This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
mktables: Calculate \p{Assigned} earlier in build
[perl5.git] / unicode_constants.h
CommitLineData
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 */
d804860b
KW
48# define VT_NATIVE 0x0B /* U+000B */
49# define ESC_NATIVE 0x1B /* U+001B */
ad88cddb
KW
50# define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0xDF /* U+00DF */
51# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0xE5 /* U+00E5 */
52# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0xC5 /* U+00C5 */
53# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xFF /* U+00FF */
54# define MICRO_SIGN_NATIVE 0xB5 /* U+00B5 */
09cc440d 55# define MAX_PRINT_A_FOR_USE_ONLY_BY_REGCOMP_DOT_C 0x7E /* The max code point that isPRINT_A */
ad88cddb
KW
56
57#endif /* ASCII/Latin1 */
58
59#if 'A' == 193 /* EBCDIC 1047 */ \
60 && '\\' == 224 && '[' == 173 && ']' == 189 && '{' == 192 && '}' == 208 \
61 && '^' == 95 && '~' == 161 && '!' == 90 && '#' == 123 && '|' == 79 \
62 && '$' == 91 && '@' == 124 && '`' == 121
63# define LATIN_SMALL_LETTER_LONG_S_UTF8 "\x8F\x73" /* U+017F */
64
65# define COMBINING_GRAVE_ACCENT_UTF8 "\xAF\x41" /* U+0300 */
66
67# define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xB3\x68" /* U+0399 */
68# define GREEK_SMALL_LETTER_MU_UTF8 "\xB4\x70" /* U+03BC */
69
70# define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xBF\x63\x72" /* U+1E9E */
71
72# define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xDD\x72\x67\x46" /* U+FB05 */
73# define LATIN_SMALL_LIGATURE_ST_UTF8 "\xDD\x72\x67\x47" /* U+FB06 */
74
75# define HYPHEN_UTF8 "\xCA\x41\x57" /* U+2010 */
76# define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xDD /* U+D800 */
77# define BOM_UTF8_FIRST_BYTE 0xDD /* U+FEFF */
78# define BOM_UTF8_TAIL "\x73\x66\x73" /* U+FEFF */
79
80# define NBSP_NATIVE 0x41 /* U+00A0 */
81# define NBSP_UTF8 "\x80\x41" /* U+00A0 */
82
83# define DEL_NATIVE 0x07 /* U+007F */
84# define CR_NATIVE 0x0D /* U+000D */
85# define LF_NATIVE 0x15 /* U+000A */
d804860b
KW
86# define VT_NATIVE 0x0B /* U+000B */
87# define ESC_NATIVE 0x27 /* U+001B */
ad88cddb
KW
88# define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0x59 /* U+00DF */
89# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x47 /* U+00E5 */
90# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x67 /* U+00C5 */
91# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xDF /* U+00FF */
92# define MICRO_SIGN_NATIVE 0xA0 /* U+00B5 */
09cc440d 93# define MAX_PRINT_A_FOR_USE_ONLY_BY_REGCOMP_DOT_C 0xF9 /* The max code point that isPRINT_A */
ad88cddb
KW
94
95#endif /* EBCDIC 1047 */
96
97#if 'A' == 193 /* EBCDIC 037 */ \
98 && '\\' == 224 && '[' == 186 && ']' == 187 && '{' == 192 && '}' == 208 \
99 && '^' == 176 && '~' == 161 && '!' == 90 && '#' == 123 && '|' == 79 \
100 && '$' == 91 && '@' == 124 && '`' == 121
101# define LATIN_SMALL_LETTER_LONG_S_UTF8 "\x8E\x72" /* U+017F */
102
103# define COMBINING_GRAVE_ACCENT_UTF8 "\xAD\x41" /* U+0300 */
104
105# define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xB2\x67" /* U+0399 */
106# define GREEK_SMALL_LETTER_MU_UTF8 "\xB3\x6A" /* U+03BC */
107
108# define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xBF\x62\x71" /* U+1E9E */
109
