| 6 |
and semantics are as close as possible to those of the Perl 5 language. |
and semantics are as close as possible to those of the Perl 5 language. |
| 7 |
|
|
| 8 |
Written by Philip Hazel |
Written by Philip Hazel |
| 9 |
Copyright (c) 1997-2007 University of Cambridge |
Copyright (c) 1997-2009 University of Cambridge |
| 10 |
|
|
| 11 |
----------------------------------------------------------------------------- |
----------------------------------------------------------------------------- |
| 12 |
Redistribution and use in source and binary forms, with or without |
Redistribution and use in source and binary forms, with or without |
| 44 |
clashes with the library. */ |
clashes with the library. */ |
| 45 |
|
|
| 46 |
|
|
| 47 |
#ifdef HAVE_CONFIG_H |
#ifdef HAVE_CONFIG_H |
| 48 |
#include <config.h> |
#include "config.h" |
| 49 |
#endif |
#endif |
| 50 |
|
|
| 51 |
#include "pcre_internal.h" |
#include "pcre_internal.h" |
| 52 |
|
|
| 87 |
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, |
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, |
| 88 |
3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5 }; |
3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5 }; |
| 89 |
|
|
| 90 |
/* This table translates Unicode property names into type and code values. It |
/* Table to translate from particular type value to the general value. */ |
| 91 |
is searched by binary chop, so must be in collating sequence of name. */ |
|
| 92 |
|
const int _pcre_ucp_gentype[] = { |
| 93 |
|
ucp_C, ucp_C, ucp_C, ucp_C, ucp_C, /* Cc, Cf, Cn, Co, Cs */ |
| 94 |
|
ucp_L, ucp_L, ucp_L, ucp_L, ucp_L, /* Ll, Lu, Lm, Lo, Lt */ |
| 95 |
|
ucp_M, ucp_M, ucp_M, /* Mc, Me, Mn */ |
| 96 |
|
ucp_N, ucp_N, ucp_N, /* Nd, Nl, No */ |
| 97 |
|
ucp_P, ucp_P, ucp_P, ucp_P, ucp_P, /* Pc, Pd, Pe, Pf, Pi */ |
| 98 |
|
ucp_P, ucp_P, /* Ps, Po */ |
| 99 |
|
ucp_S, ucp_S, ucp_S, ucp_S, /* Sc, Sk, Sm, So */ |
| 100 |
|
ucp_Z, ucp_Z, ucp_Z /* Zl, Zp, Zs */ |
| 101 |
|
}; |
| 102 |
|
|
| 103 |
|
/* The pcre_utt[] table below translates Unicode property names into type and |
| 104 |
|
code values. It is searched by binary chop, so must be in collating sequence of |
| 105 |
|
name. Originally, the table contained pointers to the name strings in the first |
| 106 |
|
field of each entry. However, that leads to a large number of relocations when |
| 107 |
|
a shared library is dynamically loaded. A significant reduction is made by |
| 108 |
|
putting all the names into a single, large string and then using offsets in the |
| 109 |
|
table itself. Maintenance is more error-prone, but frequent changes to this |
| 110 |
|
data are unlikely. |
| 111 |
|
|
| 112 |
|
July 2008: There is now a script called maint/GenerateUtt.py that can be used |
| 113 |
|
to generate this data instead of maintaining it entirely by hand. |
| 114 |
|
|
| 115 |
|
The script was updated in March 2009 to generate a new EBCDIC-compliant |
| 116 |
|
version. Like all other character and string literals that are compared against |
| 117 |
|
the regular expression pattern, we must use STR_ macros instead of literal |
| 118 |
|
strings to make sure that UTF-8 support works on EBCDIC platforms. */ |
| 119 |
|
|
| 120 |
|
#define STRING_Any0 STR_A STR_n STR_y "\0" |
| 121 |
|
