| OLD | NEW |
|---|
| (Empty) |
| 1 /******************************************************************** | |
| 2 * Copyright (c) 2001-2011,2015 International Business Machines | |
| 3 * Corporation and others. All Rights Reserved. | |
| 4 ******************************************************************** | |
| 5 * File usrchtst.c | |
| 6 * Modification History: | |
| 7 * Name Date Description | |
| 8 * synwee July 19 2001 creation | |
| 9 ********************************************************************/ | |
| 10 | |
| 11 #include "unicode/utypes.h" | |
| 12 | |
| 13 #if !UCONFIG_NO_COLLATION && !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILE_IO | |
| 14 | |
| 15 #include "unicode/usearch.h" | |
| 16 #include "unicode/ustring.h" | |
| 17 #include "ccolltst.h" | |
| 18 #include "cmemory.h" | |
| 19 #include <stdio.h> | |
| 20 #include "usrchdat.c" | |
| 21 #include "unicode/ubrk.h" | |
| 22 #include <assert.h> | |
| 23 | |
| 24 static UBool TOCLOSE_ = TRUE; | |
| 25 static UCollator *EN_US_; | |
| 26 static UCollator *FR_FR_; | |
| 27 static UCollator *DE_; | |
| 28 static UCollator *ES_; | |
| 29 | |
| 30 /** | |
| 31 * CHECK_BREAK(char *brk) | |
| 32 * Test if a break iterator is passed in AND break iteration is disabled. | |
| 33 * Skip the test if so. | |
| 34 * CHECK_BREAK_BOOL(char *brk) | |
| 35 * Same as above, but returns 'TRUE' as a passing result | |
| 36 */ | |
| 37 | |
| 38 #if !UCONFIG_NO_BREAK_ITERATION | |
| 39 static UBreakIterator *EN_WORDBREAKER_; | |
| 40 static UBreakIterator *EN_CHARACTERBREAKER_; | |
| 41 #define CHECK_BREAK(x) | |
| 42 #define CHECK_BREAK_BOOL(x) | |
| 43 #else | |
| 44 #define CHECK_BREAK(x) if(x) { log_info("Skipping test on %s:%d because UCONFIG
_NO_BREAK_ITERATION is on\n", __FILE__, __LINE__); return; } | |
| 45 #define CHECK_BREAK_BOOL(x) if(x) { log_info("Skipping test on %s:%d because UC
ONFIG_NO_BREAK_ITERATION is on\n", __FILE__, __LINE__); return TRUE; } | |
| 46 #endif | |
| 47 | |
| 48 /** | |
| 49 * Opening all static collators and break iterators | |
| 50 */ | |
| 51 static void open(UErrorCode* status) | |
| 52 { | |
| 53 if (TOCLOSE_) { | |
| 54 UChar rules[1024]; | |
| 55 int32_t rulelength = 0; | |
| 56 *status = U_ZERO_ERROR; | |
| 57 | |
| 58 EN_US_ = ucol_open("en_US", status); | |
| 59 if(U_FAILURE(*status)) { | |
| 60 log_err_status(*status, "Error opening collator\n"); | |
| 61 return; | |
| 62 } | |
| 63 FR_FR_ = ucol_open("fr_FR", status); | |
| 64 DE_ = ucol_open("de_DE", status); | |
| 65 ES_ = ucol_open("es_ES", status); | |
| 66 | |
| 67 u_strcpy(rules, ucol_getRules(DE_, &rulelength)); | |
| 68 u_unescape(EXTRACOLLATIONRULE, rules + rulelength, 1024 - rulelength); | |
| 69 | |
| 70 ucol_close(DE_); | |
| 71 | |
| 72 DE_ = ucol_openRules(rules, u_strlen(rules), UCOL_ON, UCOL_TERTIARY, | |
| 73 (UParseError *)NULL, status); | |
| 74 u_strcpy(rules, ucol_getRules(ES_, &rulelength)); | |
| 75 u_unescape(EXTRACOLLATIONRULE, rules + rulelength, 1024 - rulelength); | |
| 76 | |
| 77 ucol_close(ES_); | |
| 78 ES_ = ucol_openRules(rules, u_strlen(rules), UCOL_ON, UCOL_TERTIARY, | |
| 79 NULL, status); | |
| 80 #if !UCONFIG_NO_BREAK_ITERATION | |
| 81 EN_WORDBREAKER_ = ubrk_open(UBRK_WORD, "en_US", NULL, 0, status); | |
| 82 EN_CHARACTERBREAKER_ = ubrk_open(UBRK_CHARACTER, "en_US", NULL, 0, | |
| 83 status); | |
| 84 #endif | |
| 85 TOCLOSE_ = TRUE; | |
| 86 } | |
| 87 } | |
| 88 | |
| 89 /** | |
| 90 * Start opening all static collators and break iterators | |
| 91 */ | |
| 92 static void TestStart(void) | |
| 93 { | |
| 94 UErrorCode status = U_ZERO_ERROR; | |
| 95 open(&status); | |
| 96 if (U_FAILURE(status)) { | |
| 97 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 98 return; | |
| 99 } | |
| 100 TOCLOSE_ = FALSE; | |
| 101 } | |
| 102 | |
| 103 /** | |
| 104 * Closing all static collators and break iterators | |
| 105 */ | |
| 106 static void close(void) | |
| 107 { | |
| 108 if (TOCLOSE_) { | |
| 109 ucol_close(EN_US_); | |
| 110 ucol_close(FR_FR_); | |
| 111 ucol_close(DE_); | |
| 112 ucol_close(ES_); | |
| 113 #if !UCONFIG_NO_BREAK_ITERATION | |
| 114 ubrk_close(EN_WORDBREAKER_); | |
| 115 ubrk_close(EN_CHARACTERBREAKER_); | |
| 116 #endif | |
| 117 } | |
| 118 TOCLOSE_ = FALSE; | |
| 119 } | |
| 120 | |
| 121 /** | |
| 122 * End closing all static collators and break iterators | |
| 123 */ | |
| 124 static void TestEnd(void) | |
| 125 { | |
| 126 TOCLOSE_ = TRUE; | |
| 127 close(); | |
| 128 TOCLOSE_ = TRUE; | |
| 129 } | |
| 130 | |
| 131 /** | |
| 132 * output UChar strings for printing. | |
| 133 */ | |
| 134 static char *toCharString(const UChar* unichars) | |
| 135 { | |
| 136 static char result[1024]; | |
| 137 char *temp = result; | |
| 138 int count = 0; | |
| 139 int length = u_strlen(unichars); | |
| 140 | |
| 141 for (; count < length; count ++) { | |
| 142 UChar ch = unichars[count]; | |
| 143 if (ch >= 0x20 && ch <= 0x7e) { | |
| 144 *temp ++ = (char)ch; | |
| 145 } | |
| 146 else { | |
| 147 sprintf(temp, "\\u%04x", ch); | |
| 148 temp += 6; /* \uxxxx */ | |
| 149 } | |
| 150 } | |
| 151 *temp = 0; | |
| 152 | |
| 153 return result; | |
| 154 } | |
| 155 | |
| 156 /** | |
| 157 * Getting the collator | |
| 158 */ | |
| 159 static UCollator *getCollator(const char *collator) | |
| 160 { | |
| 161 if (collator == NULL) { | |
| 162 return EN_US_; | |
| 163 } | |
| 164 if (strcmp(collator, "fr") == 0) { | |
| 165 return FR_FR_; | |
| 166 } | |
| 167 else if (strcmp(collator, "de") == 0) { | |
| 168 return DE_; | |
| 169 } | |
| 170 else if (strcmp(collator, "es") == 0) { | |
| 171 return ES_; | |
| 172 } | |
| 173 else { | |
| 174 return EN_US_; | |
| 175 } | |
| 176 } | |
| 177 | |
| 178 /** | |
| 179 * Getting the breakiterator | |
| 180 */ | |
| 181 static UBreakIterator *getBreakIterator(const char *breaker) | |
| 182 { | |
| 183 if (breaker == NULL) { | |
| 184 return NULL; | |
| 185 } | |
| 186 #if !UCONFIG_NO_BREAK_ITERATION | |
| 187 if (strcmp(breaker, "wordbreaker") == 0) { | |
| 188 return EN_WORDBREAKER_; | |
| 189 } | |
| 190 else { | |
| 191 return EN_CHARACTERBREAKER_; | |
| 192 } | |
| 193 #else | |
| 194 return NULL; | |
| 195 #endif | |
| 196 } | |
| 197 | |
| 198 static void TestOpenClose(void) | |
| 199 { | |
| 200 UErrorCode status = U_ZERO_ERROR; | |
| 201 UStringSearch *result; | |
| 202 const UChar pattern[] = {0x61, 0x62, 0x63, 0x64, 0x65, 0x66}; | |
| 203 const UChar text[] = {0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67}; | |
| 204 #if !UCONFIG_NO_BREAK_ITERATION | |
| 205 UBreakIterator *breakiter = ubrk_open(UBRK_WORD, "en_US", | |
| 206 text, 6, &status); | |
| 207 #endif | |
| 208 /* testing null arguments */ | |
| 209 result = usearch_open(NULL, 0, NULL, 0, NULL, NULL, &status); | |
| 210 if (U_SUCCESS(status) || result != NULL) { | |
| 211 log_err("Error: NULL arguments should produce an error and a NULL result
\n"); | |
| 212 } | |
| 213 status = U_ZERO_ERROR; | |
| 214 result = usearch_openFromCollator(NULL, 0, NULL, 0, NULL, NULL, &status); | |
| 215 if (U_SUCCESS(status) || result != NULL) { | |
| 216 log_err("Error: NULL arguments should produce an error and a NULL result
\n"); | |
| 217 } | |
| 218 | |
| 219 status = U_ZERO_ERROR; | |
| 220 result = usearch_open(pattern, 3, NULL, 0, NULL, NULL, &status); | |
| 221 if (U_SUCCESS(status) || result != NULL) { | |
| 222 log_err("Error: NULL arguments should produce an error and a NULL result
\n"); | |
| 223 } | |
| 224 status = U_ZERO_ERROR; | |
| 225 result = usearch_openFromCollator(pattern, 3, NULL, 0, NULL, NULL, | |
| 226 &status); | |
| 227 if (U_SUCCESS(status) || result != NULL) { | |
| 228 log_err("Error: NULL arguments should produce an error and a NULL result
\n"); | |
| 229 } | |
| 230 | |
| 231 status = U_ZERO_ERROR; | |
| 232 result = usearch_open(pattern, 3, text, 6, NULL, NULL, &status); | |
| 233 if (U_SUCCESS(status) || result != NULL) { | |
| 234 log_err("Error: NULL arguments should produce an error and a NULL result
\n"); | |
| 235 } | |
| 236 status = U_ZERO_ERROR; | |
| 237 result = usearch_openFromCollator(pattern, 3, text, 6, NULL, NULL, | |
| 238 &status); | |
| 239 if (U_SUCCESS(status) || result != NULL) { | |
| 240 log_err("Error: NULL arguments should produce an error and a NULL result
\n"); | |
| 241 } | |
| 242 | |
| 243 status = U_ZERO_ERROR; | |
| 244 result = usearch_open(pattern, 3, text, 6, "en_US", NULL, &status); | |
| 245 if (U_FAILURE(status) || result == NULL) { | |
| 246 log_err_status(status, "Error: NULL break iterator is valid for opening
search\n"); | |
| 247 } | |
| 248 else { | |
| 249 usearch_close(result); | |
| 250 } | |
| 251 open(&status); | |
| 252 if (U_FAILURE(status)) { | |
| 253 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 254 return; | |
| 255 } | |
| 256 status = U_ZERO_ERROR; | |
| 257 result = usearch_openFromCollator(pattern, 3, text, 6, EN_US_, NULL, | |
| 258 &status); | |
| 259 if (U_FAILURE(status) || result == NULL) { | |
| 260 if (EN_US_ == NULL) { | |
| 261 log_data_err("Opening collator failed.\n"); | |
| 262 } else { | |
| 263 log_err("Error: NULL break iterator is valid for opening search\n"); | |
| 264 } | |
| 265 } | |
| 266 else { | |
| 267 usearch_close(result); | |
| 268 } | |
| 269 | |
| 270 | |
| 271 status = U_ZERO_ERROR; | |
| 272 #if !UCONFIG_NO_BREAK_ITERATION | |
| 273 | |
| 274 result = usearch_open(pattern, 3, text, 6, "en_US", breakiter, &status); | |
