1 // 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html 3 /* 4 *************************************************************************** 5 * Copyright (C) 2008-2015, International Business Machines Corporation 6 * and others. All Rights Reserved. 7 *************************************************************************** 8 * file name: uspoof.cpp 9 * encoding: UTF-8 10 * tab size: 8 (not used) 11 * indentation:4 12 * 13 * created on: 2008Feb13 14 * created by: Andy Heninger 15 * 16 * Unicode Spoof Detection 17 */ 18 #include "unicode/utypes.h" 19 #include "unicode/normalizer2.h" 20 #include "unicode/uspoof.h" 21 #include "unicode/ustring.h" 22 #include "unicode/utf16.h" 23 #include "cmemory.h" 24 #include "cstring.h" 25 #include "mutex.h" 26 #include "scriptset.h" 27 #include "uassert.h" 28 #include "ucln_in.h" 29 #include "uspoof_impl.h" 30 #include "umutex.h" 31 32 33 #if !UCONFIG_NO_NORMALIZATION 34 35 U_NAMESPACE_USE 36 37 38 // 39 // Static Objects used by the spoof impl, their thread safe initialization and their cleanup. 40 // 41 static UnicodeSet *gInclusionSet = NULL; 42 static UnicodeSet *gRecommendedSet = NULL; 43 static const Normalizer2 *gNfdNormalizer = NULL; 44 static UInitOnce gSpoofInitStaticsOnce = U_INITONCE_INITIALIZER; 45 46 static UBool U_CALLCONV 47 uspoof_cleanup(void) { 48 delete gInclusionSet; 49 gInclusionSet = NULL; 50 delete gRecommendedSet; 51 gRecommendedSet = NULL; 52 gNfdNormalizer = NULL; 53 gSpoofInitStaticsOnce.reset(); 54 return TRUE; 55 } 56 57 static void U_CALLCONV initializeStatics(UErrorCode &status) { 58 static const char *inclusionPat = 59 "['\\-.\\:\\u00B7\\u0375\\u058A\\u05F3\\u05F4\\u06FD\\u06FE\\u0F0B\\u200C\\u200D\\u2010\\u" 60 "2019\\u2027\\u30A0\\u30FB]"; 61 gInclusionSet = new UnicodeSet(UnicodeString(inclusionPat, -1, US_INV), status); 62 gInclusionSet->freeze(); 63 64 // Note: data from http://unicode.org/Public/security/9.0.0/IdentifierStatus.txt 65 // There is tooling to generate this constant in the unicodetools project: 66 // org.unicode.text.tools.RecommendedSetGenerator 67 // It will print the Java and C++ code to the console for easy copy-paste into this file. 68 // Note: concatenated string constants do not work with UNICODE_STRING_SIMPLE on all platforms. 69 static const char *recommendedPat = 70 "[0-9A-Z_a-z\\u00C0-\\u00D6\\u00D8-\\u00F6\\u00F8-\\u0131\\u0134-\\u013E\\u0141-\\u014" 71 "8\\u014A-\\u017E\\u018F\\u01A0\\u01A1\\u01AF\\u01B0\\u01CD-\\u01DC\\u01DE-\\u01E3\\u01E" 72 "6-\\u01F0\\u01F4\\u01F5\\u01F8-\\u021B\\u021E\\u021F\\u0226-\\u0233\\u0259\\u02BB\\u02B" 73 "C\\u02EC\\u0300-\\u0304\\u0306-\\u030C\\u030F-\\u0311\\u0313\\u0314\\u031B\\u0323-\\u03" 74 "28\\u032D\\u032E\\u0330\\u0331\\u0335\\u0338\\u0339\\u0342\\u0345\\u037B-\\u037D\\u0386" 75 "\\u0388-\\u038A\\u038C\\u038E-\\u03A1\\u03A3-\\u03CE\\u03FC-\\u045F\\u048A-\\u0529\\u05" 76 "2E\\u052F\\u0531-\\u0556\\u0559\\u0561-\\u0586\\u05B4\\u05D0-\\u05EA\\u05F0-\\u05F2\\u0" 77 "620-\\u063F\\u0641-\\u0655\\u0660-\\u0669\\u0670-\\u0672\\u0674\\u0679-\\u068D\\u068F-" 78 "\\u06D3\\u06D5\\u06E5\\u06E6\\u06EE-\\u06FC\\u06FF\\u0750-\\u07B1\\u08A0-\\u08AC\\u08B2" 79 "\\u08B6-\\u08BD\\u0901-\\u094D\\u094F\\u0950\\u0956\\u0957\\u0960-\\u0963\\u0966-\\u096" 80 "F\\u0971-\\u0977\\u0979-\\u097F\\u0981-\\u0983\\u0985-\\u098C\\u098F\\u0990\\u0993-\\u0" 81 "9A8\\u09AA-\\u09B0\\u09B2\\u09B6-\\u09B9\\u09BC-\\u09C4\\u09C7\\u09C8\\u09CB-\\u09CE\\u" 