#define STRING_Arabic0 STR_A STR_r STR_a STR_b STR_i STR_c "\0" |
| 122 |
|
#define STRING_Armenian0 STR_A STR_r STR_m STR_e STR_n STR_i STR_a STR_n "\0" |
| 123 |
|
#define STRING_Avestan0 STR_A STR_v STR_e STR_s STR_t STR_a STR_n "\0" |
| 124 |
|
#define STRING_Balinese0 STR_B STR_a STR_l STR_i STR_n STR_e STR_s STR_e "\0" |
| 125 |
|
#define STRING_Bamum0 STR_B STR_a STR_m STR_u STR_m "\0" |
| 126 |
|
#define STRING_Bengali0 STR_B STR_e STR_n STR_g STR_a STR_l STR_i "\0" |
| 127 |
|
#define STRING_Bopomofo0 STR_B STR_o STR_p STR_o STR_m STR_o STR_f STR_o "\0" |
| 128 |
|
#define STRING_Braille0 STR_B STR_r STR_a STR_i STR_l STR_l STR_e "\0" |
| 129 |
|
#define STRING_Buginese0 STR_B STR_u STR_g STR_i STR_n STR_e STR_s STR_e "\0" |
| 130 |
|
#define STRING_Buhid0 STR_B STR_u STR_h STR_i STR_d "\0" |
| 131 |
|
#define STRING_C0 STR_C "\0" |
| 132 |
|
#define STRING_Canadian_Aboriginal0 STR_C STR_a STR_n STR_a STR_d STR_i STR_a STR_n STR_UNDERSCORE STR_A STR_b STR_o STR_r STR_i STR_g STR_i STR_n STR_a STR_l "\0" |
| 133 |
|
#define STRING_Carian0 STR_C STR_a STR_r STR_i STR_a STR_n "\0" |
| 134 |
|
#define STRING_Cc0 STR_C STR_c "\0" |
| 135 |
|
#define STRING_Cf0 STR_C STR_f "\0" |
| 136 |
|
#define STRING_Cham0 STR_C STR_h STR_a STR_m "\0" |
| 137 |
|
#define STRING_Cherokee0 STR_C STR_h STR_e STR_r STR_o STR_k STR_e STR_e "\0" |
| 138 |
|
#define STRING_Cn0 STR_C STR_n "\0" |
| 139 |
|
#define STRING_Co0 STR_C STR_o "\0" |
| 140 |
|
#define STRING_Common0 STR_C STR_o STR_m STR_m STR_o STR_n "\0" |
| 141 |
|
#define STRING_Coptic0 STR_C STR_o STR_p STR_t STR_i STR_c "\0" |
| 142 |
|
#define STRING_Cs0 STR_C STR_s "\0" |
| 143 |
|
#define STRING_Cuneiform0 STR_C STR_u STR_n STR_e STR_i STR_f STR_o STR_r STR_m "\0" |
| 144 |
|
#define STRING_Cypriot0 STR_C STR_y STR_p STR_r STR_i STR_o STR_t "\0" |
| 145 |
|
#define STRING_Cyrillic0 STR_C STR_y STR_r STR_i STR_l STR_l STR_i STR_c "\0" |
| 146 |
|
#define STRING_Deseret0 STR_D STR_e STR_s STR_e STR_r STR_e STR_t "\0" |
| 147 |
|
#define STRING_Devanagari0 STR_D STR_e STR_v STR_a STR_n STR_a STR_g STR_a STR_r STR_i "\0" |
| 148 |
|
#define STRING_Egyptian_Hieroglyphs0 STR_E STR_g STR_y STR_p STR_t STR_i STR_a STR_n STR_UNDERSCORE STR_H STR_i STR_e STR_r STR_o STR_g STR_l STR_y STR_p STR_h STR_s "\0" |
| 149 |
|
#define STRING_Ethiopic0 STR_E STR_t STR_h STR_i STR_o STR_p STR_i STR_c "\0" |
| 150 |
|
#define STRING_Georgian0 STR_G STR_e STR_o STR_r STR_g STR_i STR_a STR_n "\0" |
| 151 |
|
#define STRING_Glagolitic0 STR_G STR_l STR_a STR_g STR_o STR_l STR_i STR_t STR_i STR_c "\0" |
| 152 |
|
#define STRING_Gothic0 STR_G STR_o STR_t STR_h STR_i STR_c "\0" |
| 153 |
|
#define STRING_Greek0 STR_G STR_r STR_e STR_e STR_k "\0" |
| 154 |
|
#define STRING_Gujarati0 STR_G STR_u STR_j STR_a STR_r STR_a STR_t STR_i "\0" |
| 155 |
|
#define STRING_Gurmukhi0 STR_G STR_u STR_r STR_m STR_u STR_k STR_h STR_i "\0" |
| 156 |
|
#define STRING_Han0 STR_H STR_a STR_n "\0" |
| 157 |
|
#define STRING_Hangul0 STR_H STR_a STR_n STR_g STR_u STR_l "\0" |
| 158 |
|
#define STRING_Hanunoo0 STR_H STR_a STR_n STR_u STR_n STR_o STR_o "\0" |
| 159 |
|
#define STRING_Hebrew0 STR_H STR_e STR_b STR_r STR_e STR_w "\0" |
| 160 |
|
#define STRING_Hiragana0 STR_H STR_i STR_r STR_a STR_g STR_a STR_n STR_a "\0" |
| 161 |
|
