https://perl5.git.perl.org / perl5.git / blame
| Commit | Line | Data |
|---|---|---|
| 61dad979 KW |
1 | /* -*- buffer-read-only: t -*- |
| 2 | * !!!!!!! DO NOT EDIT THIS FILE !!!!!!! | |
| 1b0f46bf | 3 | * This file is built by regen/unicode_constants.pl from Unicode data. |
| 61dad979 KW |
4 | * Any changes made here will be lost! |
| 5 | */ | |
| 6 | ||
| d10c72f2 | 7 | |
| 1b0f46bf KW |
8 | #ifndef H_UNICODE_CONSTANTS /* Guard against nested #includes */ |
| 9 | #define H_UNICODE_CONSTANTS 1 | |
| d10c72f2 | 10 | |
| 61dad979 | 11 | /* This file contains #defines for various Unicode code points. The values |
| 525b6419 KW |
12 | * the macros expand to are the native Unicode code point, or all or portions |
| 13 | * of the UTF-8 encoding for the code point. In the former case, the macro | |
| 14 | * name has the suffix "_NATIVE"; otherwise, the suffix "_UTF8". | |
| 61dad979 | 15 | * |
| 525b6419 KW |
16 | * The macros that have the suffix "_UTF8" may have further suffixes, as |
| 17 | * follows: | |
| 18 | * "_FIRST_BYTE" if the value is just the first byte of the UTF-8 | |
| 19 | * representation; the value will be a numeric constant. | |
| 20 | * "_TAIL" if instead it represents all but the first byte. This, and | |
| 21 | * with no additional suffix are both string constants */ | |
| 61dad979 | 22 | |
| 76837d21 | 23 | |
| ad88cddb KW |
24 | #if 'A' == 65 /* ASCII/Latin1 */ |
| 25 | # define LATIN_SMALL_LETTER_LONG_S_UTF8 "\xC5\xBF" /* U+017F */ | |
| 76837d21 | 26 | |
| ad88cddb | 27 | # define COMBINING_GRAVE_ACCENT_UTF8 "\xCC\x80" /* U+0300 */ |
| a78bc3c6 | 28 | |
| ad88cddb KW |
29 | # define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xCE\x99" /* U+0399 */ |
| 30 | # define GREEK_SMALL_LETTER_MU_UTF8 "\xCE\xBC" /* U+03BC */ | |
| f2e06375 | 31 | |
| ad88cddb | 32 | # define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xE1\xBA\x9E" /* U+1E9E */ |
| a9f50d33 | 33 | |
| ad88cddb KW |
34 | # define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xEF\xAC\x85" /* U+FB05 */ |
| 35 | # define LATIN_SMALL_LIGATURE_ST_UTF8 "\xEF\xAC\x86" /* U+FB06 */ | |
| 1dfa4f52 | 36 | |
| ad88cddb KW |
37 | # define HYPHEN_UTF8 "\xE2\x80\x90" /* U+2010 */ |
| 38 | # define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xED /* U+D800 */ | |
| 39 | # define BOM_UTF8_FIRST_BYTE 0xEF /* U+FEFF */ | |
| 40 | # define BOM_UTF8_TAIL "\xBB\xBF" /* U+FEFF */ | |
| df758df2 | 41 | |
| ad88cddb KW |
42 | # define NBSP_NATIVE 0xA0 /* U+00A0 */ |
| 43 | # define NBSP_UTF8 "\xC2\xA0" /* U+00A0 */ | |
| 44 | ||
| 45 | # define DEL_NATIVE 0x7F /* U+007F */ | |
| 46 | # define CR_NATIVE 0x0D /* U+000D */ | |
| 47 | # define LF_NATIVE 0x0A /* U+000A */ | |
| 48 | # define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0xDF /* U+00DF */ | |
| 49 | # define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0xE5 /* U+00E5 */ | |
| 50 | # define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0xC5 /* U+00C5 */ | |
| 51 | # define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xFF /* U+00FF */ | |