82 "09D7\\u09E0-\\u09E3\\u09E6-\\u09F1\\u0A01-\\u0A03\\u0A05-\\u0A0A\\u0A0F\\u0A10\\u0A13-" 83 "\\u0A28\\u0A2A-\\u0A30\\u0A32\\u0A35\\u0A38\\u0A39\\u0A3C\\u0A3E-\\u0A42\\u0A47\\u0A48\\" 84 "u0A4B-\\u0A4D\\u0A5C\\u0A66-\\u0A74\\u0A81-\\u0A83\\u0A85-\\u0A8D\\u0A8F-\\u0A91\\u0A9" 85 "3-\\u0AA8\\u0AAA-\\u0AB0\\u0AB2\\u0AB3\\u0AB5-\\u0AB9\\u0ABC-\\u0AC5\\u0AC7-\\u0AC9\\u0" 86 "ACB-\\u0ACD\\u0AD0\\u0AE0-\\u0AE3\\u0AE6-\\u0AEF\\u0B01-\\u0B03\\u0B05-\\u0B0C\\u0B0F\\" 87 "u0B10\\u0B13-\\u0B28\\u0B2A-\\u0B30\\u0B32\\u0B33\\u0B35-\\u0B39\\u0B3C-\\u0B43\\u0B47" 88 "\\u0B48\\u0B4B-\\u0B4D\\u0B56\\u0B57\\u0B5F-\\u0B61\\u0B66-\\u0B6F\\u0B71\\u0B82\\u0B83" 89 "\\u0B85-\\u0B8A\\u0B8E-\\u0B90\\u0B92-\\u0B95\\u0B99\\u0B9A\\u0B9C\\u0B9E\\u0B9F\\u0BA3" 90 "\\u0BA4\\u0BA8-\\u0BAA\\u0BAE-\\u0BB9\\u0BBE-\\u0BC2\\u0BC6-\\u0BC8\\u0BCA-\\u0BCD\\u0B" 91 "D0\\u0BD7\\u0BE6-\\u0BEF\\u0C01-\\u0C03\\u0C05-\\u0C0C\\u0C0E-\\u0C10\\u0C12-\\u0C28\\u" 92 "0C2A-\\u0C33\\u0C35-\\u0C39\\u0C3D-\\u0C44\\u0C46-\\u0C48\\u0C4A-\\u0C4D\\u0C55\\u0C56" 93 "\\u0C60\\u0C61\\u0C66-\\u0C6F\\u0C80\\u0C82\\u0C83\\u0C85-\\u0C8C\\u0C8E-\\u0C90\\u0C92" 94 "-\\u0CA8\\u0CAA-\\u0CB3\\u0CB5-\\u0CB9\\u0CBC-\\u0CC4\\u0CC6-\\u0CC8\\u0CCA-\\u0CCD\\u0" 95 "CD5\\u0CD6\\u0CE0-\\u0CE3\\u0CE6-\\u0CEF\\u0CF1\\u0CF2\\u0D02\\u0D03\\u0D05-\\u0D0C\\u0" 96 "D0E-\\u0D10\\u0D12-\\u0D3A\\u0D3D-\\u0D43\\u0D46-\\u0D48\\u0D4A-\\u0D4E\\u0D54-\\u0D57" 97 "\\u0D60\\u0D61\\u0D66-\\u0D6F\\u0D7A-\\u0D7F\\u0D82\\u0D83\\u0D85-\\u0D8E\\u0D91-\\u0D9" 98 "6\\u0D9A-\\u0DA5\\u0DA7-\\u0DB1\\u0DB3-\\u0DBB\\u0DBD\\u0DC0-\\u0DC6\\u0DCA\\u0DCF-\\u0" 99 "DD4\\u0DD6\\u0DD8-\\u0DDE\\u0DF2\\u0E01-\\u0E32\\u0E34-\\u0E3A\\u0E40-\\u0E4E\\u0E50-\\" 100 "u0E59\\u0E81\\u0E82\\u0E84\\u0E87\\u0E88\\u0E8A\\u0E8D\\u0E94-\\u0E97\\u0E99-\\u0E9F\\u" 101 "0EA1-\\u0EA3\\u0EA5\\u0EA7\\u0EAA\\u0EAB\\u0EAD-\\u0EB2\\u0EB4-\\u0EB9\\u0EBB-\\u0EBD\\" 102 "u0EC0-\\u0EC4\\u0EC6\\u0EC8-\\u0ECD\\u0ED0-\\u0ED9\\u0EDE\\u0EDF\\u0F00\\u0F20-\\u0F29" 103 "\\u0F35\\u0F37\\u0F3E-\\u0F42\\u0F44-\\u0F47\\u0F49-\\u0F4C\\u0F4E-\\u0F51\\u0F53-\\u0F" 104 "56\\u0F58-\\u0F5B\\u0F5D-\\u0F68\\u0F6A-\\u0F6C\\u0F71\\u0F72\\u0F74\\u0F7A-\\u0F80\\u0" 105 "F82-\\u0F84\\u0F86-\\u0F92\\u0F94-\\u0F97\\u0F99-\\u0F9C\\u0F9E-\\u0FA1\\u0FA3-\\u0FA6" 106 "\\u0FA8-\\u0FAB\\u0FAD-\\u0FB8\\u0FBA-\\u0FBC\\u0FC6\\u1000-\\u1049\\u1050-\\u109D\\u10" 107 "C7\\u10CD\\u10D0-\\u10F0\\u10F7-\\u10FA\\u10FD-\\u10FF\\u1200-\\u1248\\u124A-\\u124D\\u" 108 "1250-\\u1256\\u1258\\u125A-\\u125D\\u1260-\\u1288\\u128A-\\u128D\\u1290-\\u12B0\\u12B2" 109 "-\\u12B5\\u12B8-\\u12BE\\u12C0\\u12C2-\\u12C5\\u12C8-\\u12D6\\u12D8-\\u1310\\u1312-\\u1" 110 "315\\u1318-\\u135A\\u135D-\\u135F\\u1380-\\u138F\\u1780-\\u17A2\\u17A5-\\u17A7\\u17A9-" 111 "\\u17B3\\u17B6-\\u17CA\\u17D2\\u17D7\\u17DC\\u17E0-\\u17E9\\u1C80-\\u1C88\\u1E00-\\u1E9" 112 "9\\u1E9E\\u1EA0-\\u1EF9\\u1F00-\\u1F15\\u1F18-\\u1F1D\\u1F20-\\u1F45\\u1F48-\\u1F4D\\u1" 113 "F50-\\u1F57\\u1F59\\u1F5B\\u1F5D\\u1F5F-\\u1F70\\u1F72\\u1F74\\u1F76\\u1F78\\u1F7A\\u1F" 114 "7C\\u1F80-\\u1FB4\\u1FB6-\\u1FBA\\u1FBC\\u1FC2-\\u1FC4\\u1FC6-\\u1FC8\\u1FCA\\u1FCC\\u1" 115 "FD0-\\u1FD2\\u1FD6-\\u1FDA\\u1FE0-\\u1FE2\\u1FE4-\\u1FEA\\u1FEC\\u1FF2-\\u1FF4\\u1FF6-" 116 "\\u1FF8\\u1FFA\\u1FFC\\u2D27\\u2D2D\\u2D80-\\u2D96\\u2DA0-\\u2DA6\\u2DA8-\\u2DAE\\u2DB0" 117 "-\\u2DB6\\u2DB8-\\u2DBE\\u2DC0-\\u2DC6\\u2DC8-\\u2DCE\\u2DD0-\\u2DD6\\u2DD8-\\u2DDE\\u3" 118 "005-\\u3007\\u3041-\\u3096\\u3099\\u309A\\u309D\\u309E\\u30A1-\\u30FA\\u30FC-\\u30FE\\u" 119 "3105-\\u312D\\u31A0-\\u31BA\\u3400-\\u4DB5\\u4E00-\\u9FD5\\uA660\\uA661\\uA674-\\uA67B" 120 "\\uA67F\\uA69F\\uA717-\\uA71F\\uA788\\uA78D\\uA78E\\uA790-\\uA793\\uA7A0-\\uA7AA\\uA7AE" 121 "\\uA7FA\\uA9E7-\\uA9FE\\uAA60-\\uAA76\\uAA7A-\\uAA7F\\uAB01-\\uAB06\\uAB09-\\uAB0E\\uAB" 122 "11-\\uAB16\\uAB20-\\uAB26\\uAB28-\\uAB2E\\uAC00-\\uD7A3\\uFA0E\\uFA0F\\uFA11\\uFA13\\uF" 123 "A14\\uFA1F\\uFA21\\uFA23\\uFA24\\uFA27-\\uFA29\\U00020000-\\U0002A6D6\\U0002A700-\\U0" 124 "002B734\\U0002B740-\\U0002B81D\\U0002B820-\\U0002CEA1]"; 125 126 gRecommendedSet = new UnicodeSet(UnicodeString(recommendedPat, -1, US_INV), status); 127 gRecommendedSet->freeze(); 128 gNfdNormalizer = Normalizer2::getNFDInstance(status); 129 ucln_i18n_registerCleanup(UCLN_I18N_SPOOF, uspoof_cleanup); 130 } 131 132 U_CFUNC void uspoof_internalInitStatics(UErrorCode *status) { 133 umtx_initOnce(gSpoofInitStaticsOnce, &initializeStatics, *status); 134 } 135 136 U_CAPI USpoofChecker * U_EXPORT2 137 uspoof_open(UErrorCode *status) { 138 umtx_initOnce(gSpoofInitStaticsOnce, &initializeStatics, *status); 139 if (U_FAILURE(*status)) { 140 return NULL; 141 } 142 SpoofImpl *si = new SpoofImpl(*status); 143 if (U_SUCCESS(*status) && si == NULL) { 144 *status = U_MEMORY_ALLOCATION_ERROR; 145 } 146 if (U_FAILURE(*status)) { 147 delete si; 148 si = NULL; 149 } 150 return si->asUSpoofChecker(); 151 } 152 153 154 U_CAPI USpoofChecker * U_EXPORT2 155 uspoof_openFromSerialized(const void *data, int32_t length, int32_t *pActualLength, 156 UErrorCode *status) { 157 if (U_FAILURE(*status)) { 158 return NULL; 159 } 160 umtx_initOnce(gSpoofInitStaticsOnce, &initializeStatics, *status); 161 SpoofData *sd = new SpoofData(data, length, *status); 162 SpoofImpl *si = new SpoofImpl(sd, *status); 163 if (U_FAILURE(*status)) { 164 delete sd; 165 delete si; 166 return NULL; 167 } 168 if (sd == NULL || si == NULL) { 169 *status = U_MEMORY_ALLOCATION_ERROR; 170 delete sd; 171 delete si; 172 return NULL; 173 } 174 175 if (pActualLength != NULL) { 176 *pActualLength = sd->size(); 177 } 178 return si->asUSpoofChecker(); 179 } 180 181 182 U_CAPI USpoofChecker * U_EXPORT2 183 uspoof_clone(const USpoofChecker *sc, UErrorCode *status) { 184 const SpoofImpl *src = SpoofImpl::validateThis(sc, *status); 185 if (src == NULL) { 186 return NULL; 187 } 188 SpoofImpl *result = new SpoofImpl(*src, *status); // copy constructor 189 if (U_FAILURE(*status)) { 190 delete result; 191 result = NULL; 192 } 193 return result->asUSpoofChecker(); 194 } 195 196 197 U_CAPI void U_EXPORT2 198 uspoof_close(USpoofChecker *sc) { 199 UErrorCode status = U_ZERO_ERROR; 200 SpoofImpl *This = SpoofImpl::validateThis(sc, status); 201 delete This; 202 } 203 204 205 U_CAPI void U_EXPORT2 206 uspoof_setChecks(USpoofChecker *sc, int32_t checks, UErrorCode *status) { 207 SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 208 if (This == NULL) { 209 return; 210 } 211 212 // Verify that the requested checks are all ones (bits) that 213 // are acceptable, known values. 