#612: M253143 M1643126

gfx/thebes/gfxFont.cpp

@@ -2717,6 +2717,8 @@ gfxFont::ShapeTextWithoutWordCache(gfxContext *aContext,
             aTextRun->SetIsTab(aOffset + i);
         } else if (ch == '\n') {
             aTextRun->SetIsNewline(aOffset + i);
+        } else if (GetGeneralCategory(ch) == HB_UNICODE_GENERAL_CATEGORY_FORMAT) {
+            aTextRun->SetIsFormattingControl(aOffset + i);
         } else if (IsInvalidControlChar(ch) &&
             !(aTextRun->GetFlags() & gfxTextRunFactory::TEXT_HIDE_CONTROL_CHARACTERS)) {
             if (GetFontEntry()->IsUserFont() && HasCharacter(ch)) {
@@ -2931,6 +2933,8 @@ gfxFont::SplitAndInitTextRun(gfxContext *aContext,
             aTextRun->SetIsTab(aRunStart + i);
         } else if (ch == '\n') {
             aTextRun->SetIsNewline(aRunStart + i);
+        } else if (GetGeneralCategory(ch) == HB_UNICODE_GENERAL_CATEGORY_FORMAT) {
+            aTextRun->SetIsFormattingControl(aRunStart + i);
         } else if (IsInvalidControlChar(ch) &&
             !(aTextRun->GetFlags() & gfxTextRunFactory::TEXT_HIDE_CONTROL_CHARACTERS)) {
             if (GetFontEntry()->IsUserFont() && HasCharacter(ch)) {

gfx/thebes/gfxFont.h

@@ -747,7 +747,12 @@ public:
             // which is not combined with any combining characters. This flag is
             // set for all those characters except 0x20 whitespace.
             FLAG_CHAR_NO_EMPHASIS_MARK    = 0x20,
-            CHAR_TYPE_FLAGS_MASK          = 0x38,
+            // Per CSS Text, letter-spacing is not applied to formatting chars
+            // (category Cf). We mark those in the textrun so as to be able to
+            // skip them when setting up spacing in nsTextFrame.
+            FLAG_CHAR_IS_FORMATTING_CONTROL = 0x40,
+
+            CHAR_TYPE_FLAGS_MASK          = 0x78,
 
             GLYPH_COUNT_MASK = 0x00FFFF00U,
             GLYPH_COUNT_SHIFT = 8
@@ -802,6 +807,10 @@ public:
             return !CharIsSpace() &&
                 (IsSimpleGlyph() || !(mValue & FLAG_CHAR_NO_EMPHASIS_MARK));
         }
+        bool CharIsFormattingControl() const {
+            return !IsSimpleGlyph() &&
+                (mValue & FLAG_CHAR_IS_FORMATTING_CONTROL) != 0;
+        }
 
         uint32_t CharTypeFlags() const {
             return IsSimpleGlyph() ? 0 : (mValue & CHAR_TYPE_FLAGS_MASK);
@@ -880,6 +889,10 @@ public:
             NS_ASSERTION(!IsSimpleGlyph(), "Expected non-simple-glyph");
             mValue |= FLAG_CHAR_NO_EMPHASIS_MARK;
         }
+        void SetIsFormattingControl() {
+            NS_ASSERTION(!IsSimpleGlyph(), "Expected non-simple-glyph");
+            mValue |= FLAG_CHAR_IS_FORMATTING_CONTROL;
+        }
 
     private:
         uint32_t mValue;

gfx/thebes/gfxTextRun.cpp

@@ -1947,8 +1947,12 @@ gfxFontGroup::IsInvalidChar(char16_t ch)
     if (ch <= 0x9f) {
         return true;
     }
+    // Word-separating format/bidi control characters are not shaped as part
+    // of words.
     return (((ch & 0xFF00) == 0x2000 /* Unicode control character */ &&
-             (ch == 0x200B/*ZWSP*/ || ch == 0x2028/*LSEP*/ || ch == 0x2029/*PSEP*/)) ||
+             (ch == 0x200B/*ZWSP*/ || ch == 0x2028/*LSEP*/ ||
+              ch == 0x2029/*PSEP*/ || ch == 0x2060/*WJ*/)) ||
+            ch == 0xfeff/*ZWNBSP*/ ||
             IsBidiControl(ch));
 }
 

gfx/thebes/gfxTextRun.h

@@ -133,6 +133,10 @@ public:
         NS_ASSERTION(aPos < GetLength(), "aPos out of range");
         return mCharacterGlyphs[aPos].CharMayHaveEmphasisMark();
     }
+    bool CharIsFormattingControl(uint32_t aPos) const {
+        MOZ_ASSERT(aPos < GetLength());
+        return mCharacterGlyphs[aPos].CharIsFormattingControl();
+    }
 
     // All uint32_t aStart, uint32_t aLength ranges below are restricted to
     // grapheme cluster boundaries! All offsets are in terms of the string
@@ -532,6 +536,9 @@ public:
     void SetNoEmphasisMark(uint32_t aIndex) {
         EnsureComplexGlyph(aIndex).SetNoEmphasisMark();
     }
+    void SetIsFormattingControl(uint32_t aIndex) {
+        EnsureComplexGlyph(aIndex).SetIsFormattingControl();
+    }
 
     /**
      * Prefetch all the glyph extents needed to ensure that Measure calls

layout/generic/nsTextFrame.cpp

@@ -3218,12 +3218,15 @@ PropertyProvider::GetSpacing(uint32_t aStart, uint32_t aLength,
 }
 
 static bool
-CanAddSpacingAfter(gfxTextRun* aTextRun, uint32_t aOffset)
+CanAddSpacingAfter(gfxTextRun* aTextRun, uint32_t aOffset,
+                   bool aNewlineIsSignificant)
 {
   if (aOffset + 1 >= aTextRun->GetLength())
     return true;
   return aTextRun->IsClusterStart(aOffset + 1) &&
-    aTextRun->IsLigatureGroupStart(aOffset + 1);
+    aTextRun->IsLigatureGroupStart(aOffset + 1) &&
+   !aTextRun->CharIsFormattingControl(aOffset) &&
+   !(aNewlineIsSignificant && aTextRun->CharIsNewline(aOffset));
 }
 
 void
@@ -3247,11 +3250,13 @@ PropertyProvider::GetSpacingInternal(uint32_t aStart, uint32_t aLength,
     // Iterate over non-skipped characters
     nsSkipCharsRunIterator
       run(start, nsSkipCharsRunIterator::LENGTH_UNSKIPPED_ONLY, aLength);
+    bool newlineIsSignificant = mTextStyle->NewlineIsSignificant(mFrame);
     while (run.NextRun()) {
       uint32_t runOffsetInSubstring = run.GetSkippedOffset() - aStart;
       gfxSkipCharsIterator iter = run.GetPos();
       for (int32_t i = 0; i < run.GetRunLength(); ++i) {
-        if (CanAddSpacingAfter(mTextRun, run.GetSkippedOffset() + i)) {
+        if (CanAddSpacingAfter(mTextRun, run.GetSkippedOffset() + i,
+                               newlineIsSignificant)) {
           // End of a cluster, not in a ligature: put letter-spacing after it
           aSpacing[runOffsetInSubstring + i].mAfter += mLetterSpacing;
         }