icuSources/common/uchar.c

   1 /*
   2 ********************************************************************************
   3 *   Copyright (C) 1996-2008, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 ********************************************************************************
   6 *
   7 * File UCHAR.C
   8 *
   9 * Modification History:
  10 *
  11 *   Date        Name        Description
  12 *   04/02/97    aliu        Creation.
  13 *   4/15/99     Madhu       Updated all the function definitions for C Implementation
  14 *   5/20/99     Madhu       Added the function u_getVersion()
  15 *   8/19/1999   srl         Upgraded scripts to Unicode3.0
  16 *   11/11/1999  weiv        added u_isalnum(), cleaned comments
  17 *   01/11/2000  helena      Renamed u_getVersion to u_getUnicodeVersion.
  18 *   06/20/2000  helena      OS/400 port changes; mostly typecast.
  19 ******************************************************************************
  20 */
  21
  22 #include "unicode/utypes.h"
  23 #include "unicode/uchar.h"
  24 #include "unicode/uscript.h"
  25 #include "unicode/udata.h"
  26 #include "umutex.h"
  27 #include "cmemory.h"
  28 #include "ucln_cmn.h"
  29 #include "utrie.h"
  30 #include "udataswp.h"
  31 #include "unormimp.h" /* JAMO_L_BASE etc. */
  32 #include "uprops.h"
  33
  34 #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
  35
  36 /* dynamically loaded Unicode character properties -------------------------- */
  37
  38 #define UCHAR_HARDCODE_DATA 1
  39
  40 #if UCHAR_HARDCODE_DATA
  41
  42 /* uchar_props_data.c is machine-generated by genprops --csource */
  43 #include "uchar_props_data.c"
  44
  45 #else
  46
  47 /*
  48  * loaded uprops.dat -
  49  * for a description of the file format, see icu/source/tools/genprops/store.c
  50  */
  51 static const char DATA_NAME[] = "uprops";
  52 static const char DATA_TYPE[] = "icu";
  53
  54 static UDataMemory *propsData=NULL;
  55 static UErrorCode dataErrorCode=U_ZERO_ERROR;
  56
  57 static uint8_t formatVersion[4]={ 0, 0, 0, 0 };
  58 static UVersionInfo dataVersion={ 0, 0, 0, 0 };
  59
  60 static UTrie propsTrie={ 0 }, propsVectorsTrie={ 0 };
  61 static const uint32_t *pData32=NULL, *propsVectors=NULL;
  62 static int32_t countPropsVectors=0, propsVectorsColumns=0;
  63
  64 static int8_t havePropsData=0;     /*  == 0   ->  Data has not been loaded.
  65                                     *   < 0   ->  Error occured attempting to load data.
  66                                     *   > 0   ->  Data has been successfully loaded.
  67                                     */
  68
  69 /* index values loaded from uprops.dat */
  70 static int32_t indexes[UPROPS_INDEX_COUNT];
  71
  72 static UBool U_CALLCONV
  73 isAcceptable(void *context,
  74              const char *type, const char *name,
  75              const UDataInfo *pInfo) {
  76     if(
  77         pInfo->size>=20 &&
  78         pInfo->isBigEndian==U_IS_BIG_ENDIAN &&
  79         pInfo->charsetFamily==U_CHARSET_FAMILY &&
  80         pInfo->dataFormat[0]==0x55 &&   /* dataFormat="UPro" */
  81         pInfo->dataFormat[1]==0x50 &&
  82         pInfo->dataFormat[2]==0x72 &&
  83         pInfo->dataFormat[3]==0x6f &&
  84         pInfo->formatVersion[0]==4 &&
  85         pInfo->formatVersion[2]==UTRIE_SHIFT &&
  86         pInfo->formatVersion[3]==UTRIE_INDEX_SHIFT
  87     ) {
  88         uprv_memcpy(formatVersion, pInfo->formatVersion, 4);
  89         uprv_memcpy(dataVersion, pInfo->dataVersion, 4);
  90         return TRUE;
  91     } else {
  92         return FALSE;
  93     }
  94 }
  95
  96 static UBool U_CALLCONV uchar_cleanup(void)
  97 {
  98     if (propsData) {
  99         udata_close(propsData);
 100         propsData=NULL;
 101     }
 102     pData32=NULL;
 103     propsVectors=NULL;
 104     countPropsVectors=0;
 105     uprv_memset(dataVersion, 0, U_MAX_VERSION_LENGTH);
 106     dataErrorCode=U_ZERO_ERROR;
 107     havePropsData=0;
 108
 109     return TRUE;
 110 }
 111
 112 struct UCharProps {
 113     UDataMemory *propsData;
 114     UTrie propsTrie, propsVectorsTrie;
 115     const uint32_t *pData32;
 116 };
 117 typedef struct UCharProps UCharProps;
 118
 119 /* open uprops.icu */
 120 static void