214 if (checks & ~(USPOOF_ALL_CHECKS | USPOOF_AUX_INFO)) { 215 *status = U_ILLEGAL_ARGUMENT_ERROR; 216 return; 217 } 218 219 This->fChecks = checks; 220 } 221 222 223 U_CAPI int32_t U_EXPORT2 224 uspoof_getChecks(const USpoofChecker *sc, UErrorCode *status) { 225 const SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 226 if (This == NULL) { 227 return 0; 228 } 229 return This->fChecks; 230 } 231 232 U_CAPI void U_EXPORT2 233 uspoof_setRestrictionLevel(USpoofChecker *sc, URestrictionLevel restrictionLevel) { 234 UErrorCode status = U_ZERO_ERROR; 235 SpoofImpl *This = SpoofImpl::validateThis(sc, status); 236 if (This != NULL) { 237 This->fRestrictionLevel = restrictionLevel; 238 This->fChecks |= USPOOF_RESTRICTION_LEVEL; 239 } 240 } 241 242 U_CAPI URestrictionLevel U_EXPORT2 243 uspoof_getRestrictionLevel(const USpoofChecker *sc) { 244 UErrorCode status = U_ZERO_ERROR; 245 const SpoofImpl *This = SpoofImpl::validateThis(sc, status); 246 if (This == NULL) { 247 return USPOOF_UNRESTRICTIVE; 248 } 249 return This->fRestrictionLevel; 250 } 251 252 U_CAPI void U_EXPORT2 253 uspoof_setAllowedLocales(USpoofChecker *sc, const char *localesList, UErrorCode *status) { 254 SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 255 if (This == NULL) { 256 return; 257 } 258 This->setAllowedLocales(localesList, *status); 259 } 260 261 U_CAPI const char * U_EXPORT2 262 uspoof_getAllowedLocales(USpoofChecker *sc, UErrorCode *status) { 263 SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 264 if (This == NULL) { 265 return NULL; 266 } 267 return This->getAllowedLocales(*status); 268 } 269 270 271 U_CAPI const USet * U_EXPORT2 272 uspoof_getAllowedChars(const USpoofChecker *sc, UErrorCode *status) { 273 const UnicodeSet *result = uspoof_getAllowedUnicodeSet(sc, status); 274 return result->toUSet(); 275 } 276 277 U_CAPI const UnicodeSet * U_EXPORT2 278 uspoof_getAllowedUnicodeSet(const USpoofChecker *sc, UErrorCode *status) { 279 const SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 280 if (This == NULL) { 281 return NULL; 282 } 283 return This->fAllowedCharsSet; 284 } 285 286 287 U_CAPI void U_EXPORT2 288 uspoof_setAllowedChars(USpoofChecker *sc, const USet *chars, UErrorCode *status) { 289 const UnicodeSet *set = UnicodeSet::fromUSet(chars); 290 uspoof_setAllowedUnicodeSet(sc, set, status); 291 } 292 293 294 U_CAPI void U_EXPORT2 295 uspoof_setAllowedUnicodeSet(USpoofChecker *sc, const UnicodeSet *chars, UErrorCode *status) { 296 SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 297 if (This == NULL) { 298 return; 299 } 300 if (chars->isBogus()) { 301 *status = U_ILLEGAL_ARGUMENT_ERROR; 302 return; 303 } 304 UnicodeSet *clonedSet = static_cast<UnicodeSet *>(chars->clone()); 305 if (clonedSet == NULL || clonedSet->isBogus()) { 306 *status = U_MEMORY_ALLOCATION_ERROR; 307 return; 308 } 309 clonedSet->freeze(); 310 delete This->fAllowedCharsSet; 311 This->fAllowedCharsSet = clonedSet; 312 This->fChecks |= USPOOF_CHAR_LIMIT; 313 } 314 315 316 U_CAPI int32_t U_EXPORT2 317 uspoof_check(const USpoofChecker *sc, 318 const UChar *id, int32_t length, 319 int32_t *position, 320 UErrorCode *status) { 321 322 // Backwards compatibility: 323 if (position != NULL) { 324 *position = 0; 325 } 326 327 // Delegate to uspoof_check2 328 return uspoof_check2(sc, id, length, NULL, status); 329 } 330 331 332 U_CAPI int32_t U_EXPORT2 333 uspoof_check2(const USpoofChecker *sc, 334 const UChar* id, int32_t length, 335 USpoofCheckResult* checkResult, 336 UErrorCode *status) { 337 338 const SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 339 if (This == NULL) { 340 return 0; 341 } 342 if (length < -1) { 343 *status = U_ILLEGAL_ARGUMENT_ERROR; 344 return 0; 345 } 346 UnicodeString idStr((length == -1), id, length); // Aliasing constructor. 