From b7240dcf1afbe2942faccf46f386520e334cfecb Mon Sep 17 00:00:00 2001
From: "L. David Baron" <dbaron@dbaron.org>
Date: Mon, 20 Aug 2012 19:23:32 -0700
Subject: [PATCH] Only do hyphenation when the language was specified
 explicitly, rather than using an encoding-inferred language.  (Bug 702121,
 patch 2)  r=jfkthame

---
 content/base/public/nsLineBreaker.h           |  8 ++---
 content/base/src/nsLineBreaker.cpp            | 30 ++++++++++---------
 layout/generic/nsTextFrameThebes.cpp          | 10 +++++--
 layout/reftests/text/auto-hyphenation-10.html | 11 +++++++
 layout/reftests/text/auto-hyphenation-8.html  | 10 +++++++
 layout/reftests/text/auto-hyphenation-9.html  | 11 +++++++
 layout/reftests/text/reftest.list             |  4 ++-
 7 files changed, 62 insertions(+), 22 deletions(-)
 create mode 100644 layout/reftests/text/auto-hyphenation-10.html
 create mode 100644 layout/reftests/text/auto-hyphenation-8.html
 create mode 100644 layout/reftests/text/auto-hyphenation-9.html

diff --git a/content/base/public/nsLineBreaker.h b/content/base/public/nsLineBreaker.h
index d4b9723b00b..883c532d8c4 100644
--- a/content/base/public/nsLineBreaker.h
+++ b/content/base/public/nsLineBreaker.h
@@ -147,14 +147,14 @@ public:
    * @param aSink can be null if the breaks are not actually needed (we may
    * still be setting up state for later breaks)
    */
-  nsresult AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32 aLength,
+  nsresult AppendText(nsIAtom* aHyphenationLanguage, const PRUnichar* aText, PRUint32 aLength,
                       PRUint32 aFlags, nsILineBreakSink* aSink);
   /**
    * Feed 8-bit text into the linebreaker for analysis. aLength must be nonzero.
    * @param aSink can be null if the breaks are not actually needed (we may
    * still be setting up state for later breaks)
    */
-  nsresult AppendText(nsIAtom* aLangGroup, const PRUint8* aText, PRUint32 aLength,
+  nsresult AppendText(nsIAtom* aHyphenationLanguage, const PRUint8* aText, PRUint32 aLength,
                       PRUint32 aFlags, nsILineBreakSink* aSink);
   /**
    * Reset all state. This means the current run has ended; any outstanding
@@ -198,7 +198,7 @@ private:
   // appropriate sink(s). Then we clear the current word state.
   nsresult FlushCurrentWord();
 
-  void UpdateCurrentWordLangGroup(nsIAtom *aLangGroup);
+  void UpdateCurrentWordLanguage(nsIAtom *aHyphenationLanguage);
 
   void FindHyphenationPoints(nsHyphenator *aHyphenator,
                              const PRUnichar *aTextStart,
@@ -208,7 +208,7 @@ private:
   nsAutoTArray<PRUnichar,100> mCurrentWord;
   // All the items that contribute to mCurrentWord
   nsAutoTArray<TextItem,2>    mTextItems;
-  nsIAtom*                    mCurrentWordLangGroup;
+  nsIAtom*                    mCurrentWordLanguage;
   bool                        mCurrentWordContainsMixedLang;
   bool                        mCurrentWordContainsComplexChar;
 
diff --git a/content/base/src/nsLineBreaker.cpp b/content/base/src/nsLineBreaker.cpp
index ef84cd53e49..4773bd2a86c 100644
--- a/content/base/src/nsLineBreaker.cpp
+++ b/content/base/src/nsLineBreaker.cpp
@@ -11,7 +11,7 @@
 #include "nsHyphenator.h"
 
 nsLineBreaker::nsLineBreaker()
-  : mCurrentWordLangGroup(nullptr),
+  : mCurrentWordLanguage(nullptr),
     mCurrentWordContainsMixedLang(false),
     mCurrentWordContainsComplexChar(false),
     mAfterBreakableSpace(false), mBreakHere(false),
@@ -77,7 +77,7 @@ nsLineBreaker::FlushCurrentWord()
                         breakState.Elements());
   }
 