 121 _openProps(UCharProps *ucp, UErrorCode *pErrorCode) {
 122     const uint32_t *p;
 123     int32_t length;
 124
 125     ucp->propsData=udata_openChoice(NULL, DATA_TYPE, DATA_NAME, isAcceptable, NULL, pErrorCode);
 126     if(U_FAILURE(*pErrorCode)) {
 127         return;
 128     }
 129
 130     ucp->pData32=p=(const uint32_t *)udata_getMemory(ucp->propsData);
 131
 132     /* unserialize the trie; it is directly after the int32_t indexes[UPROPS_INDEX_COUNT] */
 133     length=(int32_t)p[UPROPS_PROPS32_INDEX]*4;
 134     length=utrie_unserialize(&ucp->propsTrie, (const uint8_t *)(p+UPROPS_INDEX_COUNT), length-64, pErrorCode);
 135     if(U_FAILURE(*pErrorCode)) {
 136         return;
 137     }
 138
 139     /* unserialize the properties vectors trie */
 140     length=(int32_t)(p[UPROPS_ADDITIONAL_VECTORS_INDEX]-p[UPROPS_ADDITIONAL_TRIE_INDEX])*4;
 141     if(length>0) {
 142         length=utrie_unserialize(&ucp->propsVectorsTrie, (const uint8_t *)(p+p[UPROPS_ADDITIONAL_TRIE_INDEX]), length, pErrorCode);
 143     }
 144     if(length<=0 || U_FAILURE(*pErrorCode)) {
 145         /*
 146          * length==0:
 147          * Allow the properties vectors trie to be missing -
 148          * also requires propsVectorsColumns=indexes[UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX]
 149          * to be zero so that this trie is never accessed.
 150          */
 151         uprv_memset(&ucp->propsVectorsTrie, 0, sizeof(ucp->propsVectorsTrie));
 152     }
 153 }
 154
 155 #endif
 156
 157 #if !UCHAR_HARDCODE_DATA
 158 static int8_t
 159 uprv_loadPropsData(UErrorCode *pErrorCode) {
 160     /* load Unicode character properties data from file if necessary */
 161
 162     /*
 163      * This lazy intialization with double-checked locking (without mutex protection for
 164      * haveNormData==0) is transiently unsafe under certain circumstances.
 165      * Check the readme and use u_init() if necessary.
 166      */
 167     if(havePropsData==0) {
 168         UCharProps ucp={ NULL };
 169
 170         if(U_FAILURE(*pErrorCode)) {
 171             return havePropsData;
 172         }
 173
 174         /* open the data outside the mutex block */
 175         _openProps(&ucp, pErrorCode);
 176
 177         if(U_SUCCESS(*pErrorCode)) {
 178             /* in the mutex block, set the data for this process */
 179             umtx_lock(NULL);
 180             if(propsData==NULL) {
 181                 propsData=ucp.propsData;
 182                 ucp.propsData=NULL;
 183                 pData32=ucp.pData32;
 184                 ucp.pData32=NULL;
 185                 uprv_memcpy(&propsTrie, &ucp.propsTrie, sizeof(propsTrie));
 186                 uprv_memcpy(&propsVectorsTrie, &ucp.propsVectorsTrie, sizeof(propsVectorsTrie));
 187             }
 188
 189             /* initialize some variables */
 190             uprv_memcpy(indexes, pData32, sizeof(indexes));
 191
 192             /* additional properties */
 193             if(indexes[UPROPS_ADDITIONAL_VECTORS_INDEX]!=0) {
 194                 propsVectors=pData32+indexes[UPROPS_ADDITIONAL_VECTORS_INDEX];
 195                 countPropsVectors=indexes[UPROPS_RESERVED_INDEX]-indexes[UPROPS_ADDITIONAL_VECTORS_INDEX];
 196                 propsVectorsColumns=indexes[UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX];
 197             }
 198
 199             havePropsData=1;
 200             umtx_unlock(NULL);
 201         } else {
 202             dataErrorCode=*pErrorCode;
 203             havePropsData=-1;
 204         }
 205         ucln_common_registerCleanup(UCLN_COMMON_UCHAR, uchar_cleanup);
 206
 207         /* if a different thread set it first, then close the extra data */
 208         udata_close(ucp.propsData); /* NULL if it was set correctly */
 209     }
 210
 211     return havePropsData;
 212 }
 213
 214 static int8_t
 215 loadPropsData(void) {
 216     UErrorCode   errorCode = U_ZERO_ERROR;
 217     int8_t       retVal    = uprv_loadPropsData(&errorCode);
 218     return retVal;
 219 }
 220
 221 #endif
 222
 223 /* constants and macros for access to the data ------------------------------ */
 224
 225 /* getting a uint32_t properties word from the data */
 226 #if UCHAR_HARDCODE_DATA
 227