347 int32_t result = uspoof_check2UnicodeString(sc, idStr, checkResult, status); 348 return result; 349 } 350 351 352 U_CAPI int32_t U_EXPORT2 353 uspoof_checkUTF8(const USpoofChecker *sc, 354 const char *id, int32_t length, 355 int32_t *position, 356 UErrorCode *status) { 357 358 // Backwards compatibility: 359 if (position != NULL) { 360 *position = 0; 361 } 362 363 // Delegate to uspoof_check2 364 return uspoof_check2UTF8(sc, id, length, NULL, status); 365 } 366 367 368 U_CAPI int32_t U_EXPORT2 369 uspoof_check2UTF8(const USpoofChecker *sc, 370 const char *id, int32_t length, 371 USpoofCheckResult* checkResult, 372 UErrorCode *status) { 373 374 if (U_FAILURE(*status)) { 375 return 0; 376 } 377 UnicodeString idStr = UnicodeString::fromUTF8(StringPiece(id, length>=0 ? length : static_cast<int32_t>(uprv_strlen(id)))); 378 int32_t result = uspoof_check2UnicodeString(sc, idStr, checkResult, status); 379 return result; 380 } 381 382 383 U_CAPI int32_t U_EXPORT2 384 uspoof_areConfusable(const USpoofChecker *sc, 385 const UChar *id1, int32_t length1, 386 const UChar *id2, int32_t length2, 387 UErrorCode *status) { 388 SpoofImpl::validateThis(sc, *status); 389 if (U_FAILURE(*status)) { 390 return 0; 391 } 392 if (length1 < -1 || length2 < -1) { 393 *status = U_ILLEGAL_ARGUMENT_ERROR; 394 return 0; 395 } 396 397 UnicodeString id1Str((length1==-1), id1, length1); // Aliasing constructor 398 UnicodeString id2Str((length2==-1), id2, length2); // Aliasing constructor 399 return uspoof_areConfusableUnicodeString(sc, id1Str, id2Str, status); 400 } 401 402 403 U_CAPI int32_t U_EXPORT2 404 uspoof_areConfusableUTF8(const USpoofChecker *sc, 405 const char *id1, int32_t length1, 406 const char *id2, int32_t length2, 407 UErrorCode *status) { 408 SpoofImpl::validateThis(sc, *status); 409 if (U_FAILURE(*status)) { 410 return 0; 411 } 412 if (length1 < -1 || length2 < -1) { 413 *status = U_ILLEGAL_ARGUMENT_ERROR; 414 return 0; 415 } 416 UnicodeString id1Str = UnicodeString::fromUTF8(StringPiece(id1, length1>=0? length1 : static_cast<int32_t>(uprv_strlen(id1)))); 417 UnicodeString id2Str = UnicodeString::fromUTF8(StringPiece(id2, length2>=0? length2 : static_cast<int32_t>(uprv_strlen(id2)))); 418 int32_t results = uspoof_areConfusableUnicodeString(sc, id1Str, id2Str, status); 419 return results; 420 } 421 422 423 U_CAPI int32_t U_EXPORT2 424 uspoof_areConfusableUnicodeString(const USpoofChecker *sc, 425 const icu::UnicodeString &id1, 426 const icu::UnicodeString &id2, 427 UErrorCode *status) { 428 const SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 429 if (U_FAILURE(*status)) { 430 return 0; 431 } 432 // 433 // See section 4 of UAX 39 for the algorithm for checking whether two strings are confusable, 434 // and for definitions of the types (single, whole, mixed-script) of confusables. 435 436 // We only care about a few of the check flags. Ignore the others. 437 // If no tests relavant to this function have been specified, return an error. 438 // TODO: is this really the right thing to do? It's probably an error on the caller's part, 439 // but logically we would just return 0 (no error). 440 if ((This->fChecks & USPOOF_CONFUSABLE) == 0) { 441 *status = U_INVALID_STATE_ERROR; 442 return 0; 443 } 444 445 // Compute the skeletons and check for confusability. 