dom: use the css matcher for DOM

src/dom/css.zig

@@ -0,0 +1,61 @@
+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 = NodeList.init(),
+        };
+    }
+
+    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 {
+        defer m.nl = NodeList.init();
+        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/dom/document.zig

@@ -13,6 +13,7 @@ const NodeUnion = @import("node.zig").Union;
 
 const Walker = @import("walker.zig").WalkerDepthFirst;
 const collection = @import("html_collection.zig");
+const css = @import("css.zig");
 
 const Element = @import("element.zig").Element;
 const ElementUnion = @import("element.zig").Union;
@@ -188,54 +189,18 @@ pub const Document = struct {
         return 1;
     }
 
-    // TODO netsurf doesn't handle query selectors. We have to implement a
-    // solution by ourselves.
-    // For now we handle only * and single id selector like `#foo`.
-    pub fn _querySelector(self: *parser.Document, selectors: []const u8) !?ElementUnion {
-        if (selectors.len == 0) return null;
+    pub fn _querySelector(self: *parser.Document, alloc: std.mem.Allocator, selector: []const u8) !?ElementUnion {
+        if (selector.len == 0) return null;
 
-        // catch-all, return the firstElementChild
-        if (selectors[0] == '*') return try get_firstElementChild(self);
+        const n = try css.querySelector(alloc, parser.documentToNode(self), selector);
 
-        // support only simple id selector.
-        if (selectors[0] != '#' or std.mem.indexOf(u8, selectors, " ") != null) return null;
+        if (n == null) return null;
 
-        return try _getElementById(self, selectors[1..]);
+        return try Element.toInterface(parser.nodeToElement(n.?));
     }
 
-    // TODO netsurf doesn't handle query selectors. We have to implement a
-    // solution by ourselves.
-    // We handle only * and single id selector like `#foo`.
-    pub fn _querySelectorAll(self: *parser.Document, alloc: std.mem.Allocator, selectors: []const u8) !NodeList {
-        var list = try NodeList.init();
-        errdefer list.deinit(alloc);
-
-        if (selectors.len == 0) return list;
-
-        // catch-all, return all elements
-        if (selectors[0] == '*') {
-            // walk over the node tree fo find the node by id.
-            const root = parser.documentToNode(self);
-            const walker = Walker{};
-            var next: ?*parser.Node = null;
-            while (true) {
-                next = try walker.get_next(root, next) orelse return list;
-                // ignore non-element nodes.
-                if (try parser.nodeType(next.?) != .element) {
-                    continue;
-                }
-                try list.append(alloc, next.?);
-            }
-        }
-
-        // support only simple id selector.
-        if (selectors[0] != '#' or std.mem.indexOf(u8, selectors, " ") != null) return list;
-
-        // walk over the node tree fo find the node by id.
-        const e = try parser.documentGetElementById(self, selectors[1..]) orelse return list;
-        try list.append(alloc, parser.elementToNode(e));
-
-        return list;
+    pub fn _querySelectorAll(self: *parser.Document, alloc: std.mem.Allocator, selector: []const u8) !NodeList {
+        return css.querySelectorAll(alloc, parser.documentToNode(self), selector);
     }
 
     // TODO according with https://dom.spec.whatwg.org/#parentnode, the
@@ -426,6 +391,12 @@ pub fn testExecFn(
         .{ .src = "document.querySelector('*').nodeName", .ex = "HTML" },
         .{ .src = "document.querySelector('#content').id", .ex = "content" },
         .{ .src = "document.querySelector('#para').id", .ex = "para" },
+        .{ .src = "document.querySelector('.ok').id", .ex = "link" },
+        .{ .src = "document.querySelector('a ~ p').id", .ex = "para-empty" },
+        .{ .src = "document.querySelector(':root').nodeName", .ex = "HTML" },
+
+        .{ .src = "document.querySelectorAll('p').length", .ex = "2" },
+        .{ .src = "document.querySelectorAll('.ok').item(0).id", .ex = "link" },
     };
     try checkCases(js_env, &querySelector);
 

src/dom/element.zig

@@ -9,6 +9,7 @@ const Variadic = jsruntime.Variadic;
 
 const collection = @import("html_collection.zig");
 const writeNode = @import("../browser/dump.zig").writeNode;
+const css = @import("css.zig");
 
 const Node = @import("node.zig").Node;
 const Walker = @import("walker.zig").WalkerDepthFirst;
@@ -263,56 +264,18 @@ pub const Element = struct {
         }
     }
 
