| 1 | #include <stdio.h> |
| 2 | #include <stdlib.h> |
| 3 | #include <ctype.h> |
| 4 | #include <time.h> |
| 5 | #include <assert.h> |
| 6 | |
| 7 | #include "putty.h" |
| 8 | #include "terminal.h" |
| 9 | #include "misc.h" |
| 10 | |
| 11 | /* Character conversion arrays; they are usually taken from windows, |
| 12 | * the xterm one has the four scanlines that have no unicode 2.0 |
| 13 | * equivalents mapped to their unicode 3.0 locations. |
| 14 | */ |
| 15 | static const WCHAR unitab_xterm_std[32] = { |
| 16 | 0x2666, 0x2592, 0x2409, 0x240c, 0x240d, 0x240a, 0x00b0, 0x00b1, |
| 17 | 0x2424, 0x240b, 0x2518, 0x2510, 0x250c, 0x2514, 0x253c, 0x23ba, |
| 18 | 0x23bb, 0x2500, 0x23bc, 0x23bd, 0x251c, 0x2524, 0x2534, 0x252c, |
| 19 | 0x2502, 0x2264, 0x2265, 0x03c0, 0x2260, 0x00a3, 0x00b7, 0x0020 |
| 20 | }; |
| 21 | |
| 22 | /* |
| 23 | * If the codepage is non-zero it's a window codepage, zero means use a |
| 24 | * local codepage. The name is always converted to the first of any |
| 25 | * duplicate definitions. |
| 26 | */ |
| 27 | |
| 28 | /* |
| 29 | * Tables for ISO-8859-{1-10,13-16} derived from those downloaded |
| 30 | * 2001-10-02 from <http://www.unicode.org/Public/MAPPINGS/> -- jtn |
| 31 | * Table for ISO-8859-11 derived from same on 2002-11-18. -- bjh21 |
| 32 | */ |
| 33 | |
| 34 | /* XXX: This could be done algorithmically, but I'm not sure it's |
| 35 | * worth the hassle -- jtn */ |
| 36 | /* ISO/IEC 8859-1:1998 (Latin-1, "Western", "West European") */ |
| 37 | static const wchar_t iso_8859_1[] = { |
| 38 | 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7, |
| 39 | 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF, |
| 40 | 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7, |
| 41 | 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF, |
| 42 | 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, |
| 43 | 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, |
| 44 | 0x00D0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7, |
| 45 | 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF, |
| 46 | 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, |
| 47 | 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, |
| 48 | 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7, |
| 49 | 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF |
| 50 | }; |
| 51 | |
| 52 | /* ISO/IEC 8859-2:1999 (Latin-2, "Central European", "East European") */ |
| 53 | static const wchar_t iso_8859_2[] = { |
| 54 | 0x00A0, 0x0104, 0x02D8, 0x0141, 0x00A4, 0x013D, 0x015A, 0x00A7, |
| 55 | 0x00A8, 0x0160, 0x015E, 0x0164, 0x0179, 0x00AD, 0x017D, 0x017B, |
| 56 | 0x00B0, 0x0105, 0x02DB, 0x0142, 0x00B4, 0x013E, 0x015B, 0x02C7, |
| 57 | 0x00B8, 0x0161, 0x015F, 0x0165, 0x017A, 0x02DD, 0x017E, 0x017C, |
| 58 | 0x0154, 0x00C1, 0x00C2, 0x0102, 0x00C4, 0x0139, 0x0106, 0x00C7, |
| 59 | 0x010C, 0x00C9, 0x0118, 0x00CB, 0x011A, 0x00CD, 0x00CE, 0x010E, |
| 60 | 0x0110, 0x0143, 0x0147, 0x00D3, 0x00D4, 0x0150, 0x00D6, 0x00D7, |
| 61 | 0x0158, 0x016E, 0x00DA, 0x0170, 0x00DC, 0x00DD, 0x0162, 0x00DF, |
| 62 | 0x0155, 0x00E1, 0x00E2, 0x0103, 0x00E4, 0x013A, 0x0107, 0x00E7, |
| 63 | 0x010D, 0x00E9, 0x0119, 0x00EB, 0x011B, 0x00ED, 0x00EE, 0x010F, |
| 64 | 0x0111, 0x0144, 0x0148, 0x00F3, 0x00F4, 0x0151, 0x00F6, 0x00F7, |
| 65 | 0x0159, 0x016F, 0x00FA, 0x0171, 0x00FC, 0x00FD, 0x0163, 0x02D9 |
| 66 | }; |
| 67 | |
| 68 | /* ISO/IEC 8859-3:1999 (Latin-3, "South European", "Maltese & Esperanto") */ |
| 69 | static const wchar_t iso_8859_3[] = { |
| 70 | 0x00A0, 0x0126, 0x02D8, 0x00A3, 0x00A4, 0xFFFD, 0x0124, 0x00A7, |
| 71 | 0x00A8, 0x0130, 0x015E, 0x011E, 0x0134, 0x00AD, 0xFFFD, 0x017B, |
| 72 | 0x00B0, 0x0127, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x0125, 0x00B7, |
| 73 | 0x00B8, 0x0131, 0x015F, 0x011F, 0x0135, 0x00BD, 0xFFFD, 0x017C, |
| 74 | 0x00C0, 0x00C1, 0x00C2, 0xFFFD, 0x00C4, 0x010A, 0x0108, 0x00C7, |
| 75 | 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, |
| 76 | 0xFFFD, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x0120, 0x00D6, 0x00D7, |
| 77 | 0x011C, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x016C, 0x015C, 0x00DF, |
| 78 | 0x00E0, 0x00E1, 0x00E2, 0xFFFD, 0x00E4, 0x010B, 0x0109, 0x00E7, |
| 79 | 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, |
| 80 | 0xFFFD, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x0121, 0x00F6, 0x00F7, |
| 81 | 0x011D, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x016D, 0x015D, 0x02D9 |
| 82 | }; |
| 83 | |
| 84 | /* ISO/IEC 8859-4:1998 (Latin-4, "North European") */ |
| 85 | static const wchar_t iso_8859_4[] = { |
| 86 | 0x00A0, 0x0104, 0x0138, 0x0156, 0x00A4, 0x0128, 0x013B, 0x00A7, |
| 87 | 0x00A8, 0x0160, 0x0112, 0x0122, 0x0166, 0x00AD, 0x017D, 0x00AF, |
| 88 | 0x00B0, 0x0105, 0x02DB, 0x0157, 0x00B4, 0x0129, 0x013C, 0x02C7, |
| 89 | 0x00B8, 0x0161, 0x0113, 0x0123, 0x0167, 0x014A, 0x017E, 0x014B, |
| 90 | 0x0100, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x012E, |
| 91 | 0x010C, 0x00C9, 0x0118, 0x00CB, 0x0116, 0x00CD, 0x00CE, 0x012A, |
| 92 | 0x0110, 0x0145, 0x014C, 0x0136, 0x00D4, 0x00D5, 0x00D6, 0x00D7, |
| 93 | 0x00D8, 0x0172, 0x00DA, 0x00DB, 0x00DC, 0x0168, 0x016A, 0x00DF, |
| 94 | 0x0101, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x012F, |
| 95 | 0x010D, 0x00E9, 0x0119, 0x00EB, 0x0117, 0x00ED, 0x00EE, 0x012B, |
| 96 | 0x0111, 0x0146, 0x014D, 0x0137, 0x00F4, 0x00F5, 0x00F6, 0x00F7, |
| 97 | 0x00F8, 0x0173, 0x00FA, 0x00FB, 0x00FC, 0x0169, 0x016B, 0x02D9 |
| 98 | }; |
| 99 | |
| 100 | /* ISO/IEC 8859-5:1999 (Latin/Cyrillic) */ |
| 101 | static const wchar_t iso_8859_5[] = { |
| 102 | 0x00A0, 0x0401, 0x0402, 0x0403, 0x0404, 0x0405, 0x0406, 0x0407, |
| 103 | 0x0408, 0x0409, 0x040A, 0x040B, 0x040C, 0x00AD, 0x040E, 0x040F, |
| 104 | 0x0410, 0x0411, 0x0412, 0x0413, 0x0414, 0x0415, 0x0416, 0x0417, |
| 105 | 0x0418, 0x0419, 0x041A, 0x041B, 0x041C, 0x041D, 0x041E, 0x041F, |
| 106 | 0x0420, 0x0421, 0x0422, 0x0423, 0x0424, 0x0425, 0x0426, 0x0427, |
| 107 | 0x0428, 0x0429, 0x042A, 0x042B, 0x042C, 0x042D, 0x042E, 0x042F, |
| 108 | 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437, |
| 109 | 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F, |
| 110 | 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447, |
| 111 | 0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F, |
| 112 | 0x2116, 0x0451, 0x0452, 0x0453, 0x0454, 0x0455, 0x0456, 0x0457, |
| 113 | 0x0458, 0x0459, 0x045A, 0x045B, 0x045C, 0x00A7, 0x045E, 0x045F |
| 114 | }; |
| 115 | |
| 116 | /* ISO/IEC 8859-6:1999 (Latin/Arabic) */ |
| 117 | static const wchar_t iso_8859_6[] = { |
| 118 | 0x00A0, 0xFFFD, 0xFFFD, 0xFFFD, 0x00A4, 0xFFFD, 0xFFFD, 0xFFFD, |
