browser: add a func to dump HTML

2025-10-29 15:13:28 +00:00 · 2024-01-05 16:28:33 +01:00
parent 2aefd3736c
commit 2cdbf68526
2 changed files with 103 additions and 0 deletions
--- a/src/browser/dump.zig
+++ b/src/browser/dump.zig
@@ -0,0 +1,94 @@
+const std = @import("std");
+const File = std.fs.File;
+
+const parser = @import("../netsurf.zig");
+const Walker = @import("../dom/html_collection.zig").WalkerChildren;
+
+pub fn htmlFile(root: *parser.Element, out: File) !void {
+    try out.writeAll("<!DOCTYPE html>\n<html>");
+    try nodeFile(root, out);
+    try out.writeAll("</html>\n");
+}
+
+fn nodeFile(root: *parser.Element, out: File) !void {
+    const walker = Walker{};
+    var next: ?*parser.Node = null;
+    while (true) {
+        next = try walker.get_next(parser.elementToNode(root), next) orelse break;
+        switch (try parser.nodeType(next.?)) {
+            .element => {
+                // open the tag
+                const tag = try parser.nodeLocalName(next.?);
+                try out.writeAll("<");
+                try out.writeAll(tag);
+
+                // write the attributes
+                const map = try parser.nodeGetAttributes(next.?);
+                const ln = try parser.namedNodeMapGetLength(map);
+                var i: u32 = 0;
+                while (i < ln) {
+                    const attr = try parser.namedNodeMapItem(map, i) orelse break;
+                    try out.writeAll(" ");
+                    try out.writeAll(try parser.attributeGetName(attr));
+                    try out.writeAll("=\"");
+                    try out.writeAll(try parser.attributeGetValue(attr));
+                    try out.writeAll("\"");
+                    i += 1;
+                }
+
+                try out.writeAll(">");
+
+                // write the children
+                // TODO avoid recursion
+                try nodeFile(parser.nodeToElement(next.?), out);
+
+                // close the tag
+                try out.writeAll("</");
+                try out.writeAll(tag);
+                try out.writeAll(">");
+            },
+            .text => {
+                const v = try parser.nodeValue(next.?) orelse continue;
+                try out.writeAll(v);
+            },
+            .cdata_section => {
+                const v = try parser.nodeValue(next.?) orelse continue;
+                try out.writeAll("<![CDATA[");
+                try out.writeAll(v);
+                try out.writeAll("]]>");
+            },
+            .comment => {
+                const v = try parser.nodeValue(next.?) orelse continue;
+                try out.writeAll("<!--");
+                try out.writeAll(v);
+                try out.writeAll("-->");
+            },
+            // TODO handle processing instruction dump
+            .processing_instruction => continue,
+            // document fragment is outside of the main document DOM, so we
+            // don't output it.
+            .document_fragment => continue,
+            // document will never be called, but required for completeness.
+            .document => continue,
+            // done globally instead, but required for completeness.
+            .document_type => continue,
+            // deprecated
+            .attribute => continue,
+            .entity_reference => continue,
+            .entity => continue,
+            .notation => continue,
+        }
+    }
+}
+
+// HTMLFileTestFn is run by run_tests.zig
+pub fn HTMLFileTestFn(out: File) !void {
+    const doc_html = try parser.documentHTMLParseFromFileAlloc(std.testing.allocator, "test.html");
+    // ignore close error
+    defer parser.documentHTMLClose(doc_html) catch {};
+
+    const doc = parser.documentHTMLToDocument(doc_html);
+    const root = try parser.documentGetDocumentElement(doc) orelse return error.DocumentNullRoot;
+
+    try htmlFile(root, out);
+}
--- a/src/run_tests.zig
+++ b/src/run_tests.zig
@@ -122,3 +122,12 @@ test "bug document html parsing #4" {
    doc = try parser.documentHTMLParse(file.reader(), "UTF-8");
    parser.documentHTMLClose(doc) catch {};
 }
+
+const dump = @import("browser/dump.zig");
+test "run browser tests" {
+    // const out = std.io.getStdOut();
+    const out = try std.fs.openFileAbsolute("/dev/null", .{ .mode = .write_only });
+    defer out.close();
+
+    try dump.HTMLFileTestFn(out);
+}