-  bool autoHyphenate = mCurrentWordLangGroup &&
+  bool autoHyphenate = mCurrentWordLanguage &&
     !mCurrentWordContainsMixedLang;
   PRUint32 i;
   for (i = 0; autoHyphenate && i < mTextItems.Length(); ++i) {
@@ -88,7 +88,7 @@ nsLineBreaker::FlushCurrentWord()
   }
   if (autoHyphenate) {
     nsRefPtr<nsHyphenator> hyphenator =
-      nsHyphenationManager::Instance()->GetHyphenator(mCurrentWordLangGroup);
+      nsHyphenationManager::Instance()->GetHyphenator(mCurrentWordLanguage);
     if (hyphenator) {
       FindHyphenationPoints(hyphenator,
                             mCurrentWord.Elements(),
@@ -140,12 +140,12 @@ nsLineBreaker::FlushCurrentWord()
   mTextItems.Clear();
   mCurrentWordContainsComplexChar = false;
   mCurrentWordContainsMixedLang = false;
-  mCurrentWordLangGroup = nullptr;
+  mCurrentWordLanguage = nullptr;
   return NS_OK;
 }
 
 nsresult
-nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32 aLength,
+nsLineBreaker::AppendText(nsIAtom* aHyphenationLanguage, const PRUnichar* aText, PRUint32 aLength,
                           PRUint32 aFlags, nsILineBreakSink* aSink)
 {
   NS_ASSERTION(aLength > 0, "Appending empty text...");
@@ -161,7 +161,7 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32
       if (!mCurrentWordContainsComplexChar && IsComplexChar(aText[offset])) {
         mCurrentWordContainsComplexChar = true;
       }
-      UpdateCurrentWordLangGroup(aLangGroup);
+      UpdateCurrentWordLanguage(aHyphenationLanguage);
       ++offset;
     }
 
@@ -211,8 +211,10 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32
   bool wordHasComplexChar = false;
 
   nsRefPtr<nsHyphenator> hyphenator;
-  if ((aFlags & BREAK_USE_AUTO_HYPHENATION) && !(aFlags & BREAK_SUPPRESS_INSIDE)) {
-    hyphenator = nsHyphenationManager::Instance()->GetHyphenator(aLangGroup);
+  if ((aFlags & BREAK_USE_AUTO_HYPHENATION) &&
+      !(aFlags & BREAK_SUPPRESS_INSIDE) &&
+      aHyphenationLanguage) {
+    hyphenator = nsHyphenationManager::Instance()->GetHyphenator(aHyphenationLanguage);
   }
 
   for (;;) {
@@ -275,7 +277,7 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUnichar* aText, PRUint32
         mTextItems.AppendElement(TextItem(aSink, wordStart, len, aFlags));
         // Ensure that the break-before for this word is written out
         offset = wordStart + 1;
-        UpdateCurrentWordLangGroup(aLangGroup);
+        UpdateCurrentWordLanguage(aHyphenationLanguage);
         break;
       }
     }
@@ -311,7 +313,7 @@ nsLineBreaker::FindHyphenationPoints(nsHyphenator *aHyphenator,
 }
 
 nsresult
-nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUint8* aText, PRUint32 aLength,
+nsLineBreaker::AppendText(nsIAtom* aHyphenationLanguage, const PRUint8* aText, PRUint32 aLength,
                           PRUint32 aFlags, nsILineBreakSink* aSink)
 {
   NS_ASSERTION(aLength > 0, "Appending empty text...");
@@ -321,7 +323,7 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUint8* aText, PRUint32 aL
     nsAutoString str;
     const char* cp = reinterpret_cast<const char*>(aText);
     CopyASCIItoUTF16(nsDependentCSubstring(cp, cp + aLength), str);
-    return AppendText(aLangGroup, str.get(), aLength, aFlags, aSink);
+    return AppendText(aHyphenationLanguage, str.get(), aLength, aFlags, aSink);
   }
 
   PRUint32 offset = 0;
@@ -446,12 +448,12 @@ nsLineBreaker::AppendText(nsIAtom* aLangGroup, const PRUint8* aText, PRUint32 aL
 }
 
 void
-nsLineBreaker::UpdateCurrentWordLangGroup(nsIAtom *aLangGroup)
+nsLineBreaker::UpdateCurrentWordLanguage(nsIAtom *aHyphenationLanguage)
 {
-  if (mCurrentWordLangGroup && mCurrentWordLangGroup != aLangGroup) {
+  if (mCurrentWordLanguage && mCurrentWordLanguage != aHyphenationLanguage) {
     mCurrentWordContainsMixedLang = true;
   } else {
-    mCurrentWordLangGroup = aLangGroup;
+    mCurrentWordLanguage = aHyphenationLanguage;
   }
 }
 
