1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/uchar.c Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,727 @@
1.4 +/*
1.5 +********************************************************************************
1.6 +* Copyright (C) 1996-2012, International Business Machines
1.7 +* Corporation and others. All Rights Reserved.
1.8 +********************************************************************************
1.9 +*
1.10 +* File UCHAR.C
1.11 +*
1.12 +* Modification History:
1.13 +*
1.14 +* Date Name Description
1.15 +* 04/02/97 aliu Creation.
1.16 +* 4/15/99 Madhu Updated all the function definitions for C Implementation
1.17 +* 5/20/99 Madhu Added the function u_getVersion()
1.18 +* 8/19/1999 srl Upgraded scripts to Unicode3.0
1.19 +* 11/11/1999 weiv added u_isalnum(), cleaned comments
1.20 +* 01/11/2000 helena Renamed u_getVersion to u_getUnicodeVersion.
1.21 +* 06/20/2000 helena OS/400 port changes; mostly typecast.
1.22 +******************************************************************************
1.23 +*/
1.24 +
1.25 +#include "unicode/utypes.h"
1.26 +#include "unicode/uchar.h"
1.27 +#include "unicode/uscript.h"
1.28 +#include "unicode/udata.h"
1.29 +#include "uassert.h"
1.30 +#include "cmemory.h"
1.31 +#include "ucln_cmn.h"
1.32 +#include "utrie2.h"
1.33 +#include "udataswp.h"
1.34 +#include "uprops.h"
1.35 +#include "ustr_imp.h"
1.36 +
1.37 +#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
1.38 +
1.39 +/* uchar_props_data.h is machine-generated by genprops --csource */
1.40 +#define INCLUDED_FROM_UCHAR_C
1.41 +#include "uchar_props_data.h"
1.42 +
1.43 +/* constants and macros for access to the data ------------------------------ */
1.44 +
1.45 +/* getting a uint32_t properties word from the data */
1.46 +#define GET_PROPS(c, result) ((result)=UTRIE2_GET16(&propsTrie, c));
1.47 +
1.48 +U_CFUNC UBool
1.49 +uprv_haveProperties(UErrorCode *pErrorCode) {
1.50 + if(U_FAILURE(*pErrorCode)) {
1.51 + return FALSE;
1.52 + }
1.53 + return TRUE;
1.54 +}
1.55 +
1.56 +/* API functions ------------------------------------------------------------ */
1.57 +
1.58 +/* Gets the Unicode character's general category.*/
1.59 +U_CAPI int8_t U_EXPORT2
1.60 +u_charType(UChar32 c) {
1.61 + uint32_t props;
1.62 + GET_PROPS(c, props);
1.63 + return (int8_t)GET_CATEGORY(props);
1.64 +}
1.65 +
1.66 +/* Enumerate all code points with their general categories. */
1.67 +struct _EnumTypeCallback {
1.68 + UCharEnumTypeRange *enumRange;
1.69 + const void *context;
1.70 +};
1.71 +
1.72 +static uint32_t U_CALLCONV
1.73 +_enumTypeValue(const void *context, uint32_t value) {
1.74 + return GET_CATEGORY(value);
1.75 +}
1.76 +
1.77 +static UBool U_CALLCONV
1.78 +_enumTypeRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
1.79 + /* just cast the value to UCharCategory */
1.80 + return ((struct _EnumTypeCallback *)context)->
1.81 + enumRange(((struct _EnumTypeCallback *)context)->context,
1.82 + start, end+1, (UCharCategory)value);
1.83 +}
1.84 +
1.85 +U_CAPI void U_EXPORT2
1.86 +u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context) {
