fgas/ code cleanup.

xfa/fgas/crt/fgas_codepage.cpp

 // Copyright 2014 PDFium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
 
 #include "core/fxcrt/include/fx_ext.h"
 #include "xfa/fgas/crt/fgas_codepage.h"
 #include "xfa/fgas/crt/fgas_language.h"
 
-static const FX_CHARSET_MAP g_FXCharset2CodePageTable[] = {
+namespace {
+
+struct FX_STR2CPHASH {
+  uint32_t uHash;
+  uint16_t uCodePage;
+};
+
+struct FX_CHARSET_MAP {
+  uint16_t charset;
+  uint16_t codepage;
+};
+
+struct FX_LANG2CPMAP {
+  uint16_t wLanguage;
+  uint16_t wCodepage;
+};
+
+const FX_CHARSET_MAP g_FXCharset2CodePageTable[] = {
     {0, 1252},   {1, 0},      {2, 42},     {77, 10000}, {78, 10001},
     {79, 10003}, {80, 10008}, {81, 10002}, {83, 10005}, {84, 10004},
     {85, 10006}, {86, 10081}, {87, 10021}, {88, 10029}, {89, 10007},
     {128, 932},  {129, 949},  {130, 1361}, {134, 936},  {136, 950},
     {161, 1253}, {162, 1254}, {163, 1258}, {177, 1255}, {178, 1256},
     {186, 1257}, {204, 1251}, {222, 874},  {238, 1250}, {254, 437},
     {255, 850},
 };
-uint16_t FX_GetCodePageFromCharset(uint8_t charset) {
-  int32_t iEnd = sizeof(g_FXCharset2CodePageTable) / sizeof(FX_CHARSET_MAP) - 1;
-  ASSERT(iEnd >= 0);
-  int32_t iStart = 0, iMid;
-  do {
-    iMid = (iStart + iEnd) / 2;
-    const FX_CHARSET_MAP& cp = g_FXCharset2CodePageTable[iMid];
-    if (charset == cp.charset) {
-      return cp.codepage;
-    } else if (charset < cp.charset) {
-      iEnd = iMid - 1;
-    } else {
-      iStart = iMid + 1;
-    }
-  } while (iStart <= iEnd);
-  return 0xFFFF;
-}
-static const FX_CHARSET_MAP g_FXCodepage2CharsetTable[] = {
+
+const FX_CHARSET_MAP g_FXCodepage2CharsetTable[] = {
     {1, 0},      {2, 42},     {254, 437},  {255, 850},  {222, 874},
     {128, 932},  {134, 936},  {129, 949},  {136, 950},  {238, 1250},
     {204, 1251}, {0, 1252},   {161, 1253}, {162, 1254}, {177, 1255},
     {178, 1256}, {186, 1257}, {163, 1258}, {130, 1361}, {77, 10000},
     {78, 10001}, {79, 10003}, {80, 10008}, {81, 10002}, {83, 10005},
     {84, 10004}, {85, 10006}, {86, 10081}, {87, 10021}, {88, 10029},
     {89, 10007},
 };
-uint16_t FX_GetCharsetFromCodePage(uint16_t codepage) {
-  int32_t iEnd = sizeof(g_FXCodepage2CharsetTable) / sizeof(FX_CHARSET_MAP) - 1;
-  ASSERT(iEnd >= 0);
-  int32_t iStart = 0, iMid;
-  do {
-    iMid = (iStart + iEnd) / 2;
-    const FX_CHARSET_MAP& cp = g_FXCodepage2CharsetTable[iMid];
-    if (codepage == cp.codepage) {
-      return cp.charset;
-    } else if (codepage < cp.codepage) {
-      iEnd = iMid - 1;
-    } else {
-      iStart = iMid + 1;
-    }
-  } while (iStart <= iEnd);
-  return 0xFFFF;
-}
+
 const FX_LANG2CPMAP g_FXLang2CodepageTable[] = {
     {FX_LANG_Arabic_SaudiArabia, FX_CODEPAGE_MSWin_Arabic},
     {FX_LANG_Bulgarian_Bulgaria, FX_CODEPAGE_MSWin_Cyrillic},
     {FX_LANG_Catalan_Catalan, FX_CODEPAGE_MSWin_WesternEuropean},
     {FX_LANG_Chinese_Taiwan, FX_CODEPAGE_ChineseTraditional},
     {FX_LANG_CzechRepublic, FX_CODEPAGE_MSWin_EasternEuropean},
     {FX_LANG_Danish_Denmark, FX_CODEPAGE_MSWin_WesternEuropean},
