src/subtitles/text_reader.h

   1 /************************************************************************/
   2 /* author: xy                                                           */
   3 /* date: 20110511                                                       */
   4 /************************************************************************/
   5 #ifndef __TEXT_READER_H_B2D9DE8F_8730_436C_9F0A_219EE6D78352__
   6 #define __TEXT_READER_H_B2D9DE8F_8730_436C_9F0A_219EE6D78352__
   7
   8 #include <afx.h>
   9
  10 namespace xy_utils
  11 {
  12
  13 template<typename ReaderType>
  14 class TextReader
  15 {
  16 public:
  17     typedef enum {ASCII, UTF8, LE16, BE16, ANSI} Encoding;
  18     TextReader(ReaderType &reader, Encoding encoding): _inner_reader(reader),_encoding(encoding)
  19     {}
  20
  21     ~TextReader() {}
  22
  23     bool ReadLine( CStringW * str )
  24     {
  25         if(str==NULL) return false;
  26
  27         bool fEOF = true;
  28
  29         str->Empty();
  30
  31         if(_encoding == ASCII || _encoding == ANSI)
  32         {
  33             CStringA stra;
  34             char c;
  35             while( _inner_reader.Read(&c, sizeof(c)) == sizeof(c) )
  36             {
  37                 fEOF = false;
  38                 if(c == '\r') continue;
  39                 if(c == '\n') break;
  40                 stra += c;
  41             }
  42             *str = CStringW(CString(stra)); // TODO: codepage
  43         }
  44         else if(_encoding == UTF8)
  45         {
  46             BYTE b;
  47             while(_inner_reader.Read(&b, sizeof(b)) == sizeof(b))
  48             {
  49                 fEOF = false;
  50                 WCHAR c = '?';
  51                 if(!(b&0x80)) // 0xxxxxxx
  52                 {
  53                     c = b&0x7f;
  54                 }
  55                 else if((b&0xe0) == 0xc0) // 110xxxxx 10xxxxxx
  56                 {
  57                     c = (b&0x1f)<<6;
  58                     if(_inner_reader.Read(&b, sizeof(b)) != sizeof(b)) break;
  59                     c |= (b&0x3f);
  60                 }
  61                 else if((b&0xf0) == 0xe0) // 1110xxxx 10xxxxxx 10xxxxxx
  62                 {
  63                     c = (b&0x0f)<<12;
  64                     if(_inner_reader.Read(&b, sizeof(b)) != sizeof(b)) break;
  65                     c |= (b&0x3f)<<6;
  66                     if(_inner_reader.Read(&b, sizeof(b)) != sizeof(b)) break;
  67                     c |= (b&0x3f);
  68                 }
  69                 if(c == '\r') continue;
  70                 if(c == '\n') break;
  71                 *str += c;
  72             }
  73         }
  74         else if(_encoding == LE16)
  75         {
  76             WCHAR wc;
  77             while(_inner_reader.Read(&wc, sizeof(wc)) == sizeof(wc))
  78             {
  79                 fEOF = false;
  80                 if(wc == '\r') continue;
  81                 if(wc == '\n') break;
  82                 *str += wc;
  83             }
  84         }
  85         else if(_encoding == BE16)
  86         {
  87             WCHAR wc;
  88             while(_inner_reader.Read(&wc, sizeof(wc)) == sizeof(wc))
  89             {
  90                 fEOF = false;
  91                 wc = ((wc>>8)&0x00ff)|((wc<<8)&0xff00);
  92                 if(wc == '\r') continue;
  93                 if(wc == '\n') break;
  94                 *str += wc;
  95             }
  96         }
  97
  98         return(!fEOF);
  99     }
 100     int GetEncoding() const { return _encoding; }
 101
 102 private:
 103     ReaderType& _inner_reader;
 104     Encoding _encoding;
 105 };
 106
 107 } //namespace xy_utils
 108
 109 #endif // end of __TEXT_READER_H_B2D9DE8F_8730_436C_9F0A_219EE6D78352__