1 /* $NetBSD: ucdata.h,v 1.3 2021/08/14 16:14:57 christos Exp $ */ 2 3 /* $OpenLDAP$ */ 4 /* This work is part of OpenLDAP Software <http://www.openldap.org/>. 5 * 6 * Copyright 1998-2021 The OpenLDAP Foundation. 7 * All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted only as authorized by the OpenLDAP 11 * Public License. 12 * 13 * A copy of this license is available in file LICENSE in the 14 * top-level directory of the distribution or, alternatively, at 15 * <http://www.OpenLDAP.org/license.html>. 16 */ 17 /* Copyright 2001 Computing Research Labs, New Mexico State University 18 * 19 * Permission is hereby granted, free of charge, to any person obtaining a 20 * copy of this software and associated documentation files (the "Software"), 21 * to deal in the Software without restriction, including without limitation 22 * the rights to use, copy, modify, merge, publish, distribute, sublicense, 23 * and/or sell copies of the Software, and to permit persons to whom the 24 * Software is furnished to do so, subject to the following conditions: 25 * 26 * The above copyright notice and this permission notice shall be included in 27 * all copies or substantial portions of the Software. 28 * 29 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 30 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 31 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL 32 * THE COMPUTING RESEARCH LAB OR NEW MEXICO STATE UNIVERSITY BE LIABLE FOR ANY 33 * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT 34 * OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR 35 * THE USE OR OTHER DEALINGS IN THE SOFTWARE. 36 */ 37 /* Id: ucdata.h,v 1.6 2001/01/02 18:46:20 mleisher Exp */ 38 39 #ifndef _h_ucdata 40 #define _h_ucdata 41 42 LDAP_BEGIN_DECL 43 44 #define UCDATA_VERSION "2.4" 45 46 /************************************************************************** 47 * 48 * Masks and macros for character properties. 49 * 50 **************************************************************************/ 51 52 /* 53 * Values that can appear in the `mask1' parameter of the ucisprop() 54 * function. 55 */ 56 #define UC_MN 0x00000001 /* Mark, Non-Spacing */ 57 #define UC_MC 0x00000002 /* Mark, Spacing Combining */ 58 #define UC_ME 0x00000004 /* Mark, Enclosing */ 59 #define UC_ND 0x00000008 /* Number, Decimal Digit */ 60 #define UC_NL 0x00000010 /* Number, Letter */ 61 #define UC_NO 0x00000020 /* Number, Other */ 62 #define UC_ZS 0x00000040 /* Separator, Space */ 63 #define UC_ZL 0x00000080 /* Separator, Line */ 64 #define UC_ZP 0x00000100 /* Separator, Paragraph */ 65 #define UC_CC 0x00000200 /* Other, Control */ 66 #define UC_CF 0x00000400 /* Other, Format */ 67 #define UC_OS 0x00000800 /* Other, Surrogate */ 68 #define UC_CO 0x00001000 /* Other, Private Use */ 69 #define UC_CN 0x00002000 /* Other, Not Assigned */ 70 #define UC_LU 0x00004000 /* Letter, Uppercase */ 71 #define UC_LL 0x00008000 /* Letter, Lowercase */ 72 #define UC_LT 0x00010000 /* Letter, Titlecase */ 73 #define UC_LM 0x00020000 /* Letter, Modifier */ 74 #define UC_LO 0x00040000 /* Letter, Other */ 75 #define UC_PC 0x00080000 /* Punctuation, Connector */ 76 #define UC_PD 0x00100000 /* Punctuation, Dash */ 77 #define UC_PS 0x00200000 /* Punctuation, Open */ 78 #define UC_PE 0x00400000 /* Punctuation, Close */ 79 #define UC_PO 0x00800000 /* Punctuation, Other */ 80 #define UC_SM 0x01000000 /* Symbol, Math */ 81 #define UC_SC 0x02000000 /* Symbol, Currency */ 82 #define UC_SK 0x04000000 /* Symbol, Modifier */ 83 #define UC_SO 0x08000000 /* Symbol, Other */ 84 #define UC_L 0x10000000 /* Left-To-Right */ 85 #define UC_R 0x20000000 /* Right-To-Left */ 86 #define UC_EN 0x40000000 /* European Number */ 87 #define UC_ES 0x80000000 /* European Number Separator */ 88 89 /* 90 * Values that can appear in the `mask2' parameter of the ucisprop() 91 * function. 