Merge pull request #71 from Browsercore/domdoc-getelementsbyclassname

document.getElementsByClassName

src/dom/document.zig

@@ -7,7 +7,8 @@ const Case = jsruntime.test_utils.Case;
 const checkCases = jsruntime.test_utils.checkCases;
 
 const Node = @import("node.zig").Node;
-const HTMLCollection = @import("html_collection.zig").HTMLCollection;
+
+const collection = @import("html_collection.zig");
 
 const Element = @import("element.zig").Element;
 const ElementUnion = @import("element.zig").Union;
@@ -49,12 +50,14 @@ pub const Document = struct {
     // 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: []const u8) HTMLCollection {
+    pub fn _getElementsByTagName(self: *parser.Document, tag_name: []const u8) collection.HTMLCollection {
         const root = parser.documentGetDocumentElement(self);
-        return HTMLCollection{
-            .root = parser.elementToNode(root),
-            .match = tag_name,
-        };
+        return collection.HTMLCollectionByTagName(parser.elementToNode(root), tag_name);
+    }
+
+    pub fn _getElementsByClassName(self: *parser.Document, classNames: []const u8) collection.HTMLCollection {
+        const root = parser.documentGetDocumentElement(self);
+        return collection.HTMLCollectionByClassName(parser.elementToNode(root), classNames);
     }
 };
 
@@ -93,6 +96,16 @@ pub fn testExecFn(
     };
     try checkCases(js_env, &getElementsByTagName);
 
+    var getElementsByClassName = [_]Case{
+        .{ .src = "let ok = document.getElementsByClassName('ok')", .ex = "undefined" },
+        .{ .src = "ok.length", .ex = "2" },
+        .{ .src = "let empty = document.getElementsByClassName('empty')", .ex = "undefined" },
+        .{ .src = "empty.length", .ex = "1" },
+        .{ .src = "let emptyok = document.getElementsByClassName('empty ok')", .ex = "undefined" },
+        .{ .src = "emptyok.length", .ex = "1" },
+    };
+    try checkCases(js_env, &getElementsByClassName);
+
     const tags = comptime parser.Tag.all();
     comptime var createElements: [(tags.len) * 2]Case = undefined;
     inline for (tags, 0..) |tag, i| {

src/dom/html_collection.zig

@@ -5,11 +5,81 @@ const parser = @import("../netsurf.zig");
 const jsruntime = @import("jsruntime");
 const Case = jsruntime.test_utils.Case;
 const checkCases = jsruntime.test_utils.checkCases;
+const generate = @import("../generate.zig");
 
 const utils = @import("utils.z");
 const Element = @import("element.zig").Element;
 const Union = @import("element.zig").Union;
 
+const Matcher = union(enum) {
+    matchByTagName: MatchByTagName,
+    matchByClassName: MatchByClassName,
+
+    pub fn match(self: Matcher, node: *parser.Node) bool {
+        switch (self) {
+            inline else => |case| return case.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 {
+        return MatchByTagName{
+            .tag = tag_name,
+            .is_wildcard = std.mem.eql(u8, tag_name, "*"),
+        };
+    }
+
+    pub fn match(self: MatchByTagName, node: *parser.Node) bool {
+        return self.is_wildcard or std.ascii.eqlIgnoreCase(self.tag, parser.nodeName(node));
+    }
+};
+
+pub fn HTMLCollectionByTagName(root: *parser.Node, tag_name: []const u8) HTMLCollection {
+    return HTMLCollection{
+        .root = root,
+        .matcher = Matcher{
+            .matchByTagName = MatchByTagName.init(tag_name),
+        },
+    };
+}
+
+pub const MatchByClassName = struct {
+    classNames: []const u8,
+
+    fn init(classNames: []const u8) MatchByClassName {
+        return MatchByClassName{
+            .classNames = classNames,
+        };
+    }
+
+    pub fn match(self: MatchByClassName, node: *parser.Node) bool {
+        var it = std.mem.splitAny(u8, self.classNames, " ");
+        const e = parser.nodeToElement(node);
+        while (it.next()) |c| {
+            if (!parser.elementHasClass(e, c)) {
+                return false;
+            }
+        }
+
+        return true;
+    }
+};
+
+pub fn HTMLCollectionByClassName(root: *parser.Node, classNames: []const u8) HTMLCollection {
+    return HTMLCollection{
+        .root = root,
+        .matcher = Matcher{
+            .matchByClassName = MatchByClassName.init(classNames),
+        },
+    };
+}
+
 // 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 arguement.
@@ -17,10 +87,9 @@ const Union = @import("element.zig").Union;
 pub const HTMLCollection = struct {
     pub const mem_guarantied = true;
 
+    matcher: Matcher,
+
     root: *parser.Node,
-    // match is used to select node against their name.
-    // match comparison is case insensitive.
-    match: []const u8,
 
     // save a state for the collection to improve the _item speed.
     cur_idx: ?u32 = undefined,
@@ -76,20 +145,15 @@ pub const HTMLCollection = struct {
     /// 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, allocator: std.mem.Allocator) !u32 {
+    pub fn get_length(self: *HTMLCollection) u32 {
         var len: u32 = 0;
         var node: *parser.Node = self.root;
         var ntype: parser.NodeType = undefined;
 
