icuSources/common/uprops.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 2002-2016, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *   file name:  uprops.cpp
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2002feb24
  16 *   created by: Markus W. Scherer
  17 *
  18 *   Implementations for mostly non-core Unicode character properties
  19 *   stored in uprops.icu.
  20 *
  21 *   With the APIs implemented here, almost all properties files and
  22 *   their associated implementation files are used from this file,
  23 *   including those for normalization and case mappings.
  24 */
  25
  26 #include "unicode/utypes.h"
  27 #include "unicode/uchar.h"
  28 #include "unicode/unorm2.h"
  29 #include "unicode/uscript.h"
  30 #include "unicode/ustring.h"
  31 #include "cstring.h"
  32 #include "normalizer2impl.h"
  33 #include "umutex.h"
  34 #include "ubidi_props.h"
  35 #include "uprops.h"
  36 #include "ucase.h"
  37 #include "ustr_imp.h"
  38
  39 U_NAMESPACE_USE
  40
  41 /* general properties API functions ----------------------------------------- */
  42
  43 struct BinaryProperty;
  44
  45 typedef UBool BinaryPropertyContains(const BinaryProperty &prop, UChar32 c, UProperty which);
  46
  47 struct BinaryProperty {
  48     int32_t column;  // SRC_PROPSVEC column, or "source" if mask==0
  49     uint32_t mask;
  50     BinaryPropertyContains *contains;
  51 };
  52
  53 static UBool defaultContains(const BinaryProperty &prop, UChar32 c, UProperty /*which*/) {
  54     /* systematic, directly stored properties */
  55     return (u_getUnicodeProperties(c, prop.column)&prop.mask)!=0;
  56 }
  57
  58 static UBool caseBinaryPropertyContains(const BinaryProperty &/*prop*/, UChar32 c, UProperty which) {
  59     return ucase_hasBinaryProperty(c, which);
  60 }
  61
  62 static UBool isBidiControl(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
  63     return ubidi_isBidiControl(c);
  64 }
  65
  66 static UBool isMirrored(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
  67     return ubidi_isMirrored(c);
  68 }
  69
  70 static UBool isJoinControl(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
  71     return ubidi_isJoinControl(c);
  72 }
  73
  74 #if UCONFIG_NO_NORMALIZATION
  75 static UBool hasFullCompositionExclusion(const BinaryProperty &, UChar32, UProperty) {
  76     return FALSE;
  77 }
  78 #else
  79 static UBool hasFullCompositionExclusion(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
  80     // By definition, Full_Composition_Exclusion is the same as NFC_QC=No.
  81     UErrorCode errorCode=U_ZERO_ERROR;
  82     const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode);
  83     return U_SUCCESS(errorCode) && impl->isCompNo(impl->getNorm16(c));
  84 }
  85 #endif
  86
  87 // UCHAR_NF*_INERT properties
  88 #if UCONFIG_NO_NORMALIZATION
  89 static UBool isNormInert(const BinaryProperty &, UChar32, UProperty) {
  90     return FALSE;
  91 }
  92 #else
  93 static UBool isNormInert(const BinaryProperty &/*prop*/, UChar32 c, UProperty which) {
  94     UErrorCode errorCode=U_ZERO_ERROR;
  95     const Normalizer2 *norm2=Normalizer2Factory::getInstance(
  96         (UNormalizationMode)(which-UCHAR_NFD_INERT+UNORM_NFD), errorCode);
  97     return U_SUCCESS(errorCode) && norm2->isInert(c);
  98 }
  99 #endif
 100
 101 #if UCONFIG_NO_NORMALIZATION
 102 static UBool changesWhenCasefolded(const BinaryProperty &, UChar32, UProperty) {
 103     return FALSE;
 104 }
 105 #else
 106 static UBool changesWhenCasefolded(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 107     UnicodeString nfd;
 108     UErrorCode errorCode=U_ZERO_ERROR;
 109     const Normalizer2 *nfcNorm2=Normalizer2::getNFCInstance(errorCode);
 110     if(U_FAILURE(errorCode)) {
 111         return FALSE;
