool = false,
+};
+
+fatal: bool,
+ignore_bom: bool,
+stream: std.ArrayList(u8),
+
+pub fn constructor(label_: ?[]const u8, opts_: ?Options) !TextDecoder {
+    if (label_) |l| {
+        _ = std.meta.stringToEnum(SupportedLabels, l) orelse {
+            log.warn(.web_api, "not implemented", .{ .feature = "TextDecoder label", .label = l });
+            return error.NotImplemented;
+        };
+    }
+    const opts = opts_ orelse Options{};
+    return .{
+        .stream = .empty,
+        .fatal = opts.fatal,
+        .ignore_bom = opts.ignoreBOM,
+    };
+}
+
+pub fn get_encoding(_: *const TextDecoder) []const u8 {
+    return "utf-8";
+}
+
+pub fn get_ignoreBOM(self: *const TextDecoder) bool {
+    return self.ignore_bom;
+}
+
+pub fn get_fatal(self: *const TextDecoder) bool {
+    return self.fatal;
+}
+
+const DecodeOptions = struct {
+    stream: bool = false,
+};
+pub fn _decode(self: *TextDecoder, str_: ?[]const u8, opts_: ?DecodeOptions, page: *Page) ![]const u8 {
+    var str = str_ orelse return "";
+    const opts: DecodeOptions = opts_ orelse .{};
+
+    if (self.stream.items.len > 0) {
+        try self.stream.appendSlice(page.arena, str);
+        str = self.stream.items;
+    }
+
+    if (self.fatal and !std.unicode.utf8ValidateSlice(str)) {
+        if (opts.stream) {
+            if (self.stream.items.len == 0) {
+                try self.stream.appendSlice(page.arena, str);
+            }
+            return "";
+        }
+        return error.InvalidUtf8;
+    }
+
+    self.stream.clearRetainingCapacity();
+    if (self.ignore_bom == false and std.mem.startsWith(u8, str, &.{ 0xEF, 0xBB, 0xBF })) {
+        return str[3..];
+    }
+
+    return str;
+}
+
+const testing = @import("../../testing.zig");
+test "Browser: Encoding.TextDecoder" {
+    try testing.htmlRunner("encoding/decoder.html");
+}

src/browser/encoding/TextEncoder.zig

@@ -18,33 +18,31 @@
 
 const std = @import("std");
 
-const Page = @import("../Page.zig");
-const Blob = @import("Blob.zig");
 const js = @import("../js/js.zig");
 
-const File = @This();
+// https://encoding.spec.whatwg.org/#interface-textencoder
+const TextEncoder = @This();
 
-/// `File` inherits `Blob`.
-_proto: *Blob,
-
-// TODO: Implement File API.
-pub fn init(page: *Page) !*File {
-    return page._factory.blob(File{ ._proto = undefined });
+pub fn constructor() !TextEncoder {
+    return .{};
 }
 
-pub const JsApi = struct {
-    pub const bridge = js.Bridge(File);
+pub fn get_encoding(_: *const TextEncoder) []const u8 {
+    return "utf-8";
+}
 
-    pub const Meta = struct {
-        pub const name = "File";
-        pub const prototype_chain = bridge.prototypeChain();
-        pub var class_id: bridge.ClassId = undefined;
-    };
+pub fn _encode(_: *const TextEncoder, v: []const u8) !js.TypedArray(u8) {
+    // Ensure the input is a valid utf-8
+    // It seems chrome accepts invalid utf-8 sequence.
+    //
+    if (!std.unicode.utf8ValidateSlice(v)) {
+        return error.InvalidUtf8;
+    }
 
-    pub const constructor = bridge.constructor(File.init, .{});
-};
+    return .{ .values = v };
+}
 
 const testing = @import("../../testing.zig");
-test "WebApi: File" {
-    try testing.htmlRunner("file.html", .{});
+test "Browser: Encoding.TextEncoder" {
+    try testing.htmlRunner("encoding/encoder.html");
 }

