lib/base/estring.cpp

   1 #include <algorithm>
   2 #include <cctype>
   3 #include <climits>
   4 #include <string>
   5 #include <lib/base/eerror.h>
   6 #include <lib/base/encoding.h>
   7 #include <lib/base/estring.h>
   8
   9 std::string buildShortName( const std::string &str )
  10 {
  11         std::string tmp;
  12         static char stropen[3] = { 0xc2, 0x86, 0x00 };
  13         static char strclose[3] = { 0xc2, 0x87, 0x00 };
  14         size_t open=std::string::npos-1;
  15         while ( (open = str.find(stropen, open+2)) != std::string::npos )
  16         {
  17                 size_t close = str.find(strclose, open);
  18                 if ( close != std::string::npos )
  19                         tmp+=str.substr( open+2, close-(open+2) );
  20         }
  21         return tmp.length() ? tmp : str;
  22 }
  23
  24 std::string getNum(int val, int sys)
  25 {
  26 //      Returns a string that contain the value val as string
  27 //      if sys == 16 than hexadezimal if sys == 10 than decimal
  28         char buf[12];
  29
  30         if (sys == 10)
  31                 snprintf(buf, 12, "%i", val);
  32         else if (sys == 16)
  33                 snprintf(buf, 12, "%X", val);
  34
  35         std::string res;
  36         res.assign(buf);
  37         return res;
  38 }
  39
  40                 // 8859-x to ucs-16 coding tables. taken from www.unicode.org/Public/MAPPINGS/ISO8859/
  41
  42 static unsigned long c88592[96]={
  43 0x00A0, 0x0104, 0x02D8, 0x0141, 0x00A4, 0x013D, 0x015A, 0x00A7, 0x00A8, 0x0160, 0x015E, 0x0164, 0x0179, 0x00AD, 0x017D, 0x017B,
  44 0x00B0, 0x0105, 0x02DB, 0x0142, 0x00B4, 0x013E, 0x015B, 0x02C7, 0x00B8, 0x0161, 0x015F, 0x0165, 0x017A, 0x02DD, 0x017E, 0x017C,
  45 0x0154, 0x00C1, 0x00C2, 0x0102, 0x00C4, 0x0139, 0x0106, 0x00C7, 0x010C, 0x00C9, 0x0118, 0x00CB, 0x011A, 0x00CD, 0x00CE, 0x010E,
  46 0x0110, 0x0143, 0x0147, 0x00D3, 0x00D4, 0x0150, 0x00D6, 0x00D7, 0x0158, 0x016E, 0x00DA, 0x0170, 0x00DC, 0x00DD, 0x0162, 0x00DF,
  47 0x0155, 0x00E1, 0x00E2, 0x0103, 0x00E4, 0x013A, 0x0107, 0x00E7, 0x010D, 0x00E9, 0x0119, 0x00EB, 0x011B, 0x00ED, 0x00EE, 0x010F,
  48 0x0111, 0x0144, 0x0148, 0x00F3, 0x00F4, 0x0151, 0x00F6, 0x00F7, 0x0159, 0x016F, 0x00FA, 0x0171, 0x00FC, 0x00FD, 0x0163, 0x02D9};
  49
  50 static unsigned long c88593[96]={
  51 0x00A0, 0x0126, 0x02D8, 0x00A3, 0x00A4, 0x0000, 0x0124, 0x00A7, 0x00A8, 0x0130, 0x015E, 0x011E, 0x0134, 0x00AD, 0x0000, 0x017B,
  52 0x00B0, 0x0127, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x0125, 0x00B7, 0x00B8, 0x0131, 0x015F, 0x011F, 0x0135, 0x00BD, 0x0000, 0x017C,
  53 0x00C0, 0x00C1, 0x00C2, 0x0000, 0x00C4, 0x010A, 0x0108, 0x00C7, 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
  54 0x0000, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x0120, 0x00D6, 0x00D7, 0x011C, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x016C, 0x015C, 0x00DF,
  55 0x00E0, 0x00E1, 0x00E2, 0x0000, 0x00E4, 0x010B, 0x0109, 0x00E7, 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
  56 0x0000, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x0121, 0x00F6, 0x00F7, 0x011D, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x016D, 0x015D, 0x02D9};
  57
  58 static unsigned long c88594[96]={
  59 0x00A0, 0x0104, 0x0138, 0x0156, 0x00A4, 0x0128, 0x013B, 0x00A7, 0x00A8, 0x0160, 0x0112, 0x0122, 0x0166, 0x00AD, 0x017D, 0x00AF,
  60 0x00B0, 0x0105, 0x02DB, 0x0157, 0x00B4, 0x0129, 0x013C, 0x02C7, 0x00B8, 0x0161, 0x0113, 0x0123, 0x0167, 0x014A, 0x017E, 0x014B,
  61 0x0100, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x012E, 0x010C, 0x00C9, 0x0118, 0x00CB, 0x0116, 0x00CD, 0x00CE, 0x012A,
  62 0x0110, 0x0145, 0x014C, 0x0136, 0x00D4, 0x00D5, 0x00D6, 0x00D7, 0x00D8, 0x0172, 0x00DA, 0x00DB, 0x00DC, 0x0168, 0x016A, 0x00DF,
  63 0x0101, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x012F, 0x010D, 0x00E9, 0x0119, 0x00EB, 0x0117, 0x00ED, 0x00EE, 0x012B,
  64 0x0111, 0x0146, 0x014D, 0x0137, 0x00F4, 0x00F5, 0x00F6, 0x00F7, 0x00F8, 0x0173, 0x00FA, 0x00FB, 0x00FC, 0x0169, 0x016B, 0x02D9};
  65
  66 static unsigned long c88595[96]={
  67 0x00A0, 0x0401, 0x0402, 0x0403, 0x0404, 0x0405, 0x0406, 0x0407, 0x0408, 0x0409, 0x040A, 0x040B, 0x040C, 0x00AD, 0x040E, 0x040F,
  68 0x0410, 0x0411, 0x0412, 0x0413, 0x0414, 0x0415, 0x0416, 0x0417, 0x0418, 0x0419, 0x041A, 0x041B, 0x041C, 0x041D, 0x041E, 0x041F,
  69 0x0420, 0x0421, 0x0422, 0x0423, 0x0424, 0x0425, 0x0426, 0x0427, 0x0428, 0x0429, 0x042A, 0x042B, 0x042C, 0x042D, 0x042E, 0x042F,
  70 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437, 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F,
  71 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447, 0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,
  72 0x2116, 0x0451, 0x0452, 0x0453, 0x0454, 0x0455, 0x0456, 0x0457, 0x0458, 0x0459, 0x045A, 0x045B, 0x045C, 0x00A7, 0x045E, 0x045F};
  73
  74 static unsigned long c88596[96]={
  75 0x00A0, 0x0000, 0x0000, 0x0000, 0x00A4, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x060C, 0x00AD, 0x0000, 0x0000,