 112     }
 113     if(nfcNorm2->getDecomposition(c, nfd)) {
 114         /* c has a decomposition */
 115         if(nfd.length()==1) {
 116             c=nfd[0];  /* single BMP code point */
 117         } else if(nfd.length()<=U16_MAX_LENGTH &&
 118                   nfd.length()==U16_LENGTH(c=nfd.char32At(0))
 119         ) {
 120             /* single supplementary code point */
 121         } else {
 122             c=U_SENTINEL;
 123         }
 124     } else if(c<0) {
 125         return FALSE;  /* protect against bad input */
 126     }
 127     if(c>=0) {
 128         /* single code point */
 129         const UChar *resultString;
 130         return (UBool)(ucase_toFullFolding(c, &resultString, U_FOLD_CASE_DEFAULT)>=0);
 131     } else {
 132         /* guess some large but stack-friendly capacity */
 133         UChar dest[2*UCASE_MAX_STRING_LENGTH];
 134         int32_t destLength;
 135         destLength=u_strFoldCase(dest, UPRV_LENGTHOF(dest),
 136                                   nfd.getBuffer(), nfd.length(),
 137                                   U_FOLD_CASE_DEFAULT, &errorCode);
 138         return (UBool)(U_SUCCESS(errorCode) &&
 139                        0!=u_strCompare(nfd.getBuffer(), nfd.length(),
 140                                        dest, destLength, FALSE));
 141     }
 142 }
 143 #endif
 144
 145 #if UCONFIG_NO_NORMALIZATION
 146 static UBool changesWhenNFKC_Casefolded(const BinaryProperty &, UChar32, UProperty) {
 147     return FALSE;
 148 }
 149 #else
 150 static UBool changesWhenNFKC_Casefolded(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 151     UErrorCode errorCode=U_ZERO_ERROR;
 152     const Normalizer2Impl *kcf=Normalizer2Factory::getNFKC_CFImpl(errorCode);
 153     if(U_FAILURE(errorCode)) {
 154         return FALSE;
 155     }
 156     UnicodeString src(c);
 157     UnicodeString dest;
 158     {
 159         // The ReorderingBuffer must be in a block because its destructor
 160         // needs to release dest's buffer before we look at its contents.
 161         ReorderingBuffer buffer(*kcf, dest);
 162         // Small destCapacity for NFKC_CF(c).
 163         if(buffer.init(5, errorCode)) {
 164             const UChar *srcArray=src.getBuffer();
 165             kcf->compose(srcArray, srcArray+src.length(), FALSE,
 166                           TRUE, buffer, errorCode);
 167         }
 168     }
 169     return U_SUCCESS(errorCode) && dest!=src;
 170 }
 171 #endif
 172
 173 #if UCONFIG_NO_NORMALIZATION
 174 static UBool isCanonSegmentStarter(const BinaryProperty &, UChar32, UProperty) {
 175     return FALSE;
 176 }
 177 #else
 178 static UBool isCanonSegmentStarter(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 179     UErrorCode errorCode=U_ZERO_ERROR;
 180     const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode);
 181     return
 182         U_SUCCESS(errorCode) && impl->ensureCanonIterData(errorCode) &&
 183         impl->isCanonSegmentStarter(c);
 184 }
 185 #endif
 186
 187 static UBool isPOSIX_alnum(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 188     return u_isalnumPOSIX(c);
 189 }
 190
 191 static UBool isPOSIX_blank(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 192     return u_isblank(c);
 193 }
 194
 195 static UBool isPOSIX_graph(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 196     return u_isgraphPOSIX(c);
 197 }
 198
 199 static UBool isPOSIX_print(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 200     return u_isprintPOSIX(c);
 201 }
 202
 203 static UBool isPOSIX_xdigit(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 204     return u_isxdigit(c);
 205 }
 206
 207 static UBool isRegionalIndicator(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 208     // Property starts are a subset of lb=RI etc.