1.87 + struct _EnumTypeCallback callback;
1.88 +
1.89 + if(enumRange==NULL) {
1.90 + return;
1.91 + }
1.92 +
1.93 + callback.enumRange=enumRange;
1.94 + callback.context=context;
1.95 + utrie2_enum(&propsTrie, _enumTypeValue, _enumTypeRange, &callback);
1.96 +}
1.97 +
1.98 +/* Checks if ch is a lower case letter.*/
1.99 +U_CAPI UBool U_EXPORT2
1.100 +u_islower(UChar32 c) {
1.101 + uint32_t props;
1.102 + GET_PROPS(c, props);
1.103 + return (UBool)(GET_CATEGORY(props)==U_LOWERCASE_LETTER);
1.104 +}
1.105 +
1.106 +/* Checks if ch is an upper case letter.*/
1.107 +U_CAPI UBool U_EXPORT2
1.108 +u_isupper(UChar32 c) {
1.109 + uint32_t props;
1.110 + GET_PROPS(c, props);
1.111 + return (UBool)(GET_CATEGORY(props)==U_UPPERCASE_LETTER);
1.112 +}
1.113 +
1.114 +/* Checks if ch is a title case letter; usually upper case letters.*/
1.115 +U_CAPI UBool U_EXPORT2
1.116 +u_istitle(UChar32 c) {
1.117 + uint32_t props;
1.118 + GET_PROPS(c, props);
1.119 + return (UBool)(GET_CATEGORY(props)==U_TITLECASE_LETTER);
1.120 +}
1.121 +
1.122 +/* Checks if ch is a decimal digit. */
1.123 +U_CAPI UBool U_EXPORT2
1.124 +u_isdigit(UChar32 c) {
1.125 + uint32_t props;
1.126 + GET_PROPS(c, props);
1.127 + return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER);
1.128 +}
1.129 +
1.130 +U_CAPI UBool U_EXPORT2
1.131 +u_isxdigit(UChar32 c) {
1.132 + uint32_t props;
1.133 +
1.134 + /* check ASCII and Fullwidth ASCII a-fA-F */
1.135 + if(
1.136 + (c<=0x66 && c>=0x41 && (c<=0x46 || c>=0x61)) ||
1.137 + (c>=0xff21 && c<=0xff46 && (c<=0xff26 || c>=0xff41))
1.138 + ) {
1.139 + return TRUE;
1.140 + }
1.141 +
1.142 + GET_PROPS(c, props);
1.143 + return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER);
1.144 +}
1.145 +
1.146 +/* Checks if the Unicode character is a letter.*/
1.147 +U_CAPI UBool U_EXPORT2
1.148 +u_isalpha(UChar32 c) {
1.149 + uint32_t props;
1.150 + GET_PROPS(c, props);
1.151 + return (UBool)((CAT_MASK(props)&U_GC_L_MASK)!=0);
1.152 +}
1.153 +
1.154 +U_CAPI UBool U_EXPORT2
1.155 +u_isUAlphabetic(UChar32 c) {
1.156 + return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_ALPHABETIC))!=0;
1.157 +}
1.158 +
1.159 +/* Checks if c is a letter or a decimal digit */
1.160 +U_CAPI UBool U_EXPORT2
1.161 +u_isalnum(UChar32 c) {
1.162 + uint32_t props;
1.163 + GET_PROPS(c, props);
1.164 + return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_ND_MASK))!=0);
1.165 +}
1.166 +
1.167 +/**
1.168 + * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
1.169 + * @internal
1.170 + */
1.171 +U_CFUNC UBool
1.172 +u_isalnumPOSIX(UChar32 c) {
1.173 + return (UBool)(u_isUAlphabetic(c) || u_isdigit(c));
1.174 +}
1.175 +
1.176 +/* Checks if ch is a unicode character with assigned character type.*/
1.177 +U_CAPI UBool U_EXPORT2
1.178 +u_isdefined(UChar32 c) {
1.179 + uint32_t props;
1.180 + GET_PROPS(c, props);
1.181 + return (UBool)(GET_CATEGORY(props)!=0);
1.182 +}
1.183 +
1.184 +/* Checks if the Unicode character is a base form character that can take a diacritic.*/
1.185 +U_CAPI UBool U_EXPORT2
1.186 +u_isbase(UChar32 c) {
1.187 + uint32_t props;
1.188 + GET_PROPS(c, props);
