Back to index

openldap  2.4.31
ucdata.h
Go to the documentation of this file.
00001 /* $OpenLDAP$ */
00002 /* This work is part of OpenLDAP Software <http://www.openldap.org/>.
00003  *
00004  * Copyright 1998-2012 The OpenLDAP Foundation.
00005  * All rights reserved.
00006  *
00007  * Redistribution and use in source and binary forms, with or without
00008  * modification, are permitted only as authorized by the OpenLDAP
00009  * Public License.
00010  *
00011  * A copy of this license is available in file LICENSE in the
00012  * top-level directory of the distribution or, alternatively, at
00013  * <http://www.OpenLDAP.org/license.html>.
00014  */
00015 /* Copyright 2001 Computing Research Labs, New Mexico State University
00016  *
00017  * Permission is hereby granted, free of charge, to any person obtaining a
00018  * copy of this software and associated documentation files (the "Software"),
00019  * to deal in the Software without restriction, including without limitation
00020  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
00021  * and/or sell copies of the Software, and to permit persons to whom the
00022  * Software is furnished to do so, subject to the following conditions:
00023  *
00024  * The above copyright notice and this permission notice shall be included in
00025  * all copies or substantial portions of the Software.
00026  *
00027  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
00028  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
00029  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
00030  * THE COMPUTING RESEARCH LAB OR NEW MEXICO STATE UNIVERSITY BE LIABLE FOR ANY
00031  * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT
00032  * OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR
00033  * THE USE OR OTHER DEALINGS IN THE SOFTWARE.
00034  */
00035 /* $Id: ucdata.h,v 1.6 2001/01/02 18:46:20 mleisher Exp $ */
00036 
00037 #ifndef _h_ucdata
00038 #define _h_ucdata
00039 
00040 LDAP_BEGIN_DECL
00041 
00042 #define UCDATA_VERSION "2.4"
00043 
00044 /**************************************************************************
00045  *
00046  * Masks and macros for character properties.
00047  *
00048  **************************************************************************/
00049 
00050 /*
00051  * Values that can appear in the `mask1' parameter of the ucisprop()
00052  * function.
00053  */
00054 #define UC_MN 0x00000001 /* Mark, Non-Spacing          */
00055 #define UC_MC 0x00000002 /* Mark, Spacing Combining    */
00056 #define UC_ME 0x00000004 /* Mark, Enclosing            */
00057 #define UC_ND 0x00000008 /* Number, Decimal Digit      */
00058 #define UC_NL 0x00000010 /* Number, Letter             */
00059 #define UC_NO 0x00000020 /* Number, Other              */
00060 #define UC_ZS 0x00000040 /* Separator, Space           */
00061 #define UC_ZL 0x00000080 /* Separator, Line            */
00062 #define UC_ZP 0x00000100 /* Separator, Paragraph       */
00063 #define UC_CC 0x00000200 /* Other, Control             */
00064 #define UC_CF 0x00000400 /* Other, Format              */
00065 #define UC_OS 0x00000800 /* Other, Surrogate           */
00066 #define UC_CO 0x00001000 /* Other, Private Use         */
00067 #define UC_CN 0x00002000 /* Other, Not Assigned        */
00068 #define UC_LU 0x00004000 /* Letter, Uppercase          */
00069 #define UC_LL 0x00008000 /* Letter, Lowercase          */
00070 #define UC_LT 0x00010000 /* Letter, Titlecase          */
00071 #define UC_LM 0x00020000 /* Letter, Modifier           */
00072 #define UC_LO 0x00040000 /* Letter, Other              */
00073 #define UC_PC 0x00080000 /* Punctuation, Connector     */
00074 #define UC_PD 0x00100000 /* Punctuation, Dash          */
00075 #define UC_PS 0x00200000 /* Punctuation, Open          */
00076 #define UC_PE 0x00400000 /* Punctuation, Close         */
00077 #define UC_PO 0x00800000 /* Punctuation, Other         */
00078 #define UC_SM 0x01000000 /* Symbol, Math               */
00079 #define UC_SC 0x02000000 /* Symbol, Currency           */
00080 #define UC_SK 0x04000000 /* Symbol, Modifier           */
00081 #define UC_SO 0x08000000 /* Symbol, Other              */
00082 #define UC_L  0x10000000 /* Left-To-Right              */
00083 #define UC_R  0x20000000 /* Right-To-Left              */
00084 #define UC_EN 0x40000000 /* European Number            */
00085 #define UC_ES 0x80000000 /* European Number Separator  */
00086 
00087 /*
00088  * Values that can appear in the `mask2' parameter of the ucisprop()
00089  * function.
