1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/unicharutil/src/ucdata.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,306 @@
1.4 +/*
1.5 + * Copyright 1996, 1997, 1998 Computing Research Labs,
1.6 + * New Mexico State University
1.7 + *
1.8 + * Permission is hereby granted, free of charge, to any person obtaining a
1.9 + * copy of this software and associated documentation files (the "Software"),
1.10 + * to deal in the Software without restriction, including without limitation
1.11 + * the rights to use, copy, modify, merge, publish, distribute, sublicense,
1.12 + * and/or sell copies of the Software, and to permit persons to whom the
1.13 + * Software is furnished to do so, subject to the following conditions:
1.14 + *
1.15 + * The above copyright notice and this permission notice shall be included in
1.16 + * all copies or substantial portions of the Software.
1.17 + *
1.18 + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
1.19 + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
1.20 + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
1.21 + * THE COMPUTING RESEARCH LAB OR NEW MEXICO STATE UNIVERSITY BE LIABLE FOR ANY
1.22 + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT
1.23 + * OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR
1.24 + * THE USE OR OTHER DEALINGS IN THE SOFTWARE.
1.25 + */
1.26 +#ifndef _h_ucdata
1.27 +#define _h_ucdata
1.28 +
1.29 +/*
1.30 + * $Id: ucdata.h,v 1.1 1999/01/08 00:19:12 ftang%netscape.com Exp $
1.31 + */
1.32 +
1.33 +#ifdef __cplusplus
1.34 +extern "C" {
1.35 +#endif
1.36 +
1.37 +#undef __
1.38 +#ifdef __STDC__
1.39 +#define __(x) x
1.40 +#else
1.41 +#define __(x) ()
1.42 +#endif
1.43 +
1.44 +#define UCDATA_VERSION "1.9"
1.45 +
1.46 +/**************************************************************************
1.47 + *
1.48 + * Masks and macros for character properties.
1.49 + *
1.50 + **************************************************************************/
1.51 +
1.52 +/*
1.53 + * Values that can appear in the `mask1' parameter of the ucisprop()
1.54 + * function.
1.55 + */
1.56 +#define UC_MN 0x00000001 /* Mark, Non-Spacing */
1.57 +#define UC_MC 0x00000002 /* Mark, Spacing Combining */
1.58 +#define UC_ME 0x00000004 /* Mark, Enclosing */
1.59 +#define UC_ND 0x00000008 /* Number, Decimal Digit */
1.60 +#define UC_NL 0x00000010 /* Number, Letter */
1.61 +#define UC_NO 0x00000020 /* Number, Other */
1.62 +#define UC_ZS 0x00000040 /* Separator, Space */
1.63 +#define UC_ZL 0x00000080 /* Separator, Line */
1.64 +#define UC_ZP 0x00000100 /* Separator, Paragraph */
1.65 +#define UC_CC 0x00000200 /* Other, Control */
1.66 +#define UC_CF 0x00000400 /* Other, Format */
1.67 +#define UC_OS 0x00000800 /* Other, Surrogate */
1.68 +#define UC_CO 0x00001000 /* Other, Private Use */
1.69 +#define UC_CN 0x00002000 /* Other, Not Assigned */
1.70 +#define UC_LU 0x00004000 /* Letter, Uppercase */
1.71 +#define UC_LL 0x00008000 /* Letter, Lowercase */
1.72 +#define UC_LT 0x00010000 /* Letter, Titlecase */
1.73 +#define UC_LM 0x00020000 /* Letter, Modifier */
1.74 +#define UC_LO 0x00040000 /* Letter, Other */
1.75 +#define UC_PC 0x00080000 /* Punctuation, Connector */
1.76 +#define UC_PD 0x00100000 /* Punctuation, Dash */
1.77 +#define UC_PS 0x00200000 /* Punctuation, Open */
1.78 +#define UC_PE 0x00400000 /* Punctuation, Close */
1.79 +#define UC_PO 0x00800000 /* Punctuation, Other */
1.80 +#define UC_SM 0x01000000 /* Symbol, Math */
1.81 +#define UC_SC 0x02000000 /* Symbol, Currency */
1.82 +#define UC_SK 0x04000000 /* Symbol, Modifier */
1.83 +#define UC_SO 0x08000000 /* Symbol, Other */
1.84 +#define UC_L 0x10000000 /* Left-To-Right */
1.85 +#define UC_R 0x20000000 /* Right-To-Left */
1.86 +#define UC_EN 0x40000000 /* European Number */
1.87 +#define UC_ES 0x80000000 /* European Number Separator */
1.88 +
1.89 +/*
1.90 + * Values that can appear in the `mask2' parameter of the ucisprop()
1.91 + * function.
