icuSources/common/uprops.cpp

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