mbtoutf8/mbtoutf8.cpp

   1 // mbtoutf8.cpp : \83R\83\93\83\\81[\83\8b \83A\83v\83\8a\83P\81[\83V\83\87\83\93\82Ì\83G\83\93\83g\83\8a \83|\83C\83\93\83g\82ð\92è\8b`\82µ\82Ü\82·\81B\r
   2 //\r
   3 \r
   4 #include "stdafx.h"\r
   5 \r
   6 \r
   7 int _tmain(int argc, _TCHAR* argv[])\r
   8 {\r
   9         FILE* fpIn;\r
  10         FILE* fpOut;\r
  11         char InBuffer[16384];\r
  12         char OutBuffer[65536];\r
  13         int InPosition;\r
  14         int OutPosition;\r
  15         bool bEscape;\r
  16         bool bString;\r
  17         bool bEncoded;\r
  18         int Count;\r
  19         int UTF8Count;\r
  20         wchar_t UTF16Buffer[4];\r
  21         char UTF8Buffer[8];\r
  22         int i;\r
  23         _tsetlocale(LC_ALL, _T(""));\r
  24         if(argc != 3)\r
  25         {\r
  26                 _tprintf(_T("\83}\83\8b\83`\83o\83C\83g\95¶\8e\9a\81i\83R\81[\83h\83y\81[\83W932\82Ü\82½\82ÍShift JIS\81j\82Å\8f\91\82©\82ê\82½C\8c¾\8cê\83\\81[\83X\83t\83@\83C\83\8b\93à\82Ì\95¶\8e\9a\97ñ\83\8a\83e\83\89\83\8b\82ðUTF-8\82É\83G\83\93\83R\81[\83h\82µ\82Ü\82·\81B\n"));\r
  27                 _tprintf(_T("\83R\83}\83\93\83h\83\89\83C\83\93\n"));\r
  28                 _tprintf(_T("mbtoutf8 [in] [out]\n"));\r
  29                 _tprintf(_T("[in]    \8c³\82Ì\83\\81[\83X\83t\83@\83C\83\8b\82Ì\83t\83@\83C\83\8b\96¼\n"));\r
  30                 _tprintf(_T("[out]   \95Û\91¶\90æ\82Ì\83t\83@\83C\83\8b\96¼\n"));\r
  31                 return 0;\r
  32         }\r
  33         fpIn = _tfopen(argv[1], _T("rt"));\r
  34         if(!fpIn)\r
  35         {\r
  36                 _tprintf(_T("\83t\83@\83C\83\8b\"%s\"\82ª\8aJ\82¯\82Ü\82¹\82ñ\81B\n"), argv[1]);\r
  37                 return 0;\r
  38         }\r
  39         fpOut = _tfopen(argv[2], _T("wt"));\r
  40         if(!fpIn)\r
  41         {\r
  42                 _tprintf(_T("\83t\83@\83C\83\8b\"%s\"\82ª\8dì\90¬\82Å\82«\82Ü\82¹\82ñ\81B\n"), argv[2]);\r
  43                 return 0;\r
  44         }\r
  45         while(fgets(InBuffer, sizeof(InBuffer) / sizeof(char), fpIn))\r
  46         {\r
  47                 InPosition = 0;\r
  48                 OutPosition = 0;\r
  49                 bEscape = false;\r
  50                 bString = false;\r
  51                 bEncoded = false;\r
  52                 while(InBuffer[InPosition])\r
  53                 {\r
  54                         Count = max(mblen(&InBuffer[InPosition], 4), 1);\r
  55                         if(Count == 1)\r
  56                         {\r
  57                                 switch(InBuffer[InPosition])\r
  58                                 {\r
  59                                 case '\\':\r
  60                                         bEscape = !bEscape;\r
  61                                         bEncoded = false;\r
  62                                         strncpy(&OutBuffer[OutPosition], &InBuffer[InPosition], Count);\r
  63                                         InPosition += Count;\r
  64                                         OutPosition += Count;\r
  65                                         break;\r
  66                                 case '\"':\r
  67                                         bEscape = false;\r
  68                                         bString = !bString;\r
  69                                         bEncoded = false;\r
  70                                         strncpy(&OutBuffer[OutPosition], &InBuffer[InPosition], Count);\r
  71                                         InPosition += Count;\r
  72                                         OutPosition += Count;\r
  73                                         break;\r
  74                                 case '0':\r
  75                                 case '1':\r
  76                                 case '2':\r
  77                                 case '3':\r
  78                                 case '4':\r
  79                                 case '5':\r
  80                                 case '6':\r
  81                                 case '7':\r
  82                                 case '8':\r
  83                                 case '9':\r
  84                                 case 'A':\r