1.92 + */
1.93 +#define UC_ET 0x00000001 /* European Number Terminator */
1.94 +#define UC_AN 0x00000002 /* Arabic Number */
1.95 +#define UC_CS 0x00000004 /* Common Number Separator */
1.96 +#define UC_B 0x00000008 /* Block Separator */
1.97 +#define UC_S 0x00000010 /* Segment Separator */
1.98 +#define UC_WS 0x00000020 /* Whitespace */
1.99 +#define UC_ON 0x00000040 /* Other Neutrals */
1.100 +/*
1.101 + * Implementation specific character properties.
1.102 + */
1.103 +#define UC_CM 0x00000080 /* Composite */
1.104 +#define UC_NB 0x00000100 /* Non-Breaking */
1.105 +#define UC_SY 0x00000200 /* Symmetric */
1.106 +#define UC_HD 0x00000400 /* Hex Digit */
1.107 +#define UC_QM 0x00000800 /* Quote Mark */
1.108 +#define UC_MR 0x00001000 /* Mirroring */
1.109 +#define UC_SS 0x00002000 /* Space, other */
1.110 +
1.111 +#define UC_CP 0x00004000 /* Defined */
1.112 +
1.113 +/*
1.114 + * Added for UnicodeData-2.1.3.
1.115 + */
1.116 +#define UC_PI 0x00008000 /* Punctuation, Initial */
1.117 +#define UC_PF 0x00010000 /* Punctuation, Final */
1.118 +
1.119 +/*
1.120 + * This is the primary function for testing to see if a character has some set
1.121 + * of properties. The macros that test for various character properties all
1.122 + * call this function with some set of masks.
1.123 + */
1.124 +extern int ucisprop __((unsigned long code, unsigned long mask1,
1.125 + unsigned long mask2));
1.126 +
1.127 +#define ucisalpha(cc) ucisprop(cc, UC_LU|UC_LL|UC_LM|UC_LO|UC_LT, 0)
1.128 +#define ucisdigit(cc) ucisprop(cc, UC_ND, 0)
1.129 +#define ucisalnum(cc) ucisprop(cc, UC_LU|UC_LL|UC_LM|UC_LO|UC_LT|UC_ND, 0)
1.130 +#define uciscntrl(cc) ucisprop(cc, UC_CC|UC_CF, 0)
1.131 +#define ucisspace(cc) ucisprop(cc, UC_ZS|UC_SS, 0)
1.132 +#define ucisblank(cc) ucisprop(cc, UC_ZS, 0)
1.133 +#define ucispunct(cc) ucisprop(cc, UC_PD|UC_PS|UC_PE|UC_PO, UC_PI|UC_PF)
1.134 +#define ucisgraph(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME|UC_ND|UC_NL|UC_NO|\
1.135 + UC_LU|UC_LL|UC_LT|UC_LM|UC_LO|UC_PC|UC_PD|\
1.136 + UC_PS|UC_PE|UC_PO|UC_SM|UC_SM|UC_SC|UC_SK|\
1.137 + UC_SO, UC_PI|UC_PF)
1.138 +#define ucisprint(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME|UC_ND|UC_NL|UC_NO|\
1.139 + UC_LU|UC_LL|UC_LT|UC_LM|UC_LO|UC_PC|UC_PD|\
1.140 + UC_PS|UC_PE|UC_PO|UC_SM|UC_SM|UC_SC|UC_SK|\
1.141 + UC_SO|UC_ZS, UC_PI|UC_PF)
1.142 +#define ucisupper(cc) ucisprop(cc, UC_LU, 0)
1.143 +#define ucislower(cc) ucisprop(cc, UC_LL, 0)
1.144 +#define ucistitle(cc) ucisprop(cc, UC_LT, 0)
1.145 +#define ucisxdigit(cc) ucisprop(cc, 0, UC_HD)
1.146 +