| 275 if (U_FAILURE(status) || result == NULL) { | |
| 276 log_err_status(status, "Error: Break iterator is valid for opening searc
h\n"); | |
| 277 } | |
| 278 else { | |
| 279 usearch_close(result); | |
| 280 } | |
| 281 status = U_ZERO_ERROR; | |
| 282 result = usearch_openFromCollator(pattern, 3, text, 6, EN_US_, breakiter, | |
| 283 &status); | |
| 284 if (U_FAILURE(status) || result == NULL) { | |
| 285 if (EN_US_ == NULL) { | |
| 286 log_data_err("Opening collator failed.\n"); | |
| 287 } else { | |
| 288 log_err("Error: Break iterator is valid for opening search\n"); | |
| 289 } | |
| 290 } | |
| 291 else { | |
| 292 usearch_close(result); | |
| 293 } | |
| 294 ubrk_close(breakiter); | |
| 295 #endif | |
| 296 close(); | |
| 297 } | |
| 298 | |
| 299 static void TestInitialization(void) | |
| 300 { | |
| 301 UErrorCode status = U_ZERO_ERROR; | |
| 302 UChar pattern[512]; | |
| 303 const UChar text[] = {0x61, 0x62, 0x63, 0x64, 0x65, 0x66}; | |
| 304 int32_t i = 0; | |
| 305 UStringSearch *result; | |
| 306 | |
| 307 /* simple test on the pattern ce construction */ | |
| 308 pattern[0] = 0x41; | |
| 309 pattern[1] = 0x42; | |
| 310 open(&status); | |
| 311 if (U_FAILURE(status)) { | |
| 312 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 313 return; | |
| 314 } | |
| 315 result = usearch_openFromCollator(pattern, 2, text, 3, EN_US_, NULL, | |
| 316 &status); | |
| 317 if (U_FAILURE(status)) { | |
| 318 log_err("Error opening search %s\n", u_errorName(status)); | |
| 319 } | |
| 320 usearch_close(result); | |
| 321 | |
| 322 /* testing if an extremely large pattern will fail the initialization */ | |
| 323 for(i = 0; i < 512; i++) { | |
| 324 pattern[i] = 0x41; | |
| 325 } | |
| 326 /*uprv_memset(pattern, 0x41, 512);*/ | |
| 327 result = usearch_openFromCollator(pattern, 512, text, 3, EN_US_, NULL, | |
| 328 &status); | |
| 329 if (U_FAILURE(status)) { | |
| 330 log_err("Error opening search %s\n", u_errorName(status)); | |
| 331 } | |
| 332 usearch_close(result); | |
| 333 close(); | |
| 334 } | |
| 335 | |
| 336 static UBool assertEqualWithUStringSearch( UStringSearch *strsrch, | |
| 337 const SearchData search) | |
| 338 { | |
| 339 int count = 0; | |
| 340 UErrorCode status = U_ZERO_ERROR; | |
| 341 int32_t matchindex = search.offset[count]; | |
| 342 int32_t textlength; | |
| 343 UChar matchtext[128]; | |
| 344 int32_t matchlength; | |
| 345 int32_t nextStart; | |
| 346 UBool isOverlap; | |
| 347 | |
| 348 usearch_setAttribute(strsrch, USEARCH_ELEMENT_COMPARISON, search.elemCompare
, &status); | |
| 349 if (U_FAILURE(status)) { | |
| 350 log_err("Error setting USEARCH_ELEMENT_COMPARISON attribute %s\n", u_err
orName(status)); | |
| 351 return FALSE; | |
| 352 } | |
| 353 | |
| 354 if (usearch_getMatchedStart(strsrch) != USEARCH_DONE || | |
| 355 usearch_getMatchedLength(strsrch) != 0) { | |
| 356 log_err("Error with the initialization of match start and length\n"); | |
| 357 } | |
| 358 /* start of next matches */ | |
| 359 while (U_SUCCESS(status) && matchindex >= 0) { | |
| 360 matchlength = search.size[count]; | |
| 361 usearch_next(strsrch, &status); | |
| 362 if (matchindex != usearch_getMatchedStart(strsrch) || | |
| 363 matchlength != (uint32_t)usearch_getMatchedLength(strsrch)) { | |
| 364 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 365 log_err("Text: %s\n", str); | |
| 366 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 367 log_err("Pattern: %s\n", str); | |
| 368 log_err("Error next match found at idx %d (len:%d); expected %d (len
:%d)\n", | |
| 369 usearch_getMatchedStart(strsrch), usearch_getMatchedLength(s
trsrch), | |
| 370 matchindex, matchlength); | |
| 371 return FALSE; | |
| 372 } | |
| 373 count ++; | |
| 374 | |
| 375 if (usearch_getMatchedText(strsrch, matchtext, 128, &status) != | |
| 376 (int32_t) matchlength || U_FAILURE(status) || | |
| 377 memcmp(matchtext, | |
| 378 usearch_getText(strsrch, &textlength) + matchindex, | |
| 379 matchlength * sizeof(UChar)) != 0) { | |
| 380 log_err("Error getting next matched text\n"); | |
| 381 } | |
| 382 | |
| 383 matchindex = search.offset[count]; | |
| 384 } | |
| 385 usearch_next(strsrch, &status); | |
| 386 if (usearch_getMatchedStart(strsrch) != USEARCH_DONE || | |
| 387 usearch_getMatchedLength(strsrch) != 0) { | |
| 388 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 389 log_err("Text: %s\n", str); | |
| 390 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 391 log_err("Pattern: %s\n", str); | |
| 392 log_err("Error next match found at %d (len:%d); expected <NO MATCH>\n", | |
| 393 usearch_getMatchedStart(strsrch), | |
| 394 usearch_getMatchedLength(strsrch)); | |
| 395 return FALSE; | |
| 396 } | |
| 397 /* start of previous matches */ | |
| 398 count = count == 0 ? 0 : count - 1; | |
| 399 matchindex = search.offset[count]; | |
| 400 | |
| 401 while (U_SUCCESS(status) && matchindex >= 0) { | |
| 402 matchlength = search.size[count]; | |
| 403 usearch_previous(strsrch, &status); | |
| 404 if (matchindex != usearch_getMatchedStart(strsrch) || | |
| 405 matchlength != (uint32_t)usearch_getMatchedLength(strsrch)) { | |
| 406 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 407 log_err("Text: %s\n", str); | |
| 408 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 409 log_err("Pattern: %s\n", str); | |
| 410 log_err("Error previous match found at %d (len:%d); expected %d (len
:%d)\n", | |
| 411 usearch_getMatchedStart(strsrch), usearch_getMatchedLength(s
trsrch), | |
| 412 matchindex, matchlength); | |
| 413 return FALSE; | |
| 414 } | |
| 415 | |
| 416 if (usearch_getMatchedText(strsrch, matchtext, 128, &status) != | |
| 417 (int32_t) matchlength || U_FAILURE(status) || | |
| 418 memcmp(matchtext, | |
| 419 usearch_getText(strsrch, &textlength) + matchindex, | |
| 420 matchlength * sizeof(UChar)) != 0) { | |
| 421 log_err("Error getting previous matched text\n"); | |
| 422 } | |
| 423 | |
| 424 matchindex = count > 0 ? search.offset[count - 1] : -1; | |
| 425 count --; | |
| 426 } | |
| 427 usearch_previous(strsrch, &status); | |
| 428 if (usearch_getMatchedStart(strsrch) != USEARCH_DONE || | |
| 429 usearch_getMatchedLength(strsrch) != 0) { | |
| 430 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 431 log_err("Text: %s\n", str); | |
| 432 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 433 log_err("Pattern: %s\n", str); | |
| 434 log_err("Error previous match found at %d (len:%d); expected <NO MATCH>\
n", | |
| 435 usearch_getMatchedStart(strsrch), | |
| 436 usearch_getMatchedLength(strsrch)); | |
| 437 return FALSE; | |
| 438 } | |
| 439 | |
| 440 | |
| 441 isOverlap = (usearch_getAttribute(strsrch, USEARCH_OVERLAP) == USEARCH_ON); | |
| 442 | |
| 443 /* start of following matches */ | |
| 444 count = 0; | |
| 445 matchindex = search.offset[count]; | |
| 446 nextStart = 0; | |
| 447 | |
| 448 while (TRUE) { | |
| 449 usearch_following(strsrch, nextStart, &status); | |
| 450 | |
| 451 if (matchindex < 0) { | |
| 452 if (usearch_getMatchedStart(strsrch) != USEARCH_DONE || usearch_getM
atchedLength(strsrch) != 0) { | |
| 453 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 454 log_err("Text: %s\n", str); | |
| 455 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 456 log_err("Pattern: %s\n", str); | |
| 457 log_err("Error following match starting at %d (overlap:%d) found
at %d (len:%d); expected <NO MATCH>\n", | |
| 458 nextStart, isOverlap, | |
| 459 usearch_getMatchedStart(strsrch), | |
| 460 usearch_getMatchedLength(strsrch)); | |
| 461 return FALSE; | |
| 462 } | |
| 463 /* no more matches */ | |
| 464 break; | |
| 465 } | |
| 466 | |
| 467 matchlength = search.size[count]; | |
| 468 if (usearch_getMatchedStart(strsrch) != matchindex | |
| 469 || usearch_getMatchedLength(strsrch) != matchlength | |
| 470 || U_FAILURE(status)) { | |
| 471 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 472 log_err("Text: %s\n", str); | |
| 473 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 474 log_err("Pattern: %s\n", str); | |
| 475 log_err("Error following match starting at %d (overlap: %d) found at
%d (len:%d); expected %d (len:%d)\n", | |
| 476 nextStart, isOverlap, | |
| 477 usearch_getMatchedStart(strsrch), usearch_getMatchedLeng
th(strsrch), | |
| 478 matchindex, matchlength); | |
| 479 return FALSE; | |
| 480 } | |
| 481 | |
| 482 if (isOverlap || usearch_getMatchedLength(strsrch) == 0) { | |
| 483 nextStart = usearch_getMatchedStart(strsrch) + 1; | |
| 484 } else { | |
| 485 nextStart = usearch_getMatchedStart(strsrch) + usearch_getMatchedLen
gth(strsrch); | |
| 486 } | |
| 487 | |
| 488 count++; | |
| 489 matchindex = search.offset[count]; | |
| 490 } | |
| 491 | |
| 492 /* start of preceding matches */ | |
| 493 count = -1; /* last non-negative offset index, could be -1 if no match */ | |
| 494 while (search.offset[count + 1] >= 0) { | |
| 495 count++; | |
| 496 } | |
| 497 usearch_getText(strsrch, &nextStart); | |
| 498 | |
| 499 while (TRUE) { | |
| 500 usearch_preceding(strsrch, nextStart, &status); | |
| 501 | |
| 502 if (count < 0) { | |
| 503 if (usearch_getMatchedStart(strsrch) != USEARCH_DONE || usearch_getM
atchedLength(strsrch) != 0) { | |
| 504 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 505 log_err("Text: %s\n", str); | |
| 506 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 507 log_err("Pattern: %s\n", str); | |