| 119 | 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x060C, 0x00AD, 0xFFFD, 0xFFFD, |
| 120 | 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 121 | 0xFFFD, 0xFFFD, 0xFFFD, 0x061B, 0xFFFD, 0xFFFD, 0xFFFD, 0x061F, |
| 122 | 0xFFFD, 0x0621, 0x0622, 0x0623, 0x0624, 0x0625, 0x0626, 0x0627, |
| 123 | 0x0628, 0x0629, 0x062A, 0x062B, 0x062C, 0x062D, 0x062E, 0x062F, |
| 124 | 0x0630, 0x0631, 0x0632, 0x0633, 0x0634, 0x0635, 0x0636, 0x0637, |
| 125 | 0x0638, 0x0639, 0x063A, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 126 | 0x0640, 0x0641, 0x0642, 0x0643, 0x0644, 0x0645, 0x0646, 0x0647, |
| 127 | 0x0648, 0x0649, 0x064A, 0x064B, 0x064C, 0x064D, 0x064E, 0x064F, |
| 128 | 0x0650, 0x0651, 0x0652, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 129 | 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD |
| 130 | }; |
| 131 | |
| 132 | /* ISO 8859-7:1987 (Latin/Greek) */ |
| 133 | static const wchar_t iso_8859_7[] = { |
| 134 | 0x00A0, 0x2018, 0x2019, 0x00A3, 0xFFFD, 0xFFFD, 0x00A6, 0x00A7, |
| 135 | 0x00A8, 0x00A9, 0xFFFD, 0x00AB, 0x00AC, 0x00AD, 0xFFFD, 0x2015, |
| 136 | 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x0384, 0x0385, 0x0386, 0x00B7, |
| 137 | 0x0388, 0x0389, 0x038A, 0x00BB, 0x038C, 0x00BD, 0x038E, 0x038F, |
| 138 | 0x0390, 0x0391, 0x0392, 0x0393, 0x0394, 0x0395, 0x0396, 0x0397, |
| 139 | 0x0398, 0x0399, 0x039A, 0x039B, 0x039C, 0x039D, 0x039E, 0x039F, |
| 140 | 0x03A0, 0x03A1, 0xFFFD, 0x03A3, 0x03A4, 0x03A5, 0x03A6, 0x03A7, |
| 141 | 0x03A8, 0x03A9, 0x03AA, 0x03AB, 0x03AC, 0x03AD, 0x03AE, 0x03AF, |
| 142 | 0x03B0, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7, |
| 143 | 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF, |
| 144 | 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7, |
| 145 | 0x03C8, 0x03C9, 0x03CA, 0x03CB, 0x03CC, 0x03CD, 0x03CE, 0xFFFD |
| 146 | }; |
| 147 | |
| 148 | /* ISO/IEC 8859-8:1999 (Latin/Hebrew) */ |
| 149 | static const wchar_t iso_8859_8[] = { |
| 150 | 0x00A0, 0xFFFD, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7, |
| 151 | 0x00A8, 0x00A9, 0x00D7, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF, |
| 152 | 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7, |
| 153 | 0x00B8, 0x00B9, 0x00F7, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0xFFFD, |
| 154 | 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 155 | 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 156 | 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 157 | 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x2017, |
| 158 | 0x05D0, 0x05D1, 0x05D2, 0x05D3, 0x05D4, 0x05D5, 0x05D6, 0x05D7, |
| 159 | 0x05D8, 0x05D9, 0x05DA, 0x05DB, 0x05DC, 0x05DD, 0x05DE, 0x05DF, |
| 160 | 0x05E0, 0x05E1, 0x05E2, 0x05E3, 0x05E4, 0x05E5, 0x05E6, 0x05E7, |
| 161 | 0x05E8, 0x05E9, 0x05EA, 0xFFFD, 0xFFFD, 0x200E, 0x200F, 0xFFFD |
| 162 | }; |
| 163 | |
| 164 | /* ISO/IEC 8859-9:1999 (Latin-5, "Turkish") */ |
| 165 | static const wchar_t iso_8859_9[] = { |
| 166 | 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7, |
| 167 | 0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF, |
| 168 | 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7, |
| 169 | 0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF, |
| 170 | 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, |
| 171 | 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, |
| 172 | 0x011E, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7, |
| 173 | 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x0130, 0x015E, 0x00DF, |
| 174 | 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, |
| 175 | 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, |
| 176 | 0x011F, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7, |
| 177 | 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x0131, 0x015F, 0x00FF |
| 178 | }; |
| 179 | |
| 180 | /* ISO/IEC 8859-10:1998 (Latin-6, "Nordic" [Sami, Inuit, Icelandic]) */ |
| 181 | static const wchar_t iso_8859_10[] = { |
| 182 | 0x00A0, 0x0104, 0x0112, 0x0122, 0x012A, 0x0128, 0x0136, 0x00A7, |
| 183 | 0x013B, 0x0110, 0x0160, 0x0166, 0x017D, 0x00AD, 0x016A, 0x014A, |
| 184 | 0x00B0, 0x0105, 0x0113, 0x0123, 0x012B, 0x0129, 0x0137, 0x00B7, |
| 185 | 0x013C, 0x0111, 0x0161, 0x0167, 0x017E, 0x2015, 0x016B, 0x014B, |
| 186 | 0x0100, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x012E, |
| 187 | 0x010C, 0x00C9, 0x0118, 0x00CB, 0x0116, 0x00CD, 0x00CE, 0x00CF, |
| 188 | 0x00D0, 0x0145, 0x014C, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x0168, |
| 189 | 0x00D8, 0x0172, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF, |
| 190 | 0x0101, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x012F, |
| 191 | 0x010D, 0x00E9, 0x0119, 0x00EB, 0x0117, 0x00ED, 0x00EE, 0x00EF, |
| 192 | 0x00F0, 0x0146, 0x014D, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x0169, |
| 193 | 0x00F8, 0x0173, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x0138 |
| 194 | }; |
| 195 | |
| 196 | /* ISO/IEC 8859-11:2001 ("Thai", "TIS620") */ |
| 197 | static const wchar_t iso_8859_11[] = { |
| 198 | 0x00A0, 0x0E01, 0x0E02, 0x0E03, 0x0E04, 0x0E05, 0x0E06, 0x0E07, |
| 199 | 0x0E08, 0x0E09, 0x0E0A, 0x0E0B, 0x0E0C, 0x0E0D, 0x0E0E, 0x0E0F, |
| 200 | 0x0E10, 0x0E11, 0x0E12, 0x0E13, 0x0E14, 0x0E15, 0x0E16, 0x0E17, |
| 201 | 0x0E18, 0x0E19, 0x0E1A, 0x0E1B, 0x0E1C, 0x0E1D, 0x0E1E, 0x0E1F, |
| 202 | 0x0E20, 0x0E21, 0x0E22, 0x0E23, 0x0E24, 0x0E25, 0x0E26, 0x0E27, |
| 203 | 0x0E28, 0x0E29, 0x0E2A, 0x0E2B, 0x0E2C, 0x0E2D, 0x0E2E, 0x0E2F, |
| 204 | 0x0E30, 0x0E31, 0x0E32, 0x0E33, 0x0E34, 0x0E35, 0x0E36, 0x0E37, |
| 205 | 0x0E38, 0x0E39, 0x0E3A, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x0E3F, |
| 206 | 0x0E40, 0x0E41, 0x0E42, 0x0E43, 0x0E44, 0x0E45, 0x0E46, 0x0E47, |
| 207 | 0x0E48, 0x0E49, 0x0E4A, 0x0E4B, 0x0E4C, 0x0E4D, 0x0E4E, 0x0E4F, |
| 208 | 0x0E50, 0x0E51, 0x0E52, 0x0E53, 0x0E54, 0x0E55, 0x0E56, 0x0E57, |
| 209 | 0x0E58, 0x0E59, 0x0E5A, 0x0E5B, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD |
| 210 | }; |
| 211 | |
| 212 | /* ISO/IEC 8859-13:1998 (Latin-7, "Baltic Rim") */ |
| 213 | static const wchar_t iso_8859_13[] = { |
| 214 | 0x00A0, 0x201D, 0x00A2, 0x00A3, 0x00A4, 0x201E, 0x00A6, 0x00A7, |
| 215 | 0x00D8, 0x00A9, 0x0156, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00C6, |
| 216 | 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x201C, 0x00B5, 0x00B6, 0x00B7, |
| 217 | 0x00F8, 0x00B9, 0x0157, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00E6, |
| 218 | 0x0104, 0x012E, 0x0100, 0x0106, 0x00C4, 0x00C5, 0x0118, 0x0112, |
| 219 | 0x010C, 0x00C9, 0x0179, 0x0116, 0x0122, 0x0136, 0x012A, 0x013B, |