 228 #define GET_PROPS(c, result) UTRIE_GET16(&propsTrie, c, result);
 229
 230 #else
 231
 232 #define HAVE_DATA (havePropsData>0 || loadPropsData()>0)
 233 #define GET_PROPS_UNSAFE(c, result) \
 234     UTRIE_GET16(&propsTrie, c, result);
 235 #define GET_PROPS(c, result) \
 236     if(HAVE_DATA) { \
 237         GET_PROPS_UNSAFE(c, result); \
 238     } else { \
 239         (result)=0; \
 240     }
 241
 242 #endif
 243
 244 U_CFUNC UBool
 245 uprv_haveProperties(UErrorCode *pErrorCode) {
 246     if(U_FAILURE(*pErrorCode)) {
 247         return FALSE;
 248     }
 249 #if !UCHAR_HARDCODE_DATA
 250     if(havePropsData==0) {
 251         uprv_loadPropsData(pErrorCode);
 252     }
 253     if(havePropsData<0) {
 254         *pErrorCode=dataErrorCode;
 255         return FALSE;
 256     }
 257 #endif
 258     return TRUE;
 259 }
 260
 261 /* API functions ------------------------------------------------------------ */
 262
 263 /* Gets the Unicode character's general category.*/
 264 U_CAPI int8_t U_EXPORT2
 265 u_charType(UChar32 c) {
 266     uint32_t props;
 267     GET_PROPS(c, props);
 268     return (int8_t)GET_CATEGORY(props);
 269 }
 270
 271 /* Enumerate all code points with their general categories. */
 272 struct _EnumTypeCallback {
 273     UCharEnumTypeRange *enumRange;
 274     const void *context;
 275 };
 276
 277 static uint32_t U_CALLCONV
 278 _enumTypeValue(const void *context, uint32_t value) {
 279     return GET_CATEGORY(value);
 280 }
 281
 282 static UBool U_CALLCONV
 283 _enumTypeRange(const void *context, UChar32 start, UChar32 limit, uint32_t value) {
 284     /* just cast the value to UCharCategory */
 285     return ((struct _EnumTypeCallback *)context)->
 286         enumRange(((struct _EnumTypeCallback *)context)->context,
 287                   start, limit, (UCharCategory)value);
 288 }
 289
 290 U_CAPI void U_EXPORT2
 291 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context) {
 292     struct _EnumTypeCallback callback;
 293
 294     if(enumRange==NULL
 295 #if !UCHAR_HARDCODE_DATA
 296         || !HAVE_DATA
 297 #endif
 298     ) {
 299         return;
 300     }
 301
 302     callback.enumRange=enumRange;
 303     callback.context=context;
 304     utrie_enum(&propsTrie, _enumTypeValue, _enumTypeRange, &callback);
 305 }
 306
 307 /* Checks if ch is a lower case letter.*/
 308 U_CAPI UBool U_EXPORT2
 309 u_islower(UChar32 c) {
 310     uint32_t props;
 311     GET_PROPS(c, props);
 312     return (UBool)(GET_CATEGORY(props)==U_LOWERCASE_LETTER);
 313 }
 314
 315 /* Checks if ch is an upper case letter.*/
 316 U_CAPI UBool U_EXPORT2
 317 u_isupper(UChar32 c) {
 318     uint32_t props;
 319     GET_PROPS(c, props);
 320     return (UBool)(GET_CATEGORY(props)==U_UPPERCASE_LETTER);
 321 }
 322
 323 /* Checks if ch is a title case letter; usually upper case letters.*/
 324 U_CAPI UBool U_EXPORT2
 325 u_istitle(UChar32 c) {
 326     uint32_t props;
 327     GET_PROPS(c, props);
 328     return (UBool)(GET_CATEGORY(props)==U_TITLECASE_LETTER);
 329 }
 330
 331 /* Checks if ch is a decimal digit. */
 332 U_CAPI UBool U_EXPORT2
 333 u_isdigit(UChar32 c) {
 334     uint32_t props;
 335     GET_PROPS(c, props);
 336     return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER);
 337 }
 338
 339 U_CAPI UBool U_EXPORT2
 340 u_isxdigit(UChar32 c) {
 341     uint32_t props;
 342
 343     /* check ASCII and Fullwidth ASCII a-fA-F */
 344     if(
 345         (c<=0x66 && c>=0x41 && (c<=0x46 || c>=0x61)) ||
 346         (c>=0xff21 && c<=0xff46 && (c<=0xff26 || c>=0xff41))
 347     ) {
 348         return TRUE;
 349     }
 350
 351     GET_PROPS(c, props);
 352     return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER);
 353 }
 354
 355 /* Checks if the Unicode character is a letter.*/
 356 U_CAPI UBool U_EXPORT2
 357 u_isalpha(UChar32 c) {
 358     uint32_t props;
 359     GET_PROPS(c, props);
 360     return (UBool)((CAT_MASK(props)&U_GC_L_MASK)!=0);
 361 }
 362
 363 U_CAPI UBool U_EXPORT2
 364 u_isUAlphabetic(UChar32 c) {
 365     return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_ALPHABETIC))!=0;
 366 }
 367
 368 /* Checks if c is a letter or a decimal digit */
 369 U_CAPI UBool U_EXPORT2
 370 u_isalnum(UChar32 c) {
 371     uint32_t props;
 372     GET_PROPS(c, props);