446 UnicodeString id1Skeleton; 447 uspoof_getSkeletonUnicodeString(sc, 0 /* deprecated */, id1, id1Skeleton, status); 448 UnicodeString id2Skeleton; 449 uspoof_getSkeletonUnicodeString(sc, 0 /* deprecated */, id2, id2Skeleton, status); 450 if (U_FAILURE(*status)) { return 0; } 451 if (id1Skeleton != id2Skeleton) { 452 return 0; 453 } 454 455 // If we get here, the strings are confusable. Now we just need to set the flags for the appropriate classes 456 // of confusables according to UTS 39 section 4. 457 // Start by computing the resolved script sets of id1 and id2. 458 ScriptSet id1RSS; 459 This->getResolvedScriptSet(id1, id1RSS, *status); 460 ScriptSet id2RSS; 461 This->getResolvedScriptSet(id2, id2RSS, *status); 462 463 // Turn on all applicable flags 464 int32_t result = 0; 465 if (id1RSS.intersects(id2RSS)) { 466 result |= USPOOF_SINGLE_SCRIPT_CONFUSABLE; 467 } else { 468 result |= USPOOF_MIXED_SCRIPT_CONFUSABLE; 469 if (!id1RSS.isEmpty() && !id2RSS.isEmpty()) { 470 result |= USPOOF_WHOLE_SCRIPT_CONFUSABLE; 471 } 472 } 473 474 // Turn off flags that the user doesn't want 475 if ((This->fChecks & USPOOF_SINGLE_SCRIPT_CONFUSABLE) == 0) { 476 result &= ~USPOOF_SINGLE_SCRIPT_CONFUSABLE; 477 } 478 if ((This->fChecks & USPOOF_MIXED_SCRIPT_CONFUSABLE) == 0) { 479 result &= ~USPOOF_MIXED_SCRIPT_CONFUSABLE; 480 } 481 if ((This->fChecks & USPOOF_WHOLE_SCRIPT_CONFUSABLE) == 0) { 482 result &= ~USPOOF_WHOLE_SCRIPT_CONFUSABLE; 483 } 484 485 return result; 486 } 487 488 489 U_CAPI int32_t U_EXPORT2 490 uspoof_checkUnicodeString(const USpoofChecker *sc, 491 const icu::UnicodeString &id, 492 int32_t *position, 493 UErrorCode *status) { 494 495 // Backwards compatibility: 496 if (position != NULL) { 497 *position = 0; 498 } 499 500 // Delegate to uspoof_check2 501 return uspoof_check2UnicodeString(sc, id, NULL, status); 502 } 503 504 int32_t checkImpl(const SpoofImpl* This, const UnicodeString& id, CheckResult* checkResult, UErrorCode* status) { 505 U_ASSERT(This != NULL); 506 U_ASSERT(checkResult != NULL); 507 checkResult->clear(); 508 int32_t result = 0; 509 510 if (0 != (This->fChecks & USPOOF_RESTRICTION_LEVEL)) { 511 URestrictionLevel idRestrictionLevel = This->getRestrictionLevel(id, *status); 512 if (idRestrictionLevel > This->fRestrictionLevel) { 513 result |= USPOOF_RESTRICTION_LEVEL; 514 } 515 checkResult->fRestrictionLevel = idRestrictionLevel; 516 } 517 518 if (0 != (This->fChecks & USPOOF_MIXED_NUMBERS)) { 519 UnicodeSet numerics; 520 This->getNumerics(id, numerics, *status); 521 if (numerics.size() > 1) { 522 result |= USPOOF_MIXED_NUMBERS; 523 } 524 checkResult->fNumerics = numerics; // UnicodeSet::operator= 525 } 526 527 528 if (0 != (This->fChecks & USPOOF_CHAR_LIMIT)) { 529 int32_t i; 530 UChar32 c; 531 int32_t length = id.length(); 532 for (i=0; i<length ;) { 533 c = id.char32At(i); 534 i += U16_LENGTH(c); 535 if (!This->fAllowedCharsSet->contains(c)) { 536 result |= USPOOF_CHAR_LIMIT; 537 break; 538 } 539 } 540 } 541 542 if (0 != (This->fChecks & USPOOF_INVISIBLE)) { 543 // This check needs to be done on NFD input 544 UnicodeString nfdText; 545 gNfdNormalizer->normalize(id, nfdText, *status); 546 int32_t nfdLength = nfdText.length(); 547 548 // scan for more than one occurence of the same non-spacing mark 549 // in a sequence of non-spacing marks. 550 int32_t i; 551 UChar32 c; 552 UChar32 firstNonspacingMark = 0; 553 UBool haveMultipleMarks = FALSE; 554 UnicodeSet marksSeenSoFar; // Set of combining marks in a single combining sequence. 