diff --git a/layout/generic/nsTextFrameThebes.cpp b/layout/generic/nsTextFrameThebes.cpp
index 70cbe293b86..4832c9bfa7c 100644
--- a/layout/generic/nsTextFrameThebes.cpp
+++ b/layout/generic/nsTextFrameThebes.cpp
@@ -2215,7 +2215,11 @@ BuildTextRunsScanner::SetupBreakSinksForTextRun(gfxTextRun* aTextRun,
                                                 PRUint32    aFlags)
 {
   // textruns have uniform language
-  nsIAtom* language = mMappedFlows[0].mStartFrame->GetStyleFont()->mLanguage;
+  const nsStyleFont *styleFont = mMappedFlows[0].mStartFrame->GetStyleFont();
+  // We should only use a language for hyphenation if it was specified
+  // explicitly.
+  nsIAtom* hyphenationLanguage =
+    styleFont->mExplicitLanguage ? styleFont->mLanguage : nullptr;
   // We keep this pointed at the skip-chars data for the current mappedFlow.
   // This lets us cheaply check whether the flow has compressed initial
   // whitespace...
@@ -2269,11 +2273,11 @@ BuildTextRunsScanner::SetupBreakSinksForTextRun(gfxTextRun* aTextRun,
         (aFlags & SBS_SUPPRESS_SINK) ? nullptr : (*breakSink).get();
       if (aFlags & SBS_DOUBLE_BYTE) {
         const PRUnichar* text = reinterpret_cast<const PRUnichar*>(aTextPtr);
-        mLineBreaker.AppendText(language, text + offset,
+        mLineBreaker.AppendText(hyphenationLanguage, text + offset,
                                 length, flags, sink);
       } else {
         const PRUint8* text = reinterpret_cast<const PRUint8*>(aTextPtr);
-        mLineBreaker.AppendText(language, text + offset,
+        mLineBreaker.AppendText(hyphenationLanguage, text + offset,
                                 length, flags, sink);
       }
     }
diff --git a/layout/reftests/text/auto-hyphenation-10.html b/layout/reftests/text/auto-hyphenation-10.html
new file mode 100644
index 00000000000..41f35cf3d1a
--- /dev/null
+++ b/layout/reftests/text/auto-hyphenation-10.html
@@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html>
+<meta charset="UTF-8">
+<!-- check that hyphenation is not applied when language is not specified -->
+<body>
+<div style="width: 5em; -moz-hyphens: auto;">
+supercalifragilisticexpialidocious
+</div>
+</body>
+</html>
+
diff --git a/layout/reftests/text/auto-hyphenation-8.html b/layout/reftests/text/auto-hyphenation-8.html
new file mode 100644
index 00000000000..4956bce75cf
--- /dev/null
+++ b/layout/reftests/text/auto-hyphenation-8.html
@@ -0,0 +1,10 @@
+<!DOCTYPE html>
+<html>
+<!-- check that hyphenation is not applied when language is not specified -->
+<body>
+<div style="width: 5em; -moz-hyphens: auto;">
+supercalifragilisticexpialidocious
+</div>
+</body>
+</html>
+
diff --git a/layout/reftests/text/auto-hyphenation-9.html b/layout/reftests/text/auto-hyphenation-9.html
new file mode 100644
index 00000000000..53714136e6d
--- /dev/null
+++ b/layout/reftests/text/auto-hyphenation-9.html
@@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html>
+<meta charset="iso-8859-1">
+<!-- check that hyphenation is not applied when language is not specified -->
+<body>
+<div style="width: 5em; -moz-hyphens: auto;">
+supercalifragilisticexpialidocious
+</div>
+</body>
+</html>
+
diff --git a/layout/reftests/text/reftest.list b/layout/reftests/text/reftest.list
index 9d7812b0e78..d52d3fd5c45 100644
--- a/layout/reftests/text/reftest.list
+++ b/layout/reftests/text/reftest.list
@@ -180,6 +180,9 @@ HTTP(..) == graphite-05-feat.html graphite-05-ref.html
 == auto-hyphenation-5.html auto-hyphenation-5-ref.html
 == auto-hyphenation-6.html auto-hyphenation-6-ref.html
 == auto-hyphenation-7.html auto-hyphenation-7-ref.html
+== auto-hyphenation-8.html auto-hyphenation-4-ref.html
+== auto-hyphenation-9.html auto-hyphenation-4-ref.html
+== auto-hyphenation-10.html auto-hyphenation-4-ref.html
 == auto-hyphenation-af-1.html auto-hyphenation-af-1-ref.html
 == auto-hyphenation-bg-1.html auto-hyphenation-bg-1-ref.html
 == auto-hyphenation-ca-1.html auto-hyphenation-ca-1-ref.html
@@ -217,4 +220,3 @@ HTTP(..) == graphite-05-feat.html graphite-05-ref.html
 != auto-hyphenation-sv-1.html auto-hyphenation-sv-1-notref.html # verify swedish != english
 == auto-hyphenation-tr-1.html auto-hyphenation-tr-1-ref.html
 == auto-hyphenation-uk-1.html auto-hyphenation-uk-1-ref.html
-