     {FX_LANG_German_Germany, FX_CODEPAGE_MSWin_WesternEuropean},
     {FX_LANG_Greek_Greece, FX_CODEPAGE_MSWin_Greek},
     {FX_LANG_English_UnitedStates, FX_CODEPAGE_MSWin_WesternEuropean},
@@ skipping 117 matching lines @@
     {FX_LANG_Spanish_Uruguay, FX_CODEPAGE_MSWin_WesternEuropean},
     {FX_LANG_Arabic_Bahrain, FX_CODEPAGE_MSWin_Arabic},
     {FX_LANG_Spanish_Paraguay, FX_CODEPAGE_MSWin_WesternEuropean},
     {FX_LANG_Arabic_Qatar, FX_CODEPAGE_MSWin_Arabic},
     {FX_LANG_Spanish_Bolivia, FX_CODEPAGE_MSWin_WesternEuropean},
     {FX_LANG_Spanish_ElSalvador, FX_CODEPAGE_MSWin_WesternEuropean},
     {FX_LANG_Spanish_Honduras, FX_CODEPAGE_MSWin_WesternEuropean},
     {FX_LANG_Spanish_Nicaragua, FX_CODEPAGE_MSWin_WesternEuropean},
     {FX_LANG_Spanish_PuertoRico, FX_CODEPAGE_MSWin_WesternEuropean},
 };
-uint16_t FX_GetDefCodePageByLanguage(uint16_t wLanguage) {
-  int32_t iEnd = sizeof(g_FXLang2CodepageTable) / sizeof(FX_LANG2CPMAP) - 1;
-  ASSERT(iEnd >= 0);
-  int32_t iStart = 0, iMid;
-  do {
-    iMid = (iStart + iEnd) / 2;
-    const FX_LANG2CPMAP& cp = g_FXLang2CodepageTable[iMid];
-    if (wLanguage == cp.wLanguage) {
-      return cp.wCodepage;
-    } else if (wLanguage < cp.wLanguage) {
-      iEnd = iMid - 1;
-    } else {
-      iStart = iMid + 1;
-    }
-  } while (iStart <= iEnd);
-  return 0xFFFF;
-}
-static const FX_STR2CPHASH g_FXCPHashTable[] = {
+
+const FX_STR2CPHASH g_FXCPHashTable[] = {
     {0xd45, 0x6faf},      {0xd46, 0x6fb0},      {0xd47, 0x6fb1},
     {0xd48, 0x6fb2},      {0xd49, 0x4e6},       {0xd4d, 0x6fbd},
     {0xe9e, 0x4e4},       {0xc998, 0x1b5},      {0x18ef0, 0x3a8},
     {0x19f85, 0x5182},    {0x2e2335, 0x3b6},    {0x325153, 0x5182},
     {0x145bded, 0x2716},  {0x3c9a5f2, 0xc6f3},  {0x4c45f2d, 0x3a4},
     {0x4c45f4e, 0xc431},  {0x58caf51, 0x4e4},   {0x5a5cd7d, 0x3a8},
     {0x5a6c6a7, 0x4e4},   {0x5a6ca0b, 0x1b5},   {0x5a6cd68, 0x307},
     {0x5a6d8d3, 0x4e4},   {0x5a6d948, 0x354},   {0x5a6d96b, 0x362},
     {0x5a6d984, 0x366},   {0x5a90e35, 0x1b5},   {0x5e0cf00, 0x6fb5},
     {0x609c324, 0x551},   {0x617d97f, 0x5182},  {0x6a6fd91, 0xfde8},
@@ skipping 66 matching lines @@
     {0xdd80a61a, 0x3b5},  {0xdd80c0f8, 0x4e9f}, {0xdf7e46ff, 0x4fc8},
     {0xdf8680fd, 0x556a}, {0xdfb0bd6e, 0xc42d}, {0xdff05486, 0x2c4},
     {0xe3323399, 0x3a4},  {0xe60412dd, 0x3b5},  {0xeee47add, 0x4b0},
     {0xf021a186, 0x4e2},  {0xf021a187, 0x4e3},  {0xf021a188, 0x4e4},
     {0xf021a189, 0x4e5},  {0xf021a18a, 0x4e6},  {0xf021a18b, 0x4e7},
     {0xf021a18c, 0x4e8},  {0xf021a18d, 0x4e9},  {0xf021a18e, 0x4ea},
     {0xf0700456, 0x6fb3}, {0xf274f175, 0x3b5},  {0xf2a9730b, 0x3a8},
     {0xf3d463c2, 0x3a4},  {0xf52a70a3, 0xc42e}, {0xf5693147, 0x6fb3},
     {0xf637e157, 0x478},  {0xfc213f3a, 0x2717}, {0xff654d14, 0x3b5},
 };
-uint16_t FX_GetCodePageFromStringA(const FX_CHAR* pStr, int32_t iLength) {
+
+uint16_t GetCodePageFromStringA(const FX_CHAR* pStr, int32_t iLength) {
   ASSERT(pStr != NULL);
   if (iLength < 0) {
     iLength = FXSYS_strlen(pStr);