1.147 +#define ucisisocntrl(cc) ucisprop(cc, UC_CC, 0)
1.148 +#define ucisfmtcntrl(cc) ucisprop(cc, UC_CF, 0)
1.149 +
1.150 +#define ucissymbol(cc) ucisprop(cc, UC_SM|UC_SC|UC_SO|UC_SK, 0)
1.151 +#define ucisnumber(cc) ucisprop(cc, UC_ND|UC_NO|UC_NL, 0)
1.152 +#define ucisnonspacing(cc) ucisprop(cc, UC_MN, 0)
1.153 +#define ucisopenpunct(cc) ucisprop(cc, UC_PS, 0)
1.154 +#define ucisclosepunct(cc) ucisprop(cc, UC_PE, 0)
1.155 +#define ucisinitialpunct(cc) ucisprop(cc, 0, UC_PI)
1.156 +#define ucisfinalpunct(cc) ucisprop(cc, 0, UC_PF)
1.157 +
1.158 +#define uciscomposite(cc) ucisprop(cc, 0, UC_CM)
1.159 +#define ucishex(cc) ucisprop(cc, 0, UC_HD)
1.160 +#define ucisquote(cc) ucisprop(cc, 0, UC_QM)
1.161 +#define ucissymmetric(cc) ucisprop(cc, 0, UC_SY)
1.162 +#define ucismirroring(cc) ucisprop(cc, 0, UC_MR)
1.163 +#define ucisnonbreaking(cc) ucisprop(cc, 0, UC_NB)
1.164 +
1.165 +/*
1.166 + * Directionality macros.
1.167 + */
1.168 +#define ucisrtl(cc) ucisprop(cc, UC_R, 0)
1.169 +#define ucisltr(cc) ucisprop(cc, UC_L, 0)
1.170 +#define ucisstrong(cc) ucisprop(cc, UC_L|UC_R, 0)
1.171 +#define ucisweak(cc) ucisprop(cc, UC_EN|UC_ES, UC_ET|UC_AN|UC_CS)
1.172 +#define ucisneutral(cc) ucisprop(cc, 0, UC_B|UC_S|UC_WS|UC_ON)
1.173 +#define ucisseparator(cc) ucisprop(cc, 0, UC_B|UC_S)
1.174 +
1.175 +/*
1.176 + * Other macros inspired by John Cowan.
1.177 + */
1.178 +#define ucismark(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME, 0)
1.179 +#define ucismodif(cc) ucisprop(cc, UC_LM, 0)
1.180 +#define ucisletnum(cc) ucisprop(cc, UC_NL, 0)
1.181 +#define ucisconnect(cc) ucisprop(cc, UC_PC, 0)
1.182 +#define ucisdash(cc) ucisprop(cc, UC_PD, 0)
1.183 +#define ucismath(cc) ucisprop(cc, UC_SM, 0)
1.184 +#define uciscurrency(cc) ucisprop(cc, UC_SC, 0)
1.185 +#define ucismodifsymbol(cc) ucisprop(cc, UC_SK, 0)
1.186 +#define ucisnsmark(cc) ucisprop(cc, UC_MN, 0)
1.187 +#define ucisspmark(cc) ucisprop(cc, UC_MC, 0)
1.188 +#define ucisenclosing(cc) ucisprop(cc, UC_ME, 0)
1.189 +#define ucisprivate(cc) ucisprop(cc, UC_CO, 0)
1.190 +#define ucissurrogate(cc) ucisprop(cc, UC_OS, 0)
1.191 +#define ucislsep(cc) ucisprop(cc, UC_ZL, 0)
1.192 +#define ucispsep(cc) ucisprop(cc, UC_ZP, 0)
1.193 +
1.194 +#define ucisidentstart(cc) ucisprop(cc, UC_LU|UC_LL|UC_LT|UC_LO|UC_NL, 0)
1.195 +#define ucisidentpart(cc) ucisprop(cc, UC_LU|UC_LL|UC_LT|UC_LO|UC_NL|\
1.196 + UC_MN|UC_MC|UC_ND|UC_PC|UC_CF, 0)
1.197 +
1.198 +#define ucisdefined(cc) ucisprop(cc, 0, UC_CP)
1.199 +#define ucisundefined(cc) !ucisprop(cc, 0, UC_CP)
1.200 +
1.201 +/*
1.202 + * Other miscellaneous character property macros.