| 508 log_err("Error preceding match starting at %d (overlap: %d) foun
d at %d (len:%d); expected <NO MATCH>\n", | |
| 509 nextStart, isOverlap, | |
| 510 usearch_getMatchedStart(strsrch), | |
| 511 usearch_getMatchedLength(strsrch)); | |
| 512 return FALSE; | |
| 513 } | |
| 514 /* no more matches */ | |
| 515 break; | |
| 516 } | |
| 517 | |
| 518 matchindex = search.offset[count]; | |
| 519 matchlength = search.size[count]; | |
| 520 if (usearch_getMatchedStart(strsrch) != matchindex | |
| 521 || usearch_getMatchedLength(strsrch) != matchlength | |
| 522 || U_FAILURE(status)) { | |
| 523 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 524 log_err("Text: %s\n", str); | |
| 525 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 526 log_err("Pattern: %s\n", str); | |
| 527 log_err("Error preceding match starting at %d (overlap: %d) found at
%d (len:%d); expected %d (len:%d)\n", | |
| 528 nextStart, isOverlap, | |
| 529 usearch_getMatchedStart(strsrch), usearch_getMatchedLeng
th(strsrch), | |
| 530 matchindex, matchlength); | |
| 531 return FALSE; | |
| 532 } | |
| 533 | |
| 534 nextStart = matchindex; | |
| 535 count--; | |
| 536 } | |
| 537 | |
| 538 usearch_setAttribute(strsrch, USEARCH_ELEMENT_COMPARISON, USEARCH_STANDARD_E
LEMENT_COMPARISON, &status); | |
| 539 return TRUE; | |
| 540 } | |
| 541 | |
| 542 static UBool assertEqual(const SearchData search) | |
| 543 { | |
| 544 UErrorCode status = U_ZERO_ERROR; | |
| 545 UChar pattern[32]; | |
| 546 UChar text[128]; | |
| 547 UCollator *collator = getCollator(search.collator); | |
| 548 UBreakIterator *breaker = getBreakIterator(search.breaker); | |
| 549 UStringSearch *strsrch; | |
| 550 | |
| 551 CHECK_BREAK_BOOL(search.breaker); | |
| 552 | |
| 553 u_unescape(search.text, text, 128); | |
| 554 u_unescape(search.pattern, pattern, 32); | |
| 555 ucol_setStrength(collator, search.strength); | |
| 556 strsrch = usearch_openFromCollator(pattern, -1, text, -1, collator, | |
| 557 breaker, &status); | |
| 558 if (U_FAILURE(status)) { | |
| 559 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 560 return FALSE; | |
| 561 } | |
| 562 | |
| 563 if (!assertEqualWithUStringSearch(strsrch, search)) { | |
| 564 ucol_setStrength(collator, UCOL_TERTIARY); | |
| 565 usearch_close(strsrch); | |
| 566 return FALSE; | |
| 567 } | |
| 568 ucol_setStrength(collator, UCOL_TERTIARY); | |
| 569 usearch_close(strsrch); | |
| 570 return TRUE; | |
| 571 } | |
| 572 | |
| 573 static UBool assertCanonicalEqual(const SearchData search) | |
| 574 { | |
| 575 UErrorCode status = U_ZERO_ERROR; | |
| 576 UChar pattern[32]; | |
| 577 UChar text[128]; | |
| 578 UCollator *collator = getCollator(search.collator); | |
| 579 UBreakIterator *breaker = getBreakIterator(search.breaker); | |
| 580 UStringSearch *strsrch; | |
| 581 UBool result = TRUE; | |
| 582 | |
| 583 CHECK_BREAK_BOOL(search.breaker); | |
| 584 u_unescape(search.text, text, 128); | |
| 585 u_unescape(search.pattern, pattern, 32); | |
| 586 ucol_setStrength(collator, search.strength); | |
| 587 ucol_setAttribute(collator, UCOL_NORMALIZATION_MODE, UCOL_ON, &status); | |
| 588 strsrch = usearch_openFromCollator(pattern, -1, text, -1, collator, | |
| 589 breaker, &status);
| |
| 590 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 591 &status); | |
| 592 if (U_FAILURE(status)) { | |
| 593 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 594 result = FALSE; | |
| 595 goto bail; | |
| 596 } | |
| 597 | |
| 598 if (!assertEqualWithUStringSearch(strsrch, search)) { | |
| 599 ucol_setStrength(collator, UCOL_TERTIARY); | |
| 600 usearch_close(strsrch); | |
| 601 result = FALSE; | |
| 602 goto bail; | |
| 603 } | |
| 604 | |
| 605 bail: | |
| 606 ucol_setAttribute(collator, UCOL_NORMALIZATION_MODE, UCOL_OFF, &status); | |
| 607 ucol_setStrength(collator, UCOL_TERTIARY); | |
| 608 usearch_close(strsrch); | |
| 609 return result; | |
| 610 } | |
| 611 | |
| 612 static UBool assertEqualWithAttribute(const SearchData search, | |
| 613 USearchAttributeValue canonical, | |
| 614 USearchAttributeValue overlap) | |
| 615 { | |
| 616 UErrorCode status = U_ZERO_ERROR; | |
| 617 UChar pattern[32]; | |
| 618 UChar text[128]; | |
| 619 UCollator *collator = getCollator(search.collator); | |
| 620 UBreakIterator *breaker = getBreakIterator(search.breaker); | |
| 621 UStringSearch *strsrch; | |
| 622 | |
| 623 CHECK_BREAK_BOOL(search.breaker); | |
| 624 u_unescape(search.text, text, 128); | |
| 625 u_unescape(search.pattern, pattern, 32); | |
| 626 ucol_setStrength(collator, search.strength); | |
| 627 strsrch = usearch_openFromCollator(pattern, -1, text, -1, collator, | |
| 628 breaker, &status); | |
| 629 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, canonical, | |
| 630 &status); | |
| 631 usearch_setAttribute(strsrch, USEARCH_OVERLAP, overlap, &status); | |
| 632 | |
| 633 if (U_FAILURE(status)) { | |
| 634 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 635 return FALSE; | |
| 636 } | |
| 637 | |
| 638 if (!assertEqualWithUStringSearch(strsrch, search)) { | |
| 639 ucol_setStrength(collator, UCOL_TERTIARY); | |
| 640 usearch_close(strsrch); | |
| 641 return FALSE; | |
| 642 } | |
| 643 ucol_setStrength(collator, UCOL_TERTIARY); | |
| 644 usearch_close(strsrch); | |
| 645 return TRUE; | |
| 646 } | |
| 647 | |
| 648 static void TestBasic(void) | |
| 649 { | |
| 650 int count = 0; | |
| 651 UErrorCode status = U_ZERO_ERROR; | |
| 652 open(&status); | |
| 653 if (U_FAILURE(status)) { | |
| 654 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 655 return; | |
| 656 } | |
| 657 while (BASIC[count].text != NULL) { | |
| 658 if (!assertEqual(BASIC[count])) { | |
| 659 log_err("Error at test number %d\n", count); | |
| 660 } | |
| 661 count ++; | |
| 662 } | |
| 663 close(); | |
| 664 } | |
| 665 | |
| 666 static void TestNormExact(void) | |
| 667 { | |
| 668 int count = 0; | |
| 669 UErrorCode status = U_ZERO_ERROR; | |
| 670 open(&status); | |
| 671 if (U_FAILURE(status)) { | |
| 672 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 673 return; | |
| 674 } | |
| 675 ucol_setAttribute(EN_US_, UCOL_NORMALIZATION_MODE, UCOL_ON, &status); | |
| 676 if (U_FAILURE(status)) { | |
| 677 log_err("Error setting collation normalization %s\n", | |
| 678 u_errorName(status)); | |
| 679 } | |
| 680 while (BASIC[count].text != NULL) { | |
| 681 if (!assertEqual(BASIC[count])) { | |
| 682 log_err("Error at test number %d\n", count); | |
| 683 } | |
| 684 count ++; | |
| 685 } | |
| 686 count = 0; | |
| 687 while (NORMEXACT[count].text != NULL) { | |
| 688 if (!assertEqual(NORMEXACT[count])) { | |
| 689 log_err("Error at test number %d\n", count); | |
| 690 } | |
| 691 count ++; | |
| 692 } | |
| 693 ucol_setAttribute(EN_US_, UCOL_NORMALIZATION_MODE, UCOL_OFF, &status); | |
| 694 count = 0; | |
| 695 while (NONNORMEXACT[count].text != NULL) { | |
| 696 if (!assertEqual(NONNORMEXACT[count])) { | |
| 697 log_err("Error at test number %d\n", count); | |
| 698 } | |
| 699 count ++; | |
| 700 } | |
| 701 close(); | |
| 702 } | |
| 703 | |
| 704 static void TestStrength(void) | |
| 705 { | |
| 706 int count = 0; | |
| 707 UErrorCode status = U_ZERO_ERROR; | |
| 708 open(&status); | |
| 709 if (U_FAILURE(status)) { | |
| 710 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 711 return; | |
| 712 } | |
| 713 while (STRENGTH[count].text != NULL) { | |
| 714 if (!assertEqual(STRENGTH[count])) { | |
| 715 log_err("Error at test number %d\n", count); | |
| 716 } | |
| 717 count ++; | |
| 718 } | |
| 719 close(); | |
| 720 } | |
| 721 | |
| 722 static void TestBreakIterator(void) { | |
| 723 UErrorCode status = U_ZERO_ERROR; | |
| 724 UStringSearch *strsrch; | |
| 725 UChar text[128]; | |
| 726 UChar pattern[32]; | |
| 727 int count = 0; | |
| 728 | |
| 729 CHECK_BREAK("x"); | |
| 730 | |
| 731 #if !UCONFIG_NO_BREAK_ITERATION | |
| 732 open(&status); | |
| 733 if (U_FAILURE(status)) { | |
| 734 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 735 return; | |
| 736 } | |
| 737 if (usearch_getBreakIterator(NULL) != NULL) { | |
| 738 log_err("Expected NULL breakiterator from NULL string search\n"); | |
| 739 } | |
| 740 u_unescape(BREAKITERATOREXACT[0].text, text, 128); | |
| 741 u_unescape(BREAKITERATOREXACT[0].pattern, pattern, 32); | |
| 742 strsrch = usearch_openFromCollator(pattern, -1, text, -1, EN_US_, NULL, | |
| 743 &status); | |
| 744 if (U_FAILURE(status)) { | |
| 745 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 746 goto ENDTESTBREAKITERATOR; | |
| 747 } | |
| 748 | |
| 749 usearch_setBreakIterator(strsrch, NULL, &status); | |
| 750 if (U_FAILURE(status) || usearch_getBreakIterator(strsrch) != NULL) { | |
| 751 log_err("Error usearch_getBreakIterator returned wrong object"); | |
| 752 goto ENDTESTBREAKITERATOR; | |
| 753 } | |
| 754 | |
| 755 usearch_setBreakIterator(strsrch, EN_CHARACTERBREAKER_, &status); | |
| 756 if (U_FAILURE(status) || | |
| 757 usearch_getBreakIterator(strsrch) != EN_CHARACTERBREAKER_) { | |
| 758 log_err("Error usearch_getBreakIterator returned wrong object"); | |
| 759 goto ENDTESTBREAKITERATOR; | |
| 760 } | |
| 761 | |
| 762 usearch_setBreakIterator(strsrch, EN_WORDBREAKER_, &status); | |
| 763 if (U_FAILURE(status) || | |
| 764 usearch_getBreakIterator(strsrch) != EN_WORDBREAKER_) { | |
| 765 log_err("Error usearch_getBreakIterator returned wrong object"); | |
| 766 goto ENDTESTBREAKITERATOR; | |
| 767 } | |
| 768 | |