| 220 | 0x0160, 0x0143, 0x0145, 0x00D3, 0x014C, 0x00D5, 0x00D6, 0x00D7, |
| 221 | 0x0172, 0x0141, 0x015A, 0x016A, 0x00DC, 0x017B, 0x017D, 0x00DF, |
| 222 | 0x0105, 0x012F, 0x0101, 0x0107, 0x00E4, 0x00E5, 0x0119, 0x0113, |
| 223 | 0x010D, 0x00E9, 0x017A, 0x0117, 0x0123, 0x0137, 0x012B, 0x013C, |
| 224 | 0x0161, 0x0144, 0x0146, 0x00F3, 0x014D, 0x00F5, 0x00F6, 0x00F7, |
| 225 | 0x0173, 0x0142, 0x015B, 0x016B, 0x00FC, 0x017C, 0x017E, 0x2019 |
| 226 | }; |
| 227 | |
| 228 | /* ISO/IEC 8859-14:1998 (Latin-8, "Celtic", "Gaelic/Welsh") */ |
| 229 | static const wchar_t iso_8859_14[] = { |
| 230 | 0x00A0, 0x1E02, 0x1E03, 0x00A3, 0x010A, 0x010B, 0x1E0A, 0x00A7, |
| 231 | 0x1E80, 0x00A9, 0x1E82, 0x1E0B, 0x1EF2, 0x00AD, 0x00AE, 0x0178, |
| 232 | 0x1E1E, 0x1E1F, 0x0120, 0x0121, 0x1E40, 0x1E41, 0x00B6, 0x1E56, |
| 233 | 0x1E81, 0x1E57, 0x1E83, 0x1E60, 0x1EF3, 0x1E84, 0x1E85, 0x1E61, |
| 234 | 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, |
| 235 | 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, |
| 236 | 0x0174, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x1E6A, |
| 237 | 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x0176, 0x00DF, |
| 238 | 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, |
| 239 | 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, |
| 240 | 0x0175, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x1E6B, |
| 241 | 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x0177, 0x00FF |
| 242 | }; |
| 243 | |
| 244 | /* ISO/IEC 8859-15:1999 (Latin-9 aka -0, "euro") */ |
| 245 | static const wchar_t iso_8859_15[] = { |
| 246 | 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x20AC, 0x00A5, 0x0160, 0x00A7, |
| 247 | 0x0161, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF, |
| 248 | 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x017D, 0x00B5, 0x00B6, 0x00B7, |
| 249 | 0x017E, 0x00B9, 0x00BA, 0x00BB, 0x0152, 0x0153, 0x0178, 0x00BF, |
| 250 | 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, |
| 251 | 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, |
| 252 | 0x00D0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x00D7, |
| 253 | 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE, 0x00DF, |
| 254 | 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, |
| 255 | 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, |
| 256 | 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x00F7, |
| 257 | 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE, 0x00FF |
| 258 | }; |
| 259 | |
| 260 | /* ISO/IEC 8859-16:2001 (Latin-10, "Balkan") */ |
| 261 | static const wchar_t iso_8859_16[] = { |
| 262 | 0x00A0, 0x0104, 0x0105, 0x0141, 0x20AC, 0x201E, 0x0160, 0x00A7, |
| 263 | 0x0161, 0x00A9, 0x0218, 0x00AB, 0x0179, 0x00AD, 0x017A, 0x017B, |
| 264 | 0x00B0, 0x00B1, 0x010C, 0x0142, 0x017D, 0x201D, 0x00B6, 0x00B7, |
| 265 | 0x017E, 0x010D, 0x0219, 0x00BB, 0x0152, 0x0153, 0x0178, 0x017C, |
| 266 | 0x00C0, 0x00C1, 0x00C2, 0x0102, 0x00C4, 0x0106, 0x00C6, 0x00C7, |
| 267 | 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, |
| 268 | 0x0110, 0x0143, 0x00D2, 0x00D3, 0x00D4, 0x0150, 0x00D6, 0x015A, |
| 269 | 0x0170, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x0118, 0x021A, 0x00DF, |
| 270 | 0x00E0, 0x00E1, 0x00E2, 0x0103, 0x00E4, 0x0107, 0x00E6, 0x00E7, |
| 271 | 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, |
| 272 | 0x0111, 0x0144, 0x00F2, 0x00F3, 0x00F4, 0x0151, 0x00F6, 0x015B, |
| 273 | 0x0171, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x0119, 0x021B, 0x00FF |
| 274 | }; |
| 275 | |
| 276 | static const wchar_t roman8[] = { |
| 277 | 0x00A0, 0x00C0, 0x00C2, 0x00C8, 0x00CA, 0x00CB, 0x00CE, 0x00CF, |
| 278 | 0x00B4, 0x02CB, 0x02C6, 0x00A8, 0x02DC, 0x00D9, 0x00DB, 0x20A4, |
| 279 | 0x00AF, 0x00DD, 0x00FD, 0x00B0, 0x00C7, 0x00E7, 0x00D1, 0x00F1, |
| 280 | 0x00A1, 0x00BF, 0x00A4, 0x00A3, 0x00A5, 0x00A7, 0x0192, 0x00A2, |
| 281 | 0x00E2, 0x00EA, 0x00F4, 0x00FB, 0x00E1, 0x00E9, 0x00F3, 0x00FA, |
| 282 | 0x00E0, 0x00E8, 0x00F2, 0x00F9, 0x00E4, 0x00EB, 0x00F6, 0x00FC, |
| 283 | 0x00C5, 0x00EE, 0x00D8, 0x00C6, 0x00E5, 0x00ED, 0x00F8, 0x00E6, |
| 284 | 0x00C4, 0x00EC, 0x00D6, 0x00DC, 0x00C9, 0x00EF, 0x00DF, 0x00D4, |
| 285 | 0x00C1, 0x00C3, 0x00E3, 0x00D0, 0x00F0, 0x00CD, 0x00CC, 0x00D3, |
| 286 | 0x00D2, 0x00D5, 0x00F5, 0x0160, 0x0161, 0x00DA, 0x0178, 0x00FF, |
| 287 | 0x00DE, 0x00FE, 0x00B7, 0x00B5, 0x00B6, 0x00BE, 0x2014, 0x00BC, |
| 288 | 0x00BD, 0x00AA, 0x00BA, 0x00AB, 0x25A0, 0x00BB, 0x00B1, 0xFFFD |
| 289 | }; |
| 290 | |
| 291 | static const wchar_t koi8_u[] = { |
| 292 | 0x2500, 0x2502, 0x250C, 0x2510, 0x2514, 0x2518, 0x251C, 0x2524, |
| 293 | 0x252C, 0x2534, 0x253C, 0x2580, 0x2584, 0x2588, 0x258C, 0x2590, |
| 294 | 0x2591, 0x2592, 0x2593, 0x2320, 0x25A0, 0x2022, 0x221A, 0x2248, |
| 295 | 0x2264, 0x2265, 0x00A0, 0x2321, 0x00B0, 0x00B2, 0x00B7, 0x00F7, |
| 296 | 0x2550, 0x2551, 0x2552, 0x0451, 0x0454, 0x2554, 0x0456, 0x0457, |
| 297 | 0x2557, 0x2558, 0x2559, 0x255A, 0x255B, 0x0491, 0x255D, 0x255E, |
| 298 | 0x255F, 0x2560, 0x2561, 0x0401, 0x0404, 0x2563, 0x0406, 0x0407, |
| 299 | 0x2566, 0x2567, 0x2568, 0x2569, 0x256A, 0x0490, 0x256C, 0x00A9, |
| 300 | 0x044E, 0x0430, 0x0431, 0x0446, 0x0434, 0x0435, 0x0444, 0x0433, |
| 301 | 0x0445, 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, |
| 302 | 0x043F, 0x044F, 0x0440, 0x0441, 0x0442, 0x0443, 0x0436, 0x0432, |
| 303 | 0x044C, 0x044B, 0x0437, 0x0448, 0x044D, 0x0449, 0x0447, 0x044A, |
| 304 | 0x042E, 0x0410, 0x0411, 0x0426, 0x0414, 0x0415, 0x0424, 0x0413, |
| 305 | 0x0425, 0x0418, 0x0419, 0x041A, 0x041B, 0x041C, 0x041D, 0x041E, |
| 306 | 0x041F, 0x042F, 0x0420, 0x0421, 0x0422, 0x0423, 0x0416, 0x0412, |
| 307 | 0x042C, 0x042B, 0x0417, 0x0428, 0x042D, 0x0429, 0x0427, 0x042A |
| 308 | }; |
| 309 | |
| 310 | static const wchar_t vscii[] = { |
| 311 | 0x0000, 0x0001, 0x1EB2, 0x0003, 0x0004, 0x1EB4, 0x1EAA, 0x0007, |
| 312 | 0x0008, 0x0009, 0x000a, 0x000b, 0x000c, 0x000d, 0x000e, 0x000f, |
| 313 | 0x0010, 0x0011, 0x0012, 0x0013, 0x1EF6, 0x0015, 0x0016, 0x0017, |
| 314 | 0x0018, 0x1EF8, 0x001a, 0x001b, 0x001c, 0x001d, 0x1EF4, 0x001f, |
| 315 | 0x0020, 0x0021, 0x0022, 0x0023, 0x0024, 0x0025, 0x0026, 0x0027, |
| 316 | 0x0028, 0x0029, 0x002A, 0x002B, 0x002C, 0x002D, 0x002E, 0x002F, |
| 317 | 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037, |
| 318 | 0x0038, 0x0039, 0x003A, 0x003B, 0x003C, 0x003D, 0x003E, 0x003F, |
| 319 | 0x0040, 0x0041, 0x0042, 0x0043, 0x0044, 0x0045, 0x0046, 0x0047, |
| 320 | 0x0048, 0x0049, 0x004A, 0x004B, 0x004C, 0x004D, 0x004E, 0x004F, |
| 321 | 0x0050, 0x0051, 0x0052, 0x0053, 0x0054, 0x0055, 0x0056, 0x0057, |
| 322 | 0x0058, 0x0059, 0x005A, 0x005B, 0x005C, 0x005D, 0x005E, 0x005F, |