 373     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_ND_MASK))!=0);
 374 }
 375
 376 /**
 377  * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
 378  * @internal
 379  */
 380 U_CFUNC UBool
 381 u_isalnumPOSIX(UChar32 c) {
 382     return (UBool)(u_isUAlphabetic(c) || u_isdigit(c));
 383 }
 384
 385 /* Checks if ch is a unicode character with assigned character type.*/
 386 U_CAPI UBool U_EXPORT2
 387 u_isdefined(UChar32 c) {
 388     uint32_t props;
 389     GET_PROPS(c, props);
 390     return (UBool)(GET_CATEGORY(props)!=0);
 391 }
 392
 393 /* Checks if the Unicode character is a base form character that can take a diacritic.*/
 394 U_CAPI UBool U_EXPORT2
 395 u_isbase(UChar32 c) {
 396     uint32_t props;
 397     GET_PROPS(c, props);
 398     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_N_MASK|U_GC_MC_MASK|U_GC_ME_MASK))!=0);
 399 }
 400
 401 /* Checks if the Unicode character is a control character.*/
 402 U_CAPI UBool U_EXPORT2
 403 u_iscntrl(UChar32 c) {
 404     uint32_t props;
 405     GET_PROPS(c, props);
 406     return (UBool)((CAT_MASK(props)&(U_GC_CC_MASK|U_GC_CF_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK))!=0);
 407 }
 408
 409 U_CAPI UBool U_EXPORT2
 410 u_isISOControl(UChar32 c) {
 411     return (uint32_t)c<=0x9f && (c<=0x1f || c>=0x7f);
 412 }
 413
 414 /* Some control characters that are used as space. */
 415 #define IS_THAT_CONTROL_SPACE(c) \
 416     (c<=0x9f && ((c>=TAB && c<=CR) || (c>=0x1c && c <=0x1f) || c==NL))
 417
 418 /* Checks if the Unicode character is a space character.*/
 419 U_CAPI UBool U_EXPORT2
 420 u_isspace(UChar32 c) {
 421     uint32_t props;
 422     GET_PROPS(c, props);
 423     return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0 || IS_THAT_CONTROL_SPACE(c));
 424 }
 425
 426 U_CAPI UBool U_EXPORT2
 427 u_isJavaSpaceChar(UChar32 c) {
 428     uint32_t props;
 429     GET_PROPS(c, props);
 430     return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0);
 431 }
 432
 433 /* Checks if the Unicode character is a whitespace character.*/
 434 U_CAPI UBool U_EXPORT2
 435 u_isWhitespace(UChar32 c) {
 436     uint32_t props;
 437     GET_PROPS(c, props);
 438     return (UBool)(
 439                 ((CAT_MASK(props)&U_GC_Z_MASK)!=0 &&
 440                     c!=NBSP && c!=FIGURESP && c!=NNBSP) || /* exclude no-break spaces */
 441                 IS_THAT_CONTROL_SPACE(c)
 442            );
 443 }
 444
 445 U_CAPI UBool U_EXPORT2
 446 u_isblank(UChar32 c) {
 447     if((uint32_t)c<=0x9f) {
 448         return c==9 || c==0x20; /* TAB or SPACE */
 449     } else {
 450         /* Zs */
 451         uint32_t props;
 452         GET_PROPS(c, props);
 453         return (UBool)(GET_CATEGORY(props)==U_SPACE_SEPARATOR);
 454     }
 455 }
 456
 457 U_CAPI UBool U_EXPORT2
 458 u_isUWhiteSpace(UChar32 c) {
 459     return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_WHITE_SPACE))!=0;
 460 }
 461
 462 /* Checks if the Unicode character is printable.*/
 463 U_CAPI UBool U_EXPORT2
 464 u_isprint(UChar32 c) {
 465     uint32_t props;
 466     GET_PROPS(c, props);
 467     /* comparing ==0 returns FALSE for the categories mentioned */
 468     return (UBool)((CAT_MASK(props)&U_GC_C_MASK)==0);
 469 }
 470
 471 /**
 472  * Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
 473  * Implements UCHAR_POSIX_PRINT.
 474  * @internal
 475  */
 476 U_CFUNC UBool
 477 u_isprintPOSIX(UChar32 c) {
 478     uint32_t props;
 479     GET_PROPS(c, props);
 480     /*
 481      * The only cntrl character in graph+blank is TAB (in blank).
 482      * Here we implement (blank-TAB)=Zs instead of calling u_isblank().
 483      */
 484     return (UBool)((GET_CATEGORY(props)==U_SPACE_SEPARATOR) || u_isgraphPOSIX(c));
 485 }
 486
 487 U_CAPI UBool U_EXPORT2
 488 u_isgraph(UChar32 c) {
 489     uint32_t props;
 490     GET_PROPS(c, props);
 491     /* comparing ==0 returns FALSE for the categories mentioned */
 492     return (UBool)((CAT_MASK(props)&
 493                     (U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK))
 494                    ==0);
 495 }
 496
 497 /**
 498  * Checks if c is in
 499  * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
 500  * with space=\p{Whitespace} and Control=Cc.