  76 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x061B, 0x0000, 0x0000, 0x0000, 0x061F,
  77 0x0000, 0x0621, 0x0622, 0x0623, 0x0624, 0x0625, 0x0626, 0x0627, 0x0628, 0x0629, 0x062A, 0x062B, 0x062C, 0x062D, 0x062E, 0x062F,
  78 0x0630, 0x0631, 0x0632, 0x0633, 0x0634, 0x0635, 0x0636, 0x0637, 0x0638, 0x0639, 0x063A, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
  79 0x0640, 0x0641, 0x0642, 0x0643, 0x0644, 0x0645, 0x0646, 0x0647, 0x0648, 0x0649, 0x064A, 0x064B, 0x064C, 0x064D, 0x064E, 0x064F,
  80 0x0650, 0x0651, 0x0652, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000};
  81
  82 static unsigned long c88597[96]={
  83 0x00A0, 0x2018, 0x2019, 0x00A3, 0x20AC, 0x20AF, 0x00A6, 0x00A7, 0x00A8, 0x00A9, 0x037A, 0x00AB, 0x00AC, 0x00AD, 0x0000, 0x2015,
  84 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x0384, 0x0385, 0x0386, 0x00B7, 0x0388, 0x0389, 0x038A, 0x00BB, 0x038C, 0x00BD, 0x038E, 0x038F,
  85 0x0390, 0x0391, 0x0392, 0x0393, 0x0394, 0x0395, 0x0396, 0x0397, 0x0398, 0x0399, 0x039A, 0x039B, 0x039C, 0x039D, 0x039E, 0x039F,
  86 0x03A0, 0x03A1, 0x0000, 0x03A3, 0x03A4, 0x03A5, 0x03A6, 0x03A7, 0x03A8, 0x03A9, 0x03AA, 0x03AB, 0x03AC, 0x03AD, 0x03AE, 0x03AF,
  87 0x03B0, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7, 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF,
  88 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7, 0x03C8, 0x03C9, 0x03CA, 0x03CB, 0x03CC, 0x03CD, 0x03CE, 0x0000};
  89
  90 static unsigned long c88598[96]={
  91 0x00A0, 0x0000, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7, 0x00A8, 0x00A9, 0x00D7, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
  92 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7, 0x00B8, 0x00B9, 0x00F7, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x0000,
  93 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
  94 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x2017,
  95 0x05D0, 0x05D1, 0x05D2, 0x05D3, 0x05D4, 0x05D5, 0x05D6, 0x05D7, 0x05D8, 0x05D9, 0x05DA, 0x05DB, 0x05DC, 0x05DD, 0x05DE, 0x05DF,
  96 0x05E0, 0x05E1, 0x05E2, 0x05E3, 0x05E4, 0x05E5, 0x05E6, 0x05E7, 0x05E8, 0x05E9, 0x05EA, 0x0000, 0x0000, 0x200E, 0x200F, 0x0000};
  97
  98 static unsigned long c88599[96]={
  99 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7, 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
 100 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7, 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF,
 101 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
 102 0x011E, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7, 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x0130, 0x015E, 0x00DF,
 103 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
 104 0x011F, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7, 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x0131, 0x015F, 0x00FF};
 105
 106 static unsigned long c885910[96]={
 107 0x00A0, 0x0104, 0x0112, 0x0122, 0x012A, 0x0128, 0x0136, 0x00A7, 0x013B, 0x0110, 0x0160, 0x0166, 0x017D, 0x00AD, 0x016A, 0x014A,
 108 0x00B0, 0x0105, 0x0113, 0x0123, 0x012B, 0x0129, 0x0137, 0x00B7, 0x013C, 0x0111, 0x0161, 0x0167, 0x017E, 0x2015, 0x016B, 0x014B,
 109 0x0100, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x012E, 0x010C, 0x00C9, 0x0118, 0x00CB, 0x0116, 0x00CD, 0x00CE, 0x00CF,
 110 0x00D0, 0x0145, 0x014C, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x0168, 0x00D8, 0x0172, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF,
 111 0x0101, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x012F, 0x010D, 0x00E9, 0x0119, 0x00EB, 0x0117, 0x00ED, 0x00EE, 0x00EF,
 112 0x00F0, 0x0146, 0x014D, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x0169, 0x00F8, 0x0173, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x0138};
 113
 114 static unsigned long c885911[96]={
 115 0x00A0, 0x0E01, 0x0E02, 0x0E03, 0x0E04, 0x0E05, 0x0E06, 0x0E07, 0x0E08, 0x0E09, 0x0E0A, 0x0E0B, 0x0E0C, 0x0E0D, 0x0E0E, 0x0E0F,
 116 0x0E10, 0x0E11, 0x0E12, 0x0E13, 0x0E14, 0x0E15, 0x0E16, 0x0E17, 0x0E18, 0x0E19, 0x0E1A, 0x0E1B, 0x0E1C, 0x0E1D, 0x0E1E, 0x0E1F,
 117 0x0E20, 0x0E21, 0x0E22, 0x0E23, 0x0E24, 0x0E25, 0x0E26, 0x0E27, 0x0E28, 0x0E29, 0x0E2A, 0x0E2B, 0x0E2C, 0x0E2D, 0x0E2E, 0x0E2F,
 118 0x0E30, 0x0E31, 0x0E32, 0x0E33, 0x0E34, 0x0E35, 0x0E36, 0x0E37, 0x0E38, 0x0E39, 0x0E3A, 0x0000, 0x0000, 0x0000, 0x0000, 0x0E3F,
 119 0x0E40, 0x0E41, 0x0E42, 0x0E43, 0x0E44, 0x0E45, 0x0E46, 0x0E47, 0x0E48, 0x0E49, 0x0E4A, 0x0E4B, 0x0E4C, 0x0E4D, 0x0E4E, 0x0E4F,
 120 0x0E50, 0x0E51, 0x0E52, 0x0E53, 0x0E54, 0x0E55, 0x0E56, 0x0E57, 0x0E58, 0x0E59, 0x0E5A, 0x0E5B, 0x0000, 0x0000, 0x0000, 0x0000};
 121
 122 static unsigned long c885913[96]={
 123 0x00A0, 0x201D, 0x00A2, 0x00A3, 0x00A4, 0x201E, 0x00A6, 0x00A7, 0x00D8, 0x00A9, 0x0156, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00C6,
 124 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x201C, 0x00B5, 0x00B6, 0x00B7, 0x00F8, 0x00B9, 0x0157, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00E6,
