michael@0: /* michael@0: * Copyright 1996, 1997, 1998 Computing Research Labs, michael@0: * New Mexico State University michael@0: * michael@0: * Permission is hereby granted, free of charge, to any person obtaining a michael@0: * copy of this software and associated documentation files (the "Software"), michael@0: * to deal in the Software without restriction, including without limitation michael@0: * the rights to use, copy, modify, merge, publish, distribute, sublicense, michael@0: * and/or sell copies of the Software, and to permit persons to whom the michael@0: * Software is furnished to do so, subject to the following conditions: michael@0: * michael@0: * The above copyright notice and this permission notice shall be included in michael@0: * all copies or substantial portions of the Software. michael@0: * michael@0: * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR michael@0: * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, michael@0: * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL michael@0: * THE COMPUTING RESEARCH LAB OR NEW MEXICO STATE UNIVERSITY BE LIABLE FOR ANY michael@0: * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT michael@0: * OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR michael@0: * THE USE OR OTHER DEALINGS IN THE SOFTWARE. michael@0: */ michael@0: #ifndef _h_ucdata michael@0: #define _h_ucdata michael@0: michael@0: /* michael@0: * $Id: ucdata.h,v 1.1 1999/01/08 00:19:12 ftang%netscape.com Exp $ michael@0: */ michael@0: michael@0: #ifdef __cplusplus michael@0: extern "C" { michael@0: #endif michael@0: michael@0: #undef __ michael@0: #ifdef __STDC__ michael@0: #define __(x) x michael@0: #else michael@0: #define __(x) () michael@0: #endif michael@0: michael@0: #define UCDATA_VERSION "1.9" michael@0: michael@0: /************************************************************************** michael@0: * michael@0: * Masks and macros for character properties. michael@0: * michael@0: **************************************************************************/ michael@0: michael@0: /* michael@0: * Values that can appear in the `mask1' parameter of the ucisprop() michael@0: * function. michael@0: */ michael@0: #define UC_MN 0x00000001 /* Mark, Non-Spacing */ michael@0: #define UC_MC 0x00000002 /* Mark, Spacing Combining */ michael@0: #define UC_ME 0x00000004 /* Mark, Enclosing */ michael@0: #define UC_ND 0x00000008 /* Number, Decimal Digit */ michael@0: #define UC_NL 0x00000010 /* Number, Letter */ michael@0: #define UC_NO 0x00000020 /* Number, Other */ michael@0: #define UC_ZS 0x00000040 /* Separator, Space */ michael@0: #define UC_ZL 0x00000080 /* Separator, Line */ michael@0: #define UC_ZP 0x00000100 /* Separator, Paragraph */ michael@0: #define UC_CC 0x00000200 /* Other, Control */ michael@0: #define UC_CF 0x00000400 /* Other, Format */ michael@0: #define UC_OS 0x00000800 /* Other, Surrogate */ michael@0: #define UC_CO 0x00001000 /* Other, Private Use */ michael@0: #define UC_CN 0x00002000 /* Other, Not Assigned */ michael@0: #define UC_LU 0x00004000 /* Letter, Uppercase */ michael@0: #define UC_LL 0x00008000 /* Letter, Lowercase */ michael@0: #define UC_LT 0x00010000 /* Letter, Titlecase */ michael@0: #define UC_LM 0x00020000 /* Letter, Modifier */ michael@0: #define UC_LO 0x00040000 /* Letter, Other */ michael@0: #define UC_PC 0x00080000 /* Punctuation, Connector */ michael@0: #define UC_PD 0x00100000 /* Punctuation, Dash */ michael@0: #define UC_PS 0x00200000 /* Punctuation, Open */ michael@0: #define UC_PE 0x00400000 /* Punctuation, Close */ michael@0: #define UC_PO 0x00800000 /* Punctuation, Other */ michael@0: #define UC_SM 0x01000000 /* Symbol, Math */ michael@0: #define UC_SC 0x02000000 /* Symbol, Currency */ michael@0: #define UC_SK 0x04000000 /* Symbol, Modifier */ michael@0: #define UC_SO 0x08000000 /* Symbol, Other */ michael@0: #define UC_L 0x10000000 /* Left-To-Right */ michael@0: #define UC_R 0x20000000 /* Right-To-Left */ michael@0: #define UC_EN 0x40000000 /* European Number */ michael@0: #define UC_ES 0x80000000 /* European Number Separator */ michael@0: michael@0: /* michael@0: * Values that can appear in the `mask2' parameter of the ucisprop() michael@0: * function. michael@0: */ michael@0: #define UC_ET 0x00000001 /* European Number Terminator */ michael@0: #define UC_AN 0x00000002 /* Arabic Number */ michael@0: #define UC_CS 0x00000004 /* Common Number Separator */ michael@0: #define UC_B 0x00000008 /* Block Separator */ michael@0: #define UC_S 0x00000010 /* Segment Separator */ michael@0: #define UC_WS 0x00000020 /* Whitespace */ michael@0: #define UC_ON 0x00000040 /* Other Neutrals */ michael@0: /* michael@0: * Implementation specific character properties. michael@0: */ michael@0: #define UC_CM 0x00000080 /* Composite */ michael@0: #define UC_NB 0x00000100 /* Non-Breaking */ michael@0: #define UC_SY 0x00000200 /* Symmetric */ michael@0: #define UC_HD 0x00000400 /* Hex Digit */ michael@0: #define UC_QM 0x00000800 /* Quote Mark */ michael@0: #define UC_MR 0x00001000 /* Mirroring */ michael@0: #define UC_SS 0x00002000 /* Space, other */ michael@0: michael@0: #define UC_CP 0x00004000 /* Defined */ michael@0: michael@0: /* michael@0: * Added for UnicodeData-2.1.3. michael@0: */ michael@0: #define UC_PI 0x00008000 /* Punctuation, Initial */ michael@0: #define UC_PF 0x00010000 /* Punctuation, Final */ michael@0: michael@0: /* michael@0: * This is the primary function for testing to see if a character has some set michael@0: * of properties. The macros that test for various character properties all michael@0: * call this function with some set of masks. michael@0: */ michael@0: extern int ucisprop __((unsigned long code, unsigned long mask1, michael@0: unsigned long mask2)); michael@0: michael@0: #define ucisalpha(cc) ucisprop(cc, UC_LU|UC_LL|UC_LM|UC_LO|UC_LT, 0) michael@0: #define ucisdigit(cc) ucisprop(cc, UC_ND, 0) michael@0: #define ucisalnum(cc) ucisprop(cc, UC_LU|UC_LL|UC_LM|UC_LO|UC_LT|UC_ND, 0) michael@0: #define uciscntrl(cc) ucisprop(cc, UC_CC|UC_CF, 0) michael@0: #define ucisspace(cc) ucisprop(cc, UC_ZS|UC_SS, 0) michael@0: #define ucisblank(cc) ucisprop(cc, UC_ZS, 0) michael@0: #define ucispunct(cc) ucisprop(cc, UC_PD|UC_PS|UC_PE|UC_PO, UC_PI|UC_PF) michael@0: #define ucisgraph(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME|UC_ND|UC_NL|UC_NO|\ michael@0: UC_LU|UC_LL|UC_LT|UC_LM|UC_LO|UC_PC|UC_PD|\ michael@0: UC_PS|UC_PE|UC_PO|UC_SM|UC_SM|UC_SC|UC_SK|\ michael@0: UC_SO, UC_PI|UC_PF) michael@0: #define ucisprint(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME|UC_ND|UC_NL|UC_NO|\ michael@0: UC_LU|UC_LL|UC_LT|UC_LM|UC_LO|UC_PC|UC_PD|\ michael@0: UC_PS|UC_PE|UC_PO|UC_SM|UC_SM|UC_SC|UC_SK|\ michael@0: UC_SO|UC_ZS, UC_PI|UC_PF) michael@0: #define ucisupper(cc) ucisprop(cc, UC_LU, 0) michael@0: #define ucislower(cc) ucisprop(cc, UC_LL, 0) michael@0: #define ucistitle(cc) ucisprop(cc, UC_LT, 0) michael@0: #define ucisxdigit(cc) ucisprop(cc, 0, UC_HD) michael@0: michael@0: #define