Move codepage/charset methods into related places.

Change-Id: I71417cc5b1bd00f77d42740198cc17487ebd686e
Reviewed-on: https://pdfium-review.googlesource.com/33330
Reviewed-by: dsinclair <dsinclair@chromium.org>
Commit-Queue: Art Snake <art-snake@yandex-team.ru>

4 files changed, 13 insertions, 147 deletions

diff --git a/fpdfsdk/cfx_systemhandler.cpp b/fpdfsdk/cfx_systemhandler.cpp
index 28170d2878..62d7c735b0 100644
--- a/fpdfsdk/cfx_systemhandler.cpp
+++ b/fpdfsdk/cfx_systemhandler.cpp
@@ -20,22 +20,6 @@
 #include "fpdfsdk/cpdfsdk_widget.h"
 #include "fpdfsdk/formfiller/cffl_formfiller.h"
 
-namespace {
-
-int CharSet2CP(int charset) {
-  if (charset == FX_CHARSET_ShiftJIS)
-    return FX_CODEPAGE_ShiftJIS;
-  if (charset == FX_CHARSET_ChineseSimplified)
-    return FX_CODEPAGE_ChineseSimplified;
-  if (charset == FX_CHARSET_Hangul)
-    return FX_CODEPAGE_Hangul;
-  if (charset == FX_CHARSET_ChineseTraditional)
-    return FX_CODEPAGE_ChineseTraditional;
-  return FX_CODEPAGE_DefANSI;
-}
-
-}  // namespace
-
 CFX_SystemHandler::CFX_SystemHandler(CPDFSDK_FormFillEnvironment* pFormFillEnv)
     : m_pFormFillEnv(pFormFillEnv) {}
 
@@ -121,7 +105,8 @@ CPDF_Font* CFX_SystemHandler::AddNativeTrueTypeFontToPDF(
     return nullptr;
 
   auto pFXFont = pdfium::MakeUnique<CFX_Font>();
-  pFXFont->LoadSubst(sFontFaceName, true, 0, 0, 0, CharSet2CP(nCharset), false);
+  pFXFont->LoadSubst(sFontFaceName, true, 0, 0, 0,
+                     FX_GetCodePageFromCharset(nCharset), false);
   return pDoc->AddFont(pFXFont.get(), nCharset, false);
 }
 
diff --git a/fpdfsdk/fpdf_sysfontinfo.cpp b/fpdfsdk/fpdf_sysfontinfo.cpp
index 3a3f9e1257..564912b85c 100644
--- a/fpdfsdk/fpdf_sysfontinfo.cpp
+++ b/fpdfsdk/fpdf_sysfontinfo.cpp
@@ -9,6 +9,7 @@
 #include <memory>
 
 #include "core/fxcrt/fx_codepage.h"
+#include "core/fxge/cfx_font.h"
 #include "core/fxge/cfx_fontmapper.h"
 #include "core/fxge/cfx_fontmgr.h"
 #include "core/fxge/cfx_gemodule.h"
@@ -30,6 +31,8 @@ static_assert(FXFONT_GB2312_CHARSET == FX_CHARSET_ChineseSimplified,
               "Charset must match");
 static_assert(FXFONT_CHINESEBIG5_CHARSET == FX_CHARSET_ChineseTraditional,
               "Charset must match");
+static_assert(sizeof(CFX_Font::CharsetFontMap) == sizeof(FPDF_CharsetFontMap),
+              "CFX_Font::CharsetFontMap should be same as FPDF_CharsetFontMap");
 
 class CFX_ExternalFontInfo final : public SystemFontInfoIface {
  public:
@@ -122,7 +125,7 @@ FPDF_SetSystemFontInfo(FPDF_SYSFONTINFO* pFontInfoExt) {
 }
 
 FPDF_EXPORT const FPDF_CharsetFontMap* FPDF_CALLCONV FPDF_GetDefaultTTFMap() {
-  return CPWL_FontMap::defaultTTFMap;
+  return reinterpret_cast<const FPDF_CharsetFontMap*>(CFX_Font::defaultTTFMap);
 }
 
 struct FPDF_SYSFONTINFO_DEFAULT : public FPDF_SYSFONTINFO {
diff --git a/fpdfsdk/pwl/cpwl_font_map.cpp b/fpdfsdk/pwl/cpwl_font_map.cpp
index 5e5556ebb0..f60e7d2fb1 100644
--- a/fpdfsdk/pwl/cpwl_font_map.cpp
+++ b/fpdfsdk/pwl/cpwl_font_map.cpp
@@ -21,8 +21,6 @@
 