 125 0x0104, 0x012E, 0x0100, 0x0106, 0x00C4, 0x00C5, 0x0118, 0x0112, 0x010C, 0x00C9, 0x0179, 0x0116, 0x0122, 0x0136, 0x012A, 0x013B,
 126 0x0160, 0x0143, 0x0145, 0x00D3, 0x014C, 0x00D5, 0x00D6, 0x00D7, 0x0172, 0x0141, 0x015A, 0x016A, 0x00DC, 0x017B, 0x017D, 0x00DF,
 127 0x0105, 0x012F, 0x0101, 0x0107, 0x00E4, 0x00E5, 0x0119, 0x0113, 0x010D, 0x00E9, 0x017A, 0x0117, 0x0123, 0x0137, 0x012B, 0x013C,
 128 0x0161, 0x0144, 0x0146, 0x00F3, 0x014D, 0x00F5, 0x00F6, 0x00F7, 0x0173, 0x0142, 0x015B, 0x016B, 0x00FC, 0x017C, 0x017E, 0x2019};
 129
 130 static unsigned long c885914[96]={
 131 0x00A0, 0x1E02, 0x1E03, 0x00A3, 0x010A, 0x010B, 0x1E0A, 0x00A7, 0x1E80, 0x00A9, 0x1E82, 0x1E0B, 0x1EF2, 0x00AD, 0x00AE, 0x0178,
 132 0x1E1E, 0x1E1F, 0x0120, 0x0121, 0x1E40, 0x1E41, 0x00B6, 0x1E56, 0x1E81, 0x1E57, 0x1E83, 0x1E60, 0x1EF3, 0x1E84, 0x1E85, 0x1E61,
 133 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
 134 0x0174, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x1E6A, 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x0176, 0x00DF,
 135 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
 136 0x0175, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x1E6B, 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x0177, 0x00FF};
 137
 138 static unsigned long c885915[96]={
 139 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x20AC, 0x00A5, 0x0160, 0x00A7, 0x0161, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
 140 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x017D, 0x00B5, 0x00B6, 0x00B7, 0x017E, 0x00B9, 0x00BA, 0x00BB, 0x0152, 0x0153, 0x0178, 0x00BF,
 141 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
 142 0x00D0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7, 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF,
 143 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
 144 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7, 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF};
 145
 146 static unsigned long c885916[96]={
 147 0x00A0, 0x0104, 0x0105, 0x0141, 0x20AC, 0x201E, 0x0160, 0x00A7, 0x0161, 0x00A9, 0x0218, 0x00AB, 0x0179, 0x00AD, 0x017A, 0x017B,
 148 0x00B0, 0x00B1, 0x010C, 0x0142, 0x017D, 0x201D, 0x00B6, 0x00B7, 0x017E, 0x010D, 0x0219, 0x00BB, 0x0152, 0x0153, 0x0178, 0x017C,
 149 0x00C0, 0x00C1, 0x00C2, 0x0102, 0x00C4, 0x0106, 0x00C6, 0x00C7, 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF,
 150 0x0110, 0x0143, 0x00D2, 0x00D3, 0x00D4, 0x0150, 0x00D6, 0x015A, 0x0170, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x0118, 0x021A, 0x00DF,
 151 0x00E0, 0x00E1, 0x00E2, 0x0103, 0x00E4, 0x0107, 0x00E6, 0x00E7, 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF,
 152 0x0111, 0x0144, 0x00F2, 0x00F3, 0x00F4, 0x0151, 0x00F6, 0x015B, 0x0171, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x0119, 0x021B, 0x00FF};
 153
 154 static unsigned long iso6397[96]={
 155 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x20AC, 0x00A5, 0x0000, 0x00A7, 0x00A4, 0x2018, 0x201C, 0x00AB, 0x2190, 0x2191, 0x2192, 0x2193,
 156 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00D7, 0x00B5, 0x00B6, 0x00B7, 0x00F7, 0x2019, 0x201D, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF,
 157 0x0000, 0xE002, 0xE003, 0xE004, 0xE005, 0xE006, 0xE007, 0xE008, 0xE009, 0xE00C, 0xE00A, 0xE00B, 0x0000, 0xE00D, 0xE00E, 0xE00F,
 158 0x2015, 0x00B9, 0x00AE, 0x00A9, 0x2122, 0x266A, 0x00AC, 0x00A6, 0x0000, 0x0000, 0x0000, 0x0000, 0x215B, 0x215C, 0x215D, 0x215E,
 159 0x2126, 0x00C6, 0x0110, 0x00AA, 0x0126, 0x0000, 0x0132, 0x013F, 0x0141, 0x00D8, 0x0152, 0x00BA, 0x00DE, 0x0166, 0x014A, 0x0149,
 160 0x0138, 0x00E6, 0x0111, 0x00F0, 0x0127, 0x0131, 0x0133, 0x0140, 0x0142, 0x00F8, 0x0153, 0x00DF, 0x00FE, 0x0167, 0x014B, 0x00AD};
 161
 162 // Two Char Mapping (aka ISO6397) ( many polish services and UPC Direct/HBO services)
 163 // get from http://mitglied.lycos.de/buran/charsets/videotex-suppl.html
 164 static inline unsigned int doVideoTexSuppl(int c1, int c2)
 165 {
 166         switch (c1)
 167         {
 168                 case 0xC1: // grave
 169                         switch (c2)
 170                         {
 171                                 case 0x61: return 224;                          case 0x41: return 192;
 172                                 case 0x65: return 232;                          case 0x45: return 200;
 173                                 case 0x69: return 236;                          case 0x49: return 204;
 174                                 case 0x6f: return 242;                          case 0x4f: return 210;
 175                                 case 0x75: return 249;                          case 0x55: return 217;
 176                                 default: return 0;
 177                         }
 178                 case 0xC2: // acute
 179                         switch (c2)
 180                         {
 181                                 case 0x20: return 180;
 182                                 case 0x61: return 225;                          case 0x41: return 193;