555 556 for (i=0; i<nfdLength ;) { 557 c = nfdText.char32At(i); 558 i += U16_LENGTH(c); 559 if (u_charType(c) != U_NON_SPACING_MARK) { 560 firstNonspacingMark = 0; 561 if (haveMultipleMarks) { 562 marksSeenSoFar.clear(); 563 haveMultipleMarks = FALSE; 564 } 565 continue; 566 } 567 if (firstNonspacingMark == 0) { 568 firstNonspacingMark = c; 569 continue; 570 } 571 if (!haveMultipleMarks) { 572 marksSeenSoFar.add(firstNonspacingMark); 573 haveMultipleMarks = TRUE; 574 } 575 if (marksSeenSoFar.contains(c)) { 576 // report the error, and stop scanning. 577 // No need to find more than the first failure. 578 result |= USPOOF_INVISIBLE; 579 break; 580 } 581 marksSeenSoFar.add(c); 582 } 583 } 584 585 checkResult->fChecks = result; 586 return checkResult->toCombinedBitmask(This->fChecks); 587 } 588 589 U_CAPI int32_t U_EXPORT2 590 uspoof_check2UnicodeString(const USpoofChecker *sc, 591 const icu::UnicodeString &id, 592 USpoofCheckResult* checkResult, 593 UErrorCode *status) { 594 const SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 595 if (This == NULL) { 596 return FALSE; 597 } 598 599 if (checkResult != NULL) { 600 CheckResult* ThisCheckResult = CheckResult::validateThis(checkResult, *status); 601 if (ThisCheckResult == NULL) { 602 return FALSE; 603 } 604 return checkImpl(This, id, ThisCheckResult, status); 605 } else { 606 // Stack-allocate the checkResult since this method doesn't return it 607 CheckResult stackCheckResult; 608 return checkImpl(This, id, &stackCheckResult, status); 609 } 610 } 611 612 613 U_CAPI int32_t U_EXPORT2 614 uspoof_getSkeleton(const USpoofChecker *sc, 615 uint32_t type, 616 const UChar *id, int32_t length, 617 UChar *dest, int32_t destCapacity, 618 UErrorCode *status) { 619 620 SpoofImpl::validateThis(sc, *status); 621 if (U_FAILURE(*status)) { 622 return 0; 623 } 624 if (length<-1 || destCapacity<0 || (destCapacity==0 && dest!=NULL)) { 625 *status = U_ILLEGAL_ARGUMENT_ERROR; 626 return 0; 627 } 628 629 UnicodeString idStr((length==-1), id, length); // Aliasing constructor 630 UnicodeString destStr; 631 uspoof_getSkeletonUnicodeString(sc, type, idStr, destStr, status); 632 destStr.extract(dest, destCapacity, *status); 633 return destStr.length(); 634 } 635 636 637 638 U_I18N_API UnicodeString & U_EXPORT2 639 uspoof_getSkeletonUnicodeString(const USpoofChecker *sc, 640 uint32_t /*type*/, 641 const UnicodeString &id, 642 UnicodeString &dest, 643 UErrorCode *status) { 644 const SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 645 if (U_FAILURE(*status)) { 646 return dest; 647 } 648 649 UnicodeString nfdId; 650 gNfdNormalizer->normalize(id, nfdId, *status); 651 652 // Apply the skeleton mapping to the NFD normalized input string 653 // Accumulate the skeleton, possibly unnormalized, in a UnicodeString. 654 int32_t inputIndex = 0; 655 UnicodeString skelStr; 656 int32_t normalizedLen = nfdId.length(); 657 for (inputIndex=0; inputIndex < normalizedLen; ) { 658 UChar32 c = nfdId.char32At(inputIndex); 659 inputIndex += U16_LENGTH(c); 660 This->fSpoofData->confusableLookup(c, skelStr); 661 } 662 663 gNfdNormalizer->normalize(skelStr, dest, *status); 664 return dest; 665 } 666 667 668 U_CAPI int32_t U_EXPORT2 669 uspoof_getSkeletonUTF8(const USpoofChecker *sc, 670 uint32_t type, 671 const char *id, int32_t length, 672 char *dest, int32_t destCapacity, 673 UErrorCode *status) { 674 SpoofImpl::validateThis(sc, *status); 675 if (U_FAILURE(*status)) { 676 return 0; 677 } 678 if (length<-1 || destCapacity<0 || (destCapacity==0 && dest!