92 */ 93 #define UC_ET 0x00000001 /* European Number Terminator */ 94 #define UC_AN 0x00000002 /* Arabic Number */ 95 #define UC_CS 0x00000004 /* Common Number Separator */ 96 #define UC_B 0x00000008 /* Block Separator */ 97 #define UC_S 0x00000010 /* Segment Separator */ 98 #define UC_WS 0x00000020 /* Whitespace */ 99 #define UC_ON 0x00000040 /* Other Neutrals */ 100 /* 101 * Implementation specific character properties. 102 */ 103 #define UC_CM 0x00000080 /* Composite */ 104 #define UC_NB 0x00000100 /* Non-Breaking */ 105 #define UC_SY 0x00000200 /* Symmetric */ 106 #define UC_HD 0x00000400 /* Hex Digit */ 107 #define UC_QM 0x00000800 /* Quote Mark */ 108 #define UC_MR 0x00001000 /* Mirroring */ 109 #define UC_SS 0x00002000 /* Space, other */ 110 111 #define UC_CP 0x00004000 /* Defined */ 112 113 /* 114 * Added for UnicodeData-2.1.3. 115 */ 116 #define UC_PI 0x00008000 /* Punctuation, Initial */ 117 #define UC_PF 0x00010000 /* Punctuation, Final */ 118 119 /* 120 * This is the primary function for testing to see if a character has some set 121 * of properties. The macros that test for various character properties all 122 * call this function with some set of masks. 123 */ 124 LDAP_LUNICODE_F (int) 125 ucisprop LDAP_P((ac_uint4 code, ac_uint4 mask1, ac_uint4 mask2)); 126 127 #define ucisalpha(cc) ucisprop(cc, UC_LU|UC_LL|UC_LM|UC_LO|UC_LT, 0) 128 #define ucisdigit(cc) ucisprop(cc, UC_ND, 0) 129 #define ucisalnum(cc) ucisprop(cc, UC_LU|UC_LL|UC_LM|UC_LO|UC_LT|UC_ND, 0) 130 #define uciscntrl(cc) ucisprop(cc, UC_CC|UC_CF, 0) 131 #define ucisspace(cc) ucisprop(cc, UC_ZS|UC_SS, 0) 132 #define ucisblank(cc) ucisprop(cc, UC_ZS, 0) 133 #define ucispunct(cc) ucisprop(cc, UC_PD|UC_PS|UC_PE|UC_PO, UC_PI|UC_PF) 134 #define ucisgraph(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME|UC_ND|UC_NL|UC_NO|\ 135 UC_LU|UC_LL|UC_LT|UC_LM|UC_LO|UC_PC|UC_PD|\ 136 UC_PS|UC_PE|UC_PO|UC_SM|UC_SM|UC_SC|UC_SK|\ 137 UC_SO, UC_PI|UC_PF) 138 #define ucisprint(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME|UC_ND|UC_NL|UC_NO|\ 139 UC_LU|UC_LL|UC_LT|UC_LM|UC_LO|UC_PC|UC_PD|\ 140 UC_PS|UC_PE|UC_PO|UC_SM|UC_SM|UC_SC|UC_SK|\ 141 UC_SO|UC_ZS, UC_PI|UC_PF) 142 #define ucisupper(cc) ucisprop(cc, UC_LU, 0) 143 #define ucislower(cc) ucisprop(cc, UC_LL, 0) 144 #define ucistitle(cc) ucisprop(cc, UC_LT, 0) 145 #define ucisxdigit(cc) ucisprop(cc, 0, UC_HD) 146 147 #define ucisisocntrl(cc) ucisprop(cc, UC_CC, 0) 148 #define ucisfmtcntrl(cc) ucisprop(cc, UC_CF, 0) 149 150 #define ucissymbol(cc) ucisprop(cc, UC_SM|UC_SC|UC_SO|UC_SK, 0) 151 #define ucisnumber(cc) ucisprop(cc, UC_ND|UC_NO|UC_NL, 0) 152 #define ucisnonspacing(cc) ucisprop(cc, UC_MN, 0) 153 #define ucisopenpunct(cc) ucisprop(cc, UC_PS, 0) 154 #define ucisclosepunct(cc) ucisprop(cc, UC_PE, 0) 155 #define ucisinitialpunct(cc) ucisprop(cc, 0, UC_PI) 156 #define ucisfinalpunct(cc) ucisprop(cc, 0, UC_PF) 157 158 #define uciscomposite(cc) ucisprop(cc, 0, UC_CM) 159 #define ucishex(cc) ucisprop(cc, 0, UC_HD) 160 #define ucisquote(cc) ucisprop(cc, 0, UC_QM) 161 #define ucissymmetric(cc) ucisprop(cc, 0, UC_SY) 162 #define ucismirroring(cc) ucisprop(cc, 0, UC_MR) 163 #define ucisnonbreaking(cc) ucisprop(cc, 0, UC_NB) 164 165 /* 166 * Directionality macros. 