 183                                 case 0x65: return 233;                          case 0x45: return 201;
 184                                 case 0x69: return 237;                          case 0x49: return 205;
 185                                 case 0x6f: return 243;                          case 0x4f: return 211;
 186                                 case 0x75: return 250;                          case 0x55: return 218;
 187                                 case 0x79: return 253;                          case 0x59: return 221;
 188                                 case 0x63: return 263;                          case 0x43: return 262;
 189                                 case 0x6c: return 314;                          case 0x4c: return 313;
 190                                 case 0x6e: return 324;                          case 0x4e: return 323;
 191                                 case 0x72: return 341;                          case 0x52: return 340;
 192                                 case 0x73: return 347;                          case 0x53: return 346;
 193                                 case 0x7a: return 378;                          case 0x5a: return 377;
 194                                 default: return 0;
 195                         }
 196                 case 0xC3: // cedilla
 197                         switch (c2)
 198                         {
 199                                 case 0x61: return 226;                          case 0x41: return 194;
 200                                 case 0x65: return 234;                          case 0x45: return 202;
 201                                 case 0x69: return 238;                          case 0x49: return 206;
 202                                 case 0x6f: return 244;                          case 0x4f: return 212;
 203                                 case 0x75: return 251;                          case 0x55: return 219;
 204                                 case 0x79: return 375;                          case 0x59: return 374;
 205                                 case 0x63: return 265;                          case 0x43: return 264;
 206                                 case 0x67: return 285;                          case 0x47: return 284;
 207                                 case 0x68: return 293;                          case 0x48: return 292;
 208                                 case 0x6a: return 309;                          case 0x4a: return 308;
 209                                 case 0x73: return 349;                          case 0x53: return 348;
 210                                 case 0x77: return 373;                          case 0x57: return 372;
 211                                 default: return 0;
 212                         }
 213                 case 0xC4: // tilde
 214                         switch (c2)
 215                         {
 216                                 case 0x61: return 227;                          case 0x41: return 195;
 217                                 case 0x6e: return 241;                          case 0x4e: return 209;
 218                                 case 0x69: return 297;                          case 0x49: return 296;
 219                                 case 0x6f: return 245;                          case 0x4f: return 213;
 220                                 case 0x75: return 361;                          case 0x55: return 360;
 221                                 default: return 0;
 222                         }
 223                 case 0xC5: // macron
 224                         switch (c2)
 225                         {
 226                                 case 0x20: return 175;
 227                                 case 0x41: return 256;                          case 0x61: return 257;
 228                                 case 0x45: return 274;                          case 0x65: return 275;
 229                                 case 0x49: return 298;                          case 0x69: return 299;
 230                                 case 0x4f: return 332;                          case 0x6f: return 333;
 231                         }
 232                 case 0xC6: // breve
 233                         switch (c2)
 234                         {
 235                                 case 0x20: return 728;
 236                                 case 0x61: return 259;                          case 0x41: return 258;
 237                                 case 0x67: return 287;                          case 0x47: return 286;
 238                                 case 0x75: return 365;                          case 0x55: return 364;
 239                                 default: return 0;
 240                         }
 241                 case 0xC7: // dot above
 242                         switch (c2)
 243                         {
 244                                 case 0x20: return 729;
 245                                 case 0x63: return 267;                          case 0x43: return 266;
 246                                 case 0x65: return 279;                          case 0x45: return 278;