 namespace {
 
-const char kDefaultFontName[] = "Helvetica";
-
 const char* const g_sDEStandardFontName[] = {"Courier",
                                              "Courier-Bold",
                                              "Courier-BoldOblique",
@@ -100,7 +98,8 @@ int32_t CPWL_FontMap::GetWordFontIndex(uint16_t word,
     if (KnowWord(nNewFontIndex, word))
       return nNewFontIndex;
   }
-  nNewFontIndex = GetFontIndex("Arial Unicode MS", FX_CHARSET_Default, false);
+  nNewFontIndex = GetFontIndex(CFX_Font::kUniversalDefaultFontName,
+                               FX_CHARSET_Default, false);
   if (nNewFontIndex >= 0) {
     if (KnowWord(nNewFontIndex, word))
       return nNewFontIndex;
@@ -145,7 +144,7 @@ void CPWL_FontMap::Empty() {
 }
 
 void CPWL_FontMap::Initialize() {
-  GetFontIndex(kDefaultFontName, FX_CHARSET_ANSI, false);
+  GetFontIndex(CFX_Font::kDefaultAnsiFontName, FX_CHARSET_ANSI, false);
 }
 
 bool CPWL_FontMap::IsStandardFont(const ByteString& sFontName) {
@@ -210,7 +209,7 @@ ByteString CPWL_FontMap::GetNativeFont(int32_t nCharset) {
   if (nCharset == FX_CHARSET_Default)
     nCharset = GetNativeCharset();
 
-  ByteString sFontName = GetDefaultFontByCharset(nCharset);
+  ByteString sFontName = CFX_Font::GetDefaultFontNameByCharset(nCharset);
   if (!m_pSystemHandler->FindNativeTrueTypeFont(sFontName))
     return ByteString();
 
@@ -274,135 +273,17 @@ const CPWL_FontMap_Data* CPWL_FontMap::GetFontMapData(int32_t nIndex) const {
 }
 
 int32_t CPWL_FontMap::GetNativeCharset() {
-  uint8_t nCharset = FX_CHARSET_ANSI;
-  int32_t iCodePage = FXSYS_GetACP();
-  switch (iCodePage) {
-    case FX_CODEPAGE_ShiftJIS:
-      nCharset = FX_CHARSET_ShiftJIS;
-      break;
-    case FX_CODEPAGE_ChineseSimplified:
-      nCharset = FX_CHARSET_ChineseSimplified;
-      break;
-    case FX_CODEPAGE_ChineseTraditional:
-      nCharset = FX_CHARSET_ChineseTraditional;
-      break;
-    case FX_CODEPAGE_MSWin_WesternEuropean:
-      nCharset = FX_CHARSET_ANSI;
-      break;
-    case FX_CODEPAGE_MSDOS_Thai:
-      nCharset = FX_CHARSET_Thai;
-      break;
-    case FX_CODEPAGE_Hangul:
-      nCharset = FX_CHARSET_Hangul;
-      break;
-    case FX_CODEPAGE_UTF16LE:
-      nCharset = FX_CHARSET_ANSI;
-      break;
-    case FX_CODEPAGE_MSWin_EasternEuropean:
-      nCharset = FX_CHARSET_MSWin_EasternEuropean;
-      break;
-    case FX_CODEPAGE_MSWin_Cyrillic:
-      nCharset = FX_CHARSET_MSWin_Cyrillic;
-      break;
-    case FX_CODEPAGE_MSWin_Greek:
-      nCharset = FX_CHARSET_MSWin_Greek;
-      break;
-    case FX_CODEPAGE_MSWin_Turkish:
-      nCharset = FX_CHARSET_MSWin_Turkish;
-      break;
-    case FX_CODEPAGE_MSWin_Hebrew:
-      nCharset = FX_CHARSET_MSWin_Hebrew;
-      break;
-    case FX_CODEPAGE_MSWin_Arabic:
-      nCharset = FX_CHARSET_MSWin_Arabic;
-      break;
-    case FX_CODEPAGE_MSWin_Baltic:
-      nCharset = FX_CHARSET_MSWin_Baltic;
-      break;
-    case FX_CODEPAGE_MSWin_Vietnamese:
-      nCharset = FX_CHARSET_MSWin_Vietnamese;
-      break;
-    case FX_CODEPAGE_Johab:
-      nCharset = FX_CHARSET_Johab;
-      break;
-  }
-  return nCharset;
-}
-
-const FPDF_CharsetFontMap CPWL_FontMap::defaultTTFMap[] = {
-    {FX_CHARSET_ANSI, "Helvetica"},
-    {FX_CHARSET_ChineseSimplified, "SimSun"},
-    {FX_CHARSET_ChineseTraditional, "MingLiU"},
-    {FX_CHARSET_ShiftJIS, "MS Gothic"},
-    {FX_CHARSET_Hangul, "Batang"},
-    {FX_CHARSET_MSWin_Cyrillic, "Arial"},
-#if _FX_PLATFORM_ == _FX_PLATFORM_LINUX_ || _FX_PLATFORM_ == _FX_PLATFORM_APPLE_
-    {FX_CHARSET_MSWin_EasternEuropean, "Arial"},
-#else
-    {FX_CHARSET_MSWin_EasternEuropean, "Tahoma"},
-#endif
-    {FX_CHARSET_MSWin_Arabic, "Arial"},
-    {-1, nullptr}};
-
-ByteString CPWL_FontMap::GetDefaultFontByCharset(int32_t nCharset) {
-  int i = 0;
-  while (defaultTTFMap[i].charset != -1) {
-    if (nCharset == defaultTTFMap[i].charset)
-      return defaultTTFMap[i].fontname;
-    ++i;
-  }
-  return "";
+  return FX_GetCharsetFromCodePage(FXSYS_GetACP());
 }
 