| 52 | # define MICRO_SIGN_NATIVE 0xB5 /* U+00B5 */ | |
| 53 | ||
| 54 | #endif /* ASCII/Latin1 */ | |
| 55 | ||
| 56 | #if 'A' == 193 /* EBCDIC 1047 */ \ | |
| 57 | && '\\' == 224 && '[' == 173 && ']' == 189 && '{' == 192 && '}' == 208 \ | |
| 58 | && '^' == 95 && '~' == 161 && '!' == 90 && '#' == 123 && '|' == 79 \ | |
| 59 | && '$' == 91 && '@' == 124 && '`' == 121 | |
| 60 | # define LATIN_SMALL_LETTER_LONG_S_UTF8 "\x8F\x73" /* U+017F */ | |
| 61 | ||
| 62 | # define COMBINING_GRAVE_ACCENT_UTF8 "\xAF\x41" /* U+0300 */ | |
| 63 | ||
| 64 | # define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xB3\x68" /* U+0399 */ | |
| 65 | # define GREEK_SMALL_LETTER_MU_UTF8 "\xB4\x70" /* U+03BC */ | |
| 66 | ||
| 67 | # define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xBF\x63\x72" /* U+1E9E */ | |
| 68 | ||
| 69 | # define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xDD\x72\x67\x46" /* U+FB05 */ | |
| 70 | # define LATIN_SMALL_LIGATURE_ST_UTF8 "\xDD\x72\x67\x47" /* U+FB06 */ | |
| 71 | ||
| 72 | # define HYPHEN_UTF8 "\xCA\x41\x57" /* U+2010 */ | |
| 73 | # define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xDD /* U+D800 */ | |
| 74 | # define BOM_UTF8_FIRST_BYTE 0xDD /* U+FEFF */ | |
| 75 | # define BOM_UTF8_TAIL "\x73\x66\x73" /* U+FEFF */ | |
| 76 | ||
| 77 | # define NBSP_NATIVE 0x41 /* U+00A0 */ | |
| 78 | # define NBSP_UTF8 "\x80\x41" /* U+00A0 */ | |
| 79 | ||
| 80 | # define DEL_NATIVE 0x07 /* U+007F */ | |
| 81 | # define CR_NATIVE 0x0D /* U+000D */ | |
| 82 | # define LF_NATIVE 0x15 /* U+000A */ | |
| 83 | # define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0x59 /* U+00DF */ | |
| 84 | # define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x47 /* U+00E5 */ | |
| 85 | # define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x67 /* U+00C5 */ | |
| 86 | # define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xDF /* U+00FF */ | |
| 87 | # define MICRO_SIGN_NATIVE 0xA0 /* U+00B5 */ | |
| 88 | ||
| 89 | #endif /* EBCDIC 1047 */ | |
| 90 | ||
| 91 | #if 'A' == 193 /* EBCDIC 037 */ \ | |
| 92 | && '\\' == 224 && '[' == 186 && ']' == 187 && '{' == 192 && '}' == 208 \ | |
| 93 | && '^' == 176 && '~' == 161 && '!' == 90 && '#' == 123 && '|' == 79 \ | |
| 94 | && '$' == 91 && '@' == 124 && '`' == 121 | |
| 95 | # define LATIN_SMALL_LETTER_LONG_S_UTF8 "\x8E\x72" /* U+017F */ | |
| 96 | ||
| 97 | # define COMBINING_GRAVE_ACCENT_UTF8 "\xAD\x41" /* U+0300 */ | |
| 98 | ||
| 99 | # define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xB2\x67" /* U+0399 */ | |
| 100 | # define GREEK_SMALL_LETTER_MU_UTF8 "\xB3\x6A" /* U+03BC */ | |
| 101 | ||
| 102 | # define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xBF\x62\x71" /* U+1E9E */ | |
| 103 | ||
| 104 | # define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xDD\x71\x66\x46" /* U+FB05 */ | |
| 105 | # define LATIN_SMALL_LIGATURE_ST_UTF8 "\xDD\x71\x66\x47" /* U+FB06 */ | |
| 106 | ||
| 107 | # define HYPHEN_UTF8 "\xCA\x41\x57" /* U+2010 */ | |
