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, 103 insertions, 34 deletions

diff --git a/core/fxge/android/cfpf_skiafontmgr.cpp b/core/fxge/android/cfpf_skiafontmgr.cpp
index d44d7d56d2..0b49f6283c 100644
--- a/core/fxge/android/cfpf_skiafontmgr.cpp
+++ b/core/fxge/android/cfpf_skiafontmgr.cpp
@@ -187,9 +187,7 @@ uint32_t FPF_SKIAGetFamilyHash(const ByteStringView& bsFamily,
 }
 
 bool FPF_SkiaIsCJK(uint8_t uCharset) {
-  return (uCharset == FX_CHARSET_ChineseSimplified) ||
-         (uCharset == FX_CHARSET_ChineseTraditional) ||
-         (uCharset == FX_CHARSET_Hangul) || (uCharset == FX_CHARSET_ShiftJIS);
+  return FX_CharSetIsCJK(uCharset);
 }
 
 bool FPF_SkiaMaybeSymbol(const ByteStringView& bsFacename) {
diff --git a/core/fxge/cfx_font.cpp b/core/fxge/cfx_font.cpp
index d04fc0dac0..bee1d789ef 100644
--- a/core/fxge/cfx_font.cpp
+++ b/core/fxge/cfx_font.cpp
@@ -209,6 +209,88 @@ const uint8_t CFX_Font::s_WeightPow_SHIFTJIS[] = {
     59, 59, 59, 59, 59, 59, 59, 60, 60, 60, 60, 60, 60, 60, 60,
 };
 
+const CFX_Font::CharsetFontMap CFX_Font::defaultTTFMap[] = {
+    {FX_CHARSET_ANSI, kDefaultAnsiFontName},
+    {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}};
+
+// static
+const char CFX_Font::kDefaultAnsiFontName[] = "Helvetica";
+// static
+const char CFX_Font::kUniversalDefaultFontName[] = "Arial Unicode MS";
+
+// static
+ByteString CFX_Font::GetDefaultFontNameByCharset(uint8_t nCharset) {
+  int i = 0;
+  while (defaultTTFMap[i].charset != -1) {
+    if (nCharset == static_cast<uint8_t>(defaultTTFMap[i].charset))
+      return defaultTTFMap[i].fontname;
+    ++i;
+  }
+  return kUniversalDefaultFontName;
+}
+
+// static
+uint8_t CFX_Font::GetCharSetFromUnicode(uint16_t word) {
+  // to avoid CJK Font to show ASCII
+  if (word < 0x7F)
+    return FX_CHARSET_ANSI;
+
+  // 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;
+}
+
 CFX_Font::CFX_Font()
     :
       m_Face(nullptr),
diff --git a/core/fxge/cfx_font.h b/core/fxge/cfx_font.h
index fbf35eaca1..a5a0057c16 100644
--- a/core/fxge/cfx_font.h
+++ b/core/fxge/cfx_font.h
@@ -30,6 +30,11 @@ class CFX_Font {
   CFX_Font();
   ~CFX_Font();
 
+  static const char kDefaultAnsiFontName[];
+  static const char kUniversalDefaultFontName[];
+  static ByteString GetDefaultFontNameByCharset(uint8_t nCharset);
+  static uint8_t GetCharSetFromUnicode(uint16_t word);
+
   void LoadSubst(const ByteString& face_name,
                  bool bTrueType,
                  uint32_t flags,
@@ -96,6 +101,19 @@ class CFX_Font {
   static const uint8_t s_WeightPow_11[kWeightPowArraySize];
   static const uint8_t s_WeightPow_SHIFTJIS[kWeightPowArraySize];
 
+  // This struct shoub same as FPDF_CharsetFontMap
+  typedef struct {
+    int charset;           // Character Set Enum value, see FX_CHARSET_XXX.
+    const char* fontname;  // Name of default font to use with that charset.
+  } CharsetFontMap;
+
+  /**
+   *    Pointer to the default character set to TT Font name map. The
+   *    map is an array of CharsetFontMap structs, with its end indicated
+   *    by a { -1, NULL } entry.
+   **/
+  static const CharsetFontMap defaultTTFMap[];
+
 #ifdef PDF_ENABLE_XFA
  protected:
   std::unique_ptr<FXFT_StreamRec> m_pOwnedStream;
diff --git a/core/fxge/cfx_fontmapper.cpp b/core/fxge/cfx_fontmapper.cpp
index b9a8eac1a1..e6fc984edc 100644
--- a/core/fxge/cfx_fontmapper.cpp
+++ b/core/fxge/cfx_fontmapper.cpp
@@ -145,19 +145,6 @@ const struct AltFontFamily {
     {"ForteMT", "Forte"},
 };
 
-const struct CODEPAGE_MAP {
-  uint16_t codepage;
-  uint8_t charset;
-} g_Codepage2CharsetTable[] = {
-    {0, 1},      {42, 2},     {437, 254},  {850, 255},  {874, 222},
-    {932, 128},  {936, 134},  {949, 129},  {950, 136},  {1250, 238},
-    {1251, 204}, {1252, 0},   {1253, 161}, {1254, 162}, {1255, 177},
-    {1256, 178}, {1257, 186}, {1258, 163}, {1361, 130}, {10000, 77},
-    {10001, 78}, {10002, 81}, {10003, 79}, {10004, 84}, {10005, 83},
-    {10006, 85}, {10007, 89}, {10008, 80}, {10021, 87}, {10029, 88},
-    {10081, 86},
-};
-
 ByteString TT_NormalizeName(const char* family) {
   ByteString norm(family);
   norm.Remove(' ');
@@ -170,19 +157,6 @@ ByteString TT_NormalizeName(const char* family) {
   return norm;
 }
 
-uint8_t GetCharsetFromCodePage(uint16_t codepage) {
-  const CODEPAGE_MAP* pEnd =
-      g_Codepage2CharsetTable + FX_ArraySize(g_Codepage2CharsetTable);
-  const CODEPAGE_MAP* pCharmap =
-      std::lower_bound(g_Codepage2CharsetTable, pEnd, codepage,
-                       [](const CODEPAGE_MAP& charset, uint16_t page) {
-                         return charset.codepage < page;
-                       });
-  if (pCharmap < pEnd && codepage == pCharmap->codepage)
-    return pCharmap->charset;
-  return FX_CHARSET_Default;
-}
-
 void GetFontFamily(uint32_t nStyle, ByteString* fontName) {
   if (fontName->Contains("Script")) {
     if (FontStyleIsBold(nStyle))
@@ -544,13 +518,10 @@ FXFT_Face CFX_FontMapper::FindSubstFont(const ByteString& name,
 
   int Charset = FX_CHARSET_ANSI;
   if (WindowCP)
-    Charset = GetCharsetFromCodePage(WindowCP);
+    Charset = FX_GetCharsetFromCodePage(WindowCP);
   else if (iBaseFont == kNumStandardFonts && FontStyleIsSymbolic(flags))
     Charset = FX_CHARSET_Symbol;
-  const bool bCJK = (Charset == FX_CHARSET_ShiftJIS ||
-                     Charset == FX_CHARSET_ChineseSimplified ||
-                     Charset == FX_CHARSET_Hangul ||
-                     Charset == FX_CHARSET_ChineseTraditional);
+  const bool bCJK = FX_CharSetIsCJK(Charset);
   bool bItalic = FontStyleIsItalic(nStyle);
 
   GetFontFamily(nStyle, &family);