#define STRING_Imperial_Aramaic0 STR_I STR_m STR_p STR_e STR_r STR_i STR_a STR_l STR_UNDERSCORE STR_A STR_r STR_a STR_m STR_a STR_i STR_c "\0" |
| 162 |
|
#define STRING_Inherited0 STR_I STR_n STR_h STR_e STR_r STR_i STR_t STR_e STR_d "\0" |
| 163 |
|
#define STRING_Inscriptional_Pahlavi0 STR_I STR_n STR_s STR_c STR_r STR_i STR_p STR_t STR_i STR_o STR_n STR_a STR_l STR_UNDERSCORE STR_P STR_a STR_h STR_l STR_a STR_v STR_i "\0" |
| 164 |
|
#define STRING_Inscriptional_Parthian0 STR_I STR_n STR_s STR_c STR_r STR_i STR_p STR_t STR_i STR_o STR_n STR_a STR_l STR_UNDERSCORE STR_P STR_a STR_r STR_t STR_h STR_i STR_a STR_n "\0" |
| 165 |
|
#define STRING_Javanese0 STR_J STR_a STR_v STR_a STR_n STR_e STR_s STR_e "\0" |
| 166 |
|
#define STRING_Kaithi0 STR_K STR_a STR_i STR_t STR_h STR_i "\0" |
| 167 |
|
#define STRING_Kannada0 STR_K STR_a STR_n STR_n STR_a STR_d STR_a "\0" |
| 168 |
|
#define STRING_Katakana0 STR_K STR_a STR_t STR_a STR_k STR_a STR_n STR_a "\0" |
| 169 |
|
#define STRING_Kayah_Li0 STR_K STR_a STR_y STR_a STR_h STR_UNDERSCORE STR_L STR_i "\0" |
| 170 |
|
#define STRING_Kharoshthi0 STR_K STR_h STR_a STR_r STR_o STR_s STR_h STR_t STR_h STR_i "\0" |
| 171 |
|
#define STRING_Khmer0 STR_K STR_h STR_m STR_e STR_r "\0" |
| 172 |
|
#define STRING_L0 STR_L "\0" |
| 173 |
|
#define STRING_L_AMPERSAND0 STR_L STR_AMPERSAND "\0" |
| 174 |
|
#define STRING_Lao0 STR_L STR_a STR_o "\0" |
| 175 |
|
#define STRING_Latin0 STR_L STR_a STR_t STR_i STR_n "\0" |
| 176 |
|
#define STRING_Lepcha0 STR_L STR_e STR_p STR_c STR_h STR_a "\0" |
| 177 |
|
#define STRING_Limbu0 STR_L STR_i STR_m STR_b STR_u "\0" |
| 178 |
|
#define STRING_Linear_B0 STR_L STR_i STR_n STR_e STR_a STR_r STR_UNDERSCORE STR_B "\0" |
| 179 |
|
#define STRING_Lisu0 STR_L STR_i STR_s STR_u "\0" |
| 180 |
|
#define STRING_Ll0 STR_L STR_l "\0" |
| 181 |
|
#define STRING_Lm0 STR_L STR_m "\0" |
| 182 |
|
#define STRING_Lo0 STR_L STR_o "\0" |
| 183 |
|
#define STRING_Lt0 STR_L STR_t "\0" |
| 184 |
|
#define STRING_Lu0 STR_L STR_u "\0" |
| 185 |
|
#define STRING_Lycian0 STR_L STR_y STR_c STR_i STR_a STR_n "\0" |
| 186 |
|
#define STRING_Lydian0 STR_L STR_y STR_d STR_i STR_a STR_n "\0" |
| 187 |
|
#define STRING_M0 STR_M "\0" |
| 188 |
|
#define STRING_Malayalam0 STR_M STR_a STR_l STR_a STR_y STR_a STR_l STR_a STR_m "\0" |
| 189 |
|
#define STRING_Mc0 STR_M STR_c "\0" |
| 190 |
|
#define STRING_Me0 STR_M STR_e "\0" |
| 191 |
|
#define STRING_Meetei_Mayek0 STR_M STR_e STR_e STR_t STR_e STR_i STR_UNDERSCORE STR_M STR_a STR_y STR_e STR_k "\0" |
| 192 |
|
#define STRING_Mn0 STR_M STR_n "\0" |
| 193 |
|
#define STRING_Mongolian0 STR_M STR_o STR_n STR_g STR_o STR_l STR_i STR_a STR_n "\0" |
| 194 |
|
#define STRING_Myanmar0 STR_M STR_y STR_a STR_n STR_m STR_a STR_r "\0" |
| 195 |
|
#define STRING_N0 STR_N "\0" |
| 196 |
|
#define STRING_Nd0 STR_N STR_d "\0" |
| 197 |
|
#define STRING_New_Tai_Lue0 STR_N STR_e STR_w STR_UNDERSCORE STR_T STR_a STR_i STR_UNDERSCORE STR_L STR_u STR_e "\0" |
| 198 |
|
#define STRING_Nko0 STR_N STR_k STR_o "\0" |
| 199 |
|
#define STRING_Nl0 STR_N STR_l "\0" |
| 200 |
|
#define STRING_No0 STR_N STR_o "\0" |
| 201 |
|
#define STRING_Ogham0 STR_O STR_g STR_h STR_a STR_m "\0" |
| 202 |
|
#define STRING_Ol_Chiki0 STR_O STR_l STR_UNDERSCORE STR_C STR_h STR_i STR_k STR_i "\0" |
| 203 |
|