| 323 | 0x0060, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067, |
| 324 | 0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F, |
| 325 | 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077, |
| 326 | 0x0078, 0x0079, 0x007A, 0x007B, 0x007C, 0x007D, 0x007E, 0x007f, |
| 327 | 0x1EA0, 0x1EAE, 0x1EB0, 0x1EB6, 0x1EA4, 0x1EA6, 0x1EA8, 0x1EAC, |
| 328 | 0x1EBC, 0x1EB8, 0x1EBE, 0x1EC0, 0x1EC2, 0x1EC4, 0x1EC6, 0x1ED0, |
| 329 | 0x1ED2, 0x1ED4, 0x1ED6, 0x1ED8, 0x1EE2, 0x1EDA, 0x1EDC, 0x1EDE, |
| 330 | 0x1ECA, 0x1ECE, 0x1ECC, 0x1EC8, 0x1EE6, 0x0168, 0x1EE4, 0x1EF2, |
| 331 | 0x00D5, 0x1EAF, 0x1EB1, 0x1EB7, 0x1EA5, 0x1EA7, 0x1EA8, 0x1EAD, |
| 332 | 0x1EBD, 0x1EB9, 0x1EBF, 0x1EC1, 0x1EC3, 0x1EC5, 0x1EC7, 0x1ED1, |
| 333 | 0x1ED3, 0x1ED5, 0x1ED7, 0x1EE0, 0x01A0, 0x1ED9, 0x1EDD, 0x1EDF, |
| 334 | 0x1ECB, 0x1EF0, 0x1EE8, 0x1EEA, 0x1EEC, 0x01A1, 0x1EDB, 0x01AF, |
| 335 | 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x1EA2, 0x0102, 0x1EB3, 0x1EB5, |
| 336 | 0x00C8, 0x00C9, 0x00CA, 0x1EBA, 0x00CC, 0x00CD, 0x0128, 0x1EF3, |
| 337 | 0x0110, 0x1EE9, 0x00D2, 0x00D3, 0x00D4, 0x1EA1, 0x1EF7, 0x1EEB, |
| 338 | 0x1EED, 0x00D9, 0x00DA, 0x1EF9, 0x1EF5, 0x00DD, 0x1EE1, 0x01B0, |
| 339 | 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x1EA3, 0x0103, 0x1EEF, 0x1EAB, |
| 340 | 0x00E8, 0x00E9, 0x00EA, 0x1EBB, 0x00EC, 0x00ED, 0x0129, 0x1EC9, |
| 341 | 0x0111, 0x1EF1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x1ECF, 0x1ECD, |
| 342 | 0x1EE5, 0x00F9, 0x00FA, 0x0169, 0x1EE7, 0x00FD, 0x1EE3, 0x1EEE |
| 343 | }; |
| 344 | |
| 345 | static const wchar_t dec_mcs[] = { |
| 346 | 0x00A0, 0x00A1, 0x00A2, 0x00A3, 0xFFFD, 0x00A5, 0xFFFD, 0x00A7, |
| 347 | 0x00A4, 0x00A9, 0x00AA, 0x00AB, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 348 | 0x00B0, 0x00B1, 0x00B2, 0x00B3, 0xFFFD, 0x00B5, 0x00B6, 0x00B7, |
| 349 | 0xFFFD, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0xFFFD, 0x00BF, |
| 350 | 0x00C0, 0x00C1, 0x00C2, 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, |
| 351 | 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, 0x00CD, 0x00CE, 0x00CF, |
| 352 | 0xFFFD, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, 0x0152, |
| 353 | 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x0178, 0xFFFD, 0x00DF, |
| 354 | 0x00E0, 0x00E1, 0x00E2, 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, |
| 355 | 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, 0x00ED, 0x00EE, 0x00EF, |
| 356 | 0xFFFD, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, 0x0153, |
| 357 | 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FF, 0xFFFD, 0xFFFD |
| 358 | }; |
| 359 | |
| 360 | struct cp_list_item { |
| 361 | char *name; |
| 362 | int codepage; |
| 363 | int cp_size; |
| 364 | const wchar_t *cp_table; |
| 365 | }; |
| 366 | |
| 367 | static const struct cp_list_item cp_list[] = { |
| 368 | {"ISO-8859-1:1998 (Latin-1, West Europe)", 0, 96, iso_8859_1}, |
| 369 | {"ISO-8859-2:1999 (Latin-2, East Europe)", 0, 96, iso_8859_2}, |
| 370 | {"ISO-8859-3:1999 (Latin-3, South Europe)", 0, 96, iso_8859_3}, |
| 371 | {"ISO-8859-4:1998 (Latin-4, North Europe)", 0, 96, iso_8859_4}, |
| 372 | {"ISO-8859-5:1999 (Latin/Cyrillic)", 0, 96, iso_8859_5}, |
| 373 | {"ISO-8859-6:1999 (Latin/Arabic)", 0, 96, iso_8859_6}, |
| 374 | {"ISO-8859-7:1987 (Latin/Greek)", 0, 96, iso_8859_7}, |
| 375 | {"ISO-8859-8:1999 (Latin/Hebrew)", 0, 96, iso_8859_8}, |
| 376 | {"ISO-8859-9:1999 (Latin-5, Turkish)", 0, 96, iso_8859_9}, |
| 377 | {"ISO-8859-10:1998 (Latin-6, Nordic)", 0, 96, iso_8859_10}, |
| 378 | {"ISO-8859-11:2001 (Latin/Thai)", 0, 96, iso_8859_11}, |
| 379 | {"ISO-8859-13:1998 (Latin-7, Baltic)", 0, 96, iso_8859_13}, |
| 380 | {"ISO-8859-14:1998 (Latin-8, Celtic)", 0, 96, iso_8859_14}, |
| 381 | {"ISO-8859-15:1999 (Latin-9, \"euro\")", 0, 96, iso_8859_15}, |
| 382 | {"ISO-8859-16:2001 (Latin-10, Balkan)", 0, 96, iso_8859_16}, |
| 383 | |
| 384 | {"UTF-8", CP_UTF8}, |
| 385 | |
| 386 | {"KOI8-U", 0, 128, koi8_u}, |
| 387 | {"KOI8-R", 20866}, |
| 388 | {"HP-ROMAN8", 0, 96, roman8}, |
| 389 | {"VSCII", 0, 256, vscii}, |
| 390 | {"DEC-MCS", 0, 96, dec_mcs}, |
| 391 | |
| 392 | {"Win1250 (Central European)", 1250}, |
| 393 | {"Win1251 (Cyrillic)", 1251}, |
| 394 | {"Win1252 (Western)", 1252}, |
| 395 | {"Win1253 (Greek)", 1253}, |
| 396 | {"Win1254 (Turkish)", 1254}, |
| 397 | {"Win1255 (Hebrew)", 1255}, |
| 398 | {"Win1256 (Arabic)", 1256}, |
| 399 | {"Win1257 (Baltic)", 1257}, |
| 400 | {"Win1258 (Vietnamese)", 1258}, |
| 401 | |
| 402 | {"CP437", 437}, |
| 403 | {"CP819", 28591}, |
| 404 | {"CP878", 20866}, |
| 405 | |
| 406 | {"Use font encoding", -1}, |
| 407 | |
| 408 | {0, 0} |
| 409 | }; |
| 410 | |
| 411 | static void link_font(WCHAR * line_tbl, WCHAR * font_tbl, WCHAR attr); |
| 412 | |
| 413 | void init_ucs(Config *cfg, struct unicode_data *ucsdata) |
| 414 | { |
| 415 | int i, j; |
| 416 | int used_dtf = 0; |
| 417 | char tbuf[256]; |
| 418 | |
| 419 | for (i = 0; i < 256; i++) |
| 420 | tbuf[i] = i; |
| 421 | |
| 422 | /* Decide on the Line and Font codepages */ |
| 423 | ucsdata->line_codepage = decode_codepage(cfg->line_codepage); |
| 424 | |
| 425 | if (ucsdata->font_codepage <= 0) { |
| 426 | ucsdata->font_codepage=0; |
| 427 | ucsdata->dbcs_screenfont=0; |
| 428 | } |
| 429 | |
| 430 | if (cfg->vtmode == VT_OEMONLY) { |
| 431 | ucsdata->font_codepage = 437; |
| 432 | ucsdata->dbcs_screenfont = 0; |
| 433 | if (ucsdata->line_codepage <= 0) |
| 434 | ucsdata->line_codepage = GetACP(); |
| 435 | } else if (ucsdata->line_codepage <= 0) |
| 436 | ucsdata->line_codepage = ucsdata->font_codepage; |
| 437 | |
| 438 | /* Collect screen font ucs table */ |
| 439 | if (ucsdata->dbcs_screenfont || ucsdata->font_codepage == 0) { |
| 440 | get_unitab(ucsdata->font_codepage, ucsdata->unitab_font, 2); |
| 441 | for (i = 128; i < 256; i++) |
| 442 | ucsdata->unitab_font[i] = (WCHAR) (ATTR_ACP + i); |
| 443 | } else { |
| 444 | get_unitab(ucsdata->font_codepage, ucsdata->unitab_font, 1); |
| 445 | |
| 446 | /* CP437 fonts are often broken ... */ |
| 447 | if (ucsdata->font_codepage == 437) |
| 448 | ucsdata->unitab_font[0] = ucsdata->unitab_font[255] = 0xFFFF; |
| 449 | } |
| 450 | if (cfg->vtmode == VT_XWINDOWS) |
| 451 | memcpy(ucsdata->unitab_font + 1, unitab_xterm_std, |
| 452 | sizeof(unitab_xterm_std)); |
| 453 | |
| 454 | /* Collect OEMCP ucs table */ |
| 455 | get_unitab(CP_OEMCP, ucsdata->unitab_oemcp, 1); |
| 456 | |
| 457 | /* Collect CP437 ucs table for SCO acs */ |
| 458 | if (cfg->vtmode == VT_OEMANSI || cfg->vtmode == VT_XWINDOWS) |
| 459 | memcpy(ucsdata->unitab_scoacs, ucsdata->unitab_oemcp, |