 501  * Implements UCHAR_POSIX_GRAPH.
 502  * @internal
 503  */
 504 U_CFUNC UBool
 505 u_isgraphPOSIX(UChar32 c) {
 506     uint32_t props;
 507     GET_PROPS(c, props);
 508     /* \p{space}\p{gc=Control} == \p{gc=Z}\p{Control} */
 509     /* comparing ==0 returns FALSE for the categories mentioned */
 510     return (UBool)((CAT_MASK(props)&
 511                     (U_GC_CC_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK))
 512                    ==0);
 513 }
 514
 515 U_CAPI UBool U_EXPORT2
 516 u_ispunct(UChar32 c) {
 517     uint32_t props;
 518     GET_PROPS(c, props);
 519     return (UBool)((CAT_MASK(props)&U_GC_P_MASK)!=0);
 520 }
 521
 522 /* Checks if the Unicode character can start a Unicode identifier.*/
 523 U_CAPI UBool U_EXPORT2
 524 u_isIDStart(UChar32 c) {
 525     /* same as u_isalpha() */
 526     uint32_t props;
 527     GET_PROPS(c, props);
 528     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_NL_MASK))!=0);
 529 }
 530
 531 /* Checks if the Unicode character can be a Unicode identifier part other than starting the
 532  identifier.*/
 533 U_CAPI UBool U_EXPORT2
 534 u_isIDPart(UChar32 c) {
 535     uint32_t props;
 536     GET_PROPS(c, props);
 537     return (UBool)(
 538            (CAT_MASK(props)&
 539             (U_GC_ND_MASK|U_GC_NL_MASK|
 540              U_GC_L_MASK|
 541              U_GC_PC_MASK|U_GC_MC_MASK|U_GC_MN_MASK)
 542            )!=0 ||
 543            u_isIDIgnorable(c));
 544 }
 545
 546 /*Checks if the Unicode character can be ignorable in a Java or Unicode identifier.*/
 547 U_CAPI UBool U_EXPORT2
 548 u_isIDIgnorable(UChar32 c) {
 549     if(c<=0x9f) {
 550         return u_isISOControl(c) && !IS_THAT_CONTROL_SPACE(c);
 551     } else {
 552         uint32_t props;
 553         GET_PROPS(c, props);
 554         return (UBool)(GET_CATEGORY(props)==U_FORMAT_CHAR);
 555     }
 556 }
 557
 558 /*Checks if the Unicode character can start a Java identifier.*/
 559 U_CAPI UBool U_EXPORT2
 560 u_isJavaIDStart(UChar32 c) {
 561     uint32_t props;
 562     GET_PROPS(c, props);
 563     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_SC_MASK|U_GC_PC_MASK))!=0);
 564 }
 565
 566 /*Checks if the Unicode character can be a Java identifier part other than starting the
 567  * identifier.
 568  */
 569 U_CAPI UBool U_EXPORT2
 570 u_isJavaIDPart(UChar32 c) {
 571     uint32_t props;
 572     GET_PROPS(c, props);
 573     return (UBool)(
 574            (CAT_MASK(props)&
 575             (U_GC_ND_MASK|U_GC_NL_MASK|
 576              U_GC_L_MASK|
 577              U_GC_SC_MASK|U_GC_PC_MASK|
 578              U_GC_MC_MASK|U_GC_MN_MASK)
 579            )!=0 ||
 580            u_isIDIgnorable(c));
 581 }
 582
 583 U_CAPI int32_t U_EXPORT2
 584 u_charDigitValue(UChar32 c) {
 585     uint32_t props;
 586     GET_PROPS(c, props);
 587
 588     if(GET_NUMERIC_TYPE(props)==1) {
 589         return GET_NUMERIC_VALUE(props);
 590     } else {
 591         return -1;
 592     }
 593 }
 594
 595 U_CAPI double U_EXPORT2
 596 u_getNumericValue(UChar32 c) {
 597     uint32_t props, numericType, numericValue;
 598     GET_PROPS(c, props);
 599     numericType=GET_NUMERIC_TYPE(props);
 600
 601     if(numericType==0 || numericType>=UPROPS_NT_COUNT) {
 602         return U_NO_NUMERIC_VALUE;
 603     }
 604
 605     numericValue=GET_NUMERIC_VALUE(props);
 606
 607     if(numericType<U_NT_COUNT) {
 608         /* normal type, the value is stored directly */
 609         return numericValue;
 610     } else if(numericType==UPROPS_NT_FRACTION) {
 611         /* fraction value */
 612         int32_t numerator;
 613         uint32_t denominator;
 614
 615         numerator=(int32_t)numericValue>>UPROPS_FRACTION_NUM_SHIFT;
 616         denominator=(numericValue&UPROPS_FRACTION_DEN_MASK)+UPROPS_FRACTION_DEN_OFFSET;