 209     return 0x1F1E6<=c && c<=0x1F1FF;
 210 }
 211
 212 static const BinaryProperty binProps[UCHAR_BINARY_LIMIT]={
 213     /*
 214      * column and mask values for binary properties from u_getUnicodeProperties().
 215      * Must be in order of corresponding UProperty,
 216      * and there must be exactly one entry per binary UProperty.
 217      *
 218      * Properties with mask==0 are handled in code.
 219      * For them, column is the UPropertySource value.
 220      */
 221     { 1,                U_MASK(UPROPS_ALPHABETIC), defaultContains },
 222     { 1,                U_MASK(UPROPS_ASCII_HEX_DIGIT), defaultContains },
 223     { UPROPS_SRC_BIDI,  0, isBidiControl },
 224     { UPROPS_SRC_BIDI,  0, isMirrored },
 225     { 1,                U_MASK(UPROPS_DASH), defaultContains },
 226     { 1,                U_MASK(UPROPS_DEFAULT_IGNORABLE_CODE_POINT), defaultContains },
 227     { 1,                U_MASK(UPROPS_DEPRECATED), defaultContains },
 228     { 1,                U_MASK(UPROPS_DIACRITIC), defaultContains },
 229     { 1,                U_MASK(UPROPS_EXTENDER), defaultContains },
 230     { UPROPS_SRC_NFC,   0, hasFullCompositionExclusion },
 231     { 1,                U_MASK(UPROPS_GRAPHEME_BASE), defaultContains },
 232     { 1,                U_MASK(UPROPS_GRAPHEME_EXTEND), defaultContains },
 233     { 1,                U_MASK(UPROPS_GRAPHEME_LINK), defaultContains },
 234     { 1,                U_MASK(UPROPS_HEX_DIGIT), defaultContains },
 235     { 1,                U_MASK(UPROPS_HYPHEN), defaultContains },
 236     { 1,                U_MASK(UPROPS_ID_CONTINUE), defaultContains },
 237     { 1,                U_MASK(UPROPS_ID_START), defaultContains },
 238     { 1,                U_MASK(UPROPS_IDEOGRAPHIC), defaultContains },
 239     { 1,                U_MASK(UPROPS_IDS_BINARY_OPERATOR), defaultContains },
 240     { 1,                U_MASK(UPROPS_IDS_TRINARY_OPERATOR), defaultContains },
 241     { UPROPS_SRC_BIDI,  0, isJoinControl },
 242     { 1,                U_MASK(UPROPS_LOGICAL_ORDER_EXCEPTION), defaultContains },
 243     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_LOWERCASE
 244     { 1,                U_MASK(UPROPS_MATH), defaultContains },
 245     { 1,                U_MASK(UPROPS_NONCHARACTER_CODE_POINT), defaultContains },
 246     { 1,                U_MASK(UPROPS_QUOTATION_MARK), defaultContains },
 247     { 1,                U_MASK(UPROPS_RADICAL), defaultContains },
 248     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_SOFT_DOTTED
 249     { 1,                U_MASK(UPROPS_TERMINAL_PUNCTUATION), defaultContains },
 250     { 1,                U_MASK(UPROPS_UNIFIED_IDEOGRAPH), defaultContains },
 251     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_UPPERCASE
 252     { 1,                U_MASK(UPROPS_WHITE_SPACE), defaultContains },
 253     { 1,                U_MASK(UPROPS_XID_CONTINUE), defaultContains },
 254     { 1,                U_MASK(UPROPS_XID_START), defaultContains },
 255     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CASE_SENSITIVE
 256     { 1,                U_MASK(UPROPS_S_TERM), defaultContains },
