Linux: Do better font substitution for Chinese characters.

content/browser/renderer_host/render_sandbox_host_linux.cc

Index: content/browser/renderer_host/render_sandbox_host_linux.cc
===================================================================
--- content/browser/renderer_host/render_sandbox_host_linux.cc	(revision 258675)
+++ content/browser/renderer_host/render_sandbox_host_linux.cc	(working copy)
@@ -407,24 +407,27 @@
     }
 
     FcLangSet* langset = FcLangSetCreate();
-    MSCharSetToFontconfig(langset, charset);
+    bool is_lgc = MSCharSetToFontconfig(langset, charset);
 
     FcPattern* pattern = FcPatternCreate();
-    // TODO(agl): FC_FAMILy needs to change
     FcPatternAddString(pattern, FC_FAMILY,
                        reinterpret_cast<const FcChar8*>(face.c_str()));
 
+    // TODO(thestig) Check if we can access Chrome's per-script font preference
+    // here and select better default fonts for non-LGC case.
     std::string generic_font_name;
-    switch (fallback_family) {
-      case PP_BROWSERFONT_TRUSTED_FAMILY_SERIF:
-        generic_font_name = "Times New Roman";
-        break;
-      case PP_BROWSERFONT_TRUSTED_FAMILY_SANSSERIF:
-        generic_font_name = "Arial";
-        break;
-      case PP_BROWSERFONT_TRUSTED_FAMILY_MONOSPACE:
-        generic_font_name = "Courier New";
-        break;
+    if (is_lgc) {
+      switch (fallback_family) {
+        case PP_BROWSERFONT_TRUSTED_FAMILY_SERIF:
+          generic_font_name = "Times New Roman";
+          break;
+        case PP_BROWSERFONT_TRUSTED_FAMILY_SANSSERIF:
+          generic_font_name = "Arial";
+          break;
+        case PP_BROWSERFONT_TRUSTED_FAMILY_MONOSPACE:
+          generic_font_name = "Courier New";
+          break;
+      }
     }
     if (!generic_font_name.empty()) {
       const FcChar8* fc_generic_font_name =
@@ -547,7 +550,8 @@
 
   // MSCharSetToFontconfig translates a Microsoft charset identifier to a
   // fontconfig language set by appending to |langset|.
-  static void MSCharSetToFontconfig(FcLangSet* langset, unsigned fdwCharSet) {
+  // Returns true if |langset| is Latin/Greek/Cyrillic.
+  static bool MSCharSetToFontconfig(FcLangSet* langset, unsigned fdwCharSet) {
     // We have need to translate raw fdwCharSet values into terms that
     // fontconfig can understand. (See the description of fdwCharSet in the MSDN
     // documentation for CreateFont:
@@ -566,6 +570,7 @@
     // So, for each of the documented fdwCharSet values I've had to take a
     // guess at the set of ISO 639-1 languages intended.
 
+    bool is_lgc = false;
     switch (fdwCharSet) {
       case NPCharsetAnsi:
       // These values I don't really know what to do with, so I'm going to map
@@ -574,23 +579,25 @@
       case NPCharsetMac:
       case NPCharsetOEM:
       case NPCharsetSymbol:
+        is_lgc = true;
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("en"));
         break;
       case NPCharsetBaltic:
         // The three baltic languages.
+        is_lgc = true;
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("et"));
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("lv"));
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("lt"));
         break;
-      // TODO(jungshik): Would we be better off mapping Big5 to zh-tw
-      // and GB2312 to zh-cn? Fontconfig has 4 separate orthography
-      // files (zh-{cn,tw,hk,mo}.
       case NPCharsetChineseBIG5:
+        FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("zh-tw"));
+        break;
       case NPCharsetGB2312:
-        FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("zh"));
+        FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("zh-cn"));
         break;
       case NPCharsetEastEurope:
         // A scattering of eastern European languages.
+        is_lgc = true;
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("pl"));
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("cs"));
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("sk"));
@@ -598,6 +605,7 @@
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("hr"));
         break;
       case NPCharsetGreek:
+        is_lgc = true;
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("el"));
         break;
       case NPCharsetHangul:
@@ -606,6 +614,7 @@
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("ko"));
         break;
       case NPCharsetRussian:
+        is_lgc = true;
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("ru"));
         break;
       case NPCharsetShiftJIS:
@@ -613,9 +622,11 @@
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("ja"));
         break;
       case NPCharsetTurkish:
+        is_lgc = true;
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("tr"));
         break;
       case NPCharsetVietnamese:
+        is_lgc = true;
         FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("vi"));
         break;
       case NPCharsetArabic:
@@ -631,6 +642,7 @@
       // Don't add any languages in that case that we don't recognise the
       // constant.
     }
+    return is_lgc;
   }
 
   void SendRendererReply(const std::vector<int>& fds, const Pickle& reply,