1.189 + return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_N_MASK|U_GC_MC_MASK|U_GC_ME_MASK))!=0);
1.190 +}
1.191 +
1.192 +/* Checks if the Unicode character is a control character.*/
1.193 +U_CAPI UBool U_EXPORT2
1.194 +u_iscntrl(UChar32 c) {
1.195 + uint32_t props;
1.196 + GET_PROPS(c, props);
1.197 + return (UBool)((CAT_MASK(props)&(U_GC_CC_MASK|U_GC_CF_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK))!=0);
1.198 +}
1.199 +
1.200 +U_CAPI UBool U_EXPORT2
1.201 +u_isISOControl(UChar32 c) {
1.202 + return (uint32_t)c<=0x9f && (c<=0x1f || c>=0x7f);
1.203 +}
1.204 +
1.205 +/* Some control characters that are used as space. */
1.206 +#define IS_THAT_CONTROL_SPACE(c) \
1.207 + (c<=0x9f && ((c>=TAB && c<=CR) || (c>=0x1c && c <=0x1f) || c==NL))
1.208 +
1.209 +/* Java has decided that U+0085 New Line is not whitespace any more. */
1.210 +#define IS_THAT_ASCII_CONTROL_SPACE(c) \
1.211 + (c<=0x1f && c>=TAB && (c<=CR || c>=0x1c))
1.212 +
1.213 +/* Checks if the Unicode character is a space character.*/
1.214 +U_CAPI UBool U_EXPORT2
1.215 +u_isspace(UChar32 c) {
1.216 + uint32_t props;
1.217 + GET_PROPS(c, props);
1.218 + return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0 || IS_THAT_CONTROL_SPACE(c));
1.219 +}
1.220 +
1.221 +U_CAPI UBool U_EXPORT2
1.222 +u_isJavaSpaceChar(UChar32 c) {
1.223 + uint32_t props;
1.224 + GET_PROPS(c, props);
1.225 + return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0);
1.226 +}
1.227 +
1.228 +/* Checks if the Unicode character is a whitespace character.*/
1.229 +U_CAPI UBool U_EXPORT2
1.230 +u_isWhitespace(UChar32 c) {
1.231 + uint32_t props;
1.232 + GET_PROPS(c, props);
1.233 + return (UBool)(
1.234 + ((CAT_MASK(props)&U_GC_Z_MASK)!=0 &&
1.235 + c!=NBSP && c!=FIGURESP && c!=NNBSP) || /* exclude no-break spaces */
1.236 + IS_THAT_ASCII_CONTROL_SPACE(c)
1.237 + );
1.238 +}
1.239 +
1.240 +U_CAPI UBool U_EXPORT2
1.241 +u_isblank(UChar32 c) {
1.242 + if((uint32_t)c<=0x9f) {
1.243 + return c==9 || c==0x20; /* TAB or SPACE */
1.244 + } else {
1.245 + /* Zs */
1.246 + uint32_t props;
1.247 + GET_PROPS(c, props);
1.248 + return (UBool)(GET_CATEGORY(props)==U_SPACE_SEPARATOR);
1.249 + }
1.250 +}
1.251 +
1.252 +U_CAPI UBool U_EXPORT2
1.253 +u_isUWhiteSpace(UChar32 c) {
1.254 + return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_WHITE_SPACE))!=0;
1.255 +}
1.256 +
1.257 +/* Checks if the Unicode character is printable.*/
1.258 +U_CAPI UBool U_EXPORT2
1.259 +u_isprint(UChar32 c) {
1.260 + uint32_t props;
1.261 + GET_PROPS(c, props);
1.262 + /* comparing ==0 returns FALSE for the categories mentioned */
1.263 + return (UBool)((CAT_MASK(props)&U_GC_C_MASK)==0);
1.264 +}
1.265 +
1.266 +/**
1.267 + * Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
1.268 + * Implements UCHAR_POSIX_PRINT.
1.269 + * @internal
1.270 + */
1.271 +U_CFUNC UBool
1.272 +u_isprintPOSIX(UChar32 c) {
1.273 + uint32_t props;
1.274 + GET_PROPS(c, props);
1.275 + /*
1.276 + * The only cntrl character in graph+blank is TAB (in blank).
1.277 + * Here we implement (blank-TAB)=Zs instead of calling u_isblank().