| 460 | sizeof(ucsdata->unitab_scoacs)); |
| 461 | else |
| 462 | get_unitab(437, ucsdata->unitab_scoacs, 1); |
| 463 | |
| 464 | /* Collect line set ucs table */ |
| 465 | if (ucsdata->line_codepage == ucsdata->font_codepage && |
| 466 | (ucsdata->dbcs_screenfont || |
| 467 | cfg->vtmode == VT_POORMAN || ucsdata->font_codepage==0)) { |
| 468 | |
| 469 | /* For DBCS and POOR fonts force direct to font */ |
| 470 | used_dtf = 1; |
| 471 | for (i = 0; i < 32; i++) |
| 472 | ucsdata->unitab_line[i] = (WCHAR) i; |
| 473 | for (i = 32; i < 256; i++) |
| 474 | ucsdata->unitab_line[i] = (WCHAR) (ATTR_ACP + i); |
| 475 | ucsdata->unitab_line[127] = (WCHAR) 127; |
| 476 | } else { |
| 477 | get_unitab(ucsdata->line_codepage, ucsdata->unitab_line, 0); |
| 478 | } |
| 479 | |
| 480 | #if 0 |
| 481 | debug( |
| 482 | ("Line cp%d, Font cp%d%s\n", ucsdata->line_codepage, |
| 483 | ucsdata->font_codepage, ucsdata->dbcs_screenfont ? " DBCS" : "")); |
| 484 | |
| 485 | for (i = 0; i < 256; i += 16) { |
| 486 | for (j = 0; j < 16; j++) { |
| 487 | debug(("%04x%s", ucsdata->unitab_line[i + j], j == 15 ? "" : ",")); |
| 488 | } |
| 489 | debug(("\n")); |
| 490 | } |
| 491 | #endif |
| 492 | |
| 493 | /* VT100 graphics - NB: Broken for non-ascii CP's */ |
| 494 | memcpy(ucsdata->unitab_xterm, ucsdata->unitab_line, |
| 495 | sizeof(ucsdata->unitab_xterm)); |
| 496 | memcpy(ucsdata->unitab_xterm + '`', unitab_xterm_std, |
| 497 | sizeof(unitab_xterm_std)); |
| 498 | ucsdata->unitab_xterm['_'] = ' '; |
| 499 | |
| 500 | /* Generate UCS ->line page table. */ |
| 501 | if (ucsdata->uni_tbl) { |
| 502 | for (i = 0; i < 256; i++) |
| 503 | if (ucsdata->uni_tbl[i]) |
| 504 | sfree(ucsdata->uni_tbl[i]); |
| 505 | sfree(ucsdata->uni_tbl); |
| 506 | ucsdata->uni_tbl = 0; |
| 507 | } |
| 508 | if (!used_dtf) { |
| 509 | for (i = 0; i < 256; i++) { |
| 510 | if (DIRECT_CHAR(ucsdata->unitab_line[i])) |
| 511 | continue; |
| 512 | if (DIRECT_FONT(ucsdata->unitab_line[i])) |
| 513 | continue; |
| 514 | if (!ucsdata->uni_tbl) { |
| 515 | ucsdata->uni_tbl = snewn(256, char *); |
| 516 | memset(ucsdata->uni_tbl, 0, 256 * sizeof(char *)); |
| 517 | } |
| 518 | j = ((ucsdata->unitab_line[i] >> 8) & 0xFF); |
| 519 | if (!ucsdata->uni_tbl[j]) { |
| 520 | ucsdata->uni_tbl[j] = snewn(256, char); |
| 521 | memset(ucsdata->uni_tbl[j], 0, 256 * sizeof(char)); |
| 522 | } |
| 523 | ucsdata->uni_tbl[j][ucsdata->unitab_line[i] & 0xFF] = i; |
| 524 | } |
| 525 | } |
| 526 | |
| 527 | /* Find the line control characters. */ |
| 528 | for (i = 0; i < 256; i++) |
| 529 | if (ucsdata->unitab_line[i] < ' ' |
| 530 | || (ucsdata->unitab_line[i] >= 0x7F && |
| 531 | ucsdata->unitab_line[i] < 0xA0)) |
| 532 | ucsdata->unitab_ctrl[i] = i; |
| 533 | else |
| 534 | ucsdata->unitab_ctrl[i] = 0xFF; |
| 535 | |
| 536 | /* Generate line->screen direct conversion links. */ |
| 537 | if (cfg->vtmode == VT_OEMANSI || cfg->vtmode == VT_XWINDOWS) |
| 538 | link_font(ucsdata->unitab_scoacs, ucsdata->unitab_oemcp, ATTR_OEMCP); |
| 539 | |
| 540 | link_font(ucsdata->unitab_line, ucsdata->unitab_font, ATTR_ACP); |
| 541 | link_font(ucsdata->unitab_scoacs, ucsdata->unitab_font, ATTR_ACP); |
| 542 | link_font(ucsdata->unitab_xterm, ucsdata->unitab_font, ATTR_ACP); |
| 543 | |
| 544 | if (cfg->vtmode == VT_OEMANSI || cfg->vtmode == VT_XWINDOWS) { |
| 545 | link_font(ucsdata->unitab_line, ucsdata->unitab_oemcp, ATTR_OEMCP); |
| 546 | link_font(ucsdata->unitab_xterm, ucsdata->unitab_oemcp, ATTR_OEMCP); |
| 547 | } |
| 548 | |
| 549 | if (ucsdata->dbcs_screenfont && |
| 550 | ucsdata->font_codepage != ucsdata->line_codepage) { |
| 551 | /* F***ing Microsoft fonts, Japanese and Korean codepage fonts |
| 552 | * have a currency symbol at 0x5C but their unicode value is |
| 553 | * still given as U+005C not the correct U+00A5. */ |
| 554 | ucsdata->unitab_line['\\'] = ATTR_OEMCP + '\\'; |
| 555 | } |
| 556 | |
| 557 | /* Last chance, if !unicode then try poorman links. */ |
| 558 | if (cfg->vtmode != VT_UNICODE) { |
| 559 | static const char poorman_scoacs[] = |
| 560 | "CueaaaaceeeiiiAAE**ooouuyOUc$YPsaiounNao?++**!<>###||||++||++++++--|-+||++--|-+----++++++++##||#aBTPEsyt******EN=+><++-=... n2* "; |
| 561 | static const char poorman_latin1[] = |
| 562 | " !cL.Y|S\"Ca<--R~o+23'u|.,1o>///?AAAAAAACEEEEIIIIDNOOOOOxOUUUUYPBaaaaaaaceeeeiiiionooooo/ouuuuypy"; |
| 563 | static const char poorman_vt100[] = "*#****o~**+++++-----++++|****L."; |
| 564 | |
| 565 | for (i = 160; i < 256; i++) |
| 566 | if (!DIRECT_FONT(ucsdata->unitab_line[i]) && |
| 567 | ucsdata->unitab_line[i] >= 160 && |
| 568 | ucsdata->unitab_line[i] < 256) { |
| 569 | ucsdata->unitab_line[i] = |
| 570 | (WCHAR) (ATTR_ACP + |
| 571 | poorman_latin1[ucsdata->unitab_line[i] - 160]); |
| 572 | } |
| 573 | for (i = 96; i < 127; i++) |
| 574 | if (!DIRECT_FONT(ucsdata->unitab_xterm[i])) |
| 575 | ucsdata->unitab_xterm[i] = |
| 576 | (WCHAR) (ATTR_ACP + poorman_vt100[i - 96]); |
| 577 | for(i=128;i<256;i++) |
| 578 | if (!DIRECT_FONT(ucsdata->unitab_scoacs[i])) |
| 579 | ucsdata->unitab_scoacs[i] = |
| 580 | (WCHAR) (ATTR_ACP + poorman_scoacs[i - 128]); |
| 581 | } |
| 582 | } |
| 583 | |
| 584 | static void link_font(WCHAR * line_tbl, WCHAR * font_tbl, WCHAR attr) |
| 585 | { |
| 586 | int font_index, line_index, i; |
| 587 | for (line_index = 0; line_index < 256; line_index++) { |
| 588 | if (DIRECT_FONT(line_tbl[line_index])) |
| 589 | continue; |
| 590 | for(i = 0; i < 256; i++) { |
| 591 | font_index = ((32 + i) & 0xFF); |
| 592 | if (line_tbl[line_index] == font_tbl[font_index]) { |
| 593 | line_tbl[line_index] = (WCHAR) (attr + font_index); |
| 594 | break; |
| 595 | } |
| 596 | } |
| 597 | } |
| 598 | } |
| 599 | |
| 600 | wchar_t xlat_uskbd2cyrllic(int ch) |
| 601 | { |
| 602 | static const wchar_t cyrtab[] = { |
| 603 | 0, 1, 2, 3, 4, 5, 6, 7, |
| 604 | 8, 9, 10, 11, 12, 13, 14, 15, |
| 605 | 16, 17, 18, 19, 20, 21, 22, 23, |
| 606 | 24, 25, 26, 27, 28, 29, 30, 31, |
| 607 | 32, 33, 0x042d, 35, 36, 37, 38, 0x044d, |
| 608 | 40, 41, 42, 0x0406, 0x0431, 0x0454, 0x044e, 0x002e, |
| 609 | 48, 49, 50, 51, 52, 53, 54, 55, |
| 610 | 56, 57, 0x0416, 0x0436, 0x0411, 0x0456, 0x042e, 0x002c, |
| 611 | 64, 0x0424, 0x0418, 0x0421, 0x0412, 0x0423, 0x0410, 0x041f, |
| 612 | 0x0420, 0x0428, 0x041e, 0x041b, 0x0414, 0x042c, 0x0422, 0x0429, |
| 613 | 0x0417, 0x0419, 0x041a, 0x042b, 0x0415, 0x0413, 0x041c, 0x0426, |
| 614 | 0x0427, 0x041d, 0x042f, 0x0445, 0x0457, 0x044a, 94, 0x0404, |
| 615 | 96, 0x0444, 0x0438, 0x0441, 0x0432, 0x0443, 0x0430, 0x043f, |
| 616 | 0x0440, 0x0448, 0x043e, 0x043b, 0x0434, 0x044c, 0x0442, 0x0449, |
| 617 | 0x0437, 0x0439, 0x043a, 0x044b, 0x0435, 0x0433, 0x043c, 0x0446, |