| 769 usearch_close(strsrch); | |
| 770 | |
| 771 count = 0; | |
| 772 while (count < 4) { | |
| 773 /* 0-3 test are fixed */ | |
| 774 const SearchData *search = &(BREAKITERATOREXACT[count]); | |
| 775 UCollator *collator = getCollator(search->collator); | |
| 776 UBreakIterator *breaker = getBreakIterator(search->breaker); | |
| 777 | |
| 778 u_unescape(search->text, text, 128); | |
| 779 u_unescape(search->pattern, pattern, 32); | |
| 780 ucol_setStrength(collator, search->strength); | |
| 781 | |
| 782 strsrch = usearch_openFromCollator(pattern, -1, text, -1, collator, | |
| 783 breaker, &status); | |
| 784 if (U_FAILURE(status) || | |
| 785 usearch_getBreakIterator(strsrch) != breaker) { | |
| 786 log_err("Error setting break iterator\n"); | |
| 787 if (strsrch != NULL) { | |
| 788 usearch_close(strsrch); | |
| 789 } | |
| 790 } | |
| 791 if (!assertEqualWithUStringSearch(strsrch, *search)) { | |
| 792 ucol_setStrength(collator, UCOL_TERTIARY); | |
| 793 usearch_close(strsrch); | |
| 794 goto ENDTESTBREAKITERATOR; | |
| 795 } | |
| 796 search = &(BREAKITERATOREXACT[count + 1]); | |
| 797 breaker = getBreakIterator(search->breaker); | |
| 798 usearch_setBreakIterator(strsrch, breaker, &status); | |
| 799 if (U_FAILURE(status) || usearch_getBreakIterator(strsrch) != breaker) { | |
| 800 log_err("Error setting break iterator\n"); | |
| 801 usearch_close(strsrch); | |
| 802 goto ENDTESTBREAKITERATOR; | |
| 803 } | |
| 804 usearch_reset(strsrch); | |
| 805 if (!assertEqualWithUStringSearch(strsrch, *search)) { | |
| 806 log_err("Error at test number %d\n", count); | |
| 807 usearch_close(strsrch); | |
| 808 goto ENDTESTBREAKITERATOR; | |
| 809 } | |
| 810 usearch_close(strsrch); | |
| 811 count += 2; | |
| 812 } | |
| 813 count = 0; | |
| 814 while (BREAKITERATOREXACT[count].text != NULL) { | |
| 815 if (!assertEqual(BREAKITERATOREXACT[count])) { | |
| 816 log_err("Error at test number %d\n", count); | |
| 817 goto ENDTESTBREAKITERATOR; | |
| 818 } | |
| 819 count ++; | |
| 820 } | |
| 821 | |
| 822 ENDTESTBREAKITERATOR: | |
| 823 close(); | |
| 824 #endif | |
| 825 } | |
| 826 | |
| 827 static void TestVariable(void) | |
| 828 { | |
| 829 int count = 0; | |
| 830 UErrorCode status = U_ZERO_ERROR; | |
| 831 open(&status); | |
| 832 if (U_FAILURE(status)) { | |
| 833 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 834 return; | |
| 835 } | |
| 836 ucol_setAttribute(EN_US_, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, &status); | |
| 837 if (U_FAILURE(status)) { | |
| 838 log_err("Error setting collation alternate attribute %s\n", | |
| 839 u_errorName(status)); | |
| 840 } | |
| 841 while (VARIABLE[count].text != NULL) { | |
| 842 log_verbose("variable %d\n", count); | |
| 843 if (!assertEqual(VARIABLE[count])) { | |
| 844 log_err("Error at test number %d\n", count); | |
| 845 } | |
| 846 count ++; | |
| 847 } | |
| 848 ucol_setAttribute(EN_US_, UCOL_ALTERNATE_HANDLING, | |
| 849 UCOL_NON_IGNORABLE, &status); | |
| 850 close(); | |
| 851 } | |
| 852 | |
| 853 static void TestOverlap(void) | |
| 854 { | |
| 855 int count = 0; | |
| 856 UErrorCode status = U_ZERO_ERROR; | |
| 857 open(&status); | |
| 858 if (U_FAILURE(status)) { | |
| 859 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 860 return; | |
| 861 } | |
| 862 while (OVERLAP[count].text != NULL) { | |
| 863 if (!assertEqualWithAttribute(OVERLAP[count], USEARCH_OFF, | |
| 864 USEARCH_ON)) { | |
| 865 log_err("Error at overlap test number %d\n", count); | |
| 866 } | |
| 867 count ++; | |
| 868 } | |
| 869 count = 0; | |
| 870 while (NONOVERLAP[count].text != NULL) { | |
| 871 if (!assertEqual(NONOVERLAP[count])) { | |
| 872 log_err("Error at non overlap test number %d\n", count); | |
| 873 } | |
| 874 count ++; | |
| 875 } | |
| 876 | |
| 877 count = 0; | |
| 878 while (count < 1) { | |
| 879 UChar pattern[32]; | |
| 880 UChar text[128]; | |
| 881 const SearchData *search = &(OVERLAP[count]); | |
| 882 UCollator *collator = getCollator(search->collator); | |
| 883 UStringSearch *strsrch; | |
| 884 status = U_ZERO_ERROR; | |
| 885 | |
| 886 u_unescape(search->text, text, 128); | |
| 887 u_unescape(search->pattern, pattern, 32); | |
| 888 strsrch = usearch_openFromCollator(pattern, -1, text, -1, collator, | |
| 889 NULL, &status); | |
| 890 if(status == U_FILE_ACCESS_ERROR) { | |
| 891 log_data_err("Is your data around?\n"); | |
| 892 return; | |
| 893 } else if(U_FAILURE(status)) { | |
| 894 log_err("Error opening searcher\n"); | |
| 895 return; | |
| 896 } | |
| 897 usearch_setAttribute(strsrch, USEARCH_OVERLAP, USEARCH_ON, &status); | |
| 898 if (U_FAILURE(status) || | |
| 899 usearch_getAttribute(strsrch, USEARCH_OVERLAP) != USEARCH_ON) { | |
| 900 log_err("Error setting overlap option\n"); | |
| 901 } | |
| 902 if (!assertEqualWithUStringSearch(strsrch, *search)) { | |
| 903 usearch_close(strsrch); | |
| 904 return; | |
| 905 } | |
| 906 search = &(NONOVERLAP[count]); | |
| 907 usearch_setAttribute(strsrch, USEARCH_OVERLAP, USEARCH_OFF, &status); | |
| 908 if (U_FAILURE(status) || | |
| 909 usearch_getAttribute(strsrch, USEARCH_OVERLAP) != USEARCH_OFF) { | |
| 910 log_err("Error setting overlap option\n"); | |
| 911 } | |
| 912 usearch_reset(strsrch); | |
| 913 if (!assertEqualWithUStringSearch(strsrch, *search)) { | |
| 914 usearch_close(strsrch); | |
| 915 log_err("Error at test number %d\n", count); | |
| 916 } | |
| 917 | |
| 918 count ++; | |
| 919 usearch_close(strsrch); | |
| 920 } | |
| 921 close(); | |
| 922 } | |
| 923 | |
| 924 static void TestCollator(void) | |
| 925 { | |
| 926 /* test collator that thinks "o" and "p" are the same thing */ | |
| 927 UChar rules[32]; | |
| 928 UCollator *tailored = NULL; | |
| 929 UErrorCode status = U_ZERO_ERROR; | |
| 930 UChar pattern[32]; | |
| 931 UChar text[128]; | |
| 932 UStringSearch *strsrch; | |
| 933 | |
| 934 text[0] = 0x41; | |
| 935 text[1] = 0x42; | |
| 936 text[2] = 0x43; | |
| 937 text[3] = 0x44; | |
| 938 text[4] = 0x45; | |
| 939 pattern[0] = 0x62; | |
| 940 pattern[1] = 0x63; | |
| 941 strsrch = usearch_open(pattern, 2, text, 5, "en_US", NULL, &status); | |
| 942 if(status == U_FILE_ACCESS_ERROR) { | |
| 943 log_data_err("Is your data around?\n"); | |
| 944 return; | |
| 945 } else if(U_FAILURE(status)) { | |
| 946 log_err("Error opening searcher\n"); | |
| 947 return; | |
| 948 } | |
| 949 tailored = usearch_getCollator(strsrch); | |
| 950 if (usearch_next(strsrch, &status) != -1) { | |
| 951 log_err("Error: Found case insensitive match, when we shouldn't\n"); | |
| 952 } | |
| 953 ucol_setStrength(tailored, UCOL_PRIMARY); | |
| 954 usearch_reset(strsrch); | |
| 955 if (usearch_next(strsrch, &status) != 1) { | |
| 956 log_err("Error: Found case insensitive match not found\n"); | |
| 957 } | |
| 958 usearch_close(strsrch); | |
| 959 | |
| 960 open(&status); | |
| 961 | |
| 962 if (usearch_getCollator(NULL) != NULL) { | |
| 963 log_err("Expected NULL collator from NULL string search\n"); | |
| 964 } | |
| 965 u_unescape(COLLATOR[0].text, text, 128); | |
| 966 u_unescape(COLLATOR[0].pattern, pattern, 32); | |
| 967 | |
| 968 strsrch = usearch_openFromCollator(pattern, -1, text, -1, EN_US_, | |
| 969 NULL, &status); | |
| 970 if (U_FAILURE(status)) { | |
| 971 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 972 } | |
| 973 if (!assertEqualWithUStringSearch(strsrch, COLLATOR[0])) { | |
| 974 goto ENDTESTCOLLATOR; | |
| 975 } | |
| 976 | |
| 977 u_unescape(TESTCOLLATORRULE, rules, 32); | |
| 978 tailored = ucol_openRules(rules, -1, UCOL_ON, COLLATOR[1].strength, | |
| 979 NULL, &status); | |
| 980 if (U_FAILURE(status)) { | |
| 981 log_err("Error opening rule based collator %s\n", u_errorName(status)); | |
| 982 } | |
| 983 | |
| 984 usearch_setCollator(strsrch, tailored, &status); | |
| 985 if (U_FAILURE(status) || usearch_getCollator(strsrch) != tailored) { | |
| 986 log_err("Error setting rule based collator\n"); | |
| 987 } | |
| 988 usearch_reset(strsrch); | |
| 989 if (!assertEqualWithUStringSearch(strsrch, COLLATOR[1])) { | |
| 990 goto ENDTESTCOLLATOR; | |
| 991 } | |
| 992 | |
| 993 usearch_setCollator(strsrch, EN_US_, &status); | |
| 994 usearch_reset(strsrch); | |
| 995 if (U_FAILURE(status) || usearch_getCollator(strsrch) != EN_US_) { | |
| 996 log_err("Error setting rule based collator\n"); | |
| 997 } | |
| 998 if (!assertEqualWithUStringSearch(strsrch, COLLATOR[0])) { | |
| 999 goto ENDTESTCOLLATOR; | |
| 1000 } | |
| 1001 | |
| 1002 ENDTESTCOLLATOR: | |
| 1003 usearch_close(strsrch); | |
| 1004 if (tailored != NULL) { | |
| 1005 ucol_close(tailored); | |
| 1006 } | |
| 1007 close(); | |
| 1008 } | |
| 1009 | |
| 1010 static void TestPattern(void) | |
| 1011 { | |
| 1012 UStringSearch *strsrch; | |
| 1013 UChar pattern[32]; | |
| 1014 UChar bigpattern[512]; | |
| 1015 UChar text[128]; | |
| 1016 const UChar *temp; | |
| 1017 int32_t templength; | |
| 1018 UErrorCode status = U_ZERO_ERROR; | |
| 1019 | |
| 1020 open(&status); | |
| 1021 if (U_FAILURE(status)) { | |
| 1022 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1023 return; | |
| 1024 } | |
| 1025 if (usearch_getPattern(NULL, &templength) != NULL) { | |
| 1026 log_err("Error NULL string search expected returning NULL pattern\n"); | |
| 1027 } | |
| 1028 usearch_setPattern(NULL, pattern, 3, &status); | |
| 1029 if (U_SUCCESS(status)) { | |
| 1030 log_err("Error expected setting pattern in NULL strings search\n"); | |