 247                                 case 0x67: return 289;                          case 0x47: return 288;
 248                                 case 0x5a: return 379;                          case 0x49: return 304;
 249                                 case 0x7a: return 380;
 250                                 default: return 0;
 251                         }
 252                 case 0xC8: // diaeresis
 253                         switch (c2)
 254                         {
 255                                 case 0x20: return 168;
 256                                 case 0x61: return 228;                          case 0x41: return 196;
 257                                 case 0x65: return 235;                          case 0x45: return 203;
 258                                 case 0x69: return 239;                          case 0x49: return 207;
 259                                 case 0x6f: return 246;                          case 0x4f: return 214;
 260                                 case 0x75: return 252;                          case 0x55: return 220;
 261                                 case 0x79: return 255;                          case 0x59: return 376;
 262                                 default: return 0;
 263                                 }
 264                 case 0xCA: // ring above
 265                         switch (c2)
 266                         {
 267                                 case 0x20: return 730;
 268                                 case 0x61: return 229;                          case 0x41: return 197;
 269                                 case 0x75: return 367;                          case 0x55: return 366;
 270                                 default: return 0;
 271                         }
 272                 case 0xCB: // cedilla
 273                         switch (c2)
 274                         {
 275                                 case 0x63: return 231;                          case 0x43: return 199;
 276                                 case 0x67: return 291;                          case 0x47: return 290;
 277                                 case 0x6b: return 311;                          case 0x4b: return 310;
 278                                 case 0x6c: return 316;                          case 0x4c: return 315;
 279                                 case 0x6e: return 326;                          case 0x4e: return 325;
 280                                 case 0x72: return 343;                          case 0x52: return 342;
 281                                 case 0x73: return 351;                          case 0x53: return 350;
 282                                 case 0x74: return 355;                          case 0x54: return 354;
 283                                 default: return 0;
 284                         }
 285                 case 0xCD: // double acute accent
 286                         switch (c2)
 287                         {
 288                                 case 0x20: return 733;
 289                                 case 0x6f: return 337;                          case 0x4f: return 336;
 290                                 case 0x75: return 369;                          case 0x55: return 368;
 291                                 default: return 0;
 292                         }
 293                 case 0xCE: // ogonek
 294                         switch (c2)
 295                         {
 296                                 case 0x20: return 731;
 297                                 case 0x61: return 261;                          case 0x41: return 260;
 298                                 case 0x65: return 281;                          case 0x45: return 280;
 299                                 case 0x69: return 303;                          case 0x49: return 302;
 300                                 case 0x75: return 371;                          case 0x55: return 370;
 301                                 default: return 0;
 302                         }
 303                 case 0xCF: // caron
 304                         switch (c2)
 305                         {
 306                                 case 0x20: return 711;
 307                                 case 0x63: return 269;                          case 0x43: return 268;
 308                                 case 0x64: return 271;                          case 0x44: return 270;
 309                                 case 0x65: return 283;                          case 0x45: return 282;
 310                                 case 0x6c: return 318;                          case 0x4c: return 317;
 311                                 case 0x6e: return 328;                          case 0x4e: return 327;
 312                                 case 0x72: return 345;                          case 0x52: return 344;
 313                                 case 0x73: return 353;                          case 0x53: return 352;