00090  */
00091 #define UC_ET 0x00000001 /* European Number Terminator */
00092 #define UC_AN 0x00000002 /* Arabic Number              */
00093 #define UC_CS 0x00000004 /* Common Number Separator    */
00094 #define UC_B  0x00000008 /* Block Separator            */
00095 #define UC_S  0x00000010 /* Segment Separator          */
00096 #define UC_WS 0x00000020 /* Whitespace                 */
00097 #define UC_ON 0x00000040 /* Other Neutrals             */
00098 /*
00099  * Implementation specific character properties.
00100  */
00101 #define UC_CM 0x00000080 /* Composite                  */
00102 #define UC_NB 0x00000100 /* Non-Breaking               */
00103 #define UC_SY 0x00000200 /* Symmetric                  */
00104 #define UC_HD 0x00000400 /* Hex Digit                  */
00105 #define UC_QM 0x00000800 /* Quote Mark                 */
00106 #define UC_MR 0x00001000 /* Mirroring                  */
00107 #define UC_SS 0x00002000 /* Space, other               */
00108 
00109 #define UC_CP 0x00004000 /* Defined                    */
00110 
00111 /*
00112  * Added for UnicodeData-2.1.3.
00113  */
00114 #define UC_PI 0x00008000 /* Punctuation, Initial       */
00115 #define UC_PF 0x00010000 /* Punctuation, Final         */
00116 
00117 /*
00118  * This is the primary function for testing to see if a character has some set
00119  * of properties.  The macros that test for various character properties all
00120  * call this function with some set of masks.
00121  */
00122 LDAP_LUNICODE_F (int)
00123 ucisprop LDAP_P((ac_uint4 code, ac_uint4 mask1, ac_uint4 mask2));
00124 
00125 #define ucisalpha(cc) ucisprop(cc, UC_LU|UC_LL|UC_LM|UC_LO|UC_LT, 0)
00126 #define ucisdigit(cc) ucisprop(cc, UC_ND, 0)
00127 #define ucisalnum(cc) ucisprop(cc, UC_LU|UC_LL|UC_LM|UC_LO|UC_LT|UC_ND, 0)
00128 #define uciscntrl(cc) ucisprop(cc, UC_CC|UC_CF, 0)
00129 #define ucisspace(cc) ucisprop(cc, UC_ZS|UC_SS, 0)
00130 #define ucisblank(cc) ucisprop(cc, UC_ZS, 0)
00131 #define ucispunct(cc) ucisprop(cc, UC_PD|UC_PS|UC_PE|UC_PO, UC_PI|UC_PF)
00132 #define ucisgraph(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME|UC_ND|UC_NL|UC_NO|\
00133                                UC_LU|UC_LL|UC_LT|UC_LM|UC_LO|UC_PC|UC_PD|\
00134                                UC_PS|UC_PE|UC_PO|UC_SM|UC_SM|UC_SC|UC_SK|\
00135                                UC_SO, UC_PI|UC_PF)
00136 #define ucisprint(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME|UC_ND|UC_NL|UC_NO|\
00137                                UC_LU|UC_LL|UC_LT|UC_LM|UC_LO|UC_PC|UC_PD|\
00138                                UC_PS|UC_PE|UC_PO|UC_SM|UC_SM|UC_SC|UC_SK|\
00139                                UC_SO|UC_ZS, UC_PI|UC_PF)
00140 #define ucisupper(cc) ucisprop(cc, UC_LU, 0)
00141 #define ucislower(cc) ucisprop(cc, UC_LL, 0)
00142 #define ucistitle(cc) ucisprop(cc, UC_LT, 0)
00143 #define ucisxdigit(cc) ucisprop(cc, 0, UC_HD)
00144 
00145 #define ucisisocntrl(cc) ucisprop(cc, UC_CC, 0)
00146 #define ucisfmtcntrl(cc) ucisprop(cc, UC_CF, 0)
00147 
00148 #define ucissymbol(cc) ucisprop(cc, UC_SM|UC_SC|UC_SO|UC_SK, 0)
00149 #define ucisnumber(cc) ucisprop(cc, UC_ND|UC_NO|UC_NL, 0)
00150 #define ucisnonspacing(cc) ucisprop(cc, UC_MN, 0)
00151 #define ucisopenpunct(cc) ucisprop(cc, UC_PS, 0)
00152 #define ucisclosepunct(cc) ucisprop(cc, UC_PE, 0)
00153 #define ucisinitialpunct(cc) ucisprop(cc, 0, UC_PI)
00154 #define ucisfinalpunct(cc) ucisprop(cc, 0, UC_PF)
00155 
00156 #define uciscomposite(cc) ucisprop(cc, 0, UC_CM)
00157 #define ucishex(cc) ucisprop(cc, 0, UC_HD)
00158 #define ucisquote(cc) ucisprop(cc, 0, UC_QM)
00159 #define ucissymmetric(cc) ucisprop(cc, 0, UC_SY)
00160 #define ucismirroring(cc) ucisprop(cc, 0, UC_MR)
00161 #define ucisnonbreaking(cc) ucisprop(cc, 0, UC_NB)
00162 
00163 /*
00164  * Directionality macros.