#define STRING_Old_Italic0 STR_O STR_l STR_d STR_UNDERSCORE STR_I STR_t STR_a STR_l STR_i STR_c "\0" |
| 204 |
|
#define STRING_Old_Persian0 STR_O STR_l STR_d STR_UNDERSCORE STR_P STR_e STR_r STR_s STR_i STR_a STR_n "\0" |
| 205 |
|
#define STRING_Old_South_Arabian0 STR_O STR_l STR_d STR_UNDERSCORE STR_S STR_o STR_u STR_t STR_h STR_UNDERSCORE STR_A STR_r STR_a STR_b STR_i STR_a STR_n "\0" |
| 206 |
|
#define STRING_Old_Turkic0 STR_O STR_l STR_d STR_UNDERSCORE STR_T STR_u STR_r STR_k STR_i STR_c "\0" |
| 207 |
|
#define STRING_Oriya0 STR_O STR_r STR_i STR_y STR_a "\0" |
| 208 |
|
#define STRING_Osmanya0 STR_O STR_s STR_m STR_a STR_n STR_y STR_a "\0" |
| 209 |
|
#define STRING_P0 STR_P "\0" |
| 210 |
|
#define STRING_Pc0 STR_P STR_c "\0" |
| 211 |
|
#define STRING_Pd0 STR_P STR_d "\0" |
| 212 |
|
#define STRING_Pe0 STR_P STR_e "\0" |
| 213 |
|
#define STRING_Pf0 STR_P STR_f "\0" |
| 214 |
|
#define STRING_Phags_Pa0 STR_P STR_h STR_a STR_g STR_s STR_UNDERSCORE STR_P STR_a "\0" |
| 215 |
|
#define STRING_Phoenician0 STR_P STR_h STR_o STR_e STR_n STR_i STR_c STR_i STR_a STR_n "\0" |
| 216 |
|
#define STRING_Pi0 STR_P STR_i "\0" |
| 217 |
|
#define STRING_Po0 STR_P STR_o "\0" |
| 218 |
|
#define STRING_Ps0 STR_P STR_s "\0" |
| 219 |
|
#define STRING_Rejang0 STR_R STR_e STR_j STR_a STR_n STR_g "\0" |
| 220 |
|
#define STRING_Runic0 STR_R STR_u STR_n STR_i STR_c "\0" |
| 221 |
|
#define STRING_S0 STR_S "\0" |
| 222 |
|
#define STRING_Samaritan0 STR_S STR_a STR_m STR_a STR_r STR_i STR_t STR_a STR_n "\0" |
| 223 |
|
#define STRING_Saurashtra0 STR_S STR_a STR_u STR_r STR_a STR_s STR_h STR_t STR_r STR_a "\0" |
| 224 |
|
#define STRING_Sc0 STR_S STR_c "\0" |
| 225 |
|
#define STRING_Shavian0 STR_S STR_h STR_a STR_v STR_i STR_a STR_n "\0" |
| 226 |
|
#define STRING_Sinhala0 STR_S STR_i STR_n STR_h STR_a STR_l STR_a "\0" |
| 227 |
|
#define STRING_Sk0 STR_S STR_k "\0" |
| 228 |
|
#define STRING_Sm0 STR_S STR_m "\0" |
| 229 |
|
#define STRING_So0 STR_S STR_o "\0" |
| 230 |
|
#define STRING_Sundanese0 STR_S STR_u STR_n STR_d STR_a STR_n STR_e STR_s STR_e "\0" |
| 231 |
|
#define STRING_Syloti_Nagri0 STR_S STR_y STR_l STR_o STR_t STR_i STR_UNDERSCORE STR_N STR_a STR_g STR_r STR_i "\0" |
| 232 |
|
#define STRING_Syriac0 STR_S STR_y STR_r STR_i STR_a STR_c "\0" |
| 233 |
|
#define STRING_Tagalog0 STR_T STR_a STR_g STR_a STR_l STR_o STR_g "\0" |
| 234 |
|
#define STRING_Tagbanwa0 STR_T STR_a STR_g STR_b STR_a STR_n STR_w STR_a "\0" |
| 235 |
|
#define STRING_Tai_Le0 STR_T STR_a STR_i STR_UNDERSCORE STR_L STR_e "\0" |
| 236 |
|
#define STRING_Tai_Tham0 STR_T STR_a STR_i STR_UNDERSCORE STR_T STR_h STR_a STR_m "\0" |
| 237 |
|
#define STRING_Tai_Viet0 STR_T STR_a STR_i STR_UNDERSCORE STR_V STR_i STR_e STR_t "\0" |
| 238 |
|
#define STRING_Tamil0 STR_T STR_a STR_m STR_i STR_l "\0" |
| 239 |
|
#define STRING_Telugu0 STR_T STR_e STR_l STR_u STR_g STR_u "\0" |
| 240 |
|
#define STRING_Thaana0 STR_T STR_h STR_a STR_a STR_n STR_a "\0" |
| 241 |
|
#define STRING_Thai0 STR_T STR_h STR_a STR_i "\0" |
| 242 |
|
#define STRING_Tibetan0 STR_T STR_i STR_b STR_e STR_t STR_a STR_n "\0" |
| 243 |
|
#define STRING_Tifinagh0 STR_T STR_i STR_f STR_i STR_n STR_a STR_g STR_h "\0" |
| 244 |
|
#define STRING_Ugaritic0 STR_U STR_g STR_a STR_r STR_i STR_t STR_i STR_c "\0" |
| 245 |
|