 314                                 case 0x74: return 357;                          case 0x54: return 356;
 315                                 case 0x7a: return 382;                          case 0x5a: return 381;
 316                                 default: return 0;
 317                         }
 318         }
 319         return 0;
 320 }
 321
 322 static inline unsigned int recode(unsigned char d, int cp)
 323 {
 324         if (d < 0xA0)
 325                 return d;
 326         switch (cp)
 327         {
 328         case 0:         // ISO6397
 329                 return iso6397[d-0xA0];
 330         case 1:         // 8859-1 <-> unicode mapping
 331                 return d;
 332         case 2:         // 8859-2 -> unicode mapping
 333                 return c88592[d-0xA0];
 334         case 3:         // 8859-3 -> unicode mapping
 335                 return c88593[d-0xA0];
 336         case 4:         // 8859-2 -> unicode mapping
 337                 return c88594[d-0xA0];
 338         case 5:         // 8859-5 -> unicode mapping
 339                 return c88595[d-0xA0];
 340         case 6:         // 8859-6 -> unicode mapping
 341                 return c88596[d-0xA0];
 342         case 7:         // 8859-7 -> unicode mapping
 343                 return c88597[d-0xA0];
 344         case 8:         // 8859-8 -> unicode mapping
 345                 return c88598[d-0xA0];
 346         case 9:         // 8859-9 -> unicode mapping
 347                 return c88599[d-0xA0];
 348         case 10:// 8859-10 -> unicode mapping
 349                 return c885910[d-0xA0];
 350         case 11:// 8859-11 -> unicode mapping
 351                 return c885911[d-0xA0];
 352 /*      case 12:// 8859-12 -> unicode mapping  // reserved for indian use..
 353                 return c885912[d-0xA0];*/
 354         case 13:// 8859-13 -> unicode mapping
 355                 return c885913[d-0xA0];
 356         case 14:// 8859-14 -> unicode mapping
 357                 return c885914[d-0xA0];
 358         case 15:// 8859-15 -> unicode mapping
 359                 return c885915[d-0xA0];
 360         case 16:// 8859-16 -> unicode mapping
 361                 return c885916[d-0xA0];
 362         default:
 363                 return d;
 364         }
 365 }
 366
 367 std::string convertDVBUTF8(const unsigned char *data, int len, int table, int tsidonid)
 368 {
 369         if (!len)
 370                 return "";
 371
 372         int i=0, t=0;
 373
 374         if ( tsidonid )
 375                 encodingHandler.getTransponderDefaultMapping(tsidonid, table);
 376
 377         switch(data[0])
 378         {
 379                 case 1 ... 11:
 380                         table=data[i++]+4;
 381 //                      eDebug("(1..11)text encoded in ISO-8859-%d",table);
 382                         break;
 383                 case 0x10:
 384                 {
 385                         int n=(data[++i]<<8);
 386                         n |= (data[++i]);
 387 //                      eDebug("(0x10)text encoded in ISO-8859-%d",n);
 388                         ++i;
 389                         switch(n)
 390                         {
 391                                 case 12:
 392                                         eDebug("unsup. ISO8859-12 enc.");
 393                                         break;
 394                                 default:
 395                                         table=n;
 396                                         break;
 397                         }
 398                         break;
 399                 }
 400                 case 0x11: //  Basic Multilingual Plane of ISO/IEC 10646-1 enc  (UTF-16... Unicode)
 401                         table = 65;
 402                         tsidonid = 0;
 403                         ++i;
 404                         break;
 405                 case 0x12:
 406                         ++i;
 407                         eDebug("unsup. KSC 5601 enc.");
 408                         break;
 409                 case 0x13:
 410                         ++i;
 411                         eDebug("unsup. GB-2312-1980 enc.");
 412                         break;
 413                 case 0x14:
 414                         ++i;
 415                         eDebug("unsup. Big5 subset of ISO/IEC 10646-1 enc.");
 416                         break;
 417                 case 0x15: // UTF-8 encoding of ISO/IEC 10646-1
 418                         return std::string((char*)data+1, len-1);
 419                 case 0x0:
 420                 case 0xC ... 0xF:
 421                 case 0x16 ... 0x1F:
 422                         eDebug("reserved %d", data[0]);
 423                         ++i;
 424                         break;
 425         }
 426
 427         bool useTwoCharMapping = !table || tsidonid && encodingHandler.getTransponderUseTwoCharMapping(tsidonid);
 428
 429         if (useTwoCharMapping && table == 5) { // i hope this dont break other transponders which realy use ISO8859-5 and two char byte mapping...