00165  */
00166 #define ucisrtl(cc) ucisprop(cc, UC_R, 0)
00167 #define ucisltr(cc) ucisprop(cc, UC_L, 0)
00168 #define ucisstrong(cc) ucisprop(cc, UC_L|UC_R, 0)
00169 #define ucisweak(cc) ucisprop(cc, UC_EN|UC_ES, UC_ET|UC_AN|UC_CS)
00170 #define ucisneutral(cc) ucisprop(cc, 0, UC_B|UC_S|UC_WS|UC_ON)
00171 #define ucisseparator(cc) ucisprop(cc, 0, UC_B|UC_S)
00172 
00173 /*
00174  * Other macros inspired by John Cowan.
00175  */
00176 #define ucismark(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME, 0)
00177 #define ucismodif(cc) ucisprop(cc, UC_LM, 0)
00178 #define ucisletnum(cc) ucisprop(cc, UC_NL, 0)
00179 #define ucisconnect(cc) ucisprop(cc, UC_PC, 0)
00180 #define ucisdash(cc) ucisprop(cc, UC_PD, 0)
00181 #define ucismath(cc) ucisprop(cc, UC_SM, 0)
00182 #define uciscurrency(cc) ucisprop(cc, UC_SC, 0)
00183 #define ucismodifsymbol(cc) ucisprop(cc, UC_SK, 0)
00184 #define ucisnsmark(cc) ucisprop(cc, UC_MN, 0)
00185 #define ucisspmark(cc) ucisprop(cc, UC_MC, 0)
00186 #define ucisenclosing(cc) ucisprop(cc, UC_ME, 0)
00187 #define ucisprivate(cc) ucisprop(cc, UC_CO, 0)
00188 #define ucissurrogate(cc) ucisprop(cc, UC_OS, 0)
00189 #define ucislsep(cc) ucisprop(cc, UC_ZL, 0)
00190 #define ucispsep(cc) ucisprop(cc, UC_ZP, 0)
00191 
00192 #define ucisidentstart(cc) ucisprop(cc, UC_LU|UC_LL|UC_LT|UC_LO|UC_NL, 0)
00193 #define ucisidentpart(cc) ucisprop(cc, UC_LU|UC_LL|UC_LT|UC_LO|UC_NL|\
00194                                    UC_MN|UC_MC|UC_ND|UC_PC|UC_CF, 0)
00195 
00196 #define ucisdefined(cc) ucisprop(cc, 0, UC_CP)
00197 #define ucisundefined(cc) !ucisprop(cc, 0, UC_CP)
00198 
00199 /*
00200  * Other miscellaneous character property macros.
00201  */
00202 #define ucishan(cc) (((cc) >= 0x4e00 && (cc) <= 0x9fff) ||\
00203                      ((cc) >= 0xf900 && (cc) <= 0xfaff))
00204 #define ucishangul(cc) ((cc) >= 0xac00 && (cc) <= 0xd7ff)
00205 
00206 /**************************************************************************
00207  *
00208  * Functions for case conversion.
00209  *
00210  **************************************************************************/
00211 
00212 LDAP_LUNICODE_F (ac_uint4) uctoupper LDAP_P((ac_uint4 code));
00213 LDAP_LUNICODE_F (ac_uint4) uctolower LDAP_P((ac_uint4 code));
00214 LDAP_LUNICODE_F (ac_uint4) uctotitle LDAP_P((ac_uint4 code));
00215 
00216 /**************************************************************************
00217  *
00218  * Functions for getting compositions.
00219  *
00220  **************************************************************************/
00221 
00222 /*
00223  * This routine determines if there exists a composition of node1 and node2.
00224  * If it returns 0, there is no composition.  Any other value indicates a
00225  * composition was returned in comp.
00226  */
00227 LDAP_LUNICODE_F (int) uccomp LDAP_P((ac_uint4 node1, ac_uint4 node2,
00228                     ac_uint4 *comp));
00229 
00230 /*
00231  * Does Hangul composition on the string str with length len, and returns
00232  * the length of the composed string.
00233  */
00234 LDAP_LUNICODE_F (int) uccomp_hangul LDAP_P((ac_uint4 *str, int len));
00235 
00236 /*
00237  * Does canonical composition on the string str with length len, and returns
00238  * the length of the composed string.
00239  */
00240 LDAP_LUNICODE_F (int) uccanoncomp LDAP_P((ac_uint4 *str, int len));
00241 
00242 /**************************************************************************
00243  *
00244  * Functions for getting decompositions.
00245  *
00246  **************************************************************************/
00247 
00248 /*
00249  * This routine determines if the code has a decomposition.  If it returns 0,
00250  * there is no decomposition.  Any other value indicates a decomposition was
00251  * returned.