1.203 + */
1.204 +#define ucishan(cc) (((cc) >= 0x4e00 && (cc) <= 0x9fff) ||\
1.205 + ((cc) >= 0xf900 && (cc) <= 0xfaff))
1.206 +#define ucishangul(cc) ((cc) >= 0xac00 && (cc) <= 0xd7ff)
1.207 +
1.208 +/**************************************************************************
1.209 + *
1.210 + * Functions for case conversion.
1.211 + *
1.212 + **************************************************************************/
1.213 +
1.214 +extern unsigned long uctoupper __((unsigned long code));
1.215 +extern unsigned long uctolower __((unsigned long code));
1.216 +extern unsigned long uctotitle __((unsigned long code));
1.217 +
1.218 +/**************************************************************************
1.219 + *
1.220 + * Functions for getting decompositions.
1.221 + *
1.222 + **************************************************************************/
1.223 +
1.224 +/*
1.225 + * This routine determines if the code has a decomposition. If it returns 0,
1.226 + * there is no decomposition. Any other value indicates a decomposition was
1.227 + * returned.
1.228 + */
1.229 +extern int ucdecomp __((unsigned long code, unsigned long *num,
1.230 +
1.231 + unsigned long **decomp));
1.232 +
1.233 +/*
1.234 + * If the code is a Hangul syllable, this routine decomposes it into the array
1.235 + * passed. The array size should be at least 3.
1.236 + */
1.237 +extern int ucdecomp_hangul __((unsigned long code, unsigned long *num,
1.238 + unsigned long decomp[]));
1.239 +
1.240 +/**************************************************************************
1.241 + *
1.242 + * Functions for getting combining classes.
1.243 + *
1.244 + **************************************************************************/
1.245 +
1.246 +/*
1.247 + * This will return the combining class for a character to be used with the
1.248 + * Canonical Ordering algorithm.
1.249 + */
1.250 +extern unsigned long uccombining_class __((unsigned long code));
1.251 +
1.252 +/**************************************************************************
1.253 + *
1.254 + * Functions for getting numbers and digits.
1.255 + *
1.256 + **************************************************************************/
1.257 +
1.258 +struct ucnumber {
1.259 + int numerator;
1.260 + int denominator;
1.261 +};
1.262 +
1.263 +extern int ucnumber_lookup __((unsigned long code, struct ucnumber *num));
1.264 +extern int ucdigit_lookup __((unsigned long code, int *digit));
1.265 +
1.266 +/*
1.267 + * For compatibility with John Cowan's "uctype" package.
1.268 + */
1.269 +extern struct ucnumber ucgetnumber __((unsigned long code));
1.270 +extern int ucgetdigit __((unsigned long code));
1.271 +
1.272 +/**************************************************************************
1.273 + *
1.274 + * Functions library initialization and cleanup.
1.275 + *
1.276 + **************************************************************************/
1.277 +
1.278 +/*
1.279 + * Macros for specifying the data tables to be loaded for ucdata_load().
1.280 + */
1.281 +#define UCDATA_CASE 0x01
1.282 +#define UCDATA_CTYPE 0x02
1.283 +#define UCDATA_DECOMP 0x04
1.284 +#define UCDATA_CMBCL 0x08
1.285 +#define UCDATA_NUM 0x10
1.286 +
1.287 +#define UCDATA_ALL (UCDATA_CASE|UCDATA_CTYPE|UCDATA_DECOMP|\
1.288 + UCDATA_CMBCL|UCDATA_NUM)
1.289 +
1.290 +/*
1.291 + * Functions to load, unload, and reload specific data files.
1.292 + */
1.293 +extern void ucdata_load __((char *paths, int mask));
1.294 +extern void ucdata_unload __((int mask));
1.295 +extern void ucdata_reload __((char *paths, int mask));
1.296 +
1.297 +/*
1.298 + * Deprecated functions, now just compatibility macros.
1.299 + */
1.300 +#define ucdata_setup(p) ucdata_load(p, UCDATA_ALL)
1.301 +#define ucdata_cleanup() ucdata_unload(UCDATA_ALL)
1.302 +
1.303 +#undef __
1.304 +
1.305 +#ifdef __cplusplus
1.306 +}
1.307 +#endif
1.308 +
1.309 +#endif /* _h_ucdata */