   }
   if (iLength == 0) {
     return 0xFFFF;
   }
   uint32_t uHash = FX_HashCode_GetA(CFX_ByteStringC(pStr, iLength), true);
   int32_t iStart = 0;
   int32_t iEnd = sizeof(g_FXCPHashTable) / sizeof(FX_STR2CPHASH) - 1;
   ASSERT(iEnd >= 0);
   do {
     int32_t iMid = (iStart + iEnd) / 2;
     const FX_STR2CPHASH& cp = g_FXCPHashTable[iMid];
     if (uHash == cp.uHash) {
       return (uint16_t)cp.uCodePage;
     } else if (uHash < cp.uHash) {
       iEnd = iMid - 1;
     } else {
       iStart = iMid + 1;
     }
   } while (iStart <= iEnd);
   return 0xFFFF;
 }
-uint16_t FX_GetCodePageFormStringW(const FX_WCHAR* pStr, int32_t iLength) {
+
+}  // namespace
+
+uint16_t FX_GetCodePageFromCharset(uint8_t charset) {
+  int32_t iEnd = sizeof(g_FXCharset2CodePageTable) / sizeof(FX_CHARSET_MAP) - 1;
+  ASSERT(iEnd >= 0);
+  int32_t iStart = 0, iMid;
+  do {
+    iMid = (iStart + iEnd) / 2;
+    const FX_CHARSET_MAP& cp = g_FXCharset2CodePageTable[iMid];
+    if (charset == cp.charset) {
+      return cp.codepage;
+    } else if (charset < cp.charset) {
+      iEnd = iMid - 1;
+    } else {
+      iStart = iMid + 1;
+    }
+  } while (iStart <= iEnd);
+  return 0xFFFF;
+}
+
+uint16_t FX_GetCharsetFromCodePage(uint16_t codepage) {
+  int32_t iEnd = sizeof(g_FXCodepage2CharsetTable) / sizeof(FX_CHARSET_MAP) - 1;
+  ASSERT(iEnd >= 0);
+  int32_t iStart = 0, iMid;
+  do {
+    iMid = (iStart + iEnd) / 2;
+    const FX_CHARSET_MAP& cp = g_FXCodepage2CharsetTable[iMid];
+    if (codepage == cp.codepage) {
+      return cp.charset;
+    } else if (codepage < cp.codepage) {
+      iEnd = iMid - 1;
+    } else {
+      iStart = iMid + 1;
+    }
+  } while (iStart <= iEnd);
+  return 0xFFFF;
+}
+
+uint16_t FX_GetDefCodePageByLanguage(uint16_t wLanguage) {
+  int32_t iEnd = sizeof(g_FXLang2CodepageTable) / sizeof(FX_LANG2CPMAP) - 1;
+  ASSERT(iEnd >= 0);
+  int32_t iStart = 0, iMid;
+  do {
+    iMid = (iStart + iEnd) / 2;
+    const FX_LANG2CPMAP& cp = g_FXLang2CodepageTable[iMid];
+    if (wLanguage == cp.wLanguage) {
+      return cp.wCodepage;
+    } else if (wLanguage < cp.wLanguage) {
+      iEnd = iMid - 1;
+    } else {
+      iStart = iMid + 1;
+    }
+  } while (iStart <= iEnd);
+  return 0xFFFF;
+}
+
+uint16_t FX_GetCodePageFromStringW(const FX_WCHAR* pStr, int32_t iLength) {
   if (iLength < 0) {
     iLength = FXSYS_wcslen(pStr);
   }
   if (iLength == 0) {
     return 0xFFFF;
   }
   CFX_ByteString csStr;
   FX_CHAR* pBuf = csStr.GetBuffer(iLength + 1);
   for (int32_t i = 0; i < iLength; ++i) {
     *pBuf++ = (FX_CHAR)*pStr++;
   }
   csStr.ReleaseBuffer(iLength);
-  return FX_GetCodePageFromStringA(csStr.c_str(), iLength);
+  return GetCodePageFromStringA(csStr.c_str(), iLength);
 }
+
+void FX_SwapByteOrder(FX_WCHAR* pStr, int32_t iLength) {
+  ASSERT(pStr != NULL);
+  if (iLength < 0) {
+    iLength = FXSYS_wcslen(pStr);
+  }
+  uint16_t wch;
+  if (sizeof(FX_WCHAR) > 2) {
+    while (iLength-- > 0) {
+      wch = (uint16_t)*pStr;
+      wch = (wch >> 8) | (wch << 8);
+      wch &= 0x00FF;