| 1031 } | |
| 1032 status = U_ZERO_ERROR; | |
| 1033 u_unescape(PATTERN[0].text, text, 128); | |
| 1034 u_unescape(PATTERN[0].pattern, pattern, 32); | |
| 1035 | |
| 1036 ucol_setStrength(EN_US_, PATTERN[0].strength); | |
| 1037 strsrch = usearch_openFromCollator(pattern, -1, text, -1, EN_US_, | |
| 1038 NULL, &status); | |
| 1039 if(status == U_FILE_ACCESS_ERROR) { | |
| 1040 log_data_err("Is your data around?\n"); | |
| 1041 return; | |
| 1042 } else if(U_FAILURE(status)) { | |
| 1043 log_err("Error opening searcher\n"); | |
| 1044 return; | |
| 1045 } | |
| 1046 | |
| 1047 status = U_ZERO_ERROR; | |
| 1048 usearch_setPattern(strsrch, NULL, 3, &status); | |
| 1049 if (U_SUCCESS(status)) { | |
| 1050 log_err("Error expected setting NULL pattern in strings search\n"); | |
| 1051 } | |
| 1052 status = U_ZERO_ERROR; | |
| 1053 usearch_setPattern(strsrch, pattern, 0, &status); | |
| 1054 if (U_SUCCESS(status)) { | |
| 1055 log_err("Error expected setting pattern with length 0 in strings search\
n"); | |
| 1056 } | |
| 1057 status = U_ZERO_ERROR; | |
| 1058 if (U_FAILURE(status)) { | |
| 1059 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 1060 goto ENDTESTPATTERN; | |
| 1061 } | |
| 1062 temp = usearch_getPattern(strsrch, &templength); | |
| 1063 if (u_strcmp(pattern, temp) != 0) { | |
| 1064 log_err("Error setting pattern\n"); | |
| 1065 } | |
| 1066 if (!assertEqualWithUStringSearch(strsrch, PATTERN[0])) { | |
| 1067 goto ENDTESTPATTERN; | |
| 1068 } | |
| 1069 | |
| 1070 u_unescape(PATTERN[1].pattern, pattern, 32); | |
| 1071 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 1072 temp = usearch_getPattern(strsrch, &templength); | |
| 1073 if (u_strcmp(pattern, temp) != 0) { | |
| 1074 log_err("Error setting pattern\n"); | |
| 1075 goto ENDTESTPATTERN; | |
| 1076 } | |
| 1077 usearch_reset(strsrch); | |
| 1078 if (U_FAILURE(status)) { | |
| 1079 log_err("Error setting pattern %s\n", u_errorName(status)); | |
| 1080 } | |
| 1081 if (!assertEqualWithUStringSearch(strsrch, PATTERN[1])) { | |
| 1082 goto ENDTESTPATTERN; | |
| 1083 } | |
| 1084 | |
| 1085 u_unescape(PATTERN[0].pattern, pattern, 32); | |
| 1086 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 1087 temp = usearch_getPattern(strsrch, &templength); | |
| 1088 if (u_strcmp(pattern, temp) != 0) { | |
| 1089 log_err("Error setting pattern\n"); | |
| 1090 goto ENDTESTPATTERN; | |
| 1091 } | |
| 1092 usearch_reset(strsrch); | |
| 1093 if (U_FAILURE(status)) { | |
| 1094 log_err("Error setting pattern %s\n", u_errorName(status)); | |
| 1095 } | |
| 1096 if (!assertEqualWithUStringSearch(strsrch, PATTERN[0])) { | |
| 1097 goto ENDTESTPATTERN; | |
| 1098 } | |
| 1099 /* enormous pattern size to see if this crashes */ | |
| 1100 for (templength = 0; templength != 512; templength ++) { | |
| 1101 bigpattern[templength] = 0x61; | |
| 1102 } | |
| 1103 bigpattern[511] = 0; | |
| 1104 usearch_setPattern(strsrch, bigpattern, -1, &status); | |
| 1105 if (U_FAILURE(status)) { | |
| 1106 log_err("Error setting pattern with size 512, %s \n", | |
| 1107 u_errorName(status)); | |
| 1108 } | |
| 1109 ENDTESTPATTERN: | |
| 1110 ucol_setStrength(EN_US_, UCOL_TERTIARY); | |
| 1111 if (strsrch != NULL) { | |
| 1112 usearch_close(strsrch); | |
| 1113 } | |
| 1114 close(); | |
| 1115 } | |
| 1116 | |
| 1117 static void TestText(void) | |
| 1118 { | |
| 1119 UStringSearch *strsrch; | |
| 1120 UChar pattern[32]; | |
| 1121 UChar text[128]; | |
| 1122 const UChar *temp; | |
| 1123 int32_t templength; | |
| 1124 UErrorCode status = U_ZERO_ERROR; | |
| 1125 | |
| 1126 u_unescape(TEXT[0].text, text, 128); | |
| 1127 u_unescape(TEXT[0].pattern, pattern, 32); | |
| 1128 | |
| 1129 open(&status); | |
| 1130 if (U_FAILURE(status)) { | |
| 1131 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1132 return; | |
| 1133 } | |
| 1134 | |
| 1135 if (usearch_getText(NULL, &templength) != NULL) { | |
| 1136 log_err("Error NULL string search should return NULL text\n"); | |
| 1137 } | |
| 1138 | |
| 1139 usearch_setText(NULL, text, 10, &status); | |
| 1140 if (U_SUCCESS(status)) { | |
| 1141 log_err("Error NULL string search should have an error when setting text
\n"); | |
| 1142 } | |
| 1143 | |
| 1144 status = U_ZERO_ERROR; | |
| 1145 strsrch = usearch_openFromCollator(pattern, -1, text, -1, EN_US_, | |
| 1146 NULL, &status); | |
| 1147 | |
| 1148 if (U_FAILURE(status)) { | |
| 1149 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 1150 goto ENDTESTPATTERN; | |
| 1151 } | |
| 1152 temp = usearch_getText(strsrch, &templength); | |
| 1153 if (u_strcmp(text, temp) != 0) { | |
| 1154 log_err("Error setting text\n"); | |
| 1155 } | |
| 1156 if (!assertEqualWithUStringSearch(strsrch, TEXT[0])) { | |
| 1157 goto ENDTESTPATTERN; | |
| 1158 } | |
| 1159 | |
| 1160 u_unescape(TEXT[1].text, text, 32); | |
| 1161 usearch_setText(strsrch, text, -1, &status); | |
| 1162 temp = usearch_getText(strsrch, &templength); | |
| 1163 if (u_strcmp(text, temp) != 0) { | |
| 1164 log_err("Error setting text\n"); | |
| 1165 goto ENDTESTPATTERN; | |
| 1166 } | |
| 1167 if (U_FAILURE(status)) { | |
| 1168 log_err("Error setting text %s\n", u_errorName(status)); | |
| 1169 } | |
| 1170 if (!assertEqualWithUStringSearch(strsrch, TEXT[1])) { | |
| 1171 goto ENDTESTPATTERN; | |
| 1172 } | |
| 1173 | |
| 1174 u_unescape(TEXT[0].text, text, 32); | |
| 1175 usearch_setText(strsrch, text, -1, &status); | |
| 1176 temp = usearch_getText(strsrch, &templength); | |
| 1177 if (u_strcmp(text, temp) != 0) { | |
| 1178 log_err("Error setting text\n"); | |
| 1179 goto ENDTESTPATTERN; | |
| 1180 } | |
| 1181 if (U_FAILURE(status)) { | |
| 1182 log_err("Error setting pattern %s\n", u_errorName(status)); | |
| 1183 } | |
| 1184 if (!assertEqualWithUStringSearch(strsrch, TEXT[0])) { | |
| 1185 goto ENDTESTPATTERN; | |
| 1186 } | |
| 1187 ENDTESTPATTERN: | |
| 1188 if (strsrch != NULL) { | |
| 1189 usearch_close(strsrch); | |
| 1190 } | |
| 1191 close(); | |
| 1192 } | |
| 1193 | |
| 1194 static void TestCompositeBoundaries(void) | |
| 1195 { | |
| 1196 int count = 0; | |
| 1197 UErrorCode status = U_ZERO_ERROR; | |
| 1198 open(&status); | |
| 1199 if (U_FAILURE(status)) { | |
| 1200 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1201 return; | |
| 1202 } | |
| 1203 while (COMPOSITEBOUNDARIES[count].text != NULL) { | |
| 1204 log_verbose("composite %d\n", count); | |
| 1205 if (!assertEqual(COMPOSITEBOUNDARIES[count])) { | |
| 1206 log_err("Error at test number %d\n", count); | |
| 1207 } | |
| 1208 count ++; | |
| 1209 } | |
| 1210 close(); | |
| 1211 } | |
| 1212 | |
| 1213 static void TestGetSetOffset(void) | |
| 1214 { | |
| 1215 int searchDataIndex = 0; | |
| 1216 UChar pattern[32]; | |
| 1217 UChar text[128]; | |
| 1218 UErrorCode status = U_ZERO_ERROR; | |
| 1219 UStringSearch *strsrch; | |
| 1220 memset(pattern, 0, 32*sizeof(UChar)); | |
| 1221 memset(text, 0, 128*sizeof(UChar)); | |
| 1222 | |
| 1223 open(&status); | |
| 1224 if (U_FAILURE(status)) { | |
| 1225 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1226 return; | |
| 1227 } | |
| 1228 if (usearch_getOffset(NULL) != USEARCH_DONE) { | |
| 1229 log_err("usearch_getOffset(NULL) expected USEARCH_DONE\n"); | |
| 1230 } | |
| 1231 strsrch = usearch_openFromCollator(pattern, 16, text, 32, EN_US_, NULL, | |
| 1232 &status); | |
| 1233 /* testing out of bounds error */ | |
| 1234 usearch_setOffset(strsrch, -1, &status); | |
| 1235 if (U_SUCCESS(status)) { | |
| 1236 log_err("Error expecting set offset error\n"); | |
| 1237 } | |
| 1238 usearch_setOffset(strsrch, 128, &status); | |
| 1239 if (U_SUCCESS(status)) { | |
| 1240 log_err("Error expecting set offset error\n"); | |
| 1241 } | |
| 1242 while (BASIC[searchDataIndex].text != NULL) { | |
| 1243 int count = 0; | |
| 1244 SearchData search = BASIC[searchDataIndex ++]; | |
| 1245 int32_t matchindex = search.offset[count]; | |
| 1246 int32_t textlength; | |
| 1247 | |
| 1248 u_unescape(search.text, text, 128); | |
| 1249 u_unescape(search.pattern, pattern, 32); | |
| 1250 status = U_ZERO_ERROR; | |
| 1251 usearch_setText(strsrch, text, -1, &status); | |
| 1252 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 1253 ucol_setStrength(usearch_getCollator(strsrch), search.strength); | |
| 1254 usearch_reset(strsrch); | |
| 1255 while (U_SUCCESS(status) && matchindex >= 0) { | |
| 1256 uint32_t matchlength = search.size[count]; | |
| 1257 usearch_next(strsrch, &status); | |
| 1258 if (matchindex != usearch_getMatchedStart(strsrch) || | |
| 1259 matchlength != (uint32_t)usearch_getMatchedLength(strsrch)) { | |
| 1260 char *str = toCharString(usearch_getText(strsrch, | |
| 1261 &textlength)); | |
| 1262 log_err("Text: %s\n", str); | |
| 1263 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 1264 log_err("Pattern: %s\n", str); | |
| 1265 log_err("Error match found at %d %d\n", | |
| 1266 usearch_getMatchedStart(strsrch), | |
| 1267 usearch_getMatchedLength(strsrch)); | |
| 1268 return; | |
| 1269 } | |
| 1270 usearch_setOffset(strsrch, matchindex + matchlength, &status); | |
| 1271 usearch_previous(strsrch, &status); | |
| 1272 if (matchindex != usearch_getMatchedStart(strsrch) || | |
| 1273 matchlength != (uint32_t)usearch_getMatchedLength(strsrch)) { | |
| 1274 char *str = toCharString(usearch_getText(strsrch, | |
| 1275 &textlength)); | |
| 1276 log_err("Text: %s\n", str); | |