 617
 618         if(numerator==0) {
 619             numerator=-1;
 620         }
 621         return (double)numerator/(double)denominator;
 622     } else /* numericType==UPROPS_NT_LARGE */ {
 623         /* large value with exponent */
 624         double numValue;
 625         int32_t mant, exp;
 626
 627         mant=(int32_t)numericValue>>UPROPS_LARGE_MANT_SHIFT;
 628         exp=(int32_t)numericValue&UPROPS_LARGE_EXP_MASK;
 629         if(mant==0) {
 630             mant=1;
 631             exp+=UPROPS_LARGE_EXP_OFFSET_EXTRA;
 632         } else if(mant>9) {
 633             return U_NO_NUMERIC_VALUE; /* reserved mantissa value */
 634         } else {
 635             exp+=UPROPS_LARGE_EXP_OFFSET;
 636         }
 637
 638         numValue=mant;
 639
 640         /* multiply by 10^exp without math.h */
 641         while(exp>=4) {
 642             numValue*=10000.;
 643             exp-=4;
 644         }
 645         switch(exp) {
 646         case 3:
 647             numValue*=1000.;
 648             break;
 649         case 2:
 650             numValue*=100.;
 651             break;
 652         case 1:
 653             numValue*=10.;
 654             break;
 655         case 0:
 656         default:
 657             break;
 658         }
 659
 660         return numValue;
 661     }
 662 }
 663
 664 /* ICU 3.4: bidi/shaping properties moved to ubidi_props.c */
 665
 666 /* ICU 2.1: u_getCombiningClass() moved to unorm.cpp */
 667
 668 U_CAPI int32_t U_EXPORT2
 669 u_digit(UChar32 ch, int8_t radix) {
 670     int8_t value;
 671     if((uint8_t)(radix-2)<=(36-2)) {
 672         value=(int8_t)u_charDigitValue(ch);
 673         if(value<0) {
 674             /* ch is not a decimal digit, try latin letters */
 675             if(ch>=0x61 && ch<=0x7A) {
 676                 value=(int8_t)(ch-0x57);  /* ch - 'a' + 10 */
 677             } else if(ch>=0x41 && ch<=0x5A) {
 678                 value=(int8_t)(ch-0x37);  /* ch - 'A' + 10 */
 679             } else if(ch>=0xFF41 && ch<=0xFF5A) {
 680                 value=(int8_t)(ch-0xFF37);  /* fullwidth ASCII a-z */
 681             } else if(ch>=0xFF21 && ch<=0xFF3A) {
 682                 value=(int8_t)(ch-0xFF17);  /* fullwidth ASCII A-Z */
 683             }
 684         }
 685     } else {
 686         value=-1;   /* invalid radix */
 687     }
 688     return (int8_t)((value<radix) ? value : -1);
 689 }
 690
 691 U_CAPI UChar32 U_EXPORT2
 692 u_forDigit(int32_t digit, int8_t radix) {
 693     if((uint8_t)(radix-2)>(36-2) || (uint32_t)digit>=(uint32_t)radix) {
 694         return 0;
 695     } else if(digit<10) {
 696         return (UChar32)(0x30+digit);
 697     } else {
 698         return (UChar32)((0x61-10)+digit);
 699     }
 700 }
 701
 702 /* miscellaneous, and support for uprops.c ---------------------------------- */
 703
 704 U_CAPI void U_EXPORT2
 705 u_getUnicodeVersion(UVersionInfo versionArray) {
 706     if(versionArray!=NULL) {
 707         uprv_memcpy(versionArray, dataVersion, U_MAX_VERSION_LENGTH);
 708     }
 709 }
 710
 711 U_CFUNC uint32_t
 712 u_getUnicodeProperties(UChar32 c, int32_t column) {
 713     uint16_t vecIndex;
 714
 715     if(column==-1) {
 716         uint32_t props;
 717         GET_PROPS(c, props);
 718         return props;
 719     } else if(
 720 #if !UCHAR_HARDCODE_DATA
 721                !HAVE_DATA || countPropsVectors==0 ||
 722 #endif
 723                column<0 || column>=propsVectorsColumns
 724     ) {
 725         return 0;
 726     } else {
 727         UTRIE_GET16(&propsVectorsTrie, c, vecIndex);
 728         return propsVectors[vecIndex+column];
 729     }
 730 }
 731
 732 U_CFUNC int32_t
 733 uprv_getMaxValues(int32_t column) {
 734 #if !UCHAR_HARDCODE_DATA
 735     if(HAVE_DATA) {
 736 #endif
 737         switch(column) {