#define STRING_Vai0 STR_V STR_a STR_i "\0" |
| 246 |
|
#define STRING_Xan0 STR_X STR_a STR_n "\0" |
| 247 |
|
#define STRING_Xps0 STR_X STR_p STR_s "\0" |
| 248 |
|
#define STRING_Xsp0 STR_X STR_s STR_p "\0" |
| 249 |
|
#define STRING_Xwd0 STR_X STR_w STR_d "\0" |
| 250 |
|
#define STRING_Yi0 STR_Y STR_i "\0" |
| 251 |
|
#define STRING_Z0 STR_Z "\0" |
| 252 |
|
#define STRING_Zl0 STR_Z STR_l "\0" |
| 253 |
|
#define STRING_Zp0 STR_Z STR_p "\0" |
| 254 |
|
#define STRING_Zs0 STR_Z STR_s "\0" |
| 255 |
|
|
| 256 |
|
const char _pcre_utt_names[] = |
| 257 |
|
STRING_Any0 |
| 258 |
|
STRING_Arabic0 |
| 259 |
|
STRING_Armenian0 |
| 260 |
|
STRING_Avestan0 |
| 261 |
|
STRING_Balinese0 |
| 262 |
|
STRING_Bamum0 |
| 263 |
|
STRING_Bengali0 |
| 264 |
|
STRING_Bopomofo0 |
| 265 |
|
STRING_Braille0 |
| 266 |
|
STRING_Buginese0 |
| 267 |
|
STRING_Buhid0 |
| 268 |
|
STRING_C0 |
| 269 |
|
STRING_Canadian_Aboriginal0 |
| 270 |
|
STRING_Carian0 |
| 271 |
|
STRING_Cc0 |
| 272 |
|
STRING_Cf0 |
| 273 |
|
STRING_Cham0 |
| 274 |
|
STRING_Cherokee0 |
| 275 |
|
STRING_Cn0 |
| 276 |
|
STRING_Co0 |
| 277 |
|
STRING_Common0 |
| 278 |
|
STRING_Coptic0 |
| 279 |
|
STRING_Cs0 |
| 280 |
|
STRING_Cuneiform0 |
| 281 |
|
STRING_Cypriot0 |
| 282 |
|
STRING_Cyrillic0 |
| 283 |
|
STRING_Deseret0 |
| 284 |
|
STRING_Devanagari0 |
| 285 |
|
STRING_Egyptian_Hieroglyphs0 |
| 286 |
|
STRING_Ethiopic0 |
| 287 |
|
STRING_Georgian0 |
| 288 |
|
STRING_Glagolitic0 |
| 289 |
|
STRING_Gothic0 |
| 290 |
|
STRING_Greek0 |
| 291 |
|
STRING_Gujarati0 |
| 292 |
|
STRING_Gurmukhi0 |
| 293 |
|
STRING_Han0 |
| 294 |
|
STRING_Hangul0 |
| 295 |
|
STRING_Hanunoo0 |
| 296 |
|
STRING_Hebrew0 |
| 297 |
|
STRING_Hiragana0 |
| 298 |
|
STRING_Imperial_Aramaic0 |
| 299 |
|
STRING_Inherited0 |
| 300 |
|
STRING_Inscriptional_Pahlavi0 |
| 301 |
|
STRING_Inscriptional_Parthian0 |
| 302 |
|
STRING_Javanese0 |
| 303 |
|
STRING_Kaithi0 |
| 304 |
|
STRING_Kannada0 |
| 305 |
|
STRING_Katakana0 |
| 306 |
|
STRING_Kayah_Li0 |
| 307 |
|
STRING_Kharoshthi0 |
| 308 |
|
STRING_Khmer0 |
| 309 |
|
STRING_L0 |
| 310 |
|
STRING_L_AMPERSAND0 |
| 311 |
|
STRING_Lao0 |
| 312 |
|
STRING_Latin0 |
| 313 |
|
STRING_Lepcha0 |
| 314 |
|
STRING_Limbu0 |
| 315 |
|
STRING_Linear_B0 |
| 316 |
|
STRING_Lisu0 |
| 317 |
|
STRING_Ll0 |
| 318 |
|
STRING_Lm0 |
| 319 |
|
STRING_Lo0 |
| 320 |
|
STRING_Lt0 |
| 321 |
|
STRING_Lu0 |
| 322 |
|
STRING_Lycian0 |
| 323 |
|
STRING_Lydian0 |
| 324 |
|
STRING_M0 |
| 325 |
|
STRING_Malayalam0 |
| 326 |
|
STRING_Mc0 |
| 327 |
|
STRING_Me0 |
| 328 |
|
STRING_Meetei_Mayek0 |
| 329 |
|
STRING_Mn0 |
| 330 |
|
STRING_Mongolian0 |
| 331 |
|
STRING_Myanmar0 |
| 332 |
|
STRING_N0 |
| 333 |
|
STRING_Nd0 |
| 334 |
|
STRING_New_Tai_Lue0 |
| 335 |
|
STRING_Nko0 |
| 336 |
|
STRING_Nl0 |
| 337 |
|
STRING_No0 |
| 338 |
|
STRING_Ogham0 |
| 339 |
|
STRING_Ol_Chiki0 |
| 340 |
|
STRING_Old_Italic0 |
| 341 |
|
STRING_Old_Persian0 |
| 342 |
|
STRING_Old_South_Arabian0 |
| 343 |
|
STRING_Old_Turkic0 |
| 344 |
|
STRING_Oriya0 |
| 345 |
|
STRING_Osmanya0 |
| 346 |
|
STRING_P0 |
| 347 |
|
STRING_Pc0 |
| 348 |
|
STRING_Pd0 |
| 349 |
|
STRING_Pe0 |
| 350 |
|
STRING_Pf0 |
| 351 |
|
STRING_Phags_Pa0 |
| 352 |
|
STRING_Phoenician0 |
| 353 |
|
STRING_Pi0 |
| 354 |
|
STRING_Po0 |
| 355 |
|
STRING_Ps0 |
| 356 |
|
STRING_Rejang0 |
| 357 |
|