| 1277 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 1278 log_err("Pattern: %s\n", str); | |
| 1279 log_err("Error match found at %d %d\n", | |
| 1280 usearch_getMatchedStart(strsrch), | |
| 1281 usearch_getMatchedLength(strsrch)); | |
| 1282 return; | |
| 1283 } | |
| 1284 usearch_setOffset(strsrch, matchindex + matchlength, &status); | |
| 1285 matchindex = search.offset[count + 1] == -1 ? -1 : | |
| 1286 search.offset[count + 2]; | |
| 1287 if (search.offset[count + 1] != -1) { | |
| 1288 usearch_setOffset(strsrch, search.offset[count + 1] + 1, | |
| 1289 &status); | |
| 1290 if (usearch_getOffset(strsrch) != search.offset[count + 1] + 1)
{ | |
| 1291 log_err("Error setting offset\n"); | |
| 1292 return; | |
| 1293 } | |
| 1294 } | |
| 1295 | |
| 1296 count += 2; | |
| 1297 } | |
| 1298 usearch_next(strsrch, &status); | |
| 1299 if (usearch_getMatchedStart(strsrch) != USEARCH_DONE) { | |
| 1300 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 1301 log_err("Text: %s\n", str); | |
| 1302 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 1303 log_err("Pattern: %s\n", str); | |
| 1304 log_err("Error match found at %d %d\n", | |
| 1305 usearch_getMatchedStart(strsrch), | |
| 1306 usearch_getMatchedLength(strsrch)); | |
| 1307 return; | |
| 1308 } | |
| 1309 } | |
| 1310 ucol_setStrength(usearch_getCollator(strsrch), UCOL_TERTIARY); | |
| 1311 usearch_close(strsrch); | |
| 1312 close(); | |
| 1313 } | |
| 1314 | |
| 1315 static void TestGetSetAttribute(void) | |
| 1316 { | |
| 1317 UErrorCode status = U_ZERO_ERROR; | |
| 1318 UChar pattern[32]; | |
| 1319 UChar text[128]; | |
| 1320 UStringSearch *strsrch; | |
| 1321 | |
| 1322 memset(pattern, 0, 32*sizeof(UChar)); | |
| 1323 memset(text, 0, 128*sizeof(UChar)); | |
| 1324 | |
| 1325 open(&status); | |
| 1326 if (U_FAILURE(status)) { | |
| 1327 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1328 return; | |
| 1329 } | |
| 1330 if (usearch_getAttribute(NULL, USEARCH_OVERLAP) != USEARCH_DEFAULT || | |
| 1331 usearch_getAttribute(NULL, USEARCH_CANONICAL_MATCH) != | |
| 1332 USEARCH_DEFAULT) { | |
| 1333 log_err( | |
| 1334 "Attributes for NULL string search should be USEARCH_DEFAULT\n"); | |
| 1335 } | |
| 1336 strsrch = usearch_openFromCollator(pattern, 16, text, 32, EN_US_, NULL, | |
| 1337 &status); | |
| 1338 if (U_FAILURE(status)) { | |
| 1339 log_err("Error opening search %s\n", u_errorName(status)); | |
| 1340 return; | |
| 1341 } | |
| 1342 | |
| 1343 usearch_setAttribute(strsrch, USEARCH_OVERLAP, USEARCH_DEFAULT, &status); | |
| 1344 if (U_FAILURE(status) || | |
| 1345 usearch_getAttribute(strsrch, USEARCH_OVERLAP) != USEARCH_OFF) { | |
| 1346 log_err("Error setting overlap to the default\n"); | |
| 1347 } | |
| 1348 usearch_setAttribute(strsrch, USEARCH_OVERLAP, USEARCH_ON, &status); | |
| 1349 if (U_FAILURE(status) || | |
| 1350 usearch_getAttribute(strsrch, USEARCH_OVERLAP) != USEARCH_ON) { | |
| 1351 log_err("Error setting overlap true\n"); | |
| 1352 } | |
| 1353 usearch_setAttribute(strsrch, USEARCH_OVERLAP, USEARCH_OFF, &status); | |
| 1354 if (U_FAILURE(status) || | |
| 1355 usearch_getAttribute(strsrch, USEARCH_OVERLAP) != USEARCH_OFF) { | |
| 1356 log_err("Error setting overlap false\n"); | |
| 1357 } | |
| 1358 usearch_setAttribute(strsrch, USEARCH_OVERLAP, | |
| 1359 USEARCH_ATTRIBUTE_VALUE_COUNT, &status); | |
| 1360 if (U_SUCCESS(status)) { | |
| 1361 log_err("Error setting overlap to illegal value\n"); | |
| 1362 } | |
| 1363 status = U_ZERO_ERROR; | |
| 1364 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_DEFAULT, | |
| 1365 &status); | |
| 1366 if (U_FAILURE(status) || | |
| 1367 usearch_getAttribute(strsrch, USEARCH_CANONICAL_MATCH) != | |
| 1368 USEARCH_OFF) { | |
| 1369 log_err("Error setting canonical match to the default\n"); | |
| 1370 } | |
| 1371 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 1372 &status); | |
| 1373 if (U_FAILURE(status) || | |
| 1374 usearch_getAttribute(strsrch, USEARCH_CANONICAL_MATCH) != | |
| 1375 USEARCH_ON) { | |
| 1376 log_err("Error setting canonical match true\n"); | |
| 1377 } | |
| 1378 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_OFF, | |
| 1379 &status); | |
| 1380 if (U_FAILURE(status) || | |
| 1381 usearch_getAttribute(strsrch, USEARCH_CANONICAL_MATCH) != | |
| 1382 USEARCH_OFF) { | |
| 1383 log_err("Error setting canonical match false\n"); | |
| 1384 } | |
| 1385 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, | |
| 1386 USEARCH_ATTRIBUTE_VALUE_COUNT, &status); | |
| 1387 if (U_SUCCESS(status)) { | |
| 1388 log_err("Error setting canonical match to illegal value\n"); | |
| 1389 } | |
| 1390 status = U_ZERO_ERROR; | |
| 1391 usearch_setAttribute(strsrch, USEARCH_ATTRIBUTE_COUNT, USEARCH_DEFAULT, | |
| 1392 &status); | |
| 1393 if (U_SUCCESS(status)) { | |
| 1394 log_err("Error setting illegal attribute success\n"); | |
| 1395 } | |
| 1396 | |
| 1397 usearch_close(strsrch); | |
| 1398 close(); | |
| 1399 } | |
| 1400 | |
| 1401 static void TestGetMatch(void) | |
| 1402 { | |
| 1403 int count = 0; | |
| 1404 UErrorCode status = U_ZERO_ERROR; | |
| 1405 UChar text[128]; | |
| 1406 UChar pattern[32]; | |
| 1407 SearchData search = MATCH[0]; | |
| 1408 int32_t matchindex = search.offset[count]; | |
| 1409 UStringSearch *strsrch; | |
| 1410 int32_t textlength; | |
| 1411 UChar matchtext[128]; | |
| 1412 | |
| 1413 open(&status); | |
| 1414 if (U_FAILURE(status)) { | |
| 1415 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1416 return; | |
| 1417 } | |
| 1418 | |
| 1419 if (usearch_getMatchedStart(NULL) != USEARCH_DONE || | |
| 1420 usearch_getMatchedLength(NULL) != USEARCH_DONE) { | |
| 1421 log_err( | |
| 1422 "Expected start and length of NULL string search should be USEARCH_DONE\n"); | |
| 1423 } | |
| 1424 | |
| 1425 u_unescape(search.text, text, 128); | |
| 1426 u_unescape(search.pattern, pattern, 32); | |
| 1427 strsrch = usearch_openFromCollator(pattern, -1, text, -1, EN_US_, | |
| 1428 NULL, &status); | |
| 1429 if (U_FAILURE(status)) { | |
| 1430 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 1431 if (strsrch != NULL) { | |
| 1432 usearch_close(strsrch); | |
| 1433 } | |
| 1434 return; | |
| 1435 } | |
| 1436 | |
| 1437 while (U_SUCCESS(status) && matchindex >= 0) { | |
| 1438 int32_t matchlength = search.size[count]; | |
| 1439 usearch_next(strsrch, &status); | |
| 1440 if (matchindex != usearch_getMatchedStart(strsrch) || | |
| 1441 matchlength != usearch_getMatchedLength(strsrch)) { | |
| 1442 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 1443 log_err("Text: %s\n", str); | |
| 1444 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 1445 log_err("Pattern: %s\n", str); | |
| 1446 log_err("Error match found at %d %d\n", | |
| 1447 usearch_getMatchedStart(strsrch), | |
| 1448 usearch_getMatchedLength(strsrch)); | |
| 1449 return; | |
| 1450 } | |
| 1451 count ++; | |
| 1452 | |
| 1453 status = U_ZERO_ERROR; | |
| 1454 if (usearch_getMatchedText(NULL, matchtext, 128, &status) != | |
| 1455 USEARCH_DONE || U_SUCCESS(status)){ | |
| 1456 log_err("Error expecting errors with NULL string search\n"); | |
| 1457 } | |
| 1458 status = U_ZERO_ERROR; | |
| 1459 if (usearch_getMatchedText(strsrch, NULL, 0, &status) != | |
| 1460 (int32_t)matchlength || U_SUCCESS(status)){ | |
| 1461 log_err("Error pre-flighting match length\n"); | |
| 1462 } | |
| 1463 status = U_ZERO_ERROR; | |
| 1464 if (usearch_getMatchedText(strsrch, matchtext, 0, &status) != | |
| 1465 (int32_t)matchlength || U_SUCCESS(status)){ | |
| 1466 log_err("Error getting match text with buffer size 0\n"); | |
| 1467 } | |
| 1468 status = U_ZERO_ERROR; | |
| 1469 if (usearch_getMatchedText(strsrch, matchtext, matchlength, &status) | |
| 1470 != (int32_t)matchlength || matchtext[matchlength - 1] == 0 || | |
| 1471 U_FAILURE(status)){ | |
| 1472 log_err("Error getting match text with exact size\n"); | |
| 1473 } | |
| 1474 status = U_ZERO_ERROR; | |
| 1475 if (usearch_getMatchedText(strsrch, matchtext, 128, &status) != | |
| 1476 (int32_t) matchlength || U_FAILURE(status) || | |
| 1477 memcmp(matchtext, | |
| 1478 usearch_getText(strsrch, &textlength) + matchindex, | |
| 1479 matchlength * sizeof(UChar)) != 0 || | |
| 1480 matchtext[matchlength] != 0) { | |
| 1481 log_err("Error getting matched text\n"); | |
| 1482 } | |
| 1483 | |
| 1484 matchindex = search.offset[count]; | |
| 1485 } | |
| 1486 status = U_ZERO_ERROR; | |
| 1487 usearch_next(strsrch, &status); | |
| 1488 if (usearch_getMatchedStart(strsrch) != USEARCH_DONE || | |
| 1489 usearch_getMatchedLength(strsrch) != 0) { | |
| 1490 log_err("Error end of match not found\n"); | |
| 1491 } | |
| 1492 status = U_ZERO_ERROR; | |
| 1493 if (usearch_getMatchedText(strsrch, matchtext, 128, &status) != | |
| 1494 USEARCH_DONE) { | |
| 1495 log_err("Error getting null matches\n"); | |
| 1496 } | |
| 1497 usearch_close(strsrch); | |
| 1498 close(); | |
| 1499 } | |
| 1500 | |
| 1501 static void TestSetMatch(void) | |
| 1502 { | |
| 1503 int count = 0; | |
| 1504 UErrorCode status = U_ZERO_ERROR; | |
| 1505 open(&status); | |
| 1506 if (U_FAILURE(status)) { | |
| 1507 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1508 return; | |
| 1509 } | |
| 1510 while (MATCH[count].text != NULL) { | |
| 1511 SearchData search = MATCH[count]; | |
| 1512 int size = 0; | |
| 1513 int offsetIndex = 0; | |
| 1514 UChar text[128]; | |
| 1515 UChar pattern[32]; | |
