ext. filtering for lang matching

components/script/dom/bindings/str.rs

@@ -113,6 +113,64 @@ pub fn is_token(s: &[u8]) -> bool {
     })
 }
 
+/// Returns whether the language is matched, as defined by
+/// [RFC 4647](https://tools.ietf.org/html/rfc4647#section-3.3.2).
+pub fn extended_filtering(tag: &str, range: &str) -> bool {
+    let lang_ranges: Vec<&str> = range.split(',').collect();
+
+    lang_ranges.iter().any(|&lang_range| {
+        // step 1
+        let range_subtags: Vec<&str> = lang_range.split('\x2d').collect();
+        let tag_subtags: Vec<&str> = tag.split('\x2d').collect();
+
+        let mut range_iter = range_subtags.iter();
+        let mut tag_iter = tag_subtags.iter();
+
+        // step 2
+        // Note: [Level-4 spec](https://drafts.csswg.org/selectors/#lang-pseudo) check for wild card
+        if let (Some(range_subtag), Some(tag_subtag)) = (range_iter.next(), tag_iter.next()) {
+            if !(range_subtag.eq_ignore_ascii_case(tag_subtag) || range_subtag.eq_ignore_ascii_case("*")) {
+                return false;
+            }
+        }
+
+        let mut current_tag_subtag = tag_iter.next();
+
+        // step 3
+        for range_subtag in range_iter {
+            // step 3a
+            if range_subtag.eq_ignore_ascii_case("*") {
+                continue;
+            }
+            match current_tag_subtag.clone() {
+                Some(tag_subtag) => {
+                    // step 3c
+                    if range_subtag.eq_ignore_ascii_case(tag_subtag) {
+                        current_tag_subtag = tag_iter.next();
+                        continue;
+                    } else {
+                        // step 3d
+                        if tag_subtag.len() == 1 {
+                            return false;
+                        } else {
+                            // else step 3e - continue with loop
+                            current_tag_subtag = tag_iter.next();
+                            if current_tag_subtag.is_none() {
+                                return false;
+                            }
+                        }
+                    }
+                },
+                // step 3b
+                None => { return false; }
+            }
+        }
+        // step 4
+        true
+    })
+}
+
+
 /// A DOMString.
 ///
 /// This type corresponds to the [`DOMString`](idl) type in WebIDL.

components/script/dom/element.rs

@@ -25,7 +25,7 @@ use dom::bindings::js::{JS, LayoutJS, MutNullableJS};
 use dom::bindings::js::{Root, RootedReference};
 use dom::bindings::refcounted::{Trusted, TrustedPromise};
 use dom::bindings::reflector::DomObject;
-use dom::bindings::str::DOMString;
+use dom::bindings::str::{DOMString, extended_filtering};
 use dom::bindings::xmlname::{namespace_from_domstring, validate_and_extract, xml_name_type};
 use dom::bindings::xmlname::XMLName::InvalidXMLName;
 use dom::characterdata::CharacterData;
@@ -2402,7 +2402,7 @@ impl<'a> ::selectors::Element for Root<Element> {
 
             // FIXME(#15746): This is wrong, we need to instead use extended filtering as per RFC4647
             //                https://tools.ietf.org/html/rfc4647#section-3.3.2
-            NonTSPseudoClass::Lang(ref lang) => lang.eq_ignore_ascii_case(&self.get_lang()),
+            NonTSPseudoClass::Lang(ref lang) => extended_filtering(&*self.get_lang(), &*lang),
 
             NonTSPseudoClass::ReadOnly =>
                 !Element::state(self).contains(pseudo_class.state_flag()),

components/script/layout_wrapper.rs

@@ -34,6 +34,7 @@ use atomic_refcell::AtomicRefCell;
 use dom::bindings::inheritance::{CharacterDataTypeId, ElementTypeId};
 use dom::bindings::inheritance::{HTMLElementTypeId, NodeTypeId};
 use dom::bindings::js::LayoutJS;
+use dom::bindings::str::extended_filtering;
 use dom::characterdata::LayoutCharacterDataHelpers;
 use dom::document::{Document, LayoutDocumentHelpers, PendingRestyle};
 use dom::element::{Element, LayoutElementHelpers, RawLayoutElementHelpers};
@@ -53,7 +54,6 @@ use selectors::matching::ElementSelectorFlags;
 use selectors::parser::{AttrSelector, NamespaceConstraint};
 use servo_atoms::Atom;
 use servo_url::ServoUrl;
-use std::ascii::AsciiExt;
 use std::fmt;
 use std::fmt::Debug;
 use std::marker::PhantomData;
@@ -621,7 +621,7 @@ impl<'le> ::selectors::Element for ServoLayoutElement<'le> {
 
             // FIXME(#15746): This is wrong, we need to instead use extended filtering as per RFC4647
             //                https://tools.ietf.org/html/rfc4647#section-3.3.2
-            NonTSPseudoClass::Lang(ref lang) => lang.eq_ignore_ascii_case(&self.element.get_lang_for_layout()),
+            NonTSPseudoClass::Lang(ref lang) => extended_filtering(&*self.element.get_lang_for_layout(), &*lang),
 
             NonTSPseudoClass::ServoNonZeroBorder => unsafe {
                 match (*self.element.unsafe_get()).get_attr_for_layout(&ns!(), &local_name!("border")) {