 430 //              eDebug("Cyfra / Cyfrowy Polsat HACK... override given ISO8859-5 with ISO6397");
 431                 table = 0;
 432         }
 433
 434         unsigned char res[2048];
 435         while (i < len)
 436         {
 437                 unsigned long code=0;
 438                 if ( useTwoCharMapping && i+1 < len && (code=doVideoTexSuppl(data[i], data[i+1])) )
 439                         i+=2;
 440                 if (!code) {
 441                         if (table == 65) { // unicode
 442                                 if (i+1 < len) {
 443                                         code=(data[i] << 8) | data[i+1];
 444                                         i += 2;
 445                                 }
 446                         }
 447                         else
 448                                 code=recode(data[i++], table);
 449                 }
 450                 if (!code)
 451                         continue;
 452                                 // Unicode->UTF8 encoding
 453                 if (code < 0x80) // identity ascii <-> utf8 mapping
 454                         res[t++]=char(code);
 455                 else if (code < 0x800) // two byte mapping
 456                 {
 457                         res[t++]=(code>>6)|0xC0;
 458                         res[t++]=(code&0x3F)|0x80;
 459                 } else if (code < 0x10000) // three bytes mapping
 460                 {
 461                         res[t++]=(code>>12)|0xE0;
 462                         res[t++]=((code>>6)&0x3F)|0x80;
 463                         res[t++]=(code&0x3F)|0x80;
 464                 } else
 465                 {
 466                         res[t++]=(code>>18)|0xF0;
 467                         res[t++]=((code>>12)&0x3F)|0x80;
 468                         res[t++]=((code>>6)&0x3F)|0x80;
 469                         res[t++]=(code&0x3F)|0x80;
 470                 }
 471                 if (t+4 > 2047)
 472                 {
 473                         eDebug("convertDVBUTF8 buffer to small.. break now");
 474                         break;
 475                 }
 476         }
 477         return std::string((char*)res, t);
 478 }
 479
 480 std::string convertUTF8DVB(const std::string &string, int table)
 481 {
 482         unsigned long *coding_table=0;
 483
 484         int len=string.length(), t=0;
 485
 486         unsigned char buf[len];
 487
 488         for(int i=0;i<len;i++)
 489         {
 490                 unsigned char c1=string[i];
 491                 unsigned int c;
 492                 if(c1<0x80)
 493                         c=c1;
 494                 else
 495                 {
 496                         i++;
 497                         unsigned char c2=string[i];
 498                         c=((c1&0x3F)<<6) + (c2&0x3F);
 499                         if (table==0||table==1||c1<0xA0)
 500                                 ;
 501                         else
 502                         {
 503                                 if (!coding_table)
 504                                 {
 505                                         switch(table)
 506                                         {
 507                                                 case 2:
 508                                                         coding_table = c88592;
 509                                                         break;
 510                                                 case 3:
 511                                                         coding_table = c88593;
 512                                                         break;
 513                                                 case 4:
 514                                                         coding_table = c88594;
 515                                                         break;
 516                                                 case 5:
 517                                                         coding_table = c88595;
 518                                                         break;
 519                                                 case 6:
 520                                                         coding_table = c88596;
 521                                                         break;
 522                                                 case 7:
 523                                                         coding_table = c88597;
 524                                                         break;
 525                                                 case 8:
 526                                                         coding_table = c88598;
 527                                                         break;
 528                                                 case 9:
 529                                                         coding_table = c88599;
 530                                                         break;
 531                                                 case 10:
 532                                                         coding_table = c885910;
 533                                                         break;
 534                                                 case 11:
 535                                                         coding_table = c885911;
 536                                                         break;
 537 /*                                              case 12:   // reserved.. for indian use
 538                                                         coding_table = c885912;
 539                                                 break;*/
 540                                                 case 13:
 541                                                         coding_table = c885913;
 542                                                         break;
