This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
uc(sharp s) is itself in very early Unicode versions
[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
4b4853d1
KW
11/* This file contains #defines for the version of Unicode being used and
12 * various Unicode code points. The values the code point macros expand to
13 * are the native Unicode code point, or all or portions of the UTF-8 encoding
14 * for the code point. In the former case, the macro name has the suffix
15 * "_NATIVE"; otherwise, the suffix "_UTF8".
61dad979 16 *
525b6419
KW
17 * The macros that have the suffix "_UTF8" may have further suffixes, as
18 * follows:
19 * "_FIRST_BYTE" if the value is just the first byte of the UTF-8
20 * representation; the value will be a numeric constant.
21 * "_TAIL" if instead it represents all but the first byte. This, and
22 * with no additional suffix are both string constants */
61dad979 23
4b4853d1
KW
24#define UNICODE_MAJOR_VERSION 8
25#define UNICODE_DOT_VERSION 0
26#define UNICODE_DOT_DOT_VERSION 0
27
76837d21 28
ad88cddb
KW
29#if 'A' == 65 /* ASCII/Latin1 */
30# define LATIN_SMALL_LETTER_LONG_S_UTF8 "\xC5\xBF" /* U+017F */
76837d21 31
ad88cddb 32# define COMBINING_GRAVE_ACCENT_UTF8 "\xCC\x80" /* U+0300 */
a78bc3c6 33
ad88cddb
KW
34# define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xCE\x99" /* U+0399 */
35# define GREEK_SMALL_LETTER_MU_UTF8 "\xCE\xBC" /* U+03BC */
f2e06375 36
ad88cddb 37# define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xE1\xBA\x9E" /* U+1E9E */
a9f50d33 38
ad88cddb
KW
39# define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xEF\xAC\x85" /* U+FB05 */
40# define LATIN_SMALL_LIGATURE_ST_UTF8 "\xEF\xAC\x86" /* U+FB06 */
1dfa4f52 41
ad88cddb
KW
42# define HYPHEN_UTF8 "\xE2\x80\x90" /* U+2010 */
43# define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xED /* U+D800 */
44# define BOM_UTF8_FIRST_BYTE 0xEF /* U+FEFF */
45# define BOM_UTF8_TAIL "\xBB\xBF" /* U+FEFF */
df758df2 46
ad88cddb
KW
47# define NBSP_NATIVE 0xA0 /* U+00A0 */
48# define NBSP_UTF8 "\xC2\xA0" /* U+00A0 */
49
50# define DEL_NATIVE 0x7F /* U+007F */
51# define CR_NATIVE 0x0D /* U+000D */
52# define LF_NATIVE 0x0A /* U+000A */
d804860b
KW
53# define VT_NATIVE 0x0B /* U+000B */
54# define ESC_NATIVE 0x1B /* U+001B */
ad88cddb
KW
55# define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0xDF /* U+00DF */
56# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0xE5 /* U+00E5 */
57# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0xC5 /* U+00C5 */
58# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xFF /* U+00FF */
59# define MICRO_SIGN_NATIVE 0xB5 /* U+00B5 */
09cc440d 60# define MAX_PRINT_A_FOR_USE_ONLY_BY_REGCOMP_DOT_C 0x7E /* The max code point that isPRINT_A */
ad88cddb
KW
61
62#endif /* ASCII/Latin1 */
63
64#if 'A' == 193 /* EBCDIC 1047 */ \
65 && '\\' == 224 && '[' == 173 && ']' == 189 && '{' == 192 && '}' == 208 \
66 && '^' == 95 && '~' == 161 && '!' == 90 && '#' == 123 && '|' == 79 \
67 && '$' == 91 && '@' == 124 && '`' == 121
68# define LATIN_SMALL_LETTER_LONG_S_UTF8 "\x8F\x73" /* U+017F */
69
70# define COMBINING_GRAVE_ACCENT_UTF8 "\xAF\x41" /* U+0300 */
71
72# define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xB3\x68" /* U+0399 */
73# define GREEK_SMALL_LETTER_MU_UTF8 "\xB4\x70" /* U+03BC */
74
75# define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xBF\x63\x72" /* U+1E9E */
76
77# define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xDD\x72\x67\x46" /* U+FB05 */
78# define LATIN_SMALL_LIGATURE_ST_UTF8 "\xDD\x72\x67\x47" /* U+FB06 */
79
80# define HYPHEN_UTF8 "\xCA\x41\x57" /* U+2010 */
81# define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xDD /* U+D800 */
82# define BOM_UTF8_FIRST_BYTE 0xDD /* U+FEFF */
83# define BOM_UTF8_TAIL "\x73\x66\x73" /* U+FEFF */
84
85# define NBSP_NATIVE 0x41 /* U+00A0 */
86# define NBSP_UTF8 "\x80\x41" /* U+00A0 */
87
88# define DEL_NATIVE 0x07 /* U+007F */
89# define CR_NATIVE 0x0D /* U+000D */
90# define LF_NATIVE 0x15 /* U+000A */
d804860b
KW
91# define VT_NATIVE 0x0B /* U+000B */
92# define ESC_NATIVE 0x27 /* U+001B */
ad88cddb
KW
93# define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0x59 /* U+00DF */
94# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x47 /* U+00E5 */
95# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x67 /* U+00C5 */
96# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xDF /* U+00FF */
97# define MICRO_SIGN_NATIVE 0xA0 /* U+00B5 */
09cc440d 98# define MAX_PRINT_A_FOR_USE_ONLY_BY_REGCOMP_DOT_C 0xF9 /* The max code point that isPRINT_A */
ad88cddb
KW
99
100#endif /* EBCDIC 1047 */
101
102#if 'A' == 193 /* EBCDIC 037 */ \
103 && '\\' == 224 && '[' == 186 && ']' == 187 && '{' == 192 && '}' == 208 \
104 && '^' == 176 && '~' == 161 && '!' == 90 && '#' == 123 && '|' == 79 \
105 && '$' == 91 && '@' == 124 && '`' == 121
106# define LATIN_SMALL_LETTER_LONG_S_UTF8 "\x8E\x72" /* U+017F */
107
108# define COMBINING_GRAVE_ACCENT_UTF8 "\xAD\x41" /* U+0300 */
109
110# define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xB2\x67" /* U+0399 */
111# define GREEK_SMALL_LETTER_MU_UTF8 "\xB3\x6A" /* U+03BC */
112
113# define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xBF\x62\x71" /* U+1E9E */
114
115# define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xDD\x71\x66\x46" /* U+FB05 */
116# define LATIN_SMALL_LIGATURE_ST_UTF8 "\xDD\x71\x66\x47" /* U+FB06 */
117
118# define HYPHEN_UTF8 "\xCA\x41\x57" /* U+2010 */
119# define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xDD /* U+D800 */
120# define BOM_UTF8_FIRST_BYTE 0xDD /* U+FEFF */
121# define BOM_UTF8_TAIL "\x72\x65\x72" /* U+FEFF */
122
123# define NBSP_NATIVE 0x41 /* U+00A0 */
124# define NBSP_UTF8 "\x78\x41" /* U+00A0 */
125
126# define DEL_NATIVE 0x07 /* U+007F */
127# define CR_NATIVE 0x0D /* U+000D */
128# define LF_NATIVE 0x25 /* U+000A */
d804860b
KW
129# define VT_NATIVE 0x0B /* U+000B */
130# define ESC_NATIVE 0x27 /* U+001B */
ad88cddb
KW
131# define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0x59 /* U+00DF */
132# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x47 /* U+00E5 */
133# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x67 /* U+00C5 */
134# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xDF /* U+00FF */
135# define MICRO_SIGN_NATIVE 0xA0 /* U+00B5 */
09cc440d 136# define MAX_PRINT_A_FOR_USE_ONLY_BY_REGCOMP_DOT_C 0xF9 /* The max code point that isPRINT_A */
ad88cddb
KW
137
138#endif /* EBCDIC 037 */
139
140#if 'A' == 193 /* EBCDIC POSIX-BC */ \
141 && '\\' == 188 && '[' == 187 && ']' == 189 && '{' == 251 && '}' == 253 \
142 && '^' == 106 && '~' == 255 && '!' == 90 && '#' == 123 && '|' == 79 \
143 && '$' == 91 && '@' == 124 && '`' == 74
144# define LATIN_SMALL_LETTER_LONG_S_UTF8 "\x90\x75" /* U+017F */
145
146# define COMBINING_GRAVE_ACCENT_UTF8 "\xAE\x41" /* U+0300 */
147
148# define GREEK_CAPITAL_LETTER_IOTA_UTF8 "\xB2\x69" /* U+0399 */
149# define GREEK_SMALL_LETTER_MU_UTF8 "\xB3\x72" /* U+03BC */
150
151# define LATIN_CAPITAL_LETTER_SHARP_S_UTF8 "\xC0\x64\x74" /* U+1E9E */
152
153# define LATIN_SMALL_LIGATURE_LONG_S_T_UTF8 "\xDC\x74\x68\x46" /* U+FB05 */
154# define LATIN_SMALL_LIGATURE_ST_UTF8 "\xDC\x74\x68\x47" /* U+FB06 */
155
156# define HYPHEN_UTF8 "\xCA\x41\x58" /* U+2010 */
157# define FIRST_SURROGATE_UTF8_FIRST_BYTE 0xDC /* U+D800 */
158# define BOM_UTF8_FIRST_BYTE 0xDC /* U+FEFF */
159# define BOM_UTF8_TAIL "\x75\x67\x75" /* U+FEFF */
160
161# define NBSP_NATIVE 0x41 /* U+00A0 */
162# define NBSP_UTF8 "\x8A\x41" /* U+00A0 */
163
164# define DEL_NATIVE 0x07 /* U+007F */
165# define CR_NATIVE 0x0D /* U+000D */
166# define LF_NATIVE 0x15 /* U+000A */
d804860b
KW
167# define VT_NATIVE 0x0B /* U+000B */
168# define ESC_NATIVE 0x27 /* U+001B */
ad88cddb
KW
169# define LATIN_SMALL_LETTER_SHARP_S_NATIVE 0x59 /* U+00DF */
170# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x47 /* U+00E5 */
171# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 0x67 /* U+00C5 */
172# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 0xDF /* U+00FF */
173# define MICRO_SIGN_NATIVE 0xA0 /* U+00B5 */
09cc440d 174# define MAX_PRINT_A_FOR_USE_ONLY_BY_REGCOMP_DOT_C 0xFF /* The max code point that isPRINT_A */
ad88cddb
KW
175
176#endif /* EBCDIC POSIX-BC */
525b6419 177
b35552de 178/* The number of code points not matching \pC */
ac71d2a0 179#define NON_OTHER_COUNT_FOR_USE_ONLY_BY_REGCOMP_DOT_C 120522
b35552de 180
1b0f46bf 181#endif /* H_UNICODE_CONSTANTS */
d10c72f2 182
61dad979 183/* ex: set ro: */