1.278 + */
1.279 + return (UBool)((GET_CATEGORY(props)==U_SPACE_SEPARATOR) || u_isgraphPOSIX(c));
1.280 +}
1.281 +
1.282 +U_CAPI UBool U_EXPORT2
1.283 +u_isgraph(UChar32 c) {
1.284 + uint32_t props;
1.285 + GET_PROPS(c, props);
1.286 + /* comparing ==0 returns FALSE for the categories mentioned */
1.287 + return (UBool)((CAT_MASK(props)&
1.288 + (U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK))
1.289 + ==0);
1.290 +}
1.291 +
1.292 +/**
1.293 + * Checks if c is in
1.294 + * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
1.295 + * with space=\p{Whitespace} and Control=Cc.
1.296 + * Implements UCHAR_POSIX_GRAPH.
1.297 + * @internal
1.298 + */
1.299 +U_CFUNC UBool
1.300 +u_isgraphPOSIX(UChar32 c) {
1.301 + uint32_t props;
1.302 + GET_PROPS(c, props);
1.303 + /* \p{space}\p{gc=Control} == \p{gc=Z}\p{Control} */
1.304 + /* comparing ==0 returns FALSE for the categories mentioned */
1.305 + return (UBool)((CAT_MASK(props)&
1.306 + (U_GC_CC_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK))
1.307 + ==0);
1.308 +}
1.309 +
1.310 +U_CAPI UBool U_EXPORT2
1.311 +u_ispunct(UChar32 c) {
1.312 + uint32_t props;
1.313 + GET_PROPS(c, props);
1.314 + return (UBool)((CAT_MASK(props)&U_GC_P_MASK)!=0);
1.315 +}
1.316 +
1.317 +/* Checks if the Unicode character can start a Unicode identifier.*/
1.318 +U_CAPI UBool U_EXPORT2
1.319 +u_isIDStart(UChar32 c) {
1.320 + /* same as u_isalpha() */
1.321 + uint32_t props;
1.322 + GET_PROPS(c, props);
1.323 + return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_NL_MASK))!=0);
1.324 +}
1.325 +
1.326 +/* Checks if the Unicode character can be a Unicode identifier part other than starting the
1.327 + identifier.*/
1.328 +U_CAPI UBool U_EXPORT2
1.329 +u_isIDPart(UChar32 c) {
1.330 + uint32_t props;
1.331 + GET_PROPS(c, props);
1.332 + return (UBool)(
1.333 + (CAT_MASK(props)&
1.334 + (U_GC_ND_MASK|U_GC_NL_MASK|
1.335 + U_GC_L_MASK|
1.336 + U_GC_PC_MASK|U_GC_MC_MASK|U_GC_MN_MASK)
1.337 + )!=0 ||
1.338 + u_isIDIgnorable(c));
1.339 +}
1.340 +
1.341 +/*Checks if the Unicode character can be ignorable in a Java or Unicode identifier.*/
1.342 +U_CAPI UBool U_EXPORT2
1.343 +u_isIDIgnorable(UChar32 c) {
1.344 + if(c<=0x9f) {
1.345 + return u_isISOControl(c) && !IS_THAT_ASCII_CONTROL_SPACE(c);
1.346 + } else {
1.347 + uint32_t props;
1.348 + GET_PROPS(c, props);
1.349 + return (UBool)(GET_CATEGORY(props)==U_FORMAT_CHAR);
1.350 + }
1.351 +}
1.352 +
1.353 +/*Checks if the Unicode character can start a Java identifier.*/
1.354 +U_CAPI UBool U_EXPORT2
1.355 +u_isJavaIDStart(UChar32 c) {
1.356 + uint32_t props;
1.357 + GET_PROPS(c, props);
1.358 + return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_SC_MASK|U_GC_PC_MASK))!=0);
1.359 +}
1.360 +
1.361 +/*Checks if the Unicode character can be a Java identifier part other than starting the
1.362 + * identifier.