STRING_Runic0 |
| 358 |
|
STRING_S0 |
| 359 |
|
STRING_Samaritan0 |
| 360 |
|
STRING_Saurashtra0 |
| 361 |
|
STRING_Sc0 |
| 362 |
|
STRING_Shavian0 |
| 363 |
|
STRING_Sinhala0 |
| 364 |
|
STRING_Sk0 |
| 365 |
|
STRING_Sm0 |
| 366 |
|
STRING_So0 |
| 367 |
|
STRING_Sundanese0 |
| 368 |
|
STRING_Syloti_Nagri0 |
| 369 |
|
STRING_Syriac0 |
| 370 |
|
STRING_Tagalog0 |
| 371 |
|
STRING_Tagbanwa0 |
| 372 |
|
STRING_Tai_Le0 |
| 373 |
|
STRING_Tai_Tham0 |
| 374 |
|
STRING_Tai_Viet0 |
| 375 |
|
STRING_Tamil0 |
| 376 |
|
STRING_Telugu0 |
| 377 |
|
STRING_Thaana0 |
| 378 |
|
STRING_Thai0 |
| 379 |
|
STRING_Tibetan0 |
| 380 |
|
STRING_Tifinagh0 |
| 381 |
|
STRING_Ugaritic0 |
| 382 |
|
STRING_Vai0 |
| 383 |
|
STRING_Xan0 |
| 384 |
|
STRING_Xps0 |
| 385 |
|
STRING_Xsp0 |
| 386 |
|
STRING_Xwd0 |
| 387 |
|
STRING_Yi0 |
| 388 |
|
STRING_Z0 |
| 389 |
|
STRING_Zl0 |
| 390 |
|
STRING_Zp0 |
| 391 |
|
STRING_Zs0; |
| 392 |
|
|
| 393 |
const ucp_type_table _pcre_utt[] = { |
const ucp_type_table _pcre_utt[] = { |
| 394 |
{ "Any", PT_ANY, 0 }, |
{ 0, PT_ANY, 0 }, |
| 395 |
{ "Arabic", PT_SC, ucp_Arabic }, |
{ 4, PT_SC, ucp_Arabic }, |
| 396 |
{ "Armenian", PT_SC, ucp_Armenian }, |
{ 11, PT_SC, ucp_Armenian }, |
| 397 |
{ "Balinese", PT_SC, ucp_Balinese }, |
{ 20, PT_SC, ucp_Avestan }, |
| 398 |
{ "Bengali", PT_SC, ucp_Bengali }, |
{ 28, PT_SC, ucp_Balinese }, |
| 399 |
{ "Bopomofo", PT_SC, ucp_Bopomofo }, |
{ 37, PT_SC, ucp_Bamum }, |
| 400 |
{ "Braille", PT_SC, ucp_Braille }, |
{ 43, PT_SC, ucp_Bengali }, |
| 401 |
{ "Buginese", PT_SC, ucp_Buginese }, |
{ 51, PT_SC, ucp_Bopomofo }, |
| 402 |
{ "Buhid", PT_SC, ucp_Buhid }, |
{ 60, PT_SC, ucp_Braille }, |
| 403 |
{ "C", PT_GC, ucp_C }, |
{ 68, PT_SC, ucp_Buginese }, |
| 404 |
{ "Canadian_Aboriginal", PT_SC, ucp_Canadian_Aboriginal }, |
{ 77, PT_SC, ucp_Buhid }, |
| 405 |
{ "Cc", PT_PC, ucp_Cc }, |
{ 83, PT_GC, ucp_C }, |
| 406 |
{ "Cf", PT_PC, ucp_Cf }, |
{ 85, PT_SC, ucp_Canadian_Aboriginal }, |
| 407 |
{ "Cherokee", PT_SC, ucp_Cherokee }, |
{ 105, PT_SC, ucp_Carian }, |
| 408 |
{ "Cn", PT_PC, ucp_Cn }, |
{ 112, PT_PC, ucp_Cc }, |
| 409 |
{ "Co", PT_PC, ucp_Co }, |
{ 115, PT_PC, ucp_Cf }, |
| 410 |
{ "Common", PT_SC, ucp_Common }, |
{ 118, PT_SC, ucp_Cham }, |
| 411 |
{ "Coptic", PT_SC, ucp_Coptic }, |
{ 123, PT_SC, ucp_Cherokee }, |
| 412 |
{ "Cs", PT_PC, ucp_Cs }, |
{ 132, PT_PC, ucp_Cn }, |
| 413 |
{ "Cuneiform", PT_SC, ucp_Cuneiform }, |
{ 135, PT_PC, ucp_Co }, |
| 414 |
{ "Cypriot", PT_SC, ucp_Cypriot }, |
{ 138, PT_SC, ucp_Common }, |
| 415 |
{ "Cyrillic", PT_SC, ucp_Cyrillic }, |
{ 145, PT_SC, ucp_Coptic }, |
| 416 |
{ "Deseret", PT_SC, ucp_Deseret }, |
{ 152, PT_PC, ucp_Cs }, |
| 417 |
{ "Devanagari", PT_SC, ucp_Devanagari }, |
{ 155, PT_SC, ucp_Cuneiform }, |
| 418 |
{ "Ethiopic", PT_SC, ucp_Ethiopic }, |
{ 165, PT_SC, ucp_Cypriot }, |
| 419 |
{ "Georgian", PT_SC, ucp_Georgian }, |
{ 173, PT_SC, ucp_Cyrillic }, |
| 420 |
{ "Glagolitic", PT_SC, ucp_Glagolitic }, |
{ 182, PT_SC, ucp_Deseret }, |
| 421 |
{ "Gothic", PT_SC, ucp_Gothic }, |
{ 190, PT_SC, ucp_Devanagari }, |
| 422 |
{ "Greek", PT_SC, ucp_Greek }, |
{ 201, PT_SC, ucp_Egyptian_Hieroglyphs }, |
| 423 |
{ "Gujarati", PT_SC, ucp_Gujarati }, |
{ 222, PT_SC, ucp_Ethiopic }, |
| 424 |
{ "Gurmukhi", PT_SC, ucp_Gurmukhi }, |
{ 231, PT_SC, ucp_Georgian }, |