-    // TODO netsurf doesn't handle query selectors. We have to implement a
-    // solution by ourselves.
-    // We handle only * and single id selector like `#foo`.
-    pub fn _querySelector(self: *parser.Element, selectors: []const u8) !?Union {
-        if (selectors.len == 0) return null;
+    pub fn _querySelector(self: *parser.Element, alloc: std.mem.Allocator, selector: []const u8) !?Union {
+        if (selector.len == 0) return null;
 
-        // catch-all, return the firstElementChild
-        if (selectors[0] == '*') return try get_firstElementChild(self);
+        const n = try css.querySelector(alloc, parser.elementToNode(self), selector);
 
-        // support only simple id selector.
-        if (selectors[0] != '#' or std.mem.indexOf(u8, selectors, " ") != null) return null;
+        if (n == null) return null;
 
-        // walk over the node tree fo find the node by id.
-        const n = try getElementById(self, selectors[1..]) orelse return null;
-        return try toInterface(parser.nodeToElement(n));
+        return try toInterface(parser.nodeToElement(n.?));
     }
 
-    // TODO netsurf doesn't handle query selectors. We have to implement a
-    // solution by ourselves.
-    // We handle only * and single id selector like `#foo`.
-    pub fn _querySelectorAll(self: *parser.Element, alloc: std.mem.Allocator, selectors: []const u8) !NodeList {
-        var list = try NodeList.init();
-        errdefer list.deinit(alloc);
-
-        if (selectors.len == 0) return list;
-
-        // catch-all, return all elements
-        if (selectors[0] == '*') {
-            // 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 list;
-                // ignore non-element nodes.
-                if (try parser.nodeType(next.?) != .element) {
-                    continue;
-                }
-                try list.append(alloc, next.?);
-            }
-        }
-
-        // support only simple id selector.
-        if (selectors[0] != '#' or std.mem.indexOf(u8, selectors, " ") != null) return list;
-
-        // walk over the node tree fo find the node by id.
-        const n = try getElementById(self, selectors[1..]) orelse return list;
-        try list.append(alloc, n);
-
-        return list;
+    pub fn _querySelectorAll(self: *parser.Element, alloc: std.mem.Allocator, selector: []const u8) !NodeList {
+        return css.querySelectorAll(alloc, parser.elementToNode(self), selector);
     }
 
     // TODO according with https://dom.spec.whatwg.org/#parentnode, the
@@ -433,6 +396,12 @@ pub fn testExecFn(
         .{ .src = "e.querySelector('#link').id", .ex = "link" },
         .{ .src = "e.querySelector('#para').id", .ex = "para" },
         .{ .src = "e.querySelector('*').id", .ex = "link" },
+        .{ .src = "e.querySelector('')", .ex = "null" },
+        .{ .src = "e.querySelector('*').id", .ex = "link" },
+        .{ .src = "e.querySelector('#content')", .ex = "null" },
+        .{ .src = "e.querySelector('#para').id", .ex = "para" },
+        .{ .src = "e.querySelector('.ok').id", .ex = "link" },
+        .{ .src = "e.querySelector('a ~ p').id", .ex = "para-empty" },
 
         .{ .src = "e.querySelectorAll('foo').length", .ex = "0" },
         .{ .src = "e.querySelectorAll('#foo').length", .ex = "0" },
@@ -441,6 +410,8 @@ pub fn testExecFn(
         .{ .src = "e.querySelectorAll('#para').length", .ex = "1" },
         .{ .src = "e.querySelectorAll('#para').item(0).id", .ex = "para" },
         .{ .src = "e.querySelectorAll('*').length", .ex = "4" },
+        .{ .src = "e.querySelectorAll('p').length", .ex = "2" },
+        .{ .src = "e.querySelectorAll('.ok').item(0).id", .ex = "link" },
     };
     try checkCases(js_env, &querySelector);
 

src/dom/node.zig

@@ -199,7 +199,7 @@ pub const Node = struct {
     }
 
     pub fn get_childNodes(self: *parser.Node, alloc: std.mem.Allocator) !NodeList {
-        var list = try NodeList.init();
+        var list = NodeList.init();
         errdefer list.deinit(alloc);
 
         var n = try parser.nodeFirstChild(self) orelse return list;

src/dom/nodelist.zig

@@ -26,7 +26,7 @@ pub const NodeList = struct {
 
     nodes: NodesArrayList,
 
-    pub fn init() !NodeList {
+    pub fn init() NodeList {
         return NodeList{
             .nodes = NodesArrayList{},
         };

src/html/document.zig

@@ -80,7 +80,7 @@ pub const HTMLDocument = struct {
     }
 
     pub fn _getElementsByName(self: *parser.DocumentHTML, alloc: std.mem.Allocator, name: []const u8) !NodeList {
-        var list = try NodeList.init();
+        var list = NodeList.init();
         errdefer list.deinit(alloc);
 
         if (name.len == 0) return list;