| 618 | 0x0447, 0x043d, 0x044f, 0x0425, 0x0407, 0x042a, 126, 127 |
| 619 | }; |
| 620 | return cyrtab[ch&0x7F]; |
| 621 | } |
| 622 | |
| 623 | int check_compose_internal(int first, int second, int recurse) |
| 624 | { |
| 625 | |
| 626 | static const struct { |
| 627 | char first, second; |
| 628 | wchar_t composed; |
| 629 | } composetbl[] = { |
| 630 | { |
| 631 | 0x2b, 0x2b, 0x0023}, { |
| 632 | 0x41, 0x41, 0x0040}, { |
| 633 | 0x28, 0x28, 0x005b}, { |
| 634 | 0x2f, 0x2f, 0x005c}, { |
| 635 | 0x29, 0x29, 0x005d}, { |
| 636 | 0x28, 0x2d, 0x007b}, { |
| 637 | 0x2d, 0x29, 0x007d}, { |
| 638 | 0x2f, 0x5e, 0x007c}, { |
| 639 | 0x21, 0x21, 0x00a1}, { |
| 640 | 0x43, 0x2f, 0x00a2}, { |
| 641 | 0x43, 0x7c, 0x00a2}, { |
| 642 | 0x4c, 0x2d, 0x00a3}, { |
| 643 | 0x4c, 0x3d, 0x20a4}, { |
| 644 | 0x58, 0x4f, 0x00a4}, { |
| 645 | 0x58, 0x30, 0x00a4}, { |
| 646 | 0x59, 0x2d, 0x00a5}, { |
| 647 | 0x59, 0x3d, 0x00a5}, { |
| 648 | 0x7c, 0x7c, 0x00a6}, { |
| 649 | 0x53, 0x4f, 0x00a7}, { |
| 650 | 0x53, 0x21, 0x00a7}, { |
| 651 | 0x53, 0x30, 0x00a7}, { |
| 652 | 0x22, 0x22, 0x00a8}, { |
| 653 | 0x43, 0x4f, 0x00a9}, { |
| 654 | 0x43, 0x30, 0x00a9}, { |
| 655 | 0x41, 0x5f, 0x00aa}, { |
| 656 | 0x3c, 0x3c, 0x00ab}, { |
| 657 | 0x2c, 0x2d, 0x00ac}, { |
| 658 | 0x2d, 0x2d, 0x00ad}, { |
| 659 | 0x52, 0x4f, 0x00ae}, { |
| 660 | 0x2d, 0x5e, 0x00af}, { |
| 661 | 0x30, 0x5e, 0x00b0}, { |
| 662 | 0x2b, 0x2d, 0x00b1}, { |
| 663 | 0x32, 0x5e, 0x00b2}, { |
| 664 | 0x33, 0x5e, 0x00b3}, { |
| 665 | 0x27, 0x27, 0x00b4}, { |
| 666 | 0x2f, 0x55, 0x00b5}, { |
| 667 | 0x50, 0x21, 0x00b6}, { |
| 668 | 0x2e, 0x5e, 0x00b7}, { |
| 669 | 0x2c, 0x2c, 0x00b8}, { |
| 670 | 0x31, 0x5e, 0x00b9}, { |
| 671 | 0x4f, 0x5f, 0x00ba}, { |
| 672 | 0x3e, 0x3e, 0x00bb}, { |
| 673 | 0x31, 0x34, 0x00bc}, { |
| 674 | 0x31, 0x32, 0x00bd}, { |
| 675 | 0x33, 0x34, 0x00be}, { |
| 676 | 0x3f, 0x3f, 0x00bf}, { |
| 677 | 0x60, 0x41, 0x00c0}, { |
| 678 | 0x27, 0x41, 0x00c1}, { |
| 679 | 0x5e, 0x41, 0x00c2}, { |
| 680 | 0x7e, 0x41, 0x00c3}, { |
| 681 | 0x22, 0x41, 0x00c4}, { |
| 682 | 0x2a, 0x41, 0x00c5}, { |
| 683 | 0x41, 0x45, 0x00c6}, { |
| 684 | 0x2c, 0x43, 0x00c7}, { |
| 685 | 0x60, 0x45, 0x00c8}, { |
| 686 | 0x27, 0x45, 0x00c9}, { |
| 687 | 0x5e, 0x45, 0x00ca}, { |
| 688 | 0x22, 0x45, 0x00cb}, { |
| 689 | 0x60, 0x49, 0x00cc}, { |
| 690 | 0x27, 0x49, 0x00cd}, { |
| 691 | 0x5e, 0x49, 0x00ce}, { |
| 692 | 0x22, 0x49, 0x00cf}, { |
| 693 | 0x2d, 0x44, 0x00d0}, { |
| 694 | 0x7e, 0x4e, 0x00d1}, { |
| 695 | 0x60, 0x4f, 0x00d2}, { |
| 696 | 0x27, 0x4f, 0x00d3}, { |
| 697 | 0x5e, 0x4f, 0x00d4}, { |
| 698 | 0x7e, 0x4f, 0x00d5}, { |
| 699 | 0x22, 0x4f, 0x00d6}, { |
| 700 | 0x58, 0x58, 0x00d7}, { |
| 701 | 0x2f, 0x4f, 0x00d8}, { |
| 702 | 0x60, 0x55, 0x00d9}, { |
| 703 | 0x27, 0x55, 0x00da}, { |
| 704 | 0x5e, 0x55, 0x00db}, { |
| 705 | 0x22, 0x55, 0x00dc}, { |
| 706 | 0x27, 0x59, 0x00dd}, { |
| 707 | 0x48, 0x54, 0x00de}, { |
| 708 | 0x73, 0x73, 0x00df}, { |
| 709 | 0x60, 0x61, 0x00e0}, { |
| 710 | 0x27, 0x61, 0x00e1}, { |
| 711 | 0x5e, 0x61, 0x00e2}, { |
| 712 | 0x7e, 0x61, 0x00e3}, { |
| 713 | 0x22, 0x61, 0x00e4}, { |
| 714 | 0x2a, 0x61, 0x00e5}, { |
| 715 | 0x61, 0x65, 0x00e6}, { |
| 716 | 0x2c, 0x63, 0x00e7}, { |
| 717 | 0x60, 0x65, 0x00e8}, { |
| 718 | 0x27, 0x65, 0x00e9}, { |
| 719 | 0x5e, 0x65, 0x00ea}, { |
| 720 | 0x22, 0x65, 0x00eb}, { |
| 721 | 0x60, 0x69, 0x00ec}, { |
| 722 | 0x27, 0x69, 0x00ed}, { |
| 723 | 0x5e, 0x69, 0x00ee}, { |
| 724 | 0x22, 0x69, 0x00ef}, { |
| 725 | 0x2d, 0x64, 0x00f0}, { |
| 726 | 0x7e, 0x6e, 0x00f1}, { |
| 727 | 0x60, 0x6f, 0x00f2}, { |
| 728 | 0x27, 0x6f, 0x00f3}, { |
| 729 | 0x5e, 0x6f, 0x00f4}, { |
| 730 | 0x7e, 0x6f, 0x00f5}, { |
| 731 | 0x22, 0x6f, 0x00f6}, { |
| 732 | 0x3a, 0x2d, 0x00f7}, { |
| 733 | 0x6f, 0x2f, 0x00f8}, { |
| 734 | 0x60, 0x75, 0x00f9}, { |
| 735 | 0x27, 0x75, 0x00fa}, { |
| 736 | 0x5e, 0x75, 0x00fb}, { |
| 737 | 0x22, 0x75, 0x00fc}, { |
| 738 | 0x27, 0x79, 0x00fd}, { |
| 739 | 0x68, 0x74, 0x00fe}, { |
| 740 | 0x22, 0x79, 0x00ff}, |
| 741 | /* Unicode extras. */ |
| 742 | { |
| 743 | 0x6f, 0x65, 0x0153}, { |
| 744 | 0x4f, 0x45, 0x0152}, |
| 745 | /* Compose pairs from UCS */ |
| 746 | { |
| 747 | 0x41, 0x2D, 0x0100}, { |
| 748 | 0x61, 0x2D, 0x0101}, { |
| 749 | 0x43, 0x27, 0x0106}, { |
| 750 | 0x63, 0x27, 0x0107}, { |
| 751 | 0x43, 0x5E, 0x0108}, { |
| 752 | 0x63, 0x5E, 0x0109}, { |
| 753 | 0x45, 0x2D, 0x0112}, { |
| 754 | 0x65, 0x2D, 0x0113}, { |
| 755 | 0x47, 0x5E, 0x011C}, { |
| 756 | 0x67, 0x5E, 0x011D}, { |
| 757 | 0x47, 0x2C, 0x0122}, { |
| 758 | 0x67, 0x2C, 0x0123}, { |
| 759 | 0x48, 0x5E, 0x0124}, { |
| 760 | 0x68, 0x5E, 0x0125}, { |
| 761 | 0x49, 0x7E, 0x0128}, { |
| 762 | 0x69, 0x7E, 0x0129}, { |
| 763 | 0x49, 0x2D, 0x012A}, { |
| 764 | 0x69, 0x2D, 0x012B}, { |
| 765 | 0x4A, 0x5E, 0x0134}, { |
| 766 | 0x6A, 0x5E, 0x0135}, { |
| 767 | 0x4B, 0x2C, 0x0136}, { |
| 768 | 0x6B, 0x2C, 0x0137}, { |
| 769 | 0x4C, 0x27, 0x0139}, { |
| 770 | 0x6C, 0x27, 0x013A}, { |
| 771 | 0x4C, 0x2C, 0x013B}, { |
| 772 | 0x6C, 0x2C, 0x013C}, { |
| 773 | 0x4E, 0x27, 0x0143}, { |
| 774 | 0x6E, 0x27, 0x0144}, { |
| 775 | 0x4E, 0x2C, 0x0145}, { |
| 776 | 0x6E, 0x2C, 0x0146}, { |
| 777 | 0x4F, 0x2D, 0x014C}, { |
| 778 | 0x6F, 0x2D, 0x014D}, { |
| 779 | 0x52, 0x27, 0x0154}, { |
| 780 | 0x72, 0x27, 0x0155}, { |
| 781 | 0x52, 0x2C, 0x0156}, { |
| 782 | 0x72, 0x2C, 0x0157}, { |
| 783 | 0x53, 0x27, 0x015A}, { |
| 784 | 0x73, 0x27, 0x015B}, { |
| 785 | 0x53, 0x5E, 0x015C}, { |
| 786 | 0x73, 0x5E, 0x015D}, { |
| 787 | 0x53, 0x2C, 0x015E}, { |
| 788 | 0x73, 0x2C, 0x015F}, { |
| 789 | 0x54, 0x2C, 0x0162}, { |
| 790 | 0x74, 0x2C, 0x0163}, { |
| 791 | 0x55, 0x7E, 0x0168}, { |
| 792 | 0x75, 0x7E, 0x0169}, { |
| 793 | 0x55, 0x2D, 0x016A}, { |
| 794 | 0x75, 0x2D, 0x016B}, { |
| 795 | 0x55, 0x2A, 0x016E}, { |
| 796 | 0x75, 0x2A, 0x016F}, { |
| 797 | 0x57, 0x5E, 0x0174}, { |
| 798 | 0x77, 0x5E, 0x0175}, { |
| 799 | 0x59, 0x5E, 0x0176}, { |
| 800 | 0x79, 0x5E, 0x0177}, { |
| 801 | 0x59, 0x22, 0x0178}, { |
| 802 | 0x5A, 0x27, 0x0179}, { |
| 803 | 0x7A, 0x27, 0x017A}, { |
| 804 | 0x47, 0x27, 0x01F4}, { |
| 805 | 0x67, 0x27, 0x01F5}, { |
| 806 | 0x4E, 0x60, 0x01F8}, { |
| 807 | 0x6E, 0x60, 0x01F9}, { |
| 808 | 0x45, 0x2C, 0x0228}, { |
| 809 | 0x65, 0x2C, 0x0229}, { |
| 810 | 0x59, 0x2D, 0x0232}, { |
| 811 | 0x79, 0x2D, 0x0233}, { |
| 812 | 0x44, 0x2C, 0x1E10}, { |
| 813 | 0x64, 0x2C, 0x1E11}, { |
| 814 | 0x47, 0x2D, 0x1E20}, { |
| 815 | 0x67, 0x2D, 0x1E21}, { |
| 816 | 0x48, 0x22, 0x1E26}, { |
| 817 | 0x68, 0x22, 0x1E27}, { |
| 818 | 0x48, 0x2C, 0x1E28}, { |
| 819 | 0x68, 0x2C, 0x1E29}, { |
| 820 | 0x4B, 0x27, 0x1E30}, { |