1.363 + */
1.364 +U_CAPI UBool U_EXPORT2
1.365 +u_isJavaIDPart(UChar32 c) {
1.366 + uint32_t props;
1.367 + GET_PROPS(c, props);
1.368 + return (UBool)(
1.369 + (CAT_MASK(props)&
1.370 + (U_GC_ND_MASK|U_GC_NL_MASK|
1.371 + U_GC_L_MASK|
1.372 + U_GC_SC_MASK|U_GC_PC_MASK|
1.373 + U_GC_MC_MASK|U_GC_MN_MASK)
1.374 + )!=0 ||
1.375 + u_isIDIgnorable(c));
1.376 +}
1.377 +
1.378 +U_CAPI int32_t U_EXPORT2
1.379 +u_charDigitValue(UChar32 c) {
1.380 + uint32_t props;
1.381 + int32_t value;
1.382 + GET_PROPS(c, props);
1.383 + value=(int32_t)GET_NUMERIC_TYPE_VALUE(props)-UPROPS_NTV_DECIMAL_START;
1.384 + if(value<=9) {
1.385 + return value;
1.386 + } else {
1.387 + return -1;
1.388 + }
1.389 +}
1.390 +
1.391 +U_CAPI double U_EXPORT2
1.392 +u_getNumericValue(UChar32 c) {
1.393 + uint32_t props;
1.394 + int32_t ntv;
1.395 + GET_PROPS(c, props);
1.396 + ntv=(int32_t)GET_NUMERIC_TYPE_VALUE(props);
1.397 +
1.398 + if(ntv==UPROPS_NTV_NONE) {
1.399 + return U_NO_NUMERIC_VALUE;
1.400 + } else if(ntv<UPROPS_NTV_DIGIT_START) {
1.401 + /* decimal digit */
1.402 + return ntv-UPROPS_NTV_DECIMAL_START;
1.403 + } else if(ntv<UPROPS_NTV_NUMERIC_START) {
1.404 + /* other digit */
1.405 + return ntv-UPROPS_NTV_DIGIT_START;
1.406 + } else if(ntv<UPROPS_NTV_FRACTION_START) {
1.407 + /* small integer */
1.408 + return ntv-UPROPS_NTV_NUMERIC_START;
1.409 + } else if(ntv<UPROPS_NTV_LARGE_START) {
1.410 + /* fraction */
1.411 + int32_t numerator=(ntv>>4)-12;
1.412 + int32_t denominator=(ntv&0xf)+1;
1.413 + return (double)numerator/denominator;
1.414 + } else if(ntv<UPROPS_NTV_BASE60_START) {
1.415 + /* large, single-significant-digit integer */
1.416 + double numValue;
1.417 + int32_t mant=(ntv>>5)-14;
1.418 + int32_t exp=(ntv&0x1f)+2;
1.419 + numValue=mant;
1.420 +
1.421 + /* multiply by 10^exp without math.h */
1.422 + while(exp>=4) {
1.423 + numValue*=10000.;
1.424 + exp-=4;
1.425 + }
1.426 + switch(exp) {
1.427 + case 3:
1.428 + numValue*=1000.;
1.429 + break;
1.430 + case 2:
1.431 + numValue*=100.;
1.432 + break;
1.433 + case 1:
1.434 + numValue*=10.;
1.435 + break;
1.436 + case 0:
1.437 + default:
1.438 + break;
1.439 + }
1.440 +
1.441 + return numValue;
1.442 + } else if(ntv<UPROPS_NTV_RESERVED_START) {
1.443 + /* sexagesimal (base 60) integer */
1.444 + int32_t numValue=(ntv>>2)-0xbf;
1.445 + int32_t exp=(ntv&3)+1;
1.446 +
1.447 + switch(exp) {
1.448 + case 4:
1.449 + numValue*=60*60*60*60;
1.450 + break;
1.451 + case 3:
1.452 + numValue*=60*60*60;
1.453 + break;
1.454 + case 2:
1.455 + numValue*=60*60;
1.456 + break;
1.457 + case 1:
1.458 + numValue*=60;
1.459 + break;
1.460 + case 0:
1.461 + default:
1.462 + break;
1.463 + }
1.464 +
1.465 + return numValue;
1.466 + } else {
1.467 + /* reserved */
1.468 + return U_NO_NUMERIC_VALUE;
1.469 + }
1.470 +}
1.471 +
1.472 +U_CAPI int32_t U_EXPORT2
1.473 +u_digit(UChar32 ch, int8_t radix) {