| 425 |
{ "Han", PT_SC, ucp_Han }, |
{ 240, PT_SC, ucp_Glagolitic }, |
| 426 |
{ "Hangul", PT_SC, ucp_Hangul }, |
{ 251, PT_SC, ucp_Gothic }, |
| 427 |
{ "Hanunoo", PT_SC, ucp_Hanunoo }, |
{ 258, PT_SC, ucp_Greek }, |
| 428 |
{ "Hebrew", PT_SC, ucp_Hebrew }, |
{ 264, PT_SC, ucp_Gujarati }, |
| 429 |
{ "Hiragana", PT_SC, ucp_Hiragana }, |
{ 273, PT_SC, ucp_Gurmukhi }, |
| 430 |
{ "Inherited", PT_SC, ucp_Inherited }, |
{ 282, PT_SC, ucp_Han }, |
| 431 |
{ "Kannada", PT_SC, ucp_Kannada }, |
{ 286, PT_SC, ucp_Hangul }, |
| 432 |
{ "Katakana", PT_SC, ucp_Katakana }, |
{ 293, PT_SC, ucp_Hanunoo }, |
| 433 |
{ "Kharoshthi", PT_SC, ucp_Kharoshthi }, |
{ 301, PT_SC, ucp_Hebrew }, |
| 434 |
{ "Khmer", PT_SC, ucp_Khmer }, |
{ 308, PT_SC, ucp_Hiragana }, |
| 435 |
{ "L", PT_GC, ucp_L }, |
{ 317, PT_SC, ucp_Imperial_Aramaic }, |
| 436 |
{ "L&", PT_LAMP, 0 }, |
{ 334, PT_SC, ucp_Inherited }, |
| 437 |
{ "Lao", PT_SC, ucp_Lao }, |
{ 344, PT_SC, ucp_Inscriptional_Pahlavi }, |
| 438 |
{ "Latin", PT_SC, ucp_Latin }, |
{ 366, PT_SC, ucp_Inscriptional_Parthian }, |
| 439 |
{ "Limbu", PT_SC, ucp_Limbu }, |
{ 389, PT_SC, ucp_Javanese }, |
| 440 |
{ "Linear_B", PT_SC, ucp_Linear_B }, |
{ 398, PT_SC, ucp_Kaithi }, |
| 441 |
{ "Ll", PT_PC, ucp_Ll }, |
{ 405, PT_SC, ucp_Kannada }, |
| 442 |
{ "Lm", PT_PC, ucp_Lm }, |
{ 413, PT_SC, ucp_Katakana }, |
| 443 |
{ "Lo", PT_PC, ucp_Lo }, |
{ 422, PT_SC, ucp_Kayah_Li }, |
| 444 |
{ "Lt", PT_PC, ucp_Lt }, |
{ 431, PT_SC, ucp_Kharoshthi }, |
| 445 |
{ "Lu", PT_PC, ucp_Lu }, |
{ 442, PT_SC, ucp_Khmer }, |
| 446 |
{ "M", PT_GC, ucp_M }, |
{ 448, PT_GC, ucp_L }, |
| 447 |
{ "Malayalam", PT_SC, ucp_Malayalam }, |
{ 450, PT_LAMP, 0 }, |
| 448 |
{ "Mc", PT_PC, ucp_Mc }, |
{ 453, PT_SC, ucp_Lao }, |
| 449 |
{ "Me", PT_PC, ucp_Me }, |
{ 457, PT_SC, ucp_Latin }, |
| 450 |
{ "Mn", PT_PC, ucp_Mn }, |
{ 463, PT_SC, ucp_Lepcha }, |
| 451 |
{ "Mongolian", PT_SC, ucp_Mongolian }, |
{ 470, PT_SC, ucp_Limbu }, |
| 452 |
{ "Myanmar", PT_SC, ucp_Myanmar }, |
{ 476, PT_SC, ucp_Linear_B }, |
| 453 |
{ "N", PT_GC, ucp_N }, |
{ 485, PT_SC, ucp_Lisu }, |
| 454 |
{ "Nd", PT_PC, ucp_Nd }, |
{ 490, PT_PC, ucp_Ll }, |
| 455 |
{ "New_Tai_Lue", PT_SC, ucp_New_Tai_Lue }, |
{ 493, PT_PC, ucp_Lm }, |
| 456 |
{ "Nko", PT_SC, ucp_Nko }, |
{ 496, PT_PC, ucp_Lo }, |
| 457 |
{ "Nl", PT_PC, ucp_Nl }, |
{ 499, PT_PC, ucp_Lt }, |
| 458 |
{ "No", PT_PC, ucp_No }, |
{ 502, PT_PC, ucp_Lu }, |
| 459 |
{ "Ogham", PT_SC, ucp_Ogham }, |
{ 505, PT_SC, ucp_Lycian }, |
| 460 |
{ "Old_Italic", PT_SC, ucp_Old_Italic }, |
{ 512, PT_SC, ucp_Lydian }, |
| 461 |
{ "Old_Persian", PT_SC, ucp_Old_Persian }, |
{ 519, PT_GC, ucp_M }, |
| 462 |
{ "Oriya", PT_SC, ucp_Oriya }, |
{ 521, PT_SC, ucp_Malayalam }, |
| 463 |
{ "Osmanya", PT_SC, ucp_Osmanya }, |
{ 531, PT_PC, ucp_Mc }, |
| 464 |
{ "P", PT_GC, ucp_P }, |
{ 534, PT_PC, ucp_Me }, |
| 465 |
{ "Pc", PT_PC, ucp_Pc }, |
{ 537, PT_SC, ucp_Meetei_Mayek }, |
| 466 |
{ "Pd", PT_PC, ucp_Pd }, |
{ 550, PT_PC, ucp_Mn }, |
| 467 |
{ "Pe", PT_PC, ucp_Pe }, |
{ 553, PT_SC, ucp_Mongolian }, |
| 468 |
{ "Pf", PT_PC, ucp_Pf }, |
{ 563, PT_SC, ucp_Myanmar }, |
| 469 |
{ "Phags_Pa", PT_SC, ucp_Phags_Pa }, |
{ 571, PT_GC, ucp_N }, |
| 470 |
{ "Phoenician", PT_SC, ucp_Phoenician }, |
{ 573, PT_PC, ucp_Nd }, |
| 471 |
{ "Pi", PT_PC, ucp_Pi }, |