  85                                 case 'B':\r
  86                                 case 'C':\r
  87                                 case 'D':\r
  88                                 case 'E':\r
  89                                 case 'F':\r
  90                                 case 'a':\r
  91                                 case 'b':\r
  92                                 case 'c':\r
  93                                 case 'd':\r
  94                                 case 'e':\r
  95                                 case 'f':\r
  96                                         bEscape = false;\r
  97                                         if(bEncoded)\r
  98                                         {\r
  99                                                 for(i = 0; i < Count; i++)\r
 100                                                 {\r
 101                                                         sprintf(&OutBuffer[OutPosition], "\\x%02X", (unsigned char)InBuffer[InPosition]);\r
 102                                                         InPosition++;\r
 103                                                         OutPosition += 4;\r
 104                                                 }\r
 105                                         }\r
 106                                         else\r
 107                                         {\r
 108                                                 strncpy(&OutBuffer[OutPosition], &InBuffer[InPosition], Count);\r
 109                                                 InPosition += Count;\r
 110                                                 OutPosition += Count;\r
 111                                                 break;\r
 112                                         }\r
 113                                         break;\r
 114                                 default:\r
 115                                         bEscape = false;\r
 116                                         bEncoded = false;\r
 117                                         strncpy(&OutBuffer[OutPosition], &InBuffer[InPosition], Count);\r
 118                                         InPosition += Count;\r
 119                                         OutPosition += Count;\r
 120                                         break;\r
 121                                 }\r
 122                         }\r
 123                         else\r
 124                         {\r
 125                                 if(bString)\r
 126                                 {\r
 127                                         bEscape = false;\r
 128                                         bEncoded = true;\r
 129                                         UTF8Count = MultiByteToWideChar(CP_ACP, 0, &InBuffer[InPosition], Count, UTF16Buffer, sizeof(UTF16Buffer) / sizeof(wchar_t));\r
 130                                         UTF8Count = WideCharToMultiByte(CP_UTF8, 0, UTF16Buffer, UTF8Count, UTF8Buffer, sizeof(UTF8Buffer) / sizeof(char), NULL, NULL);\r
 131                                         InPosition += Count;\r
 132                                         for(i = 0; i < UTF8Count; i++)\r
 133                                         {\r
 134                                                 sprintf(&OutBuffer[OutPosition], "\\x%02X", (unsigned char)UTF8Buffer[i]);\r
 135                                                 OutPosition += 4;\r
 136                                         }\r
 137                                 }\r
 138                                 else\r
 139                                 {\r
 140                                         bEscape = false;\r
 141                                         bEncoded = false;\r
 142                                         strncpy(&OutBuffer[OutPosition], &InBuffer[InPosition], Count);\r
 143                                         InPosition += Count;\r
 144                                         OutPosition += Count;\r
 145                                 }\r
 146                         }\r
 147                 }\r
 148                 OutBuffer[OutPosition] = '\0';\r
 149                 fputs(OutBuffer, fpOut);\r
 150         }\r
 151         fclose(fpIn);\r
 152         fclose(fpOut);\r
 153         return 0;\r
 154 }\r
 155 \r