1.474 + int8_t value;
1.475 + if((uint8_t)(radix-2)<=(36-2)) {
1.476 + value=(int8_t)u_charDigitValue(ch);
1.477 + if(value<0) {
1.478 + /* ch is not a decimal digit, try latin letters */
1.479 + if(ch>=0x61 && ch<=0x7A) {
1.480 + value=(int8_t)(ch-0x57); /* ch - 'a' + 10 */
1.481 + } else if(ch>=0x41 && ch<=0x5A) {
1.482 + value=(int8_t)(ch-0x37); /* ch - 'A' + 10 */
1.483 + } else if(ch>=0xFF41 && ch<=0xFF5A) {
1.484 + value=(int8_t)(ch-0xFF37); /* fullwidth ASCII a-z */
1.485 + } else if(ch>=0xFF21 && ch<=0xFF3A) {
1.486 + value=(int8_t)(ch-0xFF17); /* fullwidth ASCII A-Z */
1.487 + }
1.488 + }
1.489 + } else {
1.490 + value=-1; /* invalid radix */
1.491 + }
1.492 + return (int8_t)((value<radix) ? value : -1);
1.493 +}
1.494 +
1.495 +U_CAPI UChar32 U_EXPORT2
1.496 +u_forDigit(int32_t digit, int8_t radix) {
1.497 + if((uint8_t)(radix-2)>(36-2) || (uint32_t)digit>=(uint32_t)radix) {
1.498 + return 0;
1.499 + } else if(digit<10) {
1.500 + return (UChar32)(0x30+digit);
1.501 + } else {
1.502 + return (UChar32)((0x61-10)+digit);
1.503 + }
1.504 +}
1.505 +
1.506 +/* miscellaneous, and support for uprops.cpp -------------------------------- */
1.507 +
1.508 +U_CAPI void U_EXPORT2
1.509 +u_getUnicodeVersion(UVersionInfo versionArray) {
1.510 + if(versionArray!=NULL) {
1.511 + uprv_memcpy(versionArray, dataVersion, U_MAX_VERSION_LENGTH);
1.512 + }
1.513 +}
1.514 +
1.515 +U_CFUNC uint32_t
1.516 +u_getMainProperties(UChar32 c) {
1.517 + uint32_t props;
1.518 + GET_PROPS(c, props);
1.519 + return props;
1.520 +}
1.521 +
1.522 +U_CFUNC uint32_t
1.523 +u_getUnicodeProperties(UChar32 c, int32_t column) {
1.524 + U_ASSERT(column>=0);
1.525 + if(column>=propsVectorsColumns) {
1.526 + return 0;
1.527 + } else {
1.528 + uint16_t vecIndex=UTRIE2_GET16(&propsVectorsTrie, c);
1.529 + return propsVectors[vecIndex+column];
1.530 + }
1.531 +}
1.532 +
1.533 +U_CFUNC int32_t
1.534 +uprv_getMaxValues(int32_t column) {
1.535 + switch(column) {
1.536 + case 0:
1.537 + return indexes[UPROPS_MAX_VALUES_INDEX];
1.538 + case 2:
1.539 + return indexes[UPROPS_MAX_VALUES_2_INDEX];
1.540 + default:
1.541 + return 0;
1.542 + }
1.543 +}
1.544 +
1.545 +U_CAPI void U_EXPORT2
1.546 +u_charAge(UChar32 c, UVersionInfo versionArray) {
1.547 + if(versionArray!=NULL) {
1.548 + uint32_t version=u_getUnicodeProperties(c, 0)>>UPROPS_AGE_SHIFT;
1.549 + versionArray[0]=(uint8_t)(version>>4);
1.550 + versionArray[1]=(uint8_t)(version&0xf);
1.551 + versionArray[2]=versionArray[3]=0;
1.552 + }
1.553 +}
1.554 +
1.555 +U_CAPI UScriptCode U_EXPORT2
1.556 +uscript_getScript(UChar32 c, UErrorCode *pErrorCode) {
1.557 + uint32_t scriptX;
1.558 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.559 + return USCRIPT_INVALID_CODE;
1.560 + }
1.561 + if((uint32_t)c>0x10ffff) {
1.562 + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1.563 + return USCRIPT_INVALID_CODE;
1.564 + }