 257     { 1,                U_MASK(UPROPS_VARIATION_SELECTOR), defaultContains },
 258     { UPROPS_SRC_NFC,   0, isNormInert },  // UCHAR_NFD_INERT
 259     { UPROPS_SRC_NFKC,  0, isNormInert },  // UCHAR_NFKD_INERT
 260     { UPROPS_SRC_NFC,   0, isNormInert },  // UCHAR_NFC_INERT
 261     { UPROPS_SRC_NFKC,  0, isNormInert },  // UCHAR_NFKC_INERT
 262     { UPROPS_SRC_NFC_CANON_ITER, 0, isCanonSegmentStarter },
 263     { 1,                U_MASK(UPROPS_PATTERN_SYNTAX), defaultContains },
 264     { 1,                U_MASK(UPROPS_PATTERN_WHITE_SPACE), defaultContains },
 265     { UPROPS_SRC_CHAR_AND_PROPSVEC,  0, isPOSIX_alnum },
 266     { UPROPS_SRC_CHAR,  0, isPOSIX_blank },
 267     { UPROPS_SRC_CHAR,  0, isPOSIX_graph },
 268     { UPROPS_SRC_CHAR,  0, isPOSIX_print },
 269     { UPROPS_SRC_CHAR,  0, isPOSIX_xdigit },
 270     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CASED
 271     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CASE_IGNORABLE
 272     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CHANGES_WHEN_LOWERCASED
 273     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CHANGES_WHEN_UPPERCASED
 274     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CHANGES_WHEN_TITLECASED
 275     { UPROPS_SRC_CASE_AND_NORM,  0, changesWhenCasefolded },
 276     { UPROPS_SRC_CASE,  0, caseBinaryPropertyContains },  // UCHAR_CHANGES_WHEN_CASEMAPPED
 277     { UPROPS_SRC_NFKC_CF, 0, changesWhenNFKC_Casefolded },
 278     { 2,                U_MASK(UPROPS_2_EMOJI), defaultContains },
 279     { 2,                U_MASK(UPROPS_2_EMOJI_PRESENTATION), defaultContains },
 280     { 2,                U_MASK(UPROPS_2_EMOJI_MODIFIER), defaultContains },
 281     { 2,                U_MASK(UPROPS_2_EMOJI_MODIFIER_BASE), defaultContains },
 282     { 2,                U_MASK(UPROPS_2_EMOJI_COMPONENT), defaultContains },
 283     { 2,                0, isRegionalIndicator },
 284     { 1,                U_MASK(UPROPS_PREPENDED_CONCATENATION_MARK), defaultContains },
 285     { 2,                U_MASK(UPROPS_2_EXTENDED_PICTOGRAPHIC), defaultContains },
 286 };
 287
 288 U_CAPI UBool U_EXPORT2
 289 u_hasBinaryProperty(UChar32 c, UProperty which) {
 290     /* c is range-checked in the functions that are called from here */
 291     if(which<UCHAR_BINARY_START || UCHAR_BINARY_LIMIT<=which) {
 292         /* not a known binary property */
 293         return FALSE;
 294     } else {
 295         const BinaryProperty &prop=binProps[which];
 296         return prop.contains(prop, c, which);
 297     }
 298 }
 299
 300 // Apple-only specific version of the above
 301 U_CAPI UBool U_EXPORT2
 302 u_isEmoji(UChar32 c) {
 303     const BinaryProperty &prop=binProps[UCHAR_EMOJI];
 304     return prop.contains(prop, c, UCHAR_EMOJI);
 305 }
 306
 307 struct IntProperty;
 308
 309 typedef int32_t IntPropertyGetValue(const IntProperty &prop, UChar32 c, UProperty which);
 310 typedef int32_t IntPropertyGetMaxValue(const IntProperty &prop, UProperty which);
 311
 312 struct IntProperty {
 313     int32_t column;  // SRC_PROPSVEC column, or "source" if mask==0
 314     uint32_t mask;