ucisisocntrl(cc) ucisprop(cc, UC_CC, 0) michael@0: #define ucisfmtcntrl(cc) ucisprop(cc, UC_CF, 0) michael@0: michael@0: #define ucissymbol(cc) ucisprop(cc, UC_SM|UC_SC|UC_SO|UC_SK, 0) michael@0: #define ucisnumber(cc) ucisprop(cc, UC_ND|UC_NO|UC_NL, 0) michael@0: #define ucisnonspacing(cc) ucisprop(cc, UC_MN, 0) michael@0: #define ucisopenpunct(cc) ucisprop(cc, UC_PS, 0) michael@0: #define ucisclosepunct(cc) ucisprop(cc, UC_PE, 0) michael@0: #define ucisinitialpunct(cc) ucisprop(cc, 0, UC_PI) michael@0: #define ucisfinalpunct(cc) ucisprop(cc, 0, UC_PF) michael@0: michael@0: #define uciscomposite(cc) ucisprop(cc, 0, UC_CM) michael@0: #define ucishex(cc) ucisprop(cc, 0, UC_HD) michael@0: #define ucisquote(cc) ucisprop(cc, 0, UC_QM) michael@0: #define ucissymmetric(cc) ucisprop(cc, 0, UC_SY) michael@0: #define ucismirroring(cc) ucisprop(cc, 0, UC_MR) michael@0: #define ucisnonbreaking(cc) ucisprop(cc, 0, UC_NB) michael@0: michael@0: /* michael@0: * Directionality macros. michael@0: */ michael@0: #define ucisrtl(cc) ucisprop(cc, UC_R, 0) michael@0: #define ucisltr(cc) ucisprop(cc, UC_L, 0) michael@0: #define ucisstrong(cc) ucisprop(cc, UC_L|UC_R, 0) michael@0: #define ucisweak(cc) ucisprop(cc, UC_EN|UC_ES, UC_ET|UC_AN|UC_CS) michael@0: #define ucisneutral(cc) ucisprop(cc, 0, UC_B|UC_S|UC_WS|UC_ON) michael@0: #define ucisseparator(cc) ucisprop(cc, 0, UC_B|UC_S) michael@0: michael@0: /* michael@0: * Other macros inspired by John Cowan. michael@0: */ michael@0: #define ucismark(cc) ucisprop(cc, UC_MN|UC_MC|UC_ME, 0) michael@0: #define ucismodif(cc) ucisprop(cc, UC_LM, 0) michael@0: #define ucisletnum(cc) ucisprop(cc, UC_NL, 0) michael@0: #define ucisconnect(cc) ucisprop(cc, UC_PC, 0) michael@0: #define ucisdash(cc) ucisprop(cc, UC_PD, 0) michael@0: #define ucismath(cc) ucisprop(cc, UC_SM, 0) michael@0: #define uciscurrency(cc) ucisprop(cc, UC_SC, 0) michael@0: #define ucismodifsymbol(cc) ucisprop(cc, UC_SK, 0) michael@0: #define ucisnsmark(cc) ucisprop(cc, UC_MN, 0) michael@0: #define ucisspmark(cc) ucisprop(cc, UC_MC, 0) michael@0: #define ucisenclosing(cc) ucisprop(cc, UC_ME, 0) michael@0: #define ucisprivate(cc) ucisprop(cc, UC_CO, 0) michael@0: #define ucissurrogate(cc) ucisprop(cc, UC_OS, 0) michael@0: #define ucislsep(cc) ucisprop(cc, UC_ZL, 0) michael@0: #define ucispsep(cc) ucisprop(cc, UC_ZP, 0) michael@0: michael@0: #define ucisidentstart(cc) ucisprop(cc, UC_LU|UC_LL|UC_LT|UC_LO|UC_NL, 0) michael@0: #define ucisidentpart(cc) ucisprop(cc, UC_LU|UC_LL|UC_LT|UC_LO|UC_NL|\ michael@0: UC_MN|UC_MC|UC_ND|UC_PC|UC_CF, 0) michael@0: michael@0: #define ucisdefined(cc) ucisprop(cc, 0, UC_CP) michael@0: #define ucisundefined(cc) !ucisprop(cc, 0, UC_CP) michael@0: michael@0: /* michael@0: * Other miscellaneous character property macros. michael@0: */ michael@0: #define ucishan(cc) (((cc) >= 0x4e00 && (cc) <= 0x9fff) ||\ michael@0: ((cc) >= 0xf900 && (cc) <= 0xfaff)) michael@0: #define ucishangul(cc) ((cc) >= 0xac00 && (cc) <= 0xd7ff) michael@0: michael@0: /************************************************************************** michael@0: * michael@0: * Functions for case conversion. michael@0: * michael@0: **************************************************************************/ michael@0: michael@0: extern unsigned long uctoupper __((unsigned long code)); michael@0: extern unsigned long uctolower __((unsigned long code)); michael@0: extern unsigned long uctotitle __((unsigned long code)); michael@0: michael@0: /************************************************************************** michael@0: * michael@0: * Functions for getting decompositions. michael@0: * michael@0: **************************************************************************/ michael@0: michael@0: /* michael@0: * This routine determines if the code has a decomposition. If it returns 0, michael@0: * there is no decomposition. Any other value indicates a decomposition was michael@0: * returned. michael@0: */ michael@0: extern int ucdecomp __((unsigned long code, unsigned long *num, michael@0: michael@0: unsigned long **decomp)); michael@0: michael@0: /* michael@0: * If the code is a Hangul syllable, this routine decomposes it into the array michael@0: * passed. The array size should be at least 3. michael@0: */ michael@0: extern int ucdecomp_hangul __((unsigned long code, unsigned long *num, michael@0: unsigned long decomp[])); michael@0: michael@0: /************************************************************************** michael@0: * michael@0: * Functions for getting combining classes. michael@0: * michael@0: **************************************************************************/ michael@0: michael@0: /* michael@0: * This will return the combining class for a character to be used with the michael@0: * Canonical Ordering algorithm. michael@0: */ michael@0: extern unsigned long uccombining_class __((unsigned long code)); michael@0: michael@0: /************************************************************************** michael@0: * michael@0: * Functions for getting numbers and digits. michael@0: * michael@0: **************************************************************************/ michael@0: michael@0: struct ucnumber { michael@0: int numerator; michael@0: int denominator; michael@0: }; michael@0: michael@0: extern int ucnumber_lookup __((unsigned long code, struct ucnumber *num)); michael@0: extern int ucdigit_lookup __((unsigned long code, int *digit)); michael@0: michael@0: /* michael@0: * For compatibility with John Cowan's "uctype" package. michael@0: */ michael@0: extern struct ucnumber ucgetnumber __((unsigned long code)); michael@0: extern int ucgetdigit __((unsigned long code)); michael@0: michael@0: /************************************************************************** michael@0: * michael@0: * Functions library initialization and cleanup. michael@0: * michael@0: **************************************************************************/ michael@0: michael@0: /* michael@0: * Macros for specifying the data tables to be loaded for ucdata_load(). michael@0: */ michael@0: #define UCDATA_CASE 0x01 michael@0: #define UCDATA_CTYPE 0x02 michael@0: #define UCDATA_DECOMP 0x04 michael@0: #define UCDATA_CMBCL 0x08 michael@0: #define UCDATA_NUM 0x10 michael@0: michael@0: #define UCDATA_ALL (UCDATA_CASE|UCDATA_CTYPE|UCDATA_DECOMP|\ michael@0: UCDATA_CMBCL|UCDATA_NUM) michael@0: michael@0: /* michael@0: * Functions to load, unload, and reload specific data files. michael@0: */ michael@0: extern void ucdata_load __((char *paths, int mask)); michael@0: extern void ucdata_unload __((int mask)); michael@0: extern void ucdata_reload __((char *paths, int mask)); michael@0: michael@0: /* michael@0: * Deprecated functions, now just compatibility macros. michael@0: */ michael@0: #define ucdata_setup(p) ucdata_load(p, UCDATA_ALL) michael@0: #define ucdata_cleanup() ucdata_unload(UCDATA_ALL) michael@0: michael@0: #undef __ michael@0: michael@0: #ifdef __cplusplus michael@0: } michael@0: #endif michael@0: michael@0: #endif /* _h_ucdata */