 738         case 0:
 739             return indexes[UPROPS_MAX_VALUES_INDEX];
 740         case 2:
 741             return indexes[UPROPS_MAX_VALUES_2_INDEX];
 742         default:
 743             return 0;
 744         }
 745 #if !UCHAR_HARDCODE_DATA
 746     } else {
 747         return 0;
 748     }
 749 #endif
 750 }
 751
 752 /*
 753  * get Hangul Syllable Type
 754  * implemented here so that uchar.c (uhst_addPropertyStarts())
 755  * does not depend on uprops.c (u_getIntPropertyValue(c, UCHAR_HANGUL_SYLLABLE_TYPE))
 756  */
 757 U_CFUNC UHangulSyllableType
 758 uchar_getHST(UChar32 c) {
 759     /* purely algorithmic; hardcode known characters, check for assigned new ones */
 760     if(c<JAMO_L_BASE) {
 761         /* U_HST_NOT_APPLICABLE */
 762     } else if(c<=0x11ff) {
 763         /* Jamo range */
 764         if(c<=0x115f) {
 765             /* Jamo L range, HANGUL CHOSEONG ... */
 766             if(c==0x115f || c<=0x1159 || u_charType(c)==U_OTHER_LETTER) {
 767                 return U_HST_LEADING_JAMO;
 768             }
 769         } else if(c<=0x11a7) {
 770             /* Jamo V range, HANGUL JUNGSEONG ... */
 771             if(c<=0x11a2 || u_charType(c)==U_OTHER_LETTER) {
 772                 return U_HST_VOWEL_JAMO;
 773             }
 774         } else {
 775             /* Jamo T range */
 776             if(c<=0x11f9 || u_charType(c)==U_OTHER_LETTER) {
 777                 return U_HST_TRAILING_JAMO;
 778             }
 779         }
 780     } else if((c-=HANGUL_BASE)<0) {
 781         /* U_HST_NOT_APPLICABLE */
 782     } else if(c<HANGUL_COUNT) {
 783         /* Hangul syllable */
 784         return c%JAMO_T_COUNT==0 ? U_HST_LV_SYLLABLE : U_HST_LVT_SYLLABLE;
 785     }
 786     return U_HST_NOT_APPLICABLE;
 787 }
 788
 789 U_CAPI void U_EXPORT2
 790 u_charAge(UChar32 c, UVersionInfo versionArray) {
 791     if(versionArray!=NULL) {
 792         uint32_t version=u_getUnicodeProperties(c, 0)>>UPROPS_AGE_SHIFT;
 793         versionArray[0]=(uint8_t)(version>>4);
 794         versionArray[1]=(uint8_t)(version&0xf);
 795         versionArray[2]=versionArray[3]=0;
 796     }
 797 }
 798
 799 U_CAPI UScriptCode U_EXPORT2
 800 uscript_getScript(UChar32 c, UErrorCode *pErrorCode) {
 801     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 802         return USCRIPT_INVALID_CODE;
 803     }
 804     if((uint32_t)c>0x10ffff) {
 805         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 806         return USCRIPT_INVALID_CODE;
 807     }
 808
 809     return (UScriptCode)(u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_MASK);
 810 }
 811
 812 U_CAPI UBlockCode U_EXPORT2
 813 ublock_getCode(UChar32 c) {
 814     return (UBlockCode)((u_getUnicodeProperties(c, 0)&UPROPS_BLOCK_MASK)>>UPROPS_BLOCK_SHIFT);
 815 }
 816
 817 /* property starts for UnicodeSet ------------------------------------------- */
 818
 819 /* for Hangul_Syllable_Type */
 820 U_CFUNC void U_EXPORT2
 821 uhst_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
 822     UChar32 c;
 823     int32_t value, value2;
 824
 825     if(U_FAILURE(*pErrorCode)) {
 826         return;
 827     }
 828
 829 #if !UCHAR_HARDCODE_DATA
 830     if(!HAVE_DATA) {
 831         *pErrorCode=dataErrorCode;
 832         return;
 833     }
 834 #endif
 835
 836     /* add code points with hardcoded properties, plus the ones following them */
 837
 838     /*
 839      * Add Jamo type boundaries for UCHAR_HANGUL_SYLLABLE_TYPE.
 840      * First, we add fixed boundaries for the blocks of Jamos.
 841      * Then we check in loops to see where the current Unicode version
 842      * actually stops assigning such Jamos. We start each loop
 843      * at the end of the per-Jamo-block assignments in Unicode 4 or earlier.
 844      * (These have not changed since Unicode 2.)