{ 576, PT_SC, ucp_New_Tai_Lue }, |
| 472 |
{ "Po", PT_PC, ucp_Po }, |
{ 588, PT_SC, ucp_Nko }, |
| 473 |
{ "Ps", PT_PC, ucp_Ps }, |
{ 592, PT_PC, ucp_Nl }, |
| 474 |
{ "Runic", PT_SC, ucp_Runic }, |
{ 595, PT_PC, ucp_No }, |
| 475 |
{ "S", PT_GC, ucp_S }, |
{ 598, PT_SC, ucp_Ogham }, |
| 476 |
{ "Sc", PT_PC, ucp_Sc }, |
{ 604, PT_SC, ucp_Ol_Chiki }, |
| 477 |
{ "Shavian", PT_SC, ucp_Shavian }, |
{ 613, PT_SC, ucp_Old_Italic }, |
| 478 |
{ "Sinhala", PT_SC, ucp_Sinhala }, |
{ 624, PT_SC, ucp_Old_Persian }, |
| 479 |
{ "Sk", PT_PC, ucp_Sk }, |
{ 636, PT_SC, ucp_Old_South_Arabian }, |
| 480 |
{ "Sm", PT_PC, ucp_Sm }, |
{ 654, PT_SC, ucp_Old_Turkic }, |
| 481 |
{ "So", PT_PC, ucp_So }, |
{ 665, PT_SC, ucp_Oriya }, |
| 482 |
{ "Syloti_Nagri", PT_SC, ucp_Syloti_Nagri }, |
{ 671, PT_SC, ucp_Osmanya }, |
| 483 |
{ "Syriac", PT_SC, ucp_Syriac }, |
{ 679, PT_GC, ucp_P }, |
| 484 |
{ "Tagalog", PT_SC, ucp_Tagalog }, |
{ 681, PT_PC, ucp_Pc }, |
| 485 |
{ "Tagbanwa", PT_SC, ucp_Tagbanwa }, |
{ 684, PT_PC, ucp_Pd }, |
| 486 |
{ "Tai_Le", PT_SC, ucp_Tai_Le }, |
{ 687, PT_PC, ucp_Pe }, |
| 487 |
{ "Tamil", PT_SC, ucp_Tamil }, |
{ 690, PT_PC, ucp_Pf }, |
| 488 |
{ "Telugu", PT_SC, ucp_Telugu }, |
{ 693, PT_SC, ucp_Phags_Pa }, |
| 489 |
{ "Thaana", PT_SC, ucp_Thaana }, |
{ 702, PT_SC, ucp_Phoenician }, |
| 490 |
{ "Thai", PT_SC, ucp_Thai }, |
{ 713, PT_PC, ucp_Pi }, |
| 491 |
{ "Tibetan", PT_SC, ucp_Tibetan }, |
{ 716, PT_PC, ucp_Po }, |
| 492 |
{ "Tifinagh", PT_SC, ucp_Tifinagh }, |
{ 719, PT_PC, ucp_Ps }, |
| 493 |
{ "Ugaritic", PT_SC, ucp_Ugaritic }, |
{ 722, PT_SC, ucp_Rejang }, |
| 494 |
{ "Yi", PT_SC, ucp_Yi }, |
{ 729, PT_SC, ucp_Runic }, |
| 495 |
{ "Z", PT_GC, ucp_Z }, |
{ 735, PT_GC, ucp_S }, |
| 496 |
{ "Zl", PT_PC, ucp_Zl }, |
{ 737, PT_SC, ucp_Samaritan }, |
| 497 |
{ "Zp", PT_PC, ucp_Zp }, |
{ 747, PT_SC, ucp_Saurashtra }, |
| 498 |
{ "Zs", PT_PC, ucp_Zs } |
{ 758, PT_PC, ucp_Sc }, |
| 499 |
|
{ 761, PT_SC, ucp_Shavian }, |
| 500 |
|
{ 769, PT_SC, ucp_Sinhala }, |
| 501 |
|
{ 777, PT_PC, ucp_Sk }, |
| 502 |
|
{ 780, PT_PC, ucp_Sm }, |
| 503 |
|
{ 783, PT_PC, ucp_So }, |
| 504 |
|
{ 786, PT_SC, ucp_Sundanese }, |
| 505 |
|
{ 796, PT_SC, ucp_Syloti_Nagri }, |
| 506 |
|
{ 809, PT_SC, ucp_Syriac }, |
| 507 |
|
{ 816, PT_SC, ucp_Tagalog }, |
| 508 |
|
{ 824, PT_SC, ucp_Tagbanwa }, |
| 509 |
|
{ 833, PT_SC, ucp_Tai_Le }, |
| 510 |
|
{ 840, PT_SC, ucp_Tai_Tham }, |
| 511 |
|
{ 849, PT_SC, ucp_Tai_Viet }, |
| 512 |
|
{ 858, PT_SC, ucp_Tamil }, |
| 513 |
|
{ 864, PT_SC, ucp_Telugu }, |
| 514 |
|
{ 871, PT_SC, ucp_Thaana }, |
| 515 |
|
{ 878, PT_SC, ucp_Thai }, |
| 516 |
|
{ 883, PT_SC, ucp_Tibetan }, |
| 517 |
|
{ 891, PT_SC, ucp_Tifinagh }, |
| 518 |
|
{ 900, PT_SC, ucp_Ugaritic }, |
| 519 |
|
{ 909, PT_SC, ucp_Vai }, |
| 520 |
|
{ 913, PT_ALNUM, 0 }, |
| 521 |
|
{ 917, PT_PXSPACE, 0 }, |
| 522 |
|
{ 921, PT_SPACE, 0 }, |
| 523 |
|
{ 925, PT_WORD, 0 }, |
| 524 |
|
{ 929, PT_SC, ucp_Yi }, |
| 525 |
|
{ 932, PT_GC, ucp_Z }, |
| 526 |
|
{ 934, PT_PC, ucp_Zl }, |
| 527 |
|
{ 937, PT_PC, ucp_Zp }, |
| 528 |
|
{ 940, PT_PC, ucp_Zs } |
| 529 |
}; |
}; |
| 530 |
|
|
| 531 |
const int _pcre_utt_size = sizeof(_pcre_utt)/sizeof(ucp_type_table); |
const int _pcre_utt_size = sizeof(_pcre_utt)/sizeof(ucp_type_table); |
Parent Directory
| 
Revision Log
| 
Patch