1.565 + scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK;
1.566 + if(scriptX<UPROPS_SCRIPT_X_WITH_COMMON) {
1.567 + return (UScriptCode)scriptX;
1.568 + } else if(scriptX<UPROPS_SCRIPT_X_WITH_INHERITED) {
1.569 + return USCRIPT_COMMON;
1.570 + } else if(scriptX<UPROPS_SCRIPT_X_WITH_OTHER) {
1.571 + return USCRIPT_INHERITED;
1.572 + } else {
1.573 + return (UScriptCode)scriptExtensions[scriptX&UPROPS_SCRIPT_MASK];
1.574 + }
1.575 +}
1.576 +
1.577 +U_CAPI UBool U_EXPORT2
1.578 +uscript_hasScript(UChar32 c, UScriptCode sc) {
1.579 + const uint16_t *scx;
1.580 + uint32_t scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK;
1.581 + if(scriptX<UPROPS_SCRIPT_X_WITH_COMMON) {
1.582 + return sc==(UScriptCode)scriptX;
1.583 + }
1.584 +
1.585 + scx=scriptExtensions+(scriptX&UPROPS_SCRIPT_MASK);
1.586 + if(scriptX>=UPROPS_SCRIPT_X_WITH_OTHER) {
1.587 + scx=scriptExtensions+scx[1];
1.588 + }
1.589 + if(sc>=USCRIPT_CODE_LIMIT) {
1.590 + /* Guard against bogus input that would make us go past the Script_Extensions terminator. */
1.591 + return FALSE;
1.592 + }
1.593 + while(sc>*scx) {
1.594 + ++scx;
1.595 + }
1.596 + return sc==(*scx&0x7fff);
1.597 +}
1.598 +
1.599 +U_CAPI int32_t U_EXPORT2
1.600 +uscript_getScriptExtensions(UChar32 c,
1.601 + UScriptCode *scripts, int32_t capacity,
1.602 + UErrorCode *pErrorCode) {
1.603 + uint32_t scriptX;
1.604 + int32_t length;
1.605 + const uint16_t *scx;
1.606 + uint16_t sx;
1.607 + if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1.608 + return 0;
1.609 + }
1.610 + if(capacity<0 || (capacity>0 && scripts==NULL)) {
1.611 + *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1.612 + return 0;
1.613 + }
1.614 + scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK;
1.615 + if(scriptX<UPROPS_SCRIPT_X_WITH_COMMON) {
1.616 + if(capacity==0) {
1.617 + *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1.618 + } else {
1.619 + scripts[0]=(UScriptCode)scriptX;
1.620 + }
1.621 + return 1;
1.622 + }
1.623 +
1.624 + scx=scriptExtensions+(scriptX&UPROPS_SCRIPT_MASK);
1.625 + if(scriptX>=UPROPS_SCRIPT_X_WITH_OTHER) {
1.626 + scx=scriptExtensions+scx[1];
1.627 + }
1.628 + length=0;
1.629 + do {
1.630 + sx=*scx++;
1.631 + if(length<capacity) {
1.632 + scripts[length]=(UScriptCode)(sx&0x7fff);
1.633 + }
1.634 + ++length;
1.635 + } while(sx<0x8000);
1.636 + if(length>capacity) {
1.637 + *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1.638 + }
1.639 + return length;
1.640 +}
1.641 +
1.642 +U_CAPI UBlockCode U_EXPORT2
1.643 +ublock_getCode(UChar32 c) {
1.644 + return (UBlockCode)((u_getUnicodeProperties(c, 0)&UPROPS_BLOCK_MASK)>>UPROPS_BLOCK_SHIFT);
1.645 +}
1.646 +
1.647 +/* property starts for UnicodeSet ------------------------------------------- */
1.648 +
1.649 +static UBool U_CALLCONV
1.650 +_enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
