Bug 1331339 - Don't start a new script run when the current script appears in the next character's ScriptExtensions property, or next char is a cluster-extender. r=jrmuizel

gfx/thebes/gfxScriptItemizer.cpp

@@ -107,12 +107,17 @@ gfxScriptItemizer::fixup(Script newScriptCode)
     }
 }
 
+// We regard the current char as having the same script as the in-progress run
+// if either script code is Common or Inherited, or if the run script appears
+// in the character's ScriptExtensions, or if the char is a cluster extender.
 static inline bool
-SameScript(Script runScript, Script currCharScript)
+SameScript(Script runScript, Script currCharScript, uint32_t aCurrCh)
 {
     return runScript <= Script::INHERITED ||
            currCharScript <= Script::INHERITED ||
-           currCharScript == runScript;
+           currCharScript == runScript ||
+           IsClusterExtender(aCurrCh) ||
+           HasScript(aCurrCh, runScript);
 }
 
 gfxScriptItemizer::gfxScriptItemizer(const char16_t *src, uint32_t length)
@@ -194,7 +199,7 @@ gfxScriptItemizer::Next(uint32_t& aRunStart, uint32_t& aRunLimit,
             }
         }
 
-        if (SameScript(scriptCode, sc)) {
+        if (SameScript(scriptCode, sc, ch)) {
             if (scriptCode <= Script::INHERITED &&
                 sc > Script::INHERITED)
             {

intl/unicharutil/util/nsUnicodeProperties.h

@@ -105,6 +105,12 @@ GetScriptCode(uint32_t aCh)
   return Script(uscript_getScript(aCh, &err));
 }
 
+inline bool
+HasScript(uint32_t aCh, Script aScript)
+{
+  return uscript_hasScript(aCh, UScriptCode(aScript));
+}
+
 inline uint32_t
 GetScriptTagForCode(Script aScriptCode)
 {
@@ -189,6 +195,16 @@ uint8_t GetLineBreakClass(uint32_t aCh);
 
 Script GetScriptCode(uint32_t aCh);
 
+// We don't support ScriptExtensions.txt data when building without ICU.
+// The most important cases will still be handled in gfxScriptItemizer
+// by checking IsClusterExtender to avoid breaking script runs within
+// a cluster.
+inline bool
+HasScript(uint32_t aCh, Script aScript)
+{
+  return false;
+}
+
 uint32_t GetScriptTagForCode(Script aScriptCode);
 
 PairedBracketType GetPairedBracketType(uint32_t aCh);