 int32_t CPWL_FontMap::CharSetFromUnicode(uint16_t word, int32_t nOldCharset) {
   // to avoid CJK Font to show ASCII
   if (word < 0x7F)
     return FX_CHARSET_ANSI;
+
   // follow the old charset
   if (nOldCharset != FX_CHARSET_Default)
     return nOldCharset;
 
-  // find new charset
-  if ((word >= 0x4E00 && word <= 0x9FA5) ||
-      (word >= 0xE7C7 && word <= 0xE7F3) ||
-      (word >= 0x3000 && word <= 0x303F) ||
-      (word >= 0x2000 && word <= 0x206F)) {
-    return FX_CHARSET_ChineseSimplified;
-  }
-
-  if (((word >= 0x3040) && (word <= 0x309F)) ||
-      ((word >= 0x30A0) && (word <= 0x30FF)) ||
-      ((word >= 0x31F0) && (word <= 0x31FF)) ||
-      ((word >= 0xFF00) && (word <= 0xFFEF))) {
-    return FX_CHARSET_ShiftJIS;
-  }
-
-  if (((word >= 0xAC00) && (word <= 0xD7AF)) ||
-      ((word >= 0x1100) && (word <= 0x11FF)) ||
-      ((word >= 0x3130) && (word <= 0x318F))) {
-    return FX_CHARSET_Hangul;
-  }
-
-  if (word >= 0x0E00 && word <= 0x0E7F)
-    return FX_CHARSET_Thai;
-
-  if ((word >= 0x0370 && word <= 0x03FF) || (word >= 0x1F00 && word <= 0x1FFF))
-    return FX_CHARSET_MSWin_Greek;
-
-  if ((word >= 0x0600 && word <= 0x06FF) || (word >= 0xFB50 && word <= 0xFEFC))
-    return FX_CHARSET_MSWin_Arabic;
-
-  if (word >= 0x0590 && word <= 0x05FF)
-    return FX_CHARSET_MSWin_Hebrew;
-
-  if (word >= 0x0400 && word <= 0x04FF)
-    return FX_CHARSET_MSWin_Cyrillic;
-
-  if (word >= 0x0100 && word <= 0x024F)
-    return FX_CHARSET_MSWin_EasternEuropean;
-
-  if (word >= 0x1E00 && word <= 0x1EFF)
-    return FX_CHARSET_MSWin_Vietnamese;
-
-  return FX_CHARSET_ANSI;
+  return CFX_Font::GetCharSetFromUnicode(word);
 }
diff --git a/fpdfsdk/pwl/cpwl_font_map.h b/fpdfsdk/pwl/cpwl_font_map.h
index 592c3fafed..c465a0fc0c 100644
--- a/fpdfsdk/pwl/cpwl_font_map.h
+++ b/fpdfsdk/pwl/cpwl_font_map.h
@@ -46,9 +46,6 @@ class CPWL_FontMap : public IPVT_FontMap {
   static int32_t GetNativeCharset();
   ByteString GetNativeFontName(int32_t nCharset);
 
-  static ByteString GetDefaultFontByCharset(int32_t nCharset);
-  static const FPDF_CharsetFontMap defaultTTFMap[];
-
  protected:
   virtual void Initialize();
   virtual CPDF_Document* GetDocument();