Skip to content

Lightpanda: the headless browser designed for AI and automation

selector: match type selectors on XML and foreign elements

Jay Hemnani2e371e7e3a10
diff --git a/src/browser/tests/domparser.html b/src/browser/tests/domparser.html
index 2a67c0062..5d42993eb 100644
--- a/src/browser/tests/domparser.html
+++ b/src/browser/tests/domparser.html
@@ -433,6 +433,21 @@
}
</script>
+<script id=querySelector-xml>
+{
+ const doc = new DOMParser().parseFromString(
+ '<rss><channel><title>Feed</title><item><title>One</title><link>a</link><pubDate>Mon</pubDate></item><item><title>Two</title></item></channel></rss>',
+ 'application/xml');
+
+ testing.expectEqual(2, doc.querySelectorAll('item').length);
+ testing.expectEqual('Feed', doc.querySelector('channel > title')?.textContent);
+ testing.expectEqual(['One', 'Two'], Array.from(doc.querySelectorAll('item title')).map(e => e.textContent));
+ testing.expectEqual(1, doc.querySelectorAll('link').length);
+ testing.expectEqual(1, doc.querySelectorAll('pubDate').length);
+ testing.expectEqual(true, doc.getElementsByTagName('title')[0].matches('title'));
+}
+</script>
+
<script id=xml-namespaces>
{
const doc = new DOMParser().parseFromString(
diff --git a/src/browser/tests/element/svg/structure.html b/src/browser/tests/element/svg/structure.html
index fc96901b8..5700e3d47 100644
--- a/src/browser/tests/element/svg/structure.html
+++ b/src/browser/tests/element/svg/structure.html
@@ -106,3 +106,11 @@
testing.expectEqual(6, box.height);
}
</script>
+
+<script id=typeSelectors>
+{
+ testing.expectEqual($('#foreign'), document.querySelector('foreignObject'));
+ testing.expectEqual($('#foreign'), document.querySelector('foreignobject'));
+ testing.expectEqual('HTML content', document.querySelector('#root foreignObject div')?.textContent);
+}
+</script>
diff --git a/src/browser/webapi/selector/List.zig b/src/browser/webapi/selector/List.zig
index c805deb46..39d4ffcd1 100644
--- a/src/browser/webapi/selector/List.zig
+++ b/src/browser/webapi/selector/List.zig
@@ -406,20 +406,27 @@ fn matchesPart(el: *Node.Element, part: Part, scope: *Node, nth: ?*NthCache, fra
},
.tag => |tag| {
// Optimized: compare enum directly
- return el.getTag() == tag;
- },
- .tag_name => |tag_name| {
- // Fallback for custom/unknown tags
- // Both are lowercase, so we can use fast string comparison
- const element_tag = el.getTagNameLower();
- return std.mem.eql(u8, element_tag, tag_name);
+ const element_tag = el.getTag();
+ if (element_tag == tag) return true;
+ // Elements without a dedicated Tag (XML, other namespaces) match by name
+ return element_tag == .unknown and matchesTagName(el, @tagName(tag));
},
+ .tag_name => |tag_name| return matchesTagName(el, tag_name),
.universal => return true,
.pseudo_class => |pseudo| return matchesPseudoClass(el, pseudo, scope, nth, frame),
.attribute => |attr| return matchesAttribute(el, attr),
}
}
+// `lower` is the lowercased selector name. Foreign element names keep their
+// case (foreignObject, pubDate), so those compare case-insensitively.
+fn matchesTagName(el: *Node.Element, lower: []const u8) bool {
+ if (el._namespace == .html) {
+ return std.mem.eql(u8, el.getTagNameLower(), lower);
+ }
+ return std.ascii.eqlIgnoreCase(el.getLocalName(), lower);
+}
+
fn matchesAttribute(el: *Node.Element, attr: Selector.Attribute) bool {
// Attribute names match ASCII case-insensitively on HTML elements (both
// sides lowercased) and case-sensitively on foreign elements, whose