=NULL)) { 679 *status = U_ILLEGAL_ARGUMENT_ERROR; 680 return 0; 681 } 682 683 UnicodeString srcStr = UnicodeString::fromUTF8(StringPiece(id, length>=0 ? length : static_cast<int32_t>(uprv_strlen(id)))); 684 UnicodeString destStr; 685 uspoof_getSkeletonUnicodeString(sc, type, srcStr, destStr, status); 686 if (U_FAILURE(*status)) { 687 return 0; 688 } 689 690 int32_t lengthInUTF8 = 0; 691 u_strToUTF8(dest, destCapacity, &lengthInUTF8, 692 destStr.getBuffer(), destStr.length(), status); 693 return lengthInUTF8; 694 } 695 696 697 U_CAPI int32_t U_EXPORT2 698 uspoof_serialize(USpoofChecker *sc,void *buf, int32_t capacity, UErrorCode *status) { 699 SpoofImpl *This = SpoofImpl::validateThis(sc, *status); 700 if (This == NULL) { 701 U_ASSERT(U_FAILURE(*status)); 702 return 0; 703 } 704 705 return This->fSpoofData->serialize(buf, capacity, *status); 706 } 707 708 U_CAPI const USet * U_EXPORT2 709 uspoof_getInclusionSet(UErrorCode *status) { 710 umtx_initOnce(gSpoofInitStaticsOnce, &initializeStatics, *status); 711 return gInclusionSet->toUSet(); 712 } 713 714 U_CAPI const USet * U_EXPORT2 715 uspoof_getRecommendedSet(UErrorCode *status) { 716 umtx_initOnce(gSpoofInitStaticsOnce, &initializeStatics, *status); 717 return gRecommendedSet->toUSet(); 718 } 719 720 U_I18N_API const UnicodeSet * U_EXPORT2 721 uspoof_getInclusionUnicodeSet(UErrorCode *status) { 722 umtx_initOnce(gSpoofInitStaticsOnce, &initializeStatics, *status); 723 return gInclusionSet; 724 } 725 726 U_I18N_API const UnicodeSet * U_EXPORT2 727 uspoof_getRecommendedUnicodeSet(UErrorCode *status) { 728 umtx_initOnce(gSpoofInitStaticsOnce, &initializeStatics, *status); 729 return gRecommendedSet; 730 } 731 732 //------------------ 733 // CheckResult APIs 734 //------------------ 735 736 U_CAPI USpoofCheckResult* U_EXPORT2 737 uspoof_openCheckResult(UErrorCode *status) { 738 CheckResult* checkResult = new CheckResult(); 739 if (checkResult == NULL) { 740 *status = U_MEMORY_ALLOCATION_ERROR; 741 return NULL; 742 } 743 return checkResult->asUSpoofCheckResult(); 744 } 745 746 U_CAPI void U_EXPORT2 747 uspoof_closeCheckResult(USpoofCheckResult* checkResult) { 748 UErrorCode status = U_ZERO_ERROR; 749 CheckResult* This = CheckResult::validateThis(checkResult, status); 750 delete This; 751 } 752 753 U_CAPI int32_t U_EXPORT2 754 uspoof_getCheckResultChecks(const USpoofCheckResult *checkResult, UErrorCode *status) { 755 const CheckResult* This = CheckResult::validateThis(checkResult, *status); 756 if (U_FAILURE(*status)) { return 0; } 757 return This->fChecks; 758 } 759 760 U_CAPI URestrictionLevel U_EXPORT2 761 uspoof_getCheckResultRestrictionLevel(const USpoofCheckResult *checkResult, UErrorCode *status) { 762 const CheckResult* This = CheckResult::validateThis(checkResult, *status); 763 if (U_FAILURE(*status)) { return USPOOF_UNRESTRICTIVE; } 764 return This->fRestrictionLevel; 765 } 766 767 U_CAPI const USet* U_EXPORT2 768 uspoof_getCheckResultNumerics(const USpoofCheckResult *checkResult, UErrorCode *status) { 769 const CheckResult* This = CheckResult::validateThis(checkResult, *status); 770 if (U_FAILURE(*status)) { return NULL; } 771 return This->fNumerics.toUSet(); 772 } 773 774 775 776 #endif // !UCONFIG_NO_NORMALIZATION 777