-        const imatch = try std.ascii.allocUpperString(allocator, self.match);
-        defer allocator.free(imatch);
-
-        const is_wildcard = std.mem.eql(u8, self.match, "*");
-
         while (true) {
             ntype = parser.nodeType(node);
             if (ntype == .element) {
-                if (is_wildcard or std.mem.eql(u8, imatch, parser.nodeName(node))) {
+                if (self.matcher.match(node)) {
                     len += 1;
                 }
             }
@@ -100,26 +164,21 @@ pub const HTMLCollection = struct {
         return len;
     }
 
-    pub fn _item(self: *HTMLCollection, allocator: std.mem.Allocator, index: u32) !?Union {
+    pub fn _item(self: *HTMLCollection, index: u32) ?Union {
         var i: u32 = 0;
         var node: *parser.Node = self.root;
         var ntype: parser.NodeType = undefined;
 
-        const is_wildcard = std.mem.eql(u8, self.match, "*");
-
         // Use the current state to improve speed if possible.
         if (self.cur_idx != null and index >= self.cur_idx.?) {
             i = self.cur_idx.?;
             node = self.cur_node.?;
         }
 
-        const imatch = try std.ascii.allocUpperString(allocator, self.match);
-        defer allocator.free(imatch);
-
         while (true) {
             ntype = parser.nodeType(node);
             if (ntype == .element) {
-                if (is_wildcard or std.mem.eql(u8, imatch, parser.nodeName(node))) {
+                if (self.matcher.match(node)) {
                     // check if we found the searched element.
                     if (i == index) {
                         // save the current state
@@ -140,7 +199,7 @@ pub const HTMLCollection = struct {
         return null;
     }
 
-    pub fn _namedItem(self: *HTMLCollection, allocator: std.mem.Allocator, name: []const u8) !?Union {
+    pub fn _namedItem(self: *HTMLCollection, name: []const u8) ?Union {
         if (name.len == 0) {
             return null;
         }
@@ -148,15 +207,10 @@ pub const HTMLCollection = struct {
         var node: *parser.Node = self.root;
         var ntype: parser.NodeType = undefined;
 
-        const is_wildcard = std.mem.eql(u8, self.match, "*");
-
-        const imatch = try std.ascii.allocUpperString(allocator, self.match);
-        defer allocator.free(imatch);
-
         while (true) {
             ntype = parser.nodeType(node);
             if (ntype == .element) {
-                if (is_wildcard or std.mem.eql(u8, imatch, parser.nodeName(node))) {
+                if (self.matcher.match(node)) {
                     const elem = @as(*parser.Element, @ptrCast(node));
 
                     var attr = parser.elementGetAttribute(elem, "id");

src/netsurf.zig

@@ -65,6 +65,18 @@ inline fn stringFromData(data: []const u8) *String {
     return s.?;
 }
 
+const LWCString = c.lwc_string;
+
+// TODO implement lwcStringToData
+// inline fn lwcStringToData(s: *LWCString) []const u8 {
+// }
+
+inline fn lwcStringFromData(data: []const u8) *LWCString {
+    var s: ?*LWCString = undefined;
+    _ = c.lwc_intern_string(data.ptr, data.len, &s);
+    return s.?;
+}
+
 // Tag
 
 pub const Tag = enum(u8) {
@@ -522,6 +534,11 @@ pub fn nodeReplaceChild(node: *Node, new_child: *Node, old_child: *Node) *Node {
     return res.?;
 }
 
+// nodeToElement is an helper to convert a node to an element.
+pub inline fn nodeToElement(node: *Node) *Element {
+    return @as(*Element, @ptrCast(node));
+}
+
 // CharacterData
 pub const CharacterData = c.dom_characterdata;
 
@@ -621,6 +638,12 @@ pub fn elementGetAttribute(elem: *Element, name: []const u8) ?[]const u8 {
     return stringToData(s.?);
 }
 
+pub fn elementHasClass(elem: *Element, class: []const u8) bool {
+    var res: bool = undefined;
+    _ = elementVtable(elem).dom_element_has_class.?(elem, lwcStringFromData(class), &res);
+    return res;
+}
+
 // elementToNode is an helper to convert an element to a node.
 pub inline fn elementToNode(e: *Element) *Node {
     return @as(*Node, @ptrCast(e));

test.html

@@ -1,6 +1,6 @@
 <div id='content'>
-  <a id='link' href='foo'>OK</a>
-  <p id='para-empty'>
+  <a id='link' href='foo' class='ok'>OK</a>
+  <p id='para-empty' class='ok empty'>
     <span id='para-empty-child'></span>
   </p>
   <p id='para'> And</p>

tests/wpt/dom/Document-getElementsByClassName.html

@@ -0,0 +1,26 @@
+<!DOCTYPE html>
+<title>Document.getElementsByClassName</title>
+<link rel="author" title="Intel" href="http://www.intel.com">
+<script src="/resources/testharness.js"></script>
+<script src="/resources/testharnessreport.js"></script>
+<div id="log"></div>
+<script>
+test(function() {
+  var a = document.createElement("a"),
+      b = document.createElement("b");
+  a.className = "foo";
+  this.add_cleanup(function() {document.body.removeChild(a);});
+  document.body.appendChild(a);
+
+  var l = document.getElementsByClassName("foo");
+  assert_true(l instanceof HTMLCollection);
+  assert_equals(l.length, 1);
+
+  b.className = "foo";
+  document.body.appendChild(b);
+  assert_equals(l.length, 2);
+
+  document.body.removeChild(b);
+  assert_equals(l.length, 1);
+}, "getElementsByClassName() should be a live collection");
+</script>