| 108 | # define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xDD /* U+D800 */ | |
| 109 | # define BOM_UTF8_FIRST_BYTE 0xDD /* U+FEFF */ | |
| 110 | # define BOM_UTF8_TAIL "\x72\x65\x72" /* U+FEFF */ | |
| 111 | ||
| 112 | # define NBSP_NATIVE 0x41 /* U+00A0 */ | |
| 113 | # define NBSP_UTF8 "\x78\x41" /* U+00A0 */ | |
| 114 | ||
| 115 | # define DEL_NATIVE 0x07 /* U+007F */ | |
| 116 | # define CR_NATIVE 0x0D /* U+000D */ | |
| 117 | # define LF_NATIVE 0x25 /* U+000A */ | |
| 118 | # define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0x59 /* U+00DF */ | |
| 119 | # define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x47 /* U+00E5 */ | |
| 120 | # define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x67 /* U+00C5 */ | |
| 121 | # define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xDF /* U+00FF */ | |
| 122 | # define MICRO_SIGN_NATIVE 0xA0 /* U+00B5 */ | |
| 123 | ||
| 124 | #endif /* EBCDIC 037 */ | |
| 125 | ||
| 126 | #if 'A' == 193 /* EBCDIC POSIX-BC */ \ | |
| 127 | && '\\' == 188 && '[' == 187 && ']' == 189 && '{' == 251 && '}' == 253 \ | |
| 128 | && '^' == 106 && '~' == 255 && '!' == 90 && '#' == 123 && '|' == 79 \ | |
| 129 | && '$' == 91 && '@' == 124 && '`' == 74 | |
| 130 | # define LATIN_SMALL_LETTER_LONG_S_UTF8 "\x90\x75" /* U+017F */ | |
| 131 | ||
| 132 | # define COMBINING_GRAVE_ACCENT_UTF8 "\xAE\x41" /* U+0300 */ | |
| 133 | ||
| 134 | # define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xB2\x69" /* U+0399 */ | |
| 135 | # define GREEK_SMALL_LETTER_MU_UTF8 "\xB3\x72" /* U+03BC */ | |
| 136 | ||
| 137 | # define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xC0\x64\x74" /* U+1E9E */ | |
| 138 | ||
| 139 | # define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xDC\x74\x68\x46" /* U+FB05 */ | |
| 140 | # define LATIN_SMALL_LIGATURE_ST_UTF8 "\xDC\x74\x68\x47" /* U+FB06 */ | |
| 141 | ||
| 142 | # define HYPHEN_UTF8 "\xCA\x41\x58" /* U+2010 */ | |
| 143 | # define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xDC /* U+D800 */ | |
| 144 | # define BOM_UTF8_FIRST_BYTE 0xDC /* U+FEFF */ | |
| 145 | # define BOM_UTF8_TAIL "\x75\x67\x75" /* U+FEFF */ | |
| 146 | ||
| 147 | # define NBSP_NATIVE 0x41 /* U+00A0 */ | |
| 148 | # define NBSP_UTF8 "\x8A\x41" /* U+00A0 */ | |
| 149 | ||
| 150 | # define DEL_NATIVE 0x07 /* U+007F */ | |
| 151 | # define CR_NATIVE 0x0D /* U+000D */ | |
| 152 | # define LF_NATIVE 0x15 /* U+000A */ | |
| 153 | # define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0x59 /* U+00DF */ | |
| 154 | # define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x47 /* U+00E5 */ | |
| 155 | # define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x67 /* U+00C5 */ | |
| 156 | # define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xDF /* U+00FF */ | |
| 157 | # define MICRO_SIGN_NATIVE 0xA0 /* U+00B5 */ | |
| 158 | ||
| 159 | #endif /* EBCDIC POSIX-BC */ | |
| 525b6419 | 160 | |
| 1b0f46bf | 161 | #endif /* H_UNICODE_CONSTANTS */ |
| d10c72f2 | 162 | |
| 61dad979 | 163 | /* ex: set ro: */ |