| 821 | 0x6B, 0x27, 0x1E31}, { |
| 822 | 0x4D, 0x27, 0x1E3E}, { |
| 823 | 0x6D, 0x27, 0x1E3F}, { |
| 824 | 0x50, 0x27, 0x1E54}, { |
| 825 | 0x70, 0x27, 0x1E55}, { |
| 826 | 0x56, 0x7E, 0x1E7C}, { |
| 827 | 0x76, 0x7E, 0x1E7D}, { |
| 828 | 0x57, 0x60, 0x1E80}, { |
| 829 | 0x77, 0x60, 0x1E81}, { |
| 830 | 0x57, 0x27, 0x1E82}, { |
| 831 | 0x77, 0x27, 0x1E83}, { |
| 832 | 0x57, 0x22, 0x1E84}, { |
| 833 | 0x77, 0x22, 0x1E85}, { |
| 834 | 0x58, 0x22, 0x1E8C}, { |
| 835 | 0x78, 0x22, 0x1E8D}, { |
| 836 | 0x5A, 0x5E, 0x1E90}, { |
| 837 | 0x7A, 0x5E, 0x1E91}, { |
| 838 | 0x74, 0x22, 0x1E97}, { |
| 839 | 0x77, 0x2A, 0x1E98}, { |
| 840 | 0x79, 0x2A, 0x1E99}, { |
| 841 | 0x45, 0x7E, 0x1EBC}, { |
| 842 | 0x65, 0x7E, 0x1EBD}, { |
| 843 | 0x59, 0x60, 0x1EF2}, { |
| 844 | 0x79, 0x60, 0x1EF3}, { |
| 845 | 0x59, 0x7E, 0x1EF8}, { |
| 846 | 0x79, 0x7E, 0x1EF9}, |
| 847 | /* Compatible/possibles from UCS */ |
| 848 | { |
| 849 | 0x49, 0x4A, 0x0132}, { |
| 850 | 0x69, 0x6A, 0x0133}, { |
| 851 | 0x4C, 0x4A, 0x01C7}, { |
| 852 | 0x4C, 0x6A, 0x01C8}, { |
| 853 | 0x6C, 0x6A, 0x01C9}, { |
| 854 | 0x4E, 0x4A, 0x01CA}, { |
| 855 | 0x4E, 0x6A, 0x01CB}, { |
| 856 | 0x6E, 0x6A, 0x01CC}, { |
| 857 | 0x44, 0x5A, 0x01F1}, { |
| 858 | 0x44, 0x7A, 0x01F2}, { |
| 859 | 0x64, 0x7A, 0x01F3}, { |
| 860 | 0x2E, 0x2E, 0x2025}, { |
| 861 | 0x21, 0x21, 0x203C}, { |
| 862 | 0x3F, 0x21, 0x2048}, { |
| 863 | 0x21, 0x3F, 0x2049}, { |
| 864 | 0x52, 0x73, 0x20A8}, { |
| 865 | 0x4E, 0x6F, 0x2116}, { |
| 866 | 0x53, 0x4D, 0x2120}, { |
| 867 | 0x54, 0x4D, 0x2122}, { |
| 868 | 0x49, 0x49, 0x2161}, { |
| 869 | 0x49, 0x56, 0x2163}, { |
| 870 | 0x56, 0x49, 0x2165}, { |
| 871 | 0x49, 0x58, 0x2168}, { |
| 872 | 0x58, 0x49, 0x216A}, { |
| 873 | 0x69, 0x69, 0x2171}, { |
| 874 | 0x69, 0x76, 0x2173}, { |
| 875 | 0x76, 0x69, 0x2175}, { |
| 876 | 0x69, 0x78, 0x2178}, { |
| 877 | 0x78, 0x69, 0x217A}, { |
| 878 | 0x31, 0x30, 0x2469}, { |
| 879 | 0x31, 0x31, 0x246A}, { |
| 880 | 0x31, 0x32, 0x246B}, { |
| 881 | 0x31, 0x33, 0x246C}, { |
| 882 | 0x31, 0x34, 0x246D}, { |
| 883 | 0x31, 0x35, 0x246E}, { |
| 884 | 0x31, 0x36, 0x246F}, { |
| 885 | 0x31, 0x37, 0x2470}, { |
| 886 | 0x31, 0x38, 0x2471}, { |
| 887 | 0x31, 0x39, 0x2472}, { |
| 888 | 0x32, 0x30, 0x2473}, { |
| 889 | 0x31, 0x2E, 0x2488}, { |
| 890 | 0x32, 0x2E, 0x2489}, { |
| 891 | 0x33, 0x2E, 0x248A}, { |
| 892 | 0x34, 0x2E, 0x248B}, { |
| 893 | 0x35, 0x2E, 0x248C}, { |
| 894 | 0x36, 0x2E, 0x248D}, { |
| 895 | 0x37, 0x2E, 0x248E}, { |
| 896 | 0x38, 0x2E, 0x248F}, { |
| 897 | 0x39, 0x2E, 0x2490}, { |
| 898 | 0x64, 0x61, 0x3372}, { |
| 899 | 0x41, 0x55, 0x3373}, { |
| 900 | 0x6F, 0x56, 0x3375}, { |
| 901 | 0x70, 0x63, 0x3376}, { |
| 902 | 0x70, 0x41, 0x3380}, { |
| 903 | 0x6E, 0x41, 0x3381}, { |
| 904 | 0x6D, 0x41, 0x3383}, { |
| 905 | 0x6B, 0x41, 0x3384}, { |
| 906 | 0x4B, 0x42, 0x3385}, { |
| 907 | 0x4D, 0x42, 0x3386}, { |
| 908 | 0x47, 0x42, 0x3387}, { |
| 909 | 0x70, 0x46, 0x338A}, { |
| 910 | 0x6E, 0x46, 0x338B}, { |
| 911 | 0x6D, 0x67, 0x338E}, { |
| 912 | 0x6B, 0x67, 0x338F}, { |
| 913 | 0x48, 0x7A, 0x3390}, { |
| 914 | 0x66, 0x6D, 0x3399}, { |
| 915 | 0x6E, 0x6D, 0x339A}, { |
| 916 | 0x6D, 0x6D, 0x339C}, { |
| 917 | 0x63, 0x6D, 0x339D}, { |
| 918 | 0x6B, 0x6D, 0x339E}, { |
| 919 | 0x50, 0x61, 0x33A9}, { |
| 920 | 0x70, 0x73, 0x33B0}, { |
| 921 | 0x6E, 0x73, 0x33B1}, { |
| 922 | 0x6D, 0x73, 0x33B3}, { |
| 923 | 0x70, 0x56, 0x33B4}, { |
| 924 | 0x6E, 0x56, 0x33B5}, { |
| 925 | 0x6D, 0x56, 0x33B7}, { |
| 926 | 0x6B, 0x56, 0x33B8}, { |
| 927 | 0x4D, 0x56, 0x33B9}, { |
| 928 | 0x70, 0x57, 0x33BA}, { |
| 929 | 0x6E, 0x57, 0x33BB}, { |
| 930 | 0x6D, 0x57, 0x33BD}, { |
| 931 | 0x6B, 0x57, 0x33BE}, { |
| 932 | 0x4D, 0x57, 0x33BF}, { |
| 933 | 0x42, 0x71, 0x33C3}, { |
| 934 | 0x63, 0x63, 0x33C4}, { |
| 935 | 0x63, 0x64, 0x33C5}, { |
| 936 | 0x64, 0x42, 0x33C8}, { |
| 937 | 0x47, 0x79, 0x33C9}, { |
| 938 | 0x68, 0x61, 0x33CA}, { |
| 939 | 0x48, 0x50, 0x33CB}, { |
| 940 | 0x69, 0x6E, 0x33CC}, { |
| 941 | 0x4B, 0x4B, 0x33CD}, { |
| 942 | 0x4B, 0x4D, 0x33CE}, { |
| 943 | 0x6B, 0x74, 0x33CF}, { |
| 944 | 0x6C, 0x6D, 0x33D0}, { |
| 945 | 0x6C, 0x6E, 0x33D1}, { |
| 946 | 0x6C, 0x78, 0x33D3}, { |
| 947 | 0x6D, 0x62, 0x33D4}, { |
| 948 | 0x50, 0x48, 0x33D7}, { |
| 949 | 0x50, 0x52, 0x33DA}, { |
| 950 | 0x73, 0x72, 0x33DB}, { |
| 951 | 0x53, 0x76, 0x33DC}, { |
| 952 | 0x57, 0x62, 0x33DD}, { |
| 953 | 0x66, 0x66, 0xFB00}, { |
| 954 | 0x66, 0x69, 0xFB01}, { |
| 955 | 0x66, 0x6C, 0xFB02}, { |
| 956 | 0x73, 0x74, 0xFB06}, { |
| 957 | 0, 0, 0} |
| 958 | }, *c; |
| 959 | |
| 960 | int nc = -1; |
| 961 | |
| 962 | for (c = composetbl; c->first; c++) { |
| 963 | if (c->first == first && c->second == second) |
| 964 | return c->composed; |
| 965 | } |
| 966 | |
| 967 | if (recurse == 0) { |
| 968 | nc = check_compose_internal(second, first, 1); |
| 969 | if (nc == -1) |
| 970 | nc = check_compose_internal(toupper(first), toupper(second), 1); |
| 971 | if (nc == -1) |
| 972 | nc = check_compose_internal(toupper(second), toupper(first), 1); |
| 973 | } |
| 974 | return nc; |
| 975 | } |
| 976 | |
| 977 | int check_compose(int first, int second) |
| 978 | { |
| 979 | return check_compose_internal(first, second, 0); |
| 980 | } |
| 981 | |
| 982 | int decode_codepage(char *cp_name) |
| 983 | { |
| 984 | char *s, *d; |
| 985 | const struct cp_list_item *cpi; |
| 986 | int codepage = -1; |
| 987 | CPINFO cpinfo; |
| 988 | |
| 989 | if (!*cp_name) { |
| 990 | /* |
| 991 | * Here we select a plausible default code page based on |
| 992 | * the locale the user is in. We wish to select an ISO code |
| 993 | * page or appropriate local default _rather_ than go with |
| 994 | * the Win125* series, because it's more important to have |
| 995 | * CSI and friends enabled by default than the ghastly |
| 996 | * Windows extra quote characters, and because it's more |
| 997 | * likely the user is connecting to a remote server that |
| 998 | * does something Unixy or VMSy and hence standards- |
| 999 | * compliant than that they're connecting back to a Windows |
| 1000 | * box using horrible nonstandard charsets. |
| 1001 | * |
| 1002 | * Accordingly, Robert de Bath suggests a method for |
| 1003 | * picking a default character set that runs as follows: |
| 1004 | * first call GetACP to get the system's ANSI code page |
| 1005 | * identifier, and translate as follows: |
| 1006 | * |
| 1007 | * 1250 -> ISO 8859-2 |
| 1008 | * 1251 -> KOI8-U |
| 1009 | * 1252 -> ISO 8859-1 |
| 1010 | * 1253 -> ISO 8859-7 |
| 1011 | * 1254 -> ISO 8859-9 |
| 1012 | * 1255 -> ISO 8859-8 |
| 1013 | * 1256 -> ISO 8859-6 |
| 1014 | * 1257 -> ISO 8859-13 (changed from 8859-4 on advice of a Lithuanian) |