 315     int32_t shift;  // =maxValue if getMaxValueFromShift() is used
 316     IntPropertyGetValue *getValue;
 317     IntPropertyGetMaxValue *getMaxValue;
 318 };
 319
 320 static int32_t defaultGetValue(const IntProperty &prop, UChar32 c, UProperty /*which*/) {
 321     /* systematic, directly stored properties */
 322     return (int32_t)(u_getUnicodeProperties(c, prop.column)&prop.mask)>>prop.shift;
 323 }
 324
 325 static int32_t defaultGetMaxValue(const IntProperty &prop, UProperty /*which*/) {
 326     return (uprv_getMaxValues(prop.column)&prop.mask)>>prop.shift;
 327 }
 328
 329 static int32_t getMaxValueFromShift(const IntProperty &prop, UProperty /*which*/) {
 330     return prop.shift;
 331 }
 332
 333 static int32_t getBiDiClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 334     return (int32_t)u_charDirection(c);
 335 }
 336
 337 static int32_t getBiDiPairedBracketType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 338     return (int32_t)ubidi_getPairedBracketType(c);
 339 }
 340
 341 static int32_t biDiGetMaxValue(const IntProperty &/*prop*/, UProperty which) {
 342     return ubidi_getMaxValue(which);
 343 }
 344
 345 #if UCONFIG_NO_NORMALIZATION
 346 static int32_t getCombiningClass(const IntProperty &, UChar32, UProperty) {
 347     return 0;
 348 }
 349 #else
 350 static int32_t getCombiningClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 351     return u_getCombiningClass(c);
 352 }
 353 #endif
 354
 355 static int32_t getGeneralCategory(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 356     return (int32_t)u_charType(c);
 357 }
 358
 359 static int32_t getJoiningGroup(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 360     return ubidi_getJoiningGroup(c);
 361 }
 362
 363 static int32_t getJoiningType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 364     return ubidi_getJoiningType(c);
 365 }
 366
 367 static int32_t getNumericType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 368     int32_t ntv=(int32_t)GET_NUMERIC_TYPE_VALUE(u_getMainProperties(c));
 369     return UPROPS_NTV_GET_TYPE(ntv);
 370 }
 371
 372 static int32_t getScript(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 373     UErrorCode errorCode=U_ZERO_ERROR;
 374     return (int32_t)uscript_getScript(c, &errorCode);
 375 }
 376
 377 /*
 378  * Map some of the Grapheme Cluster Break values to Hangul Syllable Types.
 379  * Hangul_Syllable_Type is fully redundant with a subset of Grapheme_Cluster_Break.
 380  */
 381 static const UHangulSyllableType gcbToHst[]={
 382     U_HST_NOT_APPLICABLE,   /* U_GCB_OTHER */
 383     U_HST_NOT_APPLICABLE,   /* U_GCB_CONTROL */
 384     U_HST_NOT_APPLICABLE,   /* U_GCB_CR */
 385     U_HST_NOT_APPLICABLE,   /* U_GCB_EXTEND */
 386     U_HST_LEADING_JAMO,     /* U_GCB_L */
 387     U_HST_NOT_APPLICABLE,   /* U_GCB_LF */
 388     U_HST_LV_SYLLABLE,      /* U_GCB_LV */
 389     U_HST_LVT_SYLLABLE,     /* U_GCB_LVT */
 390     U_HST_TRAILING_JAMO,    /* U_GCB_T */
 391     U_HST_VOWEL_JAMO        /* U_GCB_V */
 392     /*
 393      * Omit GCB values beyond what we need for hst.
 394      * The code below checks for the array length.