1.651 + /* add the start code point to the USet */
1.652 + const USetAdder *sa=(const USetAdder *)context;
1.653 + sa->add(sa->set, start);
1.654 + return TRUE;
1.655 +}
1.656 +
1.657 +#define USET_ADD_CP_AND_NEXT(sa, cp) sa->add(sa->set, cp); sa->add(sa->set, cp+1)
1.658 +
1.659 +U_CFUNC void U_EXPORT2
1.660 +uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
1.661 + if(U_FAILURE(*pErrorCode)) {
1.662 + return;
1.663 + }
1.664 +
1.665 + /* add the start code point of each same-value range of the main trie */
1.666 + utrie2_enum(&propsTrie, NULL, _enumPropertyStartsRange, sa);
1.667 +
1.668 + /* add code points with hardcoded properties, plus the ones following them */
1.669 +
1.670 + /* add for u_isblank() */
1.671 + USET_ADD_CP_AND_NEXT(sa, TAB);
1.672 +
1.673 + /* add for IS_THAT_CONTROL_SPACE() */
1.674 + sa->add(sa->set, CR+1); /* range TAB..CR */
1.675 + sa->add(sa->set, 0x1c);
1.676 + sa->add(sa->set, 0x1f+1);
1.677 + USET_ADD_CP_AND_NEXT(sa, NL);
1.678 +
1.679 + /* add for u_isIDIgnorable() what was not added above */
1.680 + sa->add(sa->set, DEL); /* range DEL..NBSP-1, NBSP added below */
1.681 + sa->add(sa->set, HAIRSP);
1.682 + sa->add(sa->set, RLM+1);
1.683 + sa->add(sa->set, INHSWAP);
1.684 + sa->add(sa->set, NOMDIG+1);
1.685 + USET_ADD_CP_AND_NEXT(sa, ZWNBSP);
1.686 +
1.687 + /* add no-break spaces for u_isWhitespace() what was not added above */
1.688 + USET_ADD_CP_AND_NEXT(sa, NBSP);
1.689 + USET_ADD_CP_AND_NEXT(sa, FIGURESP);
1.690 + USET_ADD_CP_AND_NEXT(sa, NNBSP);
1.691 +
1.692 + /* add for u_digit() */
1.693 + sa->add(sa->set, U_a);
1.694 + sa->add(sa->set, U_z+1);
1.695 + sa->add(sa->set, U_A);
1.696 + sa->add(sa->set, U_Z+1);
1.697 + sa->add(sa->set, U_FW_a);
1.698 + sa->add(sa->set, U_FW_z+1);
1.699 + sa->add(sa->set, U_FW_A);
1.700 + sa->add(sa->set, U_FW_Z+1);
1.701 +
1.702 + /* add for u_isxdigit() */
1.703 + sa->add(sa->set, U_f+1);
1.704 + sa->add(sa->set, U_F+1);
1.705 + sa->add(sa->set, U_FW_f+1);
1.706 + sa->add(sa->set, U_FW_F+1);
1.707 +
1.708 + /* add for UCHAR_DEFAULT_IGNORABLE_CODE_POINT what was not added above */
1.709 + sa->add(sa->set, WJ); /* range WJ..NOMDIG */
1.710 + sa->add(sa->set, 0xfff0);
1.711 + sa->add(sa->set, 0xfffb+1);
1.712 + sa->add(sa->set, 0xe0000);
1.713 + sa->add(sa->set, 0xe0fff+1);
1.714 +
1.715 + /* add for UCHAR_GRAPHEME_BASE and others */
1.716 + USET_ADD_CP_AND_NEXT(sa, CGJ);
1.717 +}
1.718 +
1.719 +U_CFUNC void U_EXPORT2
1.720 +upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
1.721 + if(U_FAILURE(*pErrorCode)) {
1.722 + return;
1.723 + }
1.724 +
1.725 + /* add the start code point of each same-value range of the properties vectors trie */
1.726 + if(propsVectorsColumns>0) {
1.727 + /* if propsVectorsColumns==0 then the properties vectors trie may not be there at all */
1.728 + utrie2_enum(&propsVectorsTrie, NULL, _enumPropertyStartsRange, sa);
1.729 + }
1.730 +}