src/browser/encoding/encoding.zig

@@ -16,20 +16,7 @@
 // You should have received a copy of the GNU Affero General Public License
 // along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
-const std = @import("std");
-const js = @import("js.zig");
-
-const v8 = js.v8;
-
-const Integer = @This();
-
-handle: *const v8.Integer,
-
-pub fn init(isolate: *v8.Isolate, value: anytype) Integer {
-    const handle = switch (@TypeOf(value)) {
-        i8, i16, i32 => v8.v8__Integer__New(isolate, value).?,
-        u8, u16, u32 => v8.v8__Integer__NewFromUnsigned(isolate, value).?,
-        else => |T| @compileError("cannot create v8::Integer from: " ++ @typeName(T)),
-    };
-    return .{ .handle = handle };
-}
+pub const Interfaces = .{
+    @import("TextDecoder.zig"),
+    @import("TextEncoder.zig"),
+};

src/browser/events/PageTransitionEvent.zig

@@ -0,0 +1,84 @@
+// Copyright (C) 2023-2024  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const log = @import("../../log.zig");
+const Window = @import("../html/window.zig").Window;
+
+const parser = @import("../netsurf.zig");
+const Event = @import("../events/event.zig").Event;
+
+// https://developer.mozilla.org/en-US/docs/Web/API/PageTransitionEvent
+const PageTransitionEvent = @This();
+
+pub const prototype = *Event;
+pub const union_make_copy = true;
+
+pub const EventInit = struct {
+    persisted: ?bool,
+};
+
+proto: parser.Event,
+persisted: bool,
+
+pub fn constructor(event_type: []const u8, opts: EventInit) !PageTransitionEvent {
+    const event = try parser.eventCreate();
+    defer parser.eventDestroy(event);
+
+    try parser.eventInit(event, event_type, .{});
+    parser.eventSetInternalType(event, .page_transition_event);
+
+    return .{
+        .proto = event.*,
+        .persisted = opts.persisted orelse false,
+    };
+}
+
+const PageTransitionKind = enum { show, hide };
+
+pub fn dispatch(window: *Window, kind: PageTransitionKind, persisted: bool) void {
+    const evt_type = switch (kind) {
+        .show => "pageshow",
+        .hide => "pagehide",
+    };
+
+    log.debug(.script_event, "dispatch event", .{
+        .type = evt_type,
+        .source = "navigation",
+    });
+
+    var evt = PageTransitionEvent.constructor(evt_type, .{ .persisted = persisted }) catch |err| {
+        log.err(.app, "event constructor error", .{
+            .err = err,
+            .type = evt_type,
+            .source = "navigation",
+        });
+
+        return;
+    };
+
+    _ = parser.eventTargetDispatchEvent(
+        @as(*parser.EventTarget, @ptrCast(window)),
+        &evt.proto,
+    ) catch |err| {
+        log.err(.app, "dispatch event error", .{
+            .err = err,
+            .type = evt_type,
+            .source = "navigation",
+        });
+    };
+}

src/browser/events/composition_event.zig

@@ -0,0 +1,57 @@
+// Copyright (C) 2023-2025  Lightpanda (Selecy SAS)
+//
+// Francis Bouvier <francis@lightpanda.io>
+// Pierre Tachoire <pierre@lightpanda.io>
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as
+// published by the Free Software Foundation, either version 3 of the
+// License, or (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+const std = @import("std");
+
+const parser = @import("../netsurf.zig");
+
+// https://developer.mozilla.org/en-US/docs/Web/API/CompositionEvent
+pub const CompositionEvent = struct {
+    data: []const u8,
+    proto: parser.Event,
+
+    pub const union_make_copy = true;
+    pub const prototype = *parser.Event;
+
+    pub const ConstructorOptions = struct {
+        data: []const u8 = "",
+    };
+
+    pub fn constructor(event_type: []const u8, options_: ?ConstructorOptions) !CompositionEvent {
+        const options: ConstructorOptions = options_ orelse .{};
+
+        const event = try parser.eventCreate();
+        defer parser.eventDestroy(event);
+        try parser.eventInit(event, event_type, .{});
+        parser.eventSetInternalType(event, .composition_event);
+
+        return .{
+            .proto = event.*,
+            .data = options.data,
+        };
+    }
+
+    pub fn get_data(self: *const CompositionEvent) []const u8 {
+        return self.data;
+    }
+};
+
+const testing = @import("../../testing.zig");
+test "Browser: Events.Composition" {
+    try testing.htmlRunner("events/composition.html");
+}