 395      */
 396 };
 397
 398 static int32_t getHangulSyllableType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 399     /* see comments on gcbToHst[] above */
 400     int32_t gcb=(int32_t)(u_getUnicodeProperties(c, 2)&UPROPS_GCB_MASK)>>UPROPS_GCB_SHIFT;
 401     if(gcb<UPRV_LENGTHOF(gcbToHst)) {
 402         return gcbToHst[gcb];
 403     } else {
 404         return U_HST_NOT_APPLICABLE;
 405     }
 406 }
 407
 408 #if UCONFIG_NO_NORMALIZATION
 409 static int32_t getNormQuickCheck(const IntProperty &, UChar32, UProperty) {
 410     return 0;
 411 }
 412 #else
 413 static int32_t getNormQuickCheck(const IntProperty &/*prop*/, UChar32 c, UProperty which) {
 414     return (int32_t)unorm_getQuickCheck(c, (UNormalizationMode)(which-UCHAR_NFD_QUICK_CHECK+UNORM_NFD));
 415 }
 416 #endif
 417
 418 #if UCONFIG_NO_NORMALIZATION
 419 static int32_t getLeadCombiningClass(const IntProperty &, UChar32, UProperty) {
 420     return 0;
 421 }
 422 #else
 423 static int32_t getLeadCombiningClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 424     return unorm_getFCD16(c)>>8;
 425 }
 426 #endif
 427
 428 #if UCONFIG_NO_NORMALIZATION
 429 static int32_t getTrailCombiningClass(const IntProperty &, UChar32, UProperty) {
 430     return 0;
 431 }
 432 #else
 433 static int32_t getTrailCombiningClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
 434     return unorm_getFCD16(c)&0xff;
 435 }
 436 #endif
 437
 438 static const IntProperty intProps[UCHAR_INT_LIMIT-UCHAR_INT_START]={
 439     /*
 440      * column, mask and shift values for int-value properties from u_getUnicodeProperties().
 441      * Must be in order of corresponding UProperty,
 442      * and there must be exactly one entry per int UProperty.
 443      *
 444      * Properties with mask==0 are handled in code.
 445      * For them, column is the UPropertySource value.
 446      */
 447     { UPROPS_SRC_BIDI,  0, 0,                               getBiDiClass, biDiGetMaxValue },
 448     { 0,                UPROPS_BLOCK_MASK, UPROPS_BLOCK_SHIFT, defaultGetValue, defaultGetMaxValue },
 449     { UPROPS_SRC_NFC,   0, 0xff,                            getCombiningClass, getMaxValueFromShift },
 450     { 2,                UPROPS_DT_MASK, 0,                  defaultGetValue, defaultGetMaxValue },
 451     { 0,                UPROPS_EA_MASK, UPROPS_EA_SHIFT,    defaultGetValue, defaultGetMaxValue },
 452     { UPROPS_SRC_CHAR,  0, (int32_t)U_CHAR_CATEGORY_COUNT-1,getGeneralCategory, getMaxValueFromShift },
 453     { UPROPS_SRC_BIDI,  0, 0,                               getJoiningGroup, biDiGetMaxValue },
 454     { UPROPS_SRC_BIDI,  0, 0,                               getJoiningType, biDiGetMaxValue },
 455     { 2,                UPROPS_LB_MASK, UPROPS_LB_SHIFT,    defaultGetValue, defaultGetMaxValue },
 456     { UPROPS_SRC_CHAR,  0, (int32_t)U_NT_COUNT-1,           getNumericType, getMaxValueFromShift },
 457     { 0,                UPROPS_SCRIPT_MASK, 0,              getScript, defaultGetMaxValue },
 458     { UPROPS_SRC_PROPSVEC, 0, (int32_t)U_HST_COUNT-1,       getHangulSyllableType, getMaxValueFromShift },
 459     // UCHAR_NFD_QUICK_CHECK: max=1=YES -- never "maybe", only "no" or "yes"
 460     { UPROPS_SRC_NFC,   0, (int32_t)UNORM_YES,              getNormQuickCheck, getMaxValueFromShift },
 461     // UCHAR_NFKD_QUICK_CHECK: max=1=YES -- never "maybe", only "no" or "yes"
