src/unicode/utfconversion.cpp

   1 #include "../../include/unicode/utfconversion.h"\r
   2 \r
   3 #ifdef _WIN32\r
   4 #include "../../include/pstdint.h"\r
   5 #endif\r
   6 \r
   7 namespace UTFConversion\r
   8 {\r
   9 \r
  10 const bool FromUTF8(const std::vector<std::string::value_type> &utf8string, std::wstring &wcstring)\r
  11 {\r
  12         if(utf8string.size()==0)\r
  13         {\r
  14                 wcstring.assign(L"");\r
  15                 return true;\r
  16         }\r
  17 \r
  18         std::vector<std::wstring::value_type> dest(utf8string.size(),0);                // dest will never be bigger than the input but could be smaller\r
  19         \r
  20         const UTF8 *sourcestart=reinterpret_cast<const UTF8 *>(&utf8string[0]);\r
  21         const UTF8 *sourceend=sourcestart+utf8string.size();\r
  22         \r
  23         if(sizeof(std::wstring::value_type)==2 && sizeof(UTF16)==2)\r
  24         {       \r
  25                 UTF16 *deststart=reinterpret_cast<UTF16 *>(&dest[0]);\r
  26                 UTF16 *destend=deststart+dest.size();\r
  27                 \r
  28                 ConversionResult rval=ConvertUTF8toUTF16(&sourcestart,sourceend,&deststart,destend,lenientConversion);\r
  29                 \r
  30                 if(rval!=conversionOK)\r
  31                 {\r
  32                         return false;   \r
  33                 }\r
  34                 \r
  35                 wcstring.assign(dest.begin(),dest.end()-(destend-deststart));\r
  36                 \r
  37         }\r
  38         else if(sizeof(std::wstring::value_type)==4 && sizeof(UTF32)==4)\r
  39         {\r
  40                 UTF32 *deststart=reinterpret_cast<UTF32 *>(&dest[0]);\r
  41                 UTF32 *destend=deststart+dest.size();\r
  42                 \r
  43                 ConversionResult rval=ConvertUTF8toUTF32(&sourcestart,sourceend,&deststart,destend,lenientConversion);\r
  44 \r
  45                 if(rval!=conversionOK)\r
  46                 {\r
  47                         return false;\r
  48                 }\r
  49                 \r
  50                 wcstring.assign(dest.begin(),dest.end()-(destend-deststart));\r
  51                 \r
  52         }\r
  53         else\r
  54         {\r
  55                 std::vector<uint32_t> dest2(utf8string.size(),0);\r
  56                 UTF32 *deststart=reinterpret_cast<UTF32 *>(&dest2[0]);\r
  57                 UTF32 *destend=deststart+dest2.size();\r
  58 \r
  59                 ConversionResult rval=ConvertUTF8toUTF32(&sourcestart,sourceend,&deststart,destend,lenientConversion);\r
  60 \r
  61                 if(rval!=conversionOK)\r
  62                 {\r
  63                         return false;\r
  64                 }\r
  65 \r
  66                 wcstring.assign(dest2.begin(),dest2.end()-(destend-deststart));\r
  67 \r
  68         }\r
  69 \r
  70         return true;\r
  71 }\r
  72 \r
  73 const bool FromUTF8(const std::string &utf8string, std::wstring &wcstring)\r
  74 {\r
  75 \r
  76         if(utf8string.size()>0)\r
  77         {\r
  78                 return FromUTF8(std::vector<std::string::value_type>(utf8string.begin(),utf8string.end()),wcstring);\r
  79         }\r
  80         else\r
  81         {\r
  82                 wcstring.assign(L"");\r
  83                 return true;\r
  84         }\r
  85 \r
  86 }\r
  87 \r
  88 const bool ToUTF8(const std::wstring &wcstring, std::string &utf8string)\r
  89 {\r
  90         if(wcstring.size()==0)\r
  91         {\r
  92                 utf8string.assign("");\r
  93                 return true;\r
  94         }\r
  95 \r
  96         std::vector<std::wstring::value_type> source(wcstring.begin(),wcstring.end());\r
  97 \r
  98         if(sizeof(std::wstring::value_type)==2 && sizeof(UTF16)==2)\r
  99         {\r
 100                 std::vector<std::string::value_type> dest(wcstring.size()*2,0);\r
 101                 \r
 102                 const UTF16 *sourcestart=reinterpret_cast<const UTF16 *>(&source[0]);\r
 103                 const UTF16 *sourceend=sourcestart+source.size();\r
 104                 \r
 105                 UTF8 *deststart=reinterpret_cast<UTF8 *>(&dest[0]);\r
 106                 UTF8 *destend=deststart+dest.size();\r
 107                 \r
 108                 ConversionResult rval=ConvertUTF16toUTF8(&sourcestart,sourceend,&deststart,destend,lenientConversion);\r
 109                 \r
 110                 if(rval!=conversionOK)\r
 111                 {\r
 112                         return false;\r
 113                 }\r
 114                 \r
 115                 utf8string.assign(dest.begin(),dest.end()-(destend-deststart));\r
 116                 \r
 117         }\r
 118         else if(sizeof(std::wstring::value_type)==4 && sizeof(UTF32)==4)\r
 119         {\r
 120                 std::vector<std::string::value_type> dest(wcstring.size()*4,0);\r
 121                 \r
 122                 const UTF32 *sourcestart=reinterpret_cast<const UTF32 *>(&source[0]);\r
 123                 const UTF32 *sourceend=sourcestart+source.size();\r
 124                 \r
 125                 UTF8 *deststart=reinterpret_cast<UTF8 *>(&dest[0]);\r
 126                 UTF8 *destend=deststart+dest.size();\r
 127                 \r
 128                 ConversionResult rval=ConvertUTF32toUTF8(&sourcestart,sourceend,&deststart,destend,lenientConversion);\r
 129                 \r
 130                 if(rval!=conversionOK)\r
 131                 {\r
 132                         return false;\r
 133                 }\r
 134                 \r
 135                 utf8string.assign(dest.begin(),dest.end()-(destend-deststart));\r
 136                 \r
 137         }\r
 138         else\r
 139         {\r
 140                 std::vector<uint32_t> source2(wcstring.begin(),wcstring.end());\r
 141                 std::vector<std::string::value_type> dest(wcstring.size()*sizeof(std::wstring::value_type),0);\r
 142                 \r
 143                 const UTF32 *sourcestart=reinterpret_cast<const UTF32 *>(&source2[0]);\r
 144                 const UTF32 *sourceend=sourcestart+source2.size();\r
 145 \r
 146                 UTF8 *deststart=reinterpret_cast<UTF8 *>(&dest[0]);\r
 147                 UTF8 *destend=deststart+dest.size();\r
 148 \r
 149                 ConversionResult rval=ConvertUTF32toUTF8(&sourcestart,sourceend,&deststart,destend,lenientConversion);\r
 150 \r
 151                 if(rval!=conversionOK)\r
 152                 {\r
 153                         return false;\r
 154                 }\r
 155 \r
 156                 utf8string.assign(dest.begin(),dest.end()-(destend-deststart));\r
 157 \r
 158         }\r
 159 \r
 160         return true;\r
 161 }\r
 162 \r
 163 }       // namespace\r