167 */ 168 #define ucisrtl(cc) ucisprop(cc, UC_R, 0) 169 #define ucisltr(cc) ucisprop(cc, UC_L, 0) 170 #define ucisstrong(cc) ucisprop(cc, UC_L|UC_R, 0) 171 #define ucisweak(cc) ucisprop(cc, UC_EN|UC_ES, UC_ET|UC_AN|UC_CS) 172 #define ucisneutral(cc) ucisprop(cc, 0, UC_B|UC_S|UC_WS|UC_ON) 173 #define ucisseparator(cc) ucisprop(cc, 0, UC_B|UC_S) 174 175 /* 176 * Other macros inspired by John Cowan. 177 */ 178 #define ucismark(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME, 0) 179 #define ucismodif(cc) ucisprop(cc, UC_LM, 0) 180 #define ucisletnum(cc) ucisprop(cc, UC_NL, 0) 181 #define ucisconnect(cc) ucisprop(cc, UC_PC, 0) 182 #define ucisdash(cc) ucisprop(cc, UC_PD, 0) 183 #define ucismath(cc) ucisprop(cc, UC_SM, 0) 184 #define uciscurrency(cc) ucisprop(cc, UC_SC, 0) 185 #define ucismodifsymbol(cc) ucisprop(cc, UC_SK, 0) 186 #define ucisnsmark(cc) ucisprop(cc, UC_MN, 0) 187 #define ucisspmark(cc) ucisprop(cc, UC_MC, 0) 188 #define ucisenclosing(cc) ucisprop(cc, UC_ME, 0) 189 #define ucisprivate(cc) ucisprop(cc, UC_CO, 0) 190 #define ucissurrogate(cc) ucisprop(cc, UC_OS, 0) 191 #define ucislsep(cc) ucisprop(cc, UC_ZL, 0) 192 #define ucispsep(cc) ucisprop(cc, UC_ZP, 0) 193 194 #define ucisidentstart(cc) ucisprop(cc, UC_LU|UC_LL|UC_LT|UC_LO|UC_NL, 0) 195 #define ucisidentpart(cc) ucisprop(cc, UC_LU|UC_LL|UC_LT|UC_LO|UC_NL|\ 196 UC_MN|UC_MC|UC_ND|UC_PC|UC_CF, 0) 197 198 #define ucisdefined(cc) ucisprop(cc, 0, UC_CP) 199 #define ucisundefined(cc) !ucisprop(cc, 0, UC_CP) 200 201 /* 202 * Other miscellaneous character property macros. 203 */ 204 #define ucishan(cc) (((cc) >= 0x4e00 && (cc) <= 0x9fff) ||\ 205 ((cc) >= 0xf900 && (cc) <= 0xfaff)) 206 #define ucishangul(cc) ((cc) >= 0xac00 && (cc) <= 0xd7ff) 207 208 /************************************************************************** 209 * 210 * Functions for case conversion. 211 * 212 **************************************************************************/ 213 214 LDAP_LUNICODE_F (ac_uint4) uctoupper LDAP_P((ac_uint4 code)); 215 LDAP_LUNICODE_F (ac_uint4) uctolower LDAP_P((ac_uint4 code)); 216 LDAP_LUNICODE_F (ac_uint4) uctotitle LDAP_P((ac_uint4 code)); 217 218 /************************************************************************** 219 * 220 * Functions for getting compositions. 221 * 222 **************************************************************************/ 223 224 /* 225 * This routine determines if there exists a composition of node1 and node2. 226 * If it returns 0, there is no composition. Any other value indicates a 227 * composition was returned in comp. 228 */ 229 LDAP_LUNICODE_F (int) uccomp LDAP_P((ac_uint4 node1, ac_uint4 node2, 230 ac_uint4 *comp)); 231 232 /* 233 * Does Hangul composition on the string str with length len, and returns 234 * the length of the composed string. 235 */ 236 LDAP_LUNICODE_F (int) uccomp_hangul LDAP_P((ac_uint4 *str, int len)); 237 238 /* 239 * Does canonical composition on the string str with length len, and returns 240 * the length of the composed string. 241 */ 242 LDAP_LUNICODE_F (int) uccanoncomp LDAP_P((ac_uint4 *str, int len)); 243 244 /************************************************************************** 245 * 246 * Functions for getting decompositions. 247 * 248 **************************************************************************/ 249 250 /* 251 * This routine determines if the code has a decomposition. If it returns 0, 252 * there is no decomposition. Any other value indicates a decomposition was 253 * returned. 254 */ 255 LDAP_LUNICODE_F (int) 256 ucdecomp LDAP_P((ac_uint4 code, ac_uint4 *num, 257 ac_uint4 **decomp)); 258 259 /* 260 * Equivalent to ucdecomp() except that it includes compatibility 261 * decompositions. 