Linux: Do better font substitution for Chinese characters.

content/browser/renderer_host/sandbox_ipc_linux.cc

 // Copyright 2014 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "content/browser/renderer_host/sandbox_ipc_linux.h"
 
 #include <fcntl.h>
 #include <fontconfig/fontconfig.h>
 #include <sys/poll.h>
 #include <sys/socket.h>
@@ skipping 21 matching lines @@
 
 using blink::WebCString;
 using blink::WebFontInfo;
 using blink::WebUChar;
 using blink::WebUChar32;
 
 namespace {
 
 // 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.
+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:
   // http://msdn.microsoft.com/en-us/library/dd183499(VS.85).aspx )
   //
   // Although the argument is /called/ 'charset', the actual values conflate
   // character sets (which are sets of Unicode code points) and character
   // encodings (which are algorithms for turning a series of bits into a
   // series of code points.) Sometimes the values will name a language,
   // sometimes they'll name an encoding. In the latter case I'm assuming that
   // they mean the set of code points in the domain of that encoding.
   //
   // fontconfig deals with ISO 639-1 language codes:
   //   http://en.wikipedia.org/wiki/List_of_ISO_639-1_codes
   //
   // 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
     // them to English also.
     case NPCharsetDefault:
     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"));
       FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("hu"));
       FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("hr"));
       break;
     case NPCharsetGreek:
+      is_lgc = true;
       FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("el"));
       break;
     case NPCharsetHangul:
     case NPCharsetJohab:
       // Korean
       FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("ko"));
       break;
     case NPCharsetRussian:
+      is_lgc = true;
       FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("ru"));
       break;
     case NPCharsetShiftJIS:
       // Japanese
       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:
       FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("ar"));
       break;
     case NPCharsetHebrew:
       FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("he"));
       break;
     case NPCharsetThai:
       FcLangSetAdd(langset, reinterpret_cast<const FcChar8*>("th"));
       break;
       // default:
       // Don't add any languages in that case that we don't recognise the
       // constant.
   }
+  return is_lgc;
 }
 
 }  // namespace
 
 namespace content {
 
 SandboxIPCHandler::SandboxIPCHandler(int lifeline_fd, int browser_socket)
     : lifeline_fd_(lifeline_fd), browser_socket_(browser_socket) {
   // FontConfig doesn't provide a standard property to control subpixel
   // positioning, so we pass the current setting through to WebKit.
@@ skipping 294 matching lines @@
 
   if (!pickle.ReadString(&iter, &face) || face.empty() ||
       !pickle.ReadBool(&iter, &is_bold) ||
       !pickle.ReadBool(&iter, &is_italic) ||
       !pickle.ReadUInt32(&iter, &charset) ||
       !pickle.ReadUInt32(&iter, &fallback_family)) {
     return;
   }
 
   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 =
         reinterpret_cast<const FcChar8*>(generic_font_name.c_str());
     FcPatternAddString(pattern, FC_FAMILY, fc_generic_font_name);
   }
 
   if (is_bold)
     FcPatternAddInteger(pattern, FC_WEIGHT, FC_WEIGHT_BOLD);
   if (is_italic)
@@ skipping 155 matching lines @@
 }
 
 void SandboxIPCHandler::EnsureWebKitInitialized() {
   if (webkit_platform_support_)
     return;
   webkit_platform_support_.reset(new BlinkPlatformImpl);
   blink::initializeWithoutV8(webkit_platform_support_.get());
 }
 
 }  // namespace content