| 1516 UStringSearch *strsrch; | |
| 1517 status = U_ZERO_ERROR; | |
| 1518 | |
| 1519 if (usearch_first(NULL, &status) != USEARCH_DONE || | |
| 1520 usearch_last(NULL, &status) != USEARCH_DONE) { | |
| 1521 log_err("Error getting the first and last match of a NULL string sea
rch\n"); | |
| 1522 } | |
| 1523 u_unescape(search.text, text, 128); | |
| 1524 u_unescape(search.pattern, pattern, 32); | |
| 1525 strsrch = usearch_openFromCollator(pattern, -1, text, -1, EN_US_, | |
| 1526 NULL, &status); | |
| 1527 if (U_FAILURE(status)) { | |
| 1528 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 1529 if (strsrch != NULL) { | |
| 1530 usearch_close(strsrch); | |
| 1531 } | |
| 1532 return; | |
| 1533 } | |
| 1534 | |
| 1535 size = 0; | |
| 1536 while (search.offset[size] != -1) { | |
| 1537 size ++; | |
| 1538 } | |
| 1539 | |
| 1540 if (usearch_first(strsrch, &status) != search.offset[0] || | |
| 1541 U_FAILURE(status)) { | |
| 1542 log_err("Error getting first match\n"); | |
| 1543 } | |
| 1544 if (usearch_last(strsrch, &status) != search.offset[size -1] || | |
| 1545 U_FAILURE(status)) { | |
| 1546 log_err("Error getting last match\n"); | |
| 1547 } | |
| 1548 | |
| 1549 while (offsetIndex < size) { | |
| 1550 if (offsetIndex + 2 < size) { | |
| 1551 if (usearch_following(strsrch, search.offset[offsetIndex + 2] -
1, | |
| 1552 &status) != search.offset[offsetIndex + 2]
|| | |
| 1553 U_FAILURE(status)) { | |
| 1554 log_err("Error getting following match at index %d\n", | |
| 1555 search.offset[offsetIndex + 2] - 1); | |
| 1556 } | |
| 1557 } | |
| 1558 if (offsetIndex + 1 < size) { | |
| 1559 if (usearch_preceding(strsrch, search.offset[offsetIndex + 1] + | |
| 1560 search.size[offsetIndex + 1] + 1,
| |
| 1561 &status) != search.offset[offsetIndex + 1]
|| | |
| 1562 U_FAILURE(status)) { | |
| 1563 log_err("Error getting preceeding match at index %d\n", | |
| 1564 search.offset[offsetIndex + 1] + 1); | |
| 1565 } | |
| 1566 } | |
| 1567 offsetIndex += 2; | |
| 1568 } | |
| 1569 status = U_ZERO_ERROR; | |
| 1570 if (usearch_following(strsrch, u_strlen(text), &status) != | |
| 1571 USEARCH_DONE) { | |
| 1572 log_err("Error expecting out of bounds match\n"); | |
| 1573 } | |
| 1574 if (usearch_preceding(strsrch, 0, &status) != USEARCH_DONE) { | |
| 1575 log_err("Error expecting out of bounds match\n"); | |
| 1576 } | |
| 1577 count ++; | |
| 1578 usearch_close(strsrch); | |
| 1579 } | |
| 1580 close(); | |
| 1581 } | |
| 1582 | |
| 1583 static void TestReset(void) | |
| 1584 { | |
| 1585 UErrorCode status = U_ZERO_ERROR; | |
| 1586 UChar text[] = {0x66, 0x69, 0x73, 0x68, 0x20, | |
| 1587 0x66, 0x69, 0x73, 0x68}; | |
| 1588 UChar pattern[] = {0x73}; | |
| 1589 UStringSearch *strsrch; | |
| 1590 | |
| 1591 open(&status); | |
| 1592 if (U_FAILURE(status)) { | |
| 1593 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1594 return; | |
| 1595 } | |
| 1596 strsrch = usearch_openFromCollator(pattern, 1, text, 9, | |
| 1597 EN_US_, NULL, &status); | |
| 1598 if (U_FAILURE(status)) { | |
| 1599 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 1600 if (strsrch != NULL) { | |
| 1601 usearch_close(strsrch); | |
| 1602 } | |
| 1603 return; | |
| 1604 } | |
| 1605 usearch_setAttribute(strsrch, USEARCH_OVERLAP, USEARCH_ON, &status); | |
| 1606 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 1607 &status); | |
| 1608 usearch_setOffset(strsrch, 9, &status); | |
| 1609 if (U_FAILURE(status)) { | |
| 1610 log_err("Error setting attributes and offsets\n"); | |
| 1611 } | |
| 1612 else { | |
| 1613 usearch_reset(strsrch); | |
| 1614 if (usearch_getAttribute(strsrch, USEARCH_OVERLAP) != USEARCH_OFF || | |
| 1615 usearch_getAttribute(strsrch, USEARCH_CANONICAL_MATCH) != | |
| 1616 USEARCH_OFF || | |
| 1617 usearch_getOffset(strsrch) != 0 || | |
| 1618 usearch_getMatchedLength(strsrch) != 0 || | |
| 1619 usearch_getMatchedStart(strsrch) != USEARCH_DONE) { | |
| 1620 log_err("Error resetting string search\n"); | |
| 1621 } | |
| 1622 usearch_previous(strsrch, &status); | |
| 1623 if (usearch_getMatchedStart(strsrch) != 7 || | |
| 1624 usearch_getMatchedLength(strsrch) != 1) { | |
| 1625 log_err("Error resetting string search\n"); | |
| 1626 } | |
| 1627 } | |
| 1628 usearch_close(strsrch); | |
| 1629 close(); | |
| 1630 } | |
| 1631 | |
| 1632 static void TestSupplementary(void) | |
| 1633 { | |
| 1634 int count = 0; | |
| 1635 UErrorCode status = U_ZERO_ERROR; | |
| 1636 open(&status); | |
| 1637 if (U_FAILURE(status)) { | |
| 1638 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1639 return; | |
| 1640 } | |
| 1641 while (SUPPLEMENTARY[count].text != NULL) { | |
| 1642 if (!assertEqual(SUPPLEMENTARY[count])) { | |
| 1643 log_err("Error at test number %d\n", count); | |
| 1644 } | |
| 1645 count ++; | |
| 1646 } | |
| 1647 close(); | |
| 1648 } | |
| 1649 | |
| 1650 static void TestContraction(void) | |
| 1651 { | |
| 1652 UChar rules[128]; | |
| 1653 UChar pattern[128]; | |
| 1654 UChar text[128]; | |
| 1655 UCollator *collator; | |
| 1656 UErrorCode status = U_ZERO_ERROR; | |
| 1657 int count = 0; | |
| 1658 UStringSearch *strsrch; | |
| 1659 memset(rules, 0, 128*sizeof(UChar)); | |
| 1660 memset(pattern, 0, 128*sizeof(UChar)); | |
| 1661 memset(text, 0, 128*sizeof(UChar)); | |
| 1662 | |
| 1663 u_unescape(CONTRACTIONRULE, rules, 128); | |
| 1664 collator = ucol_openRules(rules, u_strlen(rules), UCOL_ON, | |
| 1665 UCOL_TERTIARY, NULL, &status); | |
| 1666 if(status == U_FILE_ACCESS_ERROR) { | |
| 1667 log_data_err("Is your data around?\n"); | |
| 1668 return; | |
| 1669 } else if(U_FAILURE(status)) { | |
| 1670 log_err("Error opening collator %s\n", u_errorName(status)); | |
| 1671 return; | |
| 1672 } | |
| 1673 strsrch = usearch_openFromCollator(pattern, 1, text, 1, collator, NULL, | |
| 1674 &status); | |
| 1675 if (U_FAILURE(status)) { | |
| 1676 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 1677 } | |
| 1678 | |
| 1679 while (CONTRACTION[count].text != NULL) { | |
| 1680 u_unescape(CONTRACTION[count].text, text, 128); | |
| 1681 u_unescape(CONTRACTION[count].pattern, pattern, 128); | |
| 1682 usearch_setText(strsrch, text, -1, &status); | |
| 1683 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 1684 if (!assertEqualWithUStringSearch(strsrch, CONTRACTION[count])) { | |
| 1685 log_err("Error at test number %d\n", count); | |
| 1686 } | |
| 1687 count ++; | |
| 1688 } | |
| 1689 usearch_close(strsrch); | |
| 1690 ucol_close(collator); | |
| 1691 } | |
| 1692 | |
| 1693 static void TestIgnorable(void) | |
| 1694 { | |
| 1695 UChar rules[128]; | |
| 1696 UChar pattern[128]; | |
| 1697 UChar text[128]; | |
| 1698 UCollator *collator; | |
| 1699 UErrorCode status = U_ZERO_ERROR; | |
| 1700 UStringSearch *strsrch; | |
| 1701 uint32_t count = 0; | |
| 1702 | |
| 1703 memset(rules, 0, 128*sizeof(UChar)); | |
| 1704 memset(pattern, 0, 128*sizeof(UChar)); | |
| 1705 memset(text, 0, 128*sizeof(UChar)); | |
| 1706 | |
| 1707 u_unescape(IGNORABLERULE, rules, 128); | |
| 1708 collator = ucol_openRules(rules, u_strlen(rules), UCOL_ON, | |
| 1709 IGNORABLE[count].strength, NULL, &status); | |
| 1710 if(status == U_FILE_ACCESS_ERROR) { | |
| 1711 log_data_err("Is your data around?\n"); | |
| 1712 return; | |
| 1713 } else if(U_FAILURE(status)) { | |
| 1714 log_err("Error opening collator %s\n", u_errorName(status)); | |
| 1715 return; | |
| 1716 } | |
| 1717 strsrch = usearch_openFromCollator(pattern, 1, text, 1, collator, NULL, | |
| 1718 &status); | |
| 1719 if (U_FAILURE(status)) { | |
| 1720 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 1721 } | |
| 1722 | |
| 1723 while (IGNORABLE[count].text != NULL) { | |
| 1724 u_unescape(IGNORABLE[count].text, text, 128); | |
| 1725 u_unescape(IGNORABLE[count].pattern, pattern, 128); | |
| 1726 usearch_setText(strsrch, text, -1, &status); | |
| 1727 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 1728 if (!assertEqualWithUStringSearch(strsrch, IGNORABLE[count])) { | |
| 1729 log_err("Error at test number %d\n", count); | |
| 1730 } | |
| 1731 count ++; | |
| 1732 } | |
| 1733 usearch_close(strsrch); | |
| 1734 ucol_close(collator); | |
| 1735 } | |
| 1736 | |
| 1737 static void TestDiacriticMatch(void) | |
| 1738 { | |
| 1739 UChar pattern[128]; | |
| 1740 UChar text[128]; | |
| 1741 UErrorCode status = U_ZERO_ERROR; | |
| 1742 UStringSearch *strsrch = NULL; | |
| 1743 UCollator *coll = NULL; | |
| 1744 uint32_t count = 0; | |
| 1745 SearchData search; | |
| 1746 | |
| 1747 memset(pattern, 0, 128*sizeof(UChar)); | |
| 1748 memset(text, 0, 128*sizeof(UChar)); | |
| 1749 | |
| 1750 strsrch = usearch_open(pattern, 1, text, 1, uloc_getDefault(), NULL, &status
); | |
| 1751 if (U_FAILURE(status)) { | |
| 1752 log_err_status(status, "Error opening string search %s\n", u_errorName(s
tatus)); | |
| 1753 return; | |
| 1754 } | |
| 1755 | |
| 1756 search = DIACRITICMATCH[count]; | |
| 1757 while (search.text != NULL) { | |
| 1758 if (search.collator != NULL) { | |
| 1759 coll = ucol_openFromShortString(search.collator, FALSE, NULL, &statu
s); | |
| 1760 } else { | |
| 1761 /* Always use "en_US" because some of these tests fail in Danish loc
ales. */ | |
| 1762 coll = ucol_open("en_US"/*uloc_getDefault()*/, &status); | |
| 1763 ucol_setStrength(coll, search.strength); | |
| 1764 } | |
| 1765 if (U_FAILURE(status)) { | |
| 1766 log_err("Error opening string search collator(\"%s\") %s\n", search.