| 1015 | * |
| 1016 | * and for anything else, choose direct-to-font. |
| 1017 | */ |
| 1018 | int cp = GetACP(); |
| 1019 | switch (cp) { |
| 1020 | case 1250: cp_name = "ISO-8859-2"; break; |
| 1021 | case 1251: cp_name = "KOI8-U"; break; |
| 1022 | case 1252: cp_name = "ISO-8859-1"; break; |
| 1023 | case 1253: cp_name = "ISO-8859-7"; break; |
| 1024 | case 1254: cp_name = "ISO-8859-9"; break; |
| 1025 | case 1255: cp_name = "ISO-8859-8"; break; |
| 1026 | case 1256: cp_name = "ISO-8859-6"; break; |
| 1027 | case 1257: cp_name = "ISO-8859-13"; break; |
| 1028 | /* default: leave it blank, which will select -1, direct->font */ |
| 1029 | } |
| 1030 | } |
| 1031 | |
| 1032 | if (cp_name && *cp_name) |
| 1033 | for (cpi = cp_list; cpi->name; cpi++) { |
| 1034 | s = cp_name; |
| 1035 | d = cpi->name; |
| 1036 | for (;;) { |
| 1037 | while (*s && !isalnum(*s) && *s != ':') |
| 1038 | s++; |
| 1039 | while (*d && !isalnum(*d) && *d != ':') |
| 1040 | d++; |
| 1041 | if (*s == 0) { |
| 1042 | codepage = cpi->codepage; |
| 1043 | if (codepage == CP_UTF8) |
| 1044 | goto break_break; |
| 1045 | if (codepage == -1) |
| 1046 | return codepage; |
| 1047 | if (codepage == 0) { |
| 1048 | codepage = 65536 + (cpi - cp_list); |
| 1049 | goto break_break; |
| 1050 | } |
| 1051 | |
| 1052 | if (GetCPInfo(codepage, &cpinfo) != 0) |
| 1053 | goto break_break; |
| 1054 | } |
| 1055 | if (tolower(*s++) != tolower(*d++)) |
| 1056 | break; |
| 1057 | } |
| 1058 | } |
| 1059 | |
| 1060 | if (cp_name && *cp_name) { |
| 1061 | d = cp_name; |
| 1062 | if (tolower(d[0]) == 'c' && tolower(d[1]) == 'p') |
| 1063 | d += 2; |
| 1064 | if (tolower(d[0]) == 'i' && tolower(d[1]) == 'b' |
| 1065 | && tolower(d[1]) == 'm') |
| 1066 | d += 3; |
| 1067 | for (s = d; *s >= '0' && *s <= '9'; s++); |
| 1068 | if (*s == 0 && s != d) |
| 1069 | codepage = atoi(d); /* CP999 or IBM999 */ |
| 1070 | |
| 1071 | if (codepage == CP_ACP) |
| 1072 | codepage = GetACP(); |
| 1073 | if (codepage == CP_OEMCP) |
| 1074 | codepage = GetOEMCP(); |
| 1075 | if (codepage > 65535) |
| 1076 | codepage = -2; |
| 1077 | } |
| 1078 | |
| 1079 | break_break:; |
| 1080 | if (codepage != -1) { |
| 1081 | if (codepage != CP_UTF8 && codepage < 65536) { |
| 1082 | if (GetCPInfo(codepage, &cpinfo) == 0) { |
| 1083 | codepage = -2; |
| 1084 | } else if (cpinfo.MaxCharSize > 1) |
| 1085 | codepage = -3; |
| 1086 | } |
| 1087 | } |
| 1088 | if (codepage == -1 && *cp_name) |
| 1089 | codepage = -2; |
| 1090 | return codepage; |
| 1091 | } |
| 1092 | |
| 1093 | const char *cp_name(int codepage) |
| 1094 | { |
| 1095 | const struct cp_list_item *cpi, *cpno; |
| 1096 | static char buf[32]; |
| 1097 | |
| 1098 | if (codepage == -1) { |
| 1099 | sprintf(buf, "Use font encoding"); |
| 1100 | return buf; |
| 1101 | } |
| 1102 | |
| 1103 | if (codepage > 0 && codepage < 65536) |
| 1104 | sprintf(buf, "CP%03d", codepage); |
| 1105 | else |
| 1106 | *buf = 0; |
| 1107 | |
| 1108 | if (codepage >= 65536) { |
| 1109 | cpno = 0; |
| 1110 | for (cpi = cp_list; cpi->name; cpi++) |
| 1111 | if (cpi == cp_list + (codepage - 65536)) { |
| 1112 | cpno = cpi; |
| 1113 | break; |
| 1114 | } |
| 1115 | if (cpno) |
| 1116 | for (cpi = cp_list; cpi->name; cpi++) { |
| 1117 | if (cpno->cp_table == cpi->cp_table) |
| 1118 | return cpi->name; |
| 1119 | } |
| 1120 | } else { |
| 1121 | for (cpi = cp_list; cpi->name; cpi++) { |
| 1122 | if (codepage == cpi->codepage) |
| 1123 | return cpi->name; |
| 1124 | } |
| 1125 | } |
| 1126 | return buf; |
| 1127 | } |
| 1128 | |
| 1129 | /* |
| 1130 | * Return the nth code page in the list, for use in the GUI |
| 1131 | * configurer. |
| 1132 | */ |
| 1133 | const char *cp_enumerate(int index) |
| 1134 | { |
| 1135 | if (index < 0 || index >= lenof(cp_list)) |
| 1136 | return NULL; |
| 1137 | return cp_list[index].name; |
| 1138 | } |
| 1139 | |
| 1140 | void get_unitab(int codepage, wchar_t * unitab, int ftype) |
| 1141 | { |
| 1142 | char tbuf[4]; |
| 1143 | int i, max = 256, flg = MB_ERR_INVALID_CHARS; |
| 1144 | |
| 1145 | if (ftype) |
| 1146 | flg |= MB_USEGLYPHCHARS; |
| 1147 | if (ftype == 2) |
| 1148 | max = 128; |
| 1149 | |
| 1150 | if (codepage == CP_UTF8) { |
| 1151 | for (i = 0; i < max; i++) |
| 1152 | unitab[i] = i; |
| 1153 | return; |
| 1154 | } |
| 1155 | |
| 1156 | if (codepage == CP_ACP) |
| 1157 | codepage = GetACP(); |
| 1158 | else if (codepage == CP_OEMCP) |
| 1159 | codepage = GetOEMCP(); |
| 1160 | |
| 1161 | if (codepage > 0 && codepage < 65536) { |
| 1162 | for (i = 0; i < max; i++) { |
| 1163 | tbuf[0] = i; |
| 1164 | |
| 1165 | if (mb_to_wc(codepage, flg, tbuf, 1, unitab + i, 1) |
| 1166 | != 1) |
| 1167 | unitab[i] = 0xFFFD; |
| 1168 | } |
| 1169 | } else { |
| 1170 | int j = 256 - cp_list[codepage & 0xFFFF].cp_size; |
| 1171 | for (i = 0; i < max; i++) |
| 1172 | unitab[i] = i; |
| 1173 | for (i = j; i < max; i++) |
| 1174 | unitab[i] = cp_list[codepage & 0xFFFF].cp_table[i - j]; |
| 1175 | } |
| 1176 | } |
| 1177 | |
| 1178 | int wc_to_mb(int codepage, int flags, wchar_t *wcstr, int wclen, |
| 1179 | char *mbstr, int mblen, char *defchr, int *defused, |
| 1180 | struct unicode_data *ucsdata) |
| 1181 | { |
| 1182 | char *p; |
| 1183 | int i; |
| 1184 | if (ucsdata && codepage == ucsdata->line_codepage && ucsdata->uni_tbl) { |
| 1185 | /* Do this by array lookup if we can. */ |
| 1186 | if (wclen < 0) { |
| 1187 | for (wclen = 0; wcstr[wclen++] ;); /* will include the NUL */ |
| 1188 | } |
| 1189 | for (p = mbstr, i = 0; i < wclen; i++) { |
| 1190 | wchar_t ch = wcstr[i]; |
| 1191 | int by; |
| 1192 | char *p1; |
| 1193 | if (ucsdata->uni_tbl && (p1 = ucsdata->uni_tbl[(ch >> 8) & 0xFF]) |
| 1194 | && (by = p1[ch & 0xFF])) |
| 1195 | *p++ = by; |
| 1196 | else if (ch < 0x80) |
| 1197 | *p++ = (char) ch; |
| 1198 | else if (defchr) { |
| 1199 | int j; |
| 1200 | for (j = 0; defchr[j]; j++) |
| 1201 | *p++ = defchr[j]; |
| 1202 | if (defused) *defused = 1; |
| 1203 | } |
| 1204 | #if 1 |
| 1205 | else |
| 1206 | *p++ = '.'; |
| 1207 | #endif |
| 1208 | assert(p - mbstr < mblen); |
| 1209 | } |
| 1210 | return p - mbstr; |
| 1211 | } else |
| 1212 | return WideCharToMultiByte(codepage, flags, wcstr, wclen, |
| 1213 | mbstr, mblen, defchr, defused); |
| 1214 | } |
| 1215 | |
| 1216 | int mb_to_wc(int codepage, int flags, char *mbstr, int mblen, |
| 1217 | wchar_t *wcstr, int wclen) |
| 1218 | { |
| 1219 | return MultiByteToWideChar(codepage, flags, mbstr, mblen, wcstr, wclen); |
| 1220 | } |
| 1221 | |
| 1222 | int is_dbcs_leadbyte(int codepage, char byte) |
| 1223 | { |
| 1224 | return IsDBCSLeadByteEx(codepage, byte); |
| 1225 | } |