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 | |
| f2e06375 | 23 | #define LATIN_SMALL_LETTER_LONG_S_UTF8 "\xC5\xBF" /* U+017F */ |
| 76837d21 | 24 | |
| 1dfa4f52 | 25 | #define COMBINING_GRAVE_ACCENT_UTF8 "\xCC\x80" /* U+0300 */ |
| 76837d21 | 26 | |
| a78bc3c6 KW |
27 | #define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xCE\x99" /* U+0399 */ |
| 28 | #define GREEK_SMALL_LETTER_MU_UTF8 "\xCE\xBC" /* U+03BC */ | |
| 29 | ||
| f2e06375 KW |
30 | #define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xE1\xBA\x9E" /* U+1E9E */ |
| 31 | ||
| a9f50d33 KW |
32 | #define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xEF\xAC\x85" /* U+FB05 */ |
| 33 | #define LATIN_SMALL_LIGATURE_ST_UTF8 "\xEF\xAC\x86" /* U+FB06 */ | |
| 34 | ||
| 1dfa4f52 KW |
35 | #define HYPHEN_UTF8 "\xE2\x80\x90" /* U+2010 */ |
| 36 | #define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xED /* U+D800 */ | |
| 5f0aa340 KW |
37 | #define BOM_UTF8_FIRST_BYTE 0xEF /* U+FEFF */ |
| 38 | #define BOM_UTF8_TAIL "\xBB\xBF" /* U+FEFF */ | |
| 1dfa4f52 | 39 | |
| 05016631 | 40 | #define DEL_NATIVE 0x7F /* U+007F */ |
| c5eda08a KW |
41 | #define CR_NATIVE 0x0D /* U+000D */ |
| 42 | #define LF_NATIVE 0x0A /* U+000A */ | |
| 1dfa4f52 KW |
43 | #define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0xDF /* U+00DF */ |
| 44 | #define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0xE5 /* U+00E5 */ | |
| 45 | #define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0xC5 /* U+00C5 */ | |
| 46 | #define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xFF /* U+00FF */ | |
| 47 | #define MICRO_SIGN_NATIVE 0xB5 /* U+00B5 */ | |
| 525b6419 | 48 | |
| 1b0f46bf | 49 | #endif /* H_UNICODE_CONSTANTS */ |
| d10c72f2 | 50 | |
| 61dad979 | 51 | /* ex: set ro: */ |