collator, u_errorName(status)); | |
| 1767 return; | |
| 1768 } | |
| 1769 | |
| 1770 usearch_setCollator(strsrch, coll, &status); | |
| 1771 if (U_FAILURE(status)) { | |
| 1772 log_err("Error setting string search collator %s\n", u_errorName(sta
tus)); | |
| 1773 return; | |
| 1774 } | |
| 1775 | |
| 1776 u_unescape(search.text, text, 128); | |
| 1777 u_unescape(search.pattern, pattern, 128); | |
| 1778 usearch_setText(strsrch, text, -1, &status); | |
| 1779 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 1780 if (!assertEqualWithUStringSearch(strsrch, search)) { | |
| 1781 log_err("Error at test number %d\n", count); | |
| 1782 } | |
| 1783 ucol_close(coll); | |
| 1784 | |
| 1785 search = DIACRITICMATCH[++count]; | |
| 1786 } | |
| 1787 usearch_close(strsrch); | |
| 1788 } | |
| 1789 | |
| 1790 static void TestCanonical(void) | |
| 1791 { | |
| 1792 int count = 0; | |
| 1793 UErrorCode status = U_ZERO_ERROR; | |
| 1794 open(&status); | |
| 1795 if (U_FAILURE(status)) { | |
| 1796 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1797 return; | |
| 1798 } | |
| 1799 while (BASICCANONICAL[count].text != NULL) { | |
| 1800 if (!assertCanonicalEqual(BASICCANONICAL[count])) { | |
| 1801 log_err("Error at test number %d\n", count); | |
| 1802 } | |
| 1803 count ++; | |
| 1804 } | |
| 1805 close(); | |
| 1806 } | |
| 1807 | |
| 1808 static void TestNormCanonical(void) | |
| 1809 { | |
| 1810 int count = 0; | |
| 1811 UErrorCode status = U_ZERO_ERROR; | |
| 1812 open(&status); | |
| 1813 if (U_FAILURE(status)) { | |
| 1814 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1815 return; | |
| 1816 } | |
| 1817 ucol_setAttribute(EN_US_, UCOL_NORMALIZATION_MODE, UCOL_ON, &status); | |
| 1818 count = 0; | |
| 1819 while (NORMCANONICAL[count].text != NULL) { | |
| 1820 if (!assertCanonicalEqual(NORMCANONICAL[count])) { | |
| 1821 log_err("Error at test number %d\n", count); | |
| 1822 } | |
| 1823 count ++; | |
| 1824 } | |
| 1825 ucol_setAttribute(EN_US_, UCOL_NORMALIZATION_MODE, UCOL_OFF, &status); | |
| 1826 close(); | |
| 1827 } | |
| 1828 | |
| 1829 static void TestStrengthCanonical(void) | |
| 1830 { | |
| 1831 int count = 0; | |
| 1832 UErrorCode status = U_ZERO_ERROR; | |
| 1833 open(&status); | |
| 1834 if (U_FAILURE(status)) { | |
| 1835 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1836 return; | |
| 1837 } | |
| 1838 while (STRENGTHCANONICAL[count].text != NULL) { | |
| 1839 if (!assertCanonicalEqual(STRENGTHCANONICAL[count])) { | |
| 1840 log_err("Error at test number %d\n", count); | |
| 1841 } | |
| 1842 count ++; | |
| 1843 } | |
| 1844 close(); | |
| 1845 } | |
| 1846 | |
| 1847 static void TestBreakIteratorCanonical(void) { | |
| 1848 UErrorCode status = U_ZERO_ERROR; | |
| 1849 int count = 0; | |
| 1850 | |
| 1851 CHECK_BREAK("x"); | |
| 1852 | |
| 1853 #if !UCONFIG_NO_BREAK_ITERATION | |
| 1854 | |
| 1855 open(&status); | |
| 1856 if (U_FAILURE(status)) { | |
| 1857 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1858 return; | |
| 1859 } | |
| 1860 while (count < 4) { | |
| 1861 /* 0-3 test are fixed */ | |
| 1862 UChar pattern[32]; | |
| 1863 UChar text[128]; | |
| 1864 const SearchData *search = &(BREAKITERATORCANONICAL[count]); | |
| 1865 UCollator *collator = getCollator(search->collator); | |
| 1866 UBreakIterator *breaker = getBreakIterator(search->breaker); | |
| 1867 UStringSearch *strsrch; | |
| 1868 | |
| 1869 u_unescape(search->text, text, 128); | |
| 1870 u_unescape(search->pattern, pattern, 32); | |
| 1871 ucol_setStrength(collator, search->strength); | |
| 1872 | |
| 1873 strsrch = usearch_openFromCollator(pattern, -1, text, -1, collator, | |
| 1874 breaker, &status); | |
| 1875 if(status == U_FILE_ACCESS_ERROR) { | |
| 1876 log_data_err("Is your data around?\n"); | |
| 1877 goto ENDTESTBREAKITERATOR; | |
| 1878 } else if(U_FAILURE(status)) { | |
| 1879 log_err("Error opening searcher\n"); | |
| 1880 goto ENDTESTBREAKITERATOR; | |
| 1881 } | |
| 1882 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 1883 &status); | |
| 1884 if (U_FAILURE(status) || | |
| 1885 usearch_getBreakIterator(strsrch) != breaker) { | |
| 1886 log_err("Error setting break iterator\n"); | |
| 1887 usearch_close(strsrch); | |
| 1888 goto ENDTESTBREAKITERATOR; | |
| 1889 } | |
| 1890 if (!assertEqualWithUStringSearch(strsrch, *search)) { | |
| 1891 ucol_setStrength(collator, UCOL_TERTIARY); | |
| 1892 usearch_close(strsrch); | |
| 1893 goto ENDTESTBREAKITERATOR; | |
| 1894 } | |
| 1895 search = &(BREAKITERATOREXACT[count + 1]); | |
| 1896 breaker = getBreakIterator(search->breaker); | |
| 1897 usearch_setBreakIterator(strsrch, breaker, &status); | |
| 1898 if (U_FAILURE(status) || usearch_getBreakIterator(strsrch) != breaker) { | |
| 1899 log_err("Error setting break iterator\n"); | |
| 1900 usearch_close(strsrch); | |
| 1901 goto ENDTESTBREAKITERATOR; | |
| 1902 } | |
| 1903 usearch_reset(strsrch); | |
| 1904 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 1905 &status); | |
| 1906 if (!assertEqualWithUStringSearch(strsrch, *search)) { | |
| 1907 log_err("Error at test number %d\n", count); | |
| 1908 usearch_close(strsrch); | |
| 1909 goto ENDTESTBREAKITERATOR; | |
| 1910 } | |
| 1911 usearch_close(strsrch); | |
| 1912 count += 2; | |
| 1913 } | |
| 1914 count = 0; | |
| 1915 while (BREAKITERATORCANONICAL[count].text != NULL) { | |
| 1916 if (!assertEqual(BREAKITERATORCANONICAL[count])) { | |
| 1917 log_err("Error at test number %d\n", count); | |
| 1918 goto ENDTESTBREAKITERATOR; | |
| 1919 } | |
| 1920 count ++; | |
| 1921 } | |
| 1922 | |
| 1923 ENDTESTBREAKITERATOR: | |
| 1924 close(); | |
| 1925 #endif | |
| 1926 } | |
| 1927 | |
| 1928 static void TestVariableCanonical(void) | |
| 1929 { | |
| 1930 int count = 0; | |
| 1931 UErrorCode status = U_ZERO_ERROR; | |
| 1932 open(&status); | |
| 1933 if (U_FAILURE(status)) { | |
| 1934 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1935 return; | |
| 1936 } | |
| 1937 ucol_setAttribute(EN_US_, UCOL_ALTERNATE_HANDLING, UCOL_SHIFTED, &status); | |
| 1938 if (U_FAILURE(status)) { | |
| 1939 log_err("Error setting collation alternate attribute %s\n", | |
| 1940 u_errorName(status)); | |
| 1941 } | |
| 1942 while (VARIABLE[count].text != NULL) { | |
| 1943 log_verbose("variable %d\n", count); | |
| 1944 if (!assertCanonicalEqual(VARIABLE[count])) { | |
| 1945 log_err("Error at test number %d\n", count); | |
| 1946 } | |
| 1947 count ++; | |
| 1948 } | |
| 1949 ucol_setAttribute(EN_US_, UCOL_ALTERNATE_HANDLING, | |
| 1950 UCOL_NON_IGNORABLE, &status); | |
| 1951 close(); | |
| 1952 } | |
| 1953 | |
| 1954 static void TestOverlapCanonical(void) | |
| 1955 { | |
| 1956 int count = 0; | |
| 1957 UErrorCode status = U_ZERO_ERROR; | |
| 1958 open(&status); | |
| 1959 if (U_FAILURE(status)) { | |
| 1960 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 1961 return; | |
| 1962 } | |
| 1963 while (OVERLAPCANONICAL[count].text != NULL) { | |
| 1964 if (!assertEqualWithAttribute(OVERLAPCANONICAL[count], USEARCH_ON, | |
| 1965 USEARCH_ON)) { | |
| 1966 log_err("Error at overlap test number %d\n", count); | |
| 1967 } | |
| 1968 count ++; | |
| 1969 } | |
| 1970 count = 0; | |
| 1971 while (NONOVERLAP[count].text != NULL) { | |
| 1972 if (!assertCanonicalEqual(NONOVERLAPCANONICAL[count])) { | |
| 1973 log_err("Error at non overlap test number %d\n", count); | |
| 1974 } | |
| 1975 count ++; | |
| 1976 } | |
| 1977 | |
| 1978 count = 0; | |
| 1979 while (count < 1) { | |
| 1980 UChar pattern[32]; | |
| 1981 UChar text[128]; | |
| 1982 const SearchData *search = &(OVERLAPCANONICAL[count]); | |
| 1983 UCollator *collator = getCollator(search->collator); | |
| 1984 UStringSearch *strsrch; | |
| 1985 status = U_ZERO_ERROR; | |
| 1986 | |
| 1987 u_unescape(search->text, text, 128); | |
| 1988 u_unescape(search->pattern, pattern, 32); | |
| 1989 strsrch = usearch_openFromCollator(pattern, -1, text, -1, collator, | |
| 1990 NULL, &status); | |
| 1991 if(status == U_FILE_ACCESS_ERROR) { | |
| 1992 log_data_err("Is your data around?\n"); | |
| 1993 return; | |
| 1994 } else if(U_FAILURE(status)) { | |
| 1995 log_err("Error opening searcher\n"); | |
| 1996 return; | |
| 1997 } | |
| 1998 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 1999 &status); | |
| 2000 usearch_setAttribute(strsrch, USEARCH_OVERLAP, USEARCH_ON, &status); | |
| 2001 if (U_FAILURE(status) || | |
| 2002 usearch_getAttribute(strsrch, USEARCH_OVERLAP) != USEARCH_ON) { | |
| 2003 log_err("Error setting overlap option\n"); | |
| 2004 } | |
| 2005 if (!assertEqualWithUStringSearch