 462     { UPROPS_SRC_NFKC,  0, (int32_t)UNORM_YES,              getNormQuickCheck, getMaxValueFromShift },
 463     // UCHAR_NFC_QUICK_CHECK: max=2=MAYBE
 464     { UPROPS_SRC_NFC,   0, (int32_t)UNORM_MAYBE,            getNormQuickCheck, getMaxValueFromShift },
 465     // UCHAR_NFKC_QUICK_CHECK: max=2=MAYBE
 466     { UPROPS_SRC_NFKC,  0, (int32_t)UNORM_MAYBE,            getNormQuickCheck, getMaxValueFromShift },
 467     { UPROPS_SRC_NFC,   0, 0xff,                            getLeadCombiningClass, getMaxValueFromShift },
 468     { UPROPS_SRC_NFC,   0, 0xff,                            getTrailCombiningClass, getMaxValueFromShift },
 469     { 2,                UPROPS_GCB_MASK, UPROPS_GCB_SHIFT,  defaultGetValue, defaultGetMaxValue },
 470     { 2,                UPROPS_SB_MASK, UPROPS_SB_SHIFT,    defaultGetValue, defaultGetMaxValue },
 471     { 2,                UPROPS_WB_MASK, UPROPS_WB_SHIFT,    defaultGetValue, defaultGetMaxValue },
 472     { UPROPS_SRC_BIDI,  0, 0,                               getBiDiPairedBracketType, biDiGetMaxValue },
 473 };
 474
 475 U_CAPI int32_t U_EXPORT2
 476 u_getIntPropertyValue(UChar32 c, UProperty which) {
 477     if(which<UCHAR_INT_START) {
 478         if(UCHAR_BINARY_START<=which && which<UCHAR_BINARY_LIMIT) {
 479             const BinaryProperty &prop=binProps[which];
 480             return prop.contains(prop, c, which);
 481         }
 482     } else if(which<UCHAR_INT_LIMIT) {
 483         const IntProperty &prop=intProps[which-UCHAR_INT_START];
 484         return prop.getValue(prop, c, which);
 485     } else if(which==UCHAR_GENERAL_CATEGORY_MASK) {
 486         return U_MASK(u_charType(c));
 487     }
 488     return 0;  // undefined
 489 }
 490
 491 U_CAPI int32_t U_EXPORT2
 492 u_getIntPropertyMinValue(UProperty /*which*/) {
 493     return 0; /* all binary/enum/int properties have a minimum value of 0 */
 494 }
 495
 496 U_CAPI int32_t U_EXPORT2
 497 u_getIntPropertyMaxValue(UProperty which) {
 498     if(which<UCHAR_INT_START) {
 499         if(UCHAR_BINARY_START<=which && which<UCHAR_BINARY_LIMIT) {
 500             return 1;  // maximum TRUE for all binary properties
 501         }
 502     } else if(which<UCHAR_INT_LIMIT) {
 503         const IntProperty &prop=intProps[which-UCHAR_INT_START];
 504         return prop.getMaxValue(prop, which);
 505     }
 506     return -1;  // undefined
 507 }
 508
 509 U_CFUNC UPropertySource U_EXPORT2
 510 uprops_getSource(UProperty which) {
 511     if(which<UCHAR_BINARY_START) {
 512         return UPROPS_SRC_NONE; /* undefined */
 513     } else if(which<UCHAR_BINARY_LIMIT) {
 514         const BinaryProperty &prop=binProps[which];
 515         if(prop.mask!=0) {
 516             return UPROPS_SRC_PROPSVEC;
 517         } else {
 518             return (UPropertySource)prop.column;
 519         }
 520     } else if(which<UCHAR_INT_START) {
 521         return UPROPS_SRC_NONE; /* undefined */
 522     } else if(which<UCHAR_INT_LIMIT) {
 523         const IntProperty &prop=intProps[which-UCHAR_INT_START];
 524         if(prop.mask!=0) {
 525             return UPROPS_SRC_PROPSVEC;
 526         } else {
 527             return (UPropertySource)prop.column;
 528         }
 529     } else if(which<UCHAR_STRING_START) {
 530         switch(which) {
 531         case UCHAR_GENERAL_CATEGORY_MASK:
 532         case UCHAR_NUMERIC_VALUE:
 533             return UPROPS_SRC_CHAR;
 534
 535         default:
 536             return UPROPS_SRC_NONE;