262 */ 263 LDAP_LUNICODE_F (int) 264 uckdecomp LDAP_P((ac_uint4 code, ac_uint4 *num, 265 ac_uint4 **decomp)); 266 267 /* 268 * If the code is a Hangul syllable, this routine decomposes it into the array 269 * passed. The array size should be at least 3. 270 */ 271 LDAP_LUNICODE_F (int) 272 ucdecomp_hangul LDAP_P((ac_uint4 code, ac_uint4 *num, 273 ac_uint4 decomp[])); 274 275 /* 276 * This routine does canonical decomposition of the string in of length 277 * inlen, and returns the decomposed string in out with length outlen. 278 * The memory for out is allocated by this routine. It returns the length 279 * of the decomposed string if okay, and -1 on error. 280 */ 281 LDAP_LUNICODE_F (int) 282 uccanondecomp LDAP_P((const ac_uint4 *in, int inlen, 283 ac_uint4 **out, int *outlen, void *ctx)); 284 285 /* 286 * Equivalent to uccanondecomp() except that it includes compatibility 287 * decompositions. 288 */ 289 LDAP_LUNICODE_F (int) 290 uccompatdecomp LDAP_P((const ac_uint4 *in, int inlen, 291 ac_uint4 **out, int *outlen, void *ctx)); 292 293 /************************************************************************** 294 * 295 * Functions for getting combining classes. 296 * 297 **************************************************************************/ 298 299 /* 300 * This will return the combining class for a character to be used with the 301 * Canonical Ordering algorithm. 302 */ 303 LDAP_LUNICODE_F (ac_uint4) uccombining_class LDAP_P((ac_uint4 code)); 304 305 /************************************************************************** 306 * 307 * Functions for getting numbers and digits. 308 * 309 **************************************************************************/ 310 311 struct ucnumber { 312 int numerator; 313 int denominator; 314 }; 315 316 LDAP_LUNICODE_F (int) 317 ucnumber_lookup LDAP_P((ac_uint4 code, struct ucnumber *num)); 318 319 LDAP_LUNICODE_F (int) 320 ucdigit_lookup LDAP_P((ac_uint4 code, int *digit)); 321 322 /* 323 * For compatibility with John Cowan's "uctype" package. 324 */ 325 LDAP_LUNICODE_F (struct ucnumber) ucgetnumber LDAP_P((ac_uint4 code)); 326 LDAP_LUNICODE_F (int) ucgetdigit LDAP_P((ac_uint4 code)); 327 328 /************************************************************************** 329 * 330 * Functions library initialization and cleanup. 331 * 332 **************************************************************************/ 333 334 /* 335 * Macros for specifying the data tables to be loaded, unloaded, or reloaded 336 * by the ucdata_load(), ucdata_unload(), and ucdata_reload() routines. 337 */ 338 #define UCDATA_CASE 0x01 339 #define UCDATA_CTYPE 0x02 340 #define UCDATA_DECOMP 0x04 341 #define UCDATA_CMBCL 0x08 342 #define UCDATA_NUM 0x10 343 #define UCDATA_COMP 0x20 344 #define UCDATA_KDECOMP 0x40 345 346 #define UCDATA_ALL (UCDATA_CASE|UCDATA_CTYPE|UCDATA_DECOMP|\ 347 UCDATA_CMBCL|UCDATA_NUM|UCDATA_COMP|UCDATA_KDECOMP) 348 349 /* 350 * Functions to load, unload, and reload specific data files. 351 */ 352 LDAP_LUNICODE_F (int) ucdata_load LDAP_P((char *paths, int mask)); 353 LDAP_LUNICODE_F (void) ucdata_unload LDAP_P((int mask)); 354 LDAP_LUNICODE_F (int) ucdata_reload LDAP_P((char *paths, int mask)); 355 356 #ifdef UCDATA_DEPRECATED 357 /* 358 * Deprecated functions, now just compatibility macros. 359 */ 360 #define ucdata_setup(p) ucdata_load(p, UCDATA_ALL) 361 #define ucdata_cleanup() ucdata_unload(UCDATA_ALL) 362 #endif 363 364 LDAP_END_DECL 365 366 #endif /* _h_ucdata */ 367