+      *pStr++ = wch;
+    }
+  } else {
+    while (iLength-- > 0) {
+      wch = (uint16_t)*pStr;
+      wch = (wch >> 8) | (wch << 8);
+      *pStr++ = wch;
+    }
+  }
+}
+
+void FX_UTF16ToWChar(void* pBuffer, int32_t iLength) {
+  ASSERT(pBuffer != NULL && iLength > 0);
+  if (sizeof(FX_WCHAR) == 2) {
+    return;
+  }
+  uint16_t* pSrc = (uint16_t*)pBuffer;
+  FX_WCHAR* pDst = (FX_WCHAR*)pBuffer;
+  while (--iLength >= 0) {
+    pDst[iLength] = (FX_WCHAR)pSrc[iLength];
+  }
+}
+
+void FX_WCharToUTF16(void* pBuffer, int32_t iLength) {
+  ASSERT(pBuffer != NULL && iLength > 0);
+  if (sizeof(FX_WCHAR) == 2) {
+    return;
+  }
+  const FX_WCHAR* pSrc = (const FX_WCHAR*)pBuffer;
+  uint16_t* pDst = (uint16_t*)pBuffer;
+  while (--iLength >= 0) {
+    *pDst++ = (uint16_t)*pSrc++;
+  }
+}
+
+int32_t FX_DecodeString(uint16_t wCodePage,
+                        const FX_CHAR* pSrc,
+                        int32_t* pSrcLen,
+                        FX_WCHAR* pDst,
+                        int32_t* pDstLen,
+                        FX_BOOL bErrBreak) {
+  if (wCodePage == FX_CODEPAGE_UTF8) {
+    return FX_UTF8Decode(pSrc, pSrcLen, pDst, pDstLen);
+  }
+  return -1;
+}
+int32_t FX_UTF8Decode(const FX_CHAR* pSrc,
+                      int32_t* pSrcLen,
+                      FX_WCHAR* pDst,
+                      int32_t* pDstLen) {
+  if (pSrcLen == NULL || pDstLen == NULL) {
+    return -1;
+  }
+  int32_t iSrcLen = *pSrcLen;
+  if (iSrcLen < 1) {
+    *pSrcLen = *pDstLen = 0;
+    return 1;
+  }
+  int32_t iDstLen = *pDstLen;
+  FX_BOOL bValidDst = (pDst != NULL && iDstLen > 0);
+  uint32_t dwCode = 0;
+  int32_t iPending = 0;
+  int32_t iSrcNum = 0, iDstNum = 0;
+  int32_t k = 0;
+  int32_t iIndex = 0;
+  k = 1;
+  while (iIndex < iSrcLen) {
+    uint8_t byte = (uint8_t) * (pSrc + iIndex);
+    if (byte < 0x80) {
+      iPending = 0;
+      k = 1;
+      iDstNum++;
+      iSrcNum += k;
+      if (bValidDst) {
+        *pDst++ = byte;
+        if (iDstNum >= iDstLen) {
+          break;
+        }
+      }
+    } else if (byte < 0xc0) {
+      if (iPending < 1) {
+        break;
+      }
+      iPending--;
+      dwCode |= (byte & 0x3f) << (iPending * 6);
+      if (iPending == 0) {
+        iDstNum++;
+        iSrcNum += k;
+        if (bValidDst) {
+          *pDst++ = dwCode;
+          if (iDstNum >= iDstLen) {
+            break;
+          }
+        }
+      }
+    } else if (byte < 0xe0) {
+      iPending = 1;
+      k = 2;
+      dwCode = (byte & 0x1f) << 6;
+    } else if (byte < 0xf0) {
+      iPending = 2;
+      k = 3;
+      dwCode = (byte & 0x0f) << 12;
+    } else if (byte < 0xf8) {
+      iPending = 3;
+      k = 4;
+      dwCode = (byte & 0x07) << 18;
+    } else if (byte < 0xfc) {
+      iPending = 4;
+      k = 5;
+      dwCode = (byte & 0x03) << 24;
+    } else if (byte < 0xfe) {
+      iPending = 5;
+      k = 6;
+      dwCode = (byte & 0x01) << 30;
+    } else {
+      break;
+    }
+    iIndex++;
+  }
+  *pSrcLen = iSrcNum;
+  *pDstLen = iDstNum;
+  return 1;
+}