 537         }
 538     } else if(which<UCHAR_STRING_LIMIT) {
 539         switch(which) {
 540         case UCHAR_AGE:
 541             return UPROPS_SRC_PROPSVEC;
 542
 543         case UCHAR_BIDI_MIRRORING_GLYPH:
 544             return UPROPS_SRC_BIDI;
 545
 546         case UCHAR_CASE_FOLDING:
 547         case UCHAR_LOWERCASE_MAPPING:
 548         case UCHAR_SIMPLE_CASE_FOLDING:
 549         case UCHAR_SIMPLE_LOWERCASE_MAPPING:
 550         case UCHAR_SIMPLE_TITLECASE_MAPPING:
 551         case UCHAR_SIMPLE_UPPERCASE_MAPPING:
 552         case UCHAR_TITLECASE_MAPPING:
 553         case UCHAR_UPPERCASE_MAPPING:
 554             return UPROPS_SRC_CASE;
 555
 556         case UCHAR_ISO_COMMENT:
 557         case UCHAR_NAME:
 558         case UCHAR_UNICODE_1_NAME:
 559             return UPROPS_SRC_NAMES;
 560
 561         default:
 562             return UPROPS_SRC_NONE;
 563         }
 564     } else {
 565         switch(which) {
 566         case UCHAR_SCRIPT_EXTENSIONS:
 567             return UPROPS_SRC_PROPSVEC;
 568         default:
 569             return UPROPS_SRC_NONE; /* undefined */
 570         }
 571     }
 572 }
 573
 574 #if !UCONFIG_NO_NORMALIZATION
 575
 576 U_CAPI int32_t U_EXPORT2
 577 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode) {
 578     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 579         return 0;
 580     }
 581     if(destCapacity<0 || (dest==NULL && destCapacity>0)) {
 582         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 583         return 0;
 584     }
 585     // Compute the FC_NFKC_Closure on the fly:
 586     // We have the API for complete coverage of Unicode properties, although
 587     // this value by itself is not useful via API.
 588     // (What could be useful is a custom normalization table that combines
 589     // case folding and NFKC.)
 590     // For the derivation, see Unicode's DerivedNormalizationProps.txt.
 591     const Normalizer2 *nfkc=Normalizer2::getNFKCInstance(*pErrorCode);
 592     if(U_FAILURE(*pErrorCode)) {
 593         return 0;
 594     }
 595     // first: b = NFKC(Fold(a))
 596     UnicodeString folded1String;
 597     const UChar *folded1;
 598     int32_t folded1Length=ucase_toFullFolding(c, &folded1, U_FOLD_CASE_DEFAULT);
 599     if(folded1Length<0) {
 600         const Normalizer2Impl *nfkcImpl=Normalizer2Factory::getImpl(nfkc);
 601         if(nfkcImpl->getCompQuickCheck(nfkcImpl->getNorm16(c))!=UNORM_NO) {
 602             return u_terminateUChars(dest, destCapacity, 0, pErrorCode);  // c does not change at all under CaseFolding+NFKC
 603         }
 604         folded1String.setTo(c);
 605     } else {
 606         if(folded1Length>UCASE_MAX_STRING_LENGTH) {
 607             folded1String.setTo(folded1Length);
 608         } else {
 609             folded1String.setTo(FALSE, folded1, folded1Length);
 610         }
 611     }
 612     UnicodeString kc1=nfkc->normalize(folded1String, *pErrorCode);
 613     // second: c = NFKC(Fold(b))
 614     UnicodeString folded2String(kc1);
 615     UnicodeString kc2=nfkc->normalize(folded2String.foldCase(), *pErrorCode);
 616     // if (c != b) add the mapping from a to c
 617     if(U_FAILURE(*pErrorCode) || kc1==kc2) {
 618         return u_terminateUChars(dest, destCapacity, 0, pErrorCode);
 619     } else {
 620         return kc2.extract(dest, destCapacity, *pErrorCode);
 621     }
 622 }
 623
 624 #endif