110# define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xDD\x71\x66\x46" /* U+FB05 */
111# define LATIN_SMALL_LIGATURE_ST_UTF8 "\xDD\x71\x66\x47" /* U+FB06 */
112
113# define HYPHEN_UTF8 "\xCA\x41\x57" /* U+2010 */
114# define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xDD /* U+D800 */
115# define BOM_UTF8_FIRST_BYTE 0xDD /* U+FEFF */
116# define BOM_UTF8_TAIL "\x72\x65\x72" /* U+FEFF */
117
118# define NBSP_NATIVE 0x41 /* U+00A0 */
119# define NBSP_UTF8 "\x78\x41" /* U+00A0 */
120
121# define DEL_NATIVE 0x07 /* U+007F */
122# define CR_NATIVE 0x0D /* U+000D */
123# define LF_NATIVE 0x25 /* U+000A */
d804860b
KW
124# define VT_NATIVE 0x0B /* U+000B */
125# define ESC_NATIVE 0x27 /* U+001B */
ad88cddb
KW
126# define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0x59 /* U+00DF */
127# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x47 /* U+00E5 */
128# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x67 /* U+00C5 */
129# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xDF /* U+00FF */
130# define MICRO_SIGN_NATIVE 0xA0 /* U+00B5 */
09cc440d 131# define MAX_PRINT_A_FOR_USE_ONLY_BY_REGCOMP_DOT_C 0xF9 /* The max code point that isPRINT_A */
ad88cddb
KW
132
133#endif /* EBCDIC 037 */
134
135#if 'A' == 193 /* EBCDIC POSIX-BC */ \
136 && '\\' == 188 && '[' == 187 && ']' == 189 && '{' == 251 && '}' == 253 \
137 && '^' == 106 && '~' == 255 && '!' == 90 && '#' == 123 && '|' == 79 \
138 && '$' == 91 && '@' == 124 && '`' == 74
139# define LATIN_SMALL_LETTER_LONG_S_UTF8 "\x90\x75" /* U+017F */
140
141# define COMBINING_GRAVE_ACCENT_UTF8 "\xAE\x41" /* U+0300 */
142
143# define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xB2\x69" /* U+0399 */
144# define GREEK_SMALL_LETTER_MU_UTF8 "\xB3\x72" /* U+03BC */
145
146# define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xC0\x64\x74" /* U+1E9E */
147
148# define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xDC\x74\x68\x46" /* U+FB05 */
149# define LATIN_SMALL_LIGATURE_ST_UTF8 "\xDC\x74\x68\x47" /* U+FB06 */
150
151# define HYPHEN_UTF8 "\xCA\x41\x58" /* U+2010 */
152# define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xDC /* U+D800 */
153# define BOM_UTF8_FIRST_BYTE 0xDC /* U+FEFF */
154# define BOM_UTF8_TAIL "\x75\x67\x75" /* U+FEFF */
155
156# define NBSP_NATIVE 0x41 /* U+00A0 */
157# define NBSP_UTF8 "\x8A\x41" /* U+00A0 */
158
159# define DEL_NATIVE 0x07 /* U+007F */
160# define CR_NATIVE 0x0D /* U+000D */
161# define LF_NATIVE 0x15 /* U+000A */
d804860b
KW
162# define VT_NATIVE 0x0B /* U+000B */
163# define ESC_NATIVE 0x27 /* U+001B */
ad88cddb
KW
164# define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0x59 /* U+00DF */
165# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x47 /* U+00E5 */
166# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x67 /* U+00C5 */
167# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xDF /* U+00FF */
168# define MICRO_SIGN_NATIVE 0xA0 /* U+00B5 */
09cc440d 169# define MAX_PRINT_A_FOR_USE_ONLY_BY_REGCOMP_DOT_C 0xFF /* The max code point that isPRINT_A */
ad88cddb
KW
170
171#endif /* EBCDIC POSIX-BC */
525b6419 172
b35552de 173/* The number of code points not matching \pC */
ac71d2a0 174#define NON_OTHER_COUNT_FOR_USE_ONLY_BY_REGCOMP_DOT_C 120522
b35552de 175
1b0f46bf 176#endif /* H_UNICODE_CONSTANTS */
d10c72f2 177
61dad979 178/* ex: set ro: */