 845      */
 846     sa->add(sa->set, 0x1100);
 847     value=U_HST_LEADING_JAMO;
 848     for(c=0x115a; c<=0x115f; ++c) {
 849         value2=uchar_getHST(c);
 850         if(value!=value2) {
 851             value=value2;
 852             sa->add(sa->set, c);
 853         }
 854     }
 855
 856     sa->add(sa->set, 0x1160);
 857     value=U_HST_VOWEL_JAMO;
 858     for(c=0x11a3; c<=0x11a7; ++c) {
 859         value2=uchar_getHST(c);
 860         if(value!=value2) {
 861             value=value2;
 862             sa->add(sa->set, c);
 863         }
 864     }
 865
 866     sa->add(sa->set, 0x11a8);
 867     value=U_HST_TRAILING_JAMO;
 868     for(c=0x11fa; c<=0x11ff; ++c) {
 869         value2=uchar_getHST(c);
 870         if(value!=value2) {
 871             value=value2;
 872             sa->add(sa->set, c);
 873         }
 874     }
 875
 876     /* Add Hangul type boundaries for UCHAR_HANGUL_SYLLABLE_TYPE. */
 877     for(c=HANGUL_BASE; c<(HANGUL_BASE+HANGUL_COUNT); c+=JAMO_T_COUNT) {
 878         sa->add(sa->set, c);
 879         sa->add(sa->set, c+1);
 880     }
 881     sa->add(sa->set, c);
 882 }
 883
 884 static UBool U_CALLCONV
 885 _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 limit, uint32_t value) {
 886     /* add the start code point to the USet */
 887     const USetAdder *sa=(const USetAdder *)context;
 888     sa->add(sa->set, start);
 889     return TRUE;
 890 }
 891
 892 #define USET_ADD_CP_AND_NEXT(sa, cp) sa->add(sa->set, cp); sa->add(sa->set, cp+1)
 893
 894 U_CFUNC void U_EXPORT2
 895 uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
 896     if(U_FAILURE(*pErrorCode)) {
 897         return;
 898     }
 899
 900 #if !UCHAR_HARDCODE_DATA
 901     if(!HAVE_DATA) {
 902         *pErrorCode=dataErrorCode;
 903         return;
 904     }
 905 #endif
 906
 907     /* add the start code point of each same-value range of the main trie */
 908     utrie_enum(&propsTrie, NULL, _enumPropertyStartsRange, sa);
 909
 910     /* add code points with hardcoded properties, plus the ones following them */
 911
 912     /* add for u_isblank() */
 913     USET_ADD_CP_AND_NEXT(sa, TAB);
 914
 915     /* add for IS_THAT_CONTROL_SPACE() */
 916     sa->add(sa->set, CR+1); /* range TAB..CR */
 917     sa->add(sa->set, 0x1c);
 918     sa->add(sa->set, 0x1f+1);
 919     USET_ADD_CP_AND_NEXT(sa, NL);
 920
 921     /* add for u_isIDIgnorable() what was not added above */
 922     sa->add(sa->set, DEL); /* range DEL..NBSP-1, NBSP added below */
 923     sa->add(sa->set, HAIRSP);
 924     sa->add(sa->set, RLM+1);
 925     sa->add(sa->set, INHSWAP);
 926     sa->add(sa->set, NOMDIG+1);
 927     USET_ADD_CP_AND_NEXT(sa, ZWNBSP);
 928
 929     /* add no-break spaces for u_isWhitespace() what was not added above */
 930     USET_ADD_CP_AND_NEXT(sa, NBSP);
 931     USET_ADD_CP_AND_NEXT(sa, FIGURESP);
 932     USET_ADD_CP_AND_NEXT(sa, NNBSP);
 933
 934     /* add for u_digit() */
 935     sa->add(sa->set, U_a);
 936     sa->add(sa->set, U_z+1);
 937     sa->add(sa->set, U_A);
 938     sa->add(sa->set, U_Z+1);
 939     sa->add(sa->set, U_FW_a);
 940     sa->add(sa->set, U_FW_z+1);
 941     sa->add(sa->set, U_FW_A);
 942     sa->add(sa->set, U_FW_Z+1);
 943
 944     /* add for u_isxdigit() */
 945     sa->add(sa->set, U_f+1);
 946     sa->add(sa->set, U_F+1);
 947     sa->add(sa->set, U_FW_f+1);
 948     sa->add(sa->set, U_FW_F+1);
 949
 950     /* add for UCHAR_DEFAULT_IGNORABLE_CODE_POINT what was not added above */
 951     sa->add(sa->set, WJ); /* range WJ..NOMDIG */
 952     sa->add(sa->set, 0xfff0);
 953     sa->add(sa->set, 0xfffb+1);
 954     sa->add(sa->set, 0xe0000);
 955     sa->add(sa->set, 0xe0fff+1);
 956
 957     /* add for UCHAR_GRAPHEME_BASE and others */
 958     USET_ADD_CP_AND_NEXT(sa, CGJ);
 959 }
 960
 961 U_CFUNC void U_EXPORT2
 962 upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
 963     if(U_FAILURE(*pErrorCode)) {
 964         return;
 965     }
 966
 967 #if !UCHAR_HARDCODE_DATA
 968     if(!HAVE_DATA) {
 969         *pErrorCode=dataErrorCode;
 970         return;
 971     }
 972 #endif
 973
 974     /* add the start code point of each same-value range of the properties vectors trie */
 975     if(propsVectorsColumns>0) {
 976         /* if propsVectorsColumns==0 then the properties vectors trie may not be there at all */
 977         utrie_enum(&propsVectorsTrie, NULL, _enumPropertyStartsRange, sa);
 978     }
 979 }