00252  */
00253 LDAP_LUNICODE_F (int)
00254 ucdecomp LDAP_P((ac_uint4 code, ac_uint4 *num,
00255                ac_uint4 **decomp));
00256 
00257 /*
00258  * Equivalent to ucdecomp() except that it includes compatibility
00259  * decompositions.
00260  */
00261 LDAP_LUNICODE_F (int)
00262 uckdecomp LDAP_P((ac_uint4 code, ac_uint4 *num,
00263                ac_uint4 **decomp));
00264 
00265 /*
00266  * If the code is a Hangul syllable, this routine decomposes it into the array
00267  * passed.  The array size should be at least 3.
00268  */
00269 LDAP_LUNICODE_F (int)
00270 ucdecomp_hangul LDAP_P((ac_uint4 code, ac_uint4 *num,
00271                      ac_uint4 decomp[]));
00272 
00273 /*  
00274  * This routine does canonical decomposition of the string in of length
00275  * inlen, and returns the decomposed string in out with length outlen.
00276  * The memory for out is allocated by this routine. It returns the length
00277  * of the decomposed string if okay, and -1 on error.
00278  */
00279 LDAP_LUNICODE_F (int)
00280 uccanondecomp LDAP_P((const ac_uint4 *in, int inlen,
00281                    ac_uint4 **out, int *outlen, void *ctx));
00282   
00283 /*  
00284  * Equivalent to uccanondecomp() except that it includes compatibility
00285  * decompositions.
00286  */
00287 LDAP_LUNICODE_F (int)
00288 uccompatdecomp LDAP_P((const ac_uint4 *in, int inlen,
00289                    ac_uint4 **out, int *outlen, void *ctx));
00290   
00291 /**************************************************************************
00292  *
00293  * Functions for getting combining classes.
00294  *
00295  **************************************************************************/
00296 
00297 /*
00298  * This will return the combining class for a character to be used with the
00299  * Canonical Ordering algorithm.
00300  */
00301 LDAP_LUNICODE_F (ac_uint4) uccombining_class LDAP_P((ac_uint4 code));
00302 
00303 /**************************************************************************
00304  *
00305  * Functions for getting numbers and digits.
00306  *
00307  **************************************************************************/
00308 
00309 struct ucnumber {
00310     int numerator;
00311     int denominator;
00312 };
00313 
00314 LDAP_LUNICODE_F (int)
00315 ucnumber_lookup LDAP_P((ac_uint4 code, struct ucnumber *num));
00316 
00317 LDAP_LUNICODE_F (int)
00318 ucdigit_lookup LDAP_P((ac_uint4 code, int *digit));
00319 
00320 /*
00321  * For compatibility with John Cowan's "uctype" package.
00322  */
00323 LDAP_LUNICODE_F (struct ucnumber) ucgetnumber LDAP_P((ac_uint4 code));
00324 LDAP_LUNICODE_F (int) ucgetdigit LDAP_P((ac_uint4 code));
00325 
00326 /**************************************************************************
00327  *
00328  * Functions library initialization and cleanup.
00329  *
00330  **************************************************************************/
00331 
00332 /*
00333  * Macros for specifying the data tables to be loaded, unloaded, or reloaded
00334  * by the ucdata_load(), ucdata_unload(), and ucdata_reload() routines.
00335  */
00336 #define UCDATA_CASE   0x01
00337 #define UCDATA_CTYPE  0x02
00338 #define UCDATA_DECOMP 0x04
00339 #define UCDATA_CMBCL  0x08
00340 #define UCDATA_NUM    0x10
00341 #define UCDATA_COMP   0x20
00342 #define UCDATA_KDECOMP 0x40
00343 
00344 #define UCDATA_ALL (UCDATA_CASE|UCDATA_CTYPE|UCDATA_DECOMP|\
00345                     UCDATA_CMBCL|UCDATA_NUM|UCDATA_COMP|UCDATA_KDECOMP)
00346 
00347 /*
00348  * Functions to load, unload, and reload specific data files.
00349  */
00350 LDAP_LUNICODE_F (int) ucdata_load LDAP_P((char *paths, int mask));
00351 LDAP_LUNICODE_F (void) ucdata_unload LDAP_P((int mask));
00352 LDAP_LUNICODE_F (int) ucdata_reload LDAP_P((char *paths, int mask));
00353 
00354 #ifdef UCDATA_DEPRECATED
00355 /*
00356  * Deprecated functions, now just compatibility macros.
00357  */
00358 #define ucdata_setup(p) ucdata_load(p, UCDATA_ALL)
00359 #define ucdata_cleanup() ucdata_unload(UCDATA_ALL)
00360 #endif
00361 
00362 LDAP_END_DECL
00363 
00364 #endif /* _h_ucdata */