From 86e7ebd055dbea6531d5f7839a416f50d8ee2fb5 Mon Sep 17 00:00:00 2001
From: Jonathan Kew <jkew@mozilla.com>
Date: Wed, 12 Apr 2023 21:46:46 +0000
Subject: [PATCH] Bug 1121792 - Use more complete BCP47/RFC4647 language-range
 matching rather than simple prefix-dash match for the :lang() pseudo.
 r=emilio

This is a Selectors-4 enhancement to the spec for the :lang() pseudo-class.
It seems Safari has been shipping this behavior for some time.

Differential Revision: https://phabricator.services.mozilla.com/D174999
---
 layout/style/GeckoBindings.cpp | 10 +++---
 layout/style/nsStyleUtil.cpp   | 59 ++++++++++++++++++++++++++++++++++
 layout/style/nsStyleUtil.h     |  3 ++
 3 files changed, 66 insertions(+), 6 deletions(-)

diff --git a/layout/style/GeckoBindings.cpp b/layout/style/GeckoBindings.cpp
index 86ea37ceeb7e..5b97fc5e3008 100644
--- a/layout/style/GeckoBindings.cpp
+++ b/layout/style/GeckoBindings.cpp
@@ -774,9 +774,8 @@ bool Gecko_MatchLang(const Element* aElement, nsAtom* aOverrideLang,
   // from the parent we have to be prepared to look at all parent
   // nodes.  The language itself is encoded in the LANG attribute.
   if (auto* language = aHasOverrideLang ? aOverrideLang : aElement->GetLang()) {
-    return nsStyleUtil::DashMatchCompare(
-        nsDependentAtomString(language), nsDependentString(aValue),
-        nsASCIICaseInsensitiveStringComparator);
+    return nsStyleUtil::LangTagCompare(nsAtomCString(language),
+                                       NS_ConvertUTF16toUTF8(aValue));
   }
 
   // Try to get the language from the HTTP header or if this
@@ -786,11 +785,10 @@ bool Gecko_MatchLang(const Element* aElement, nsAtom* aOverrideLang,
   nsAutoString language;
   aElement->OwnerDoc()->GetContentLanguage(language);
 
-  nsDependentString langString(aValue);
+  NS_ConvertUTF16toUTF8 langString(aValue);
   language.StripWhitespace();
   for (auto const& lang : language.Split(char16_t(','))) {
-    if (nsStyleUtil::DashMatchCompare(lang, langString,
-                                      nsASCIICaseInsensitiveStringComparator)) {
+    if (nsStyleUtil::LangTagCompare(NS_ConvertUTF16toUTF8(lang), langString)) {
       return true;
     }
   }
diff --git a/layout/style/nsStyleUtil.cpp b/layout/style/nsStyleUtil.cpp
index 8faa926acd9b..4bfab913dc90 100644
--- a/layout/style/nsStyleUtil.cpp
+++ b/layout/style/nsStyleUtil.cpp
@@ -9,6 +9,8 @@
 
 #include "mozilla/dom/Document.h"
 #include "mozilla/ExpandedPrincipal.h"
+#include "mozilla/intl/MozLocaleBindings.h"
+#include "mozilla/TextUtils.h"
 #include "nsIContent.h"
 #include "nsCSSProps.h"
 #include "nsContentUtils.h"
@@ -51,6 +53,63 @@ bool nsStyleUtil::DashMatchCompare(const nsAString& aAttributeValue,
   return result;
 }
 
+bool nsStyleUtil::LangTagCompare(const nsACString& aAttributeValue,
+                                 const nsACString& aSelectorValue) {
+  class AutoLangId {
+   public:
+    AutoLangId() = delete;
+    AutoLangId(const AutoLangId& aOther) = delete;
+    explicit AutoLangId(const nsACString& aLangTag) : mIsValid(false) {
+      mLangId = intl::ffi::unic_langid_new(&aLangTag, &mIsValid);
+    }
+
+    ~AutoLangId() { intl::ffi::unic_langid_destroy(mLangId); }
+
+    operator intl::ffi::LanguageIdentifier*() const { return mLangId; }
+    bool IsValid() const { return mIsValid; }
+
+    void Reset(const nsACString& aLangTag) {
+      intl::ffi::unic_langid_destroy(mLangId);
+      mLangId = intl::ffi::unic_langid_new(&aLangTag, &mIsValid);
+    }
+
+   private:
+    intl::ffi::LanguageIdentifier* mLangId;
+    bool mIsValid;
+  };
+
+  if (aAttributeValue.IsEmpty() || aSelectorValue.IsEmpty()) {
+    return false;
+  }
+
+  AutoLangId attrLangId(aAttributeValue);
+  if (!attrLangId.IsValid()) {
+    return false;
+  }
+
+  AutoLangId selectorId(aSelectorValue);
+  if (!selectorId.IsValid()) {
+    // If it was "invalid" because of a wildcard language subtag, replace that
+    // with 'und' and try again.
+    // XXX Should unic_langid_new handle the wildcard internally?
+    if (aSelectorValue[0] == '*') {
+      nsAutoCString temp(aSelectorValue);
+      temp.Replace(0, 1, "und");
+      selectorId.Reset(temp);
+      if (!selectorId.IsValid()) {
+        return false;
+      }
+      intl::ffi::unic_langid_clear_language(selectorId);
+    } else {
+      return false;
+    }
+  }
+
+  return intl::ffi::unic_langid_matches(attrLangId, selectorId,
+                                        /* match addrLangId as range */ false,
+                                        /* match selectorId as range */ true);
+}
+
 bool nsStyleUtil::ValueIncludes(const nsAString& aValueList,
                                 const nsAString& aValue,
                                 const nsStringComparator& aComparator) {
diff --git a/layout/style/nsStyleUtil.h b/layout/style/nsStyleUtil.h
index 2983239bd70e..990996fdbc3e 100644
--- a/layout/style/nsStyleUtil.h
+++ b/layout/style/nsStyleUtil.h
@@ -37,6 +37,9 @@ class nsStyleUtil {
                                const nsAString& aSelectorValue,
                                const nsStringComparator& aComparator);
 
+  static bool LangTagCompare(const nsACString& aAttributeValue,
+                             const nsACString& aSelectorValue);
+
   static bool ValueIncludes(const nsAString& aValueList,
                             const nsAString& aValue,
                             const nsStringComparator& aComparator);