tor-browser

commit 527055fac7f9602a28c885dabbdd5f878ec723e6
parent a2ee9ef2cf4d68b359f8f60649b9cfe90fb0619c
Author: Jonathan Kew <jkew@mozilla.com>
Date:   Fri,  5 Dec 2025 18:22:47 +0000

Bug 2003721 - Canonicalize lang-subtag case during attribute mapping. r=layout-reviewers,emilio

Differential Revision: https://phabricator.services.mozilla.com/D275267

Diffstat:
M
dom/html/nsGenericHTMLElement.cpp
 | 
38
+++++++++++++++++++++++++++++++++++---

1 file changed, 35 insertions(+), 3 deletions(-)
diff --git a/dom/html/nsGenericHTMLElement.cpp b/dom/html/nsGenericHTMLElement.cpp
@@ -57,6 +57,7 @@
 #include "mozilla/dom/ToggleEvent.h"
 #include "mozilla/dom/TouchEvent.h"
 #include "mozilla/dom/UnbindContext.h"
+#include "mozilla/intl/Locale.h"
 #include "nsAtom.h"
 #include "nsAttrValueOrString.h"
 #include "nsCOMPtr.h"
@@ -1281,10 +1282,41 @@ static inline void MapLangAttributeInto(MappedDeclarationsBuilder& aBuilder) {
     return;
   }
   MOZ_ASSERT(langValue->Type() == nsAttrValue::eAtom);
-  aBuilder.SetIdentAtomValueIfUnset(eCSSProperty__x_lang,
-                                    langValue->GetAtomValue());
+
+  // Adaptor for nsCString to expose the Buffer interface to Locale::ToString.
+  class BufferAdaptor {
+   public:
+    using CharType = char;
+
+    explicit BufferAdaptor(nsCString& aString) : mString(aString) {}
+    char* data() { return mString.BeginWriting(); }
+    size_t capacity() const { return mString.Length(); }
+    bool reserve(size_t aLen) { return mString.SetLength(aLen, fallible); }
+    void written(size_t aLen) { mString.SetLength(aLen); }
+
+   private:
+    nsCString& mString;
+  };
+
+  // Try parsing lang as a Locale and canonicalizing the subtags; if parsing
+  // succeeds, we record the canonicalized version rather than the original,
+  // so that code checking for particular codes can assume canonical casing.
+  // Note that in some cases this will also map 3-character ISO 639-3 tags to
+  // their corresponding 2-char ISO 639-1 tags.
+  RefPtr<nsAtom> lang = langValue->GetAtomValue();
+  nsAtomCString langStr(lang);
+  intl::Locale loc;
+  if (intl::LocaleParser::TryParse(langStr, loc).isOk() &&
+      loc.Canonicalize().isOk()) {
+    nsAutoCString canonical;
+    BufferAdaptor buffer(canonical);
+    if (loc.ToString(buffer).isOk() && canonical != langStr) {
+      lang = NS_Atomize(canonical);
+    }
+  }
+
+  aBuilder.SetIdentAtomValueIfUnset(eCSSProperty__x_lang, lang);
   if (!aBuilder.PropertyIsSet(eCSSProperty_text_emphasis_position)) {
-    const nsAtom* lang = langValue->GetAtomValue();
     if (nsStyleUtil::MatchesLanguagePrefix(lang, u"zh")) {
       aBuilder.SetKeywordValue(eCSSProperty_text_emphasis_position,
                                StyleTextEmphasisPosition::UNDER._0);