 543                                                 case 14:
 544                                                         coding_table = c885914;
 545                                                         break;
 546                                                 case 15:
 547                                                         coding_table = c885915;
 548                                                         break;
 549                                                 case 16:
 550                                                         coding_table = c885916;
 551                                                         break;
 552                                                 default:
 553                                                         eFatal("unknown coding table %d", table);
 554                                                         break;
 555                                         }
 556                                 }
 557                                 for(unsigned int j=0;j<96;j++)
 558                                 {
 559                                         if(coding_table[j]==c)
 560                                         {
 561                                                 c=0xA0+j;
 562                                                 break;
 563                                         }
 564                                 }
 565                         }
 566                 }
 567                 buf[t++]=(unsigned char)c;
 568         }
 569         return std::string((char*)buf,t);
 570 }
 571
 572 std::string convertLatin1UTF8(const std::string &string)
 573 {
 574         unsigned int t=0, i=0, len=string.size();
 575
 576         unsigned char res[2048];
 577
 578         while (i < len)
 579         {
 580                 unsigned long code=(unsigned char)string[i++];
 581                                 // Unicode->UTF8 encoding
 582                 if (code < 0x80) // identity latin <-> utf8 mapping
 583                         res[t++]=char(code);
 584                 else if (code < 0x800) // two byte mapping
 585                 {
 586                         res[t++]=(code>>6)|0xC0;
 587                         res[t++]=(code&0x3F)|0x80;
 588                 } else if (code < 0x10000) // three bytes mapping
 589                 {
 590                         res[t++]=(code>>12)|0xE0;
 591                         res[t++]=((code>>6)&0x3F)|0x80;
 592                         res[t++]=(code&0x3F)|0x80;
 593                 } else
 594                 {
 595                         res[t++]=(code>>18)|0xF0;
 596                         res[t++]=((code>>12)&0x3F)|0x80;
 597                         res[t++]=((code>>6)&0x3F)|0x80;
 598                         res[t++]=(code&0x3F)|0x80;
 599                 }
 600                 if (t+4 > 2047)
 601                 {
 602                         eDebug("convertLatin1UTF8 buffer to small.. break now");
 603                         break;
 604                 }
 605         }
 606         return std::string((char*)res, t);
 607 }
 608
 609 int isUTF8(const std::string &string)
 610 {
 611         unsigned int len=string.size();
 612
 613         for (unsigned int i=0; i < len; ++i)
 614         {
 615                 if (!(string[i]&0x80)) // normal ASCII
 616                         continue;
 617                 if ((string[i] & 0xE0) == 0xC0) // one char following.
 618                 {
 619                                 // first, length check:
 620                         if (i+1 >= len)
 621                                 return 0; // certainly NOT utf-8
 622                         i++;
 623                         if ((string[i]&0xC0) != 0x80)
 624                                 return 0; // no, not UTF-8.
 625                 } else if ((string[i] & 0xF0) == 0xE0)
 626                 {
 627                         if ((i+1) >= len)
 628                                 return 0;
 629                         i++;
 630                         if ((string[i]&0xC0) != 0x80)
 631                                 return 0;
 632                         i++;
 633                         if ((string[i]&0xC0) != 0x80)
 634                                 return 0;
 635                 }
 636         }
 637         return 1; // can be UTF8 (or pure ASCII, at least no non-UTF-8 8bit characters)
 638 }
 639
 640 std::string removeDVBChars(const std::string &s)
 641 {
 642         std::string res;
 643
 644         int len = s.length();
 645
 646         for(int i = 0; i < len; i++)
 647         {
 648                 unsigned char c1 = s[i];
 649                 unsigned int c;
 650
 651                         /* UTF8? decode (but only simple) */
 652                 if((c1 > 0x80) && (i < len-1))
 653                 {
 654                         unsigned char c2 = s[i + 1];
 655                         c = ((c1&0x3F)<<6) + (c2&0x3F);
 656                         if ((c >= 0x80) && (c <= 0x9F))
 657                         {
 658                                 ++i; /* skip 2nd utf8 char */
 659                                 continue;
 660                         }
 661                 }
 662
 663                 res += s[i];
 664         }
 665
 666         return res;
 667 }
 668
 669 void makeUpper(std::string &s)
 670 {
 671         std::transform(s.begin(), s.end(), s.begin(), (int(*)(int)) toupper);
 672 }
 673
 674 std::string replace_all(const std::string &in, const std::string &entity, const std::string &symbol)
 675 {
 676         std::string out = in;
 677         std::string::size_type loc = 0;
 678         while (( loc = out.find(entity, loc)) != std::string::npos )
 679         out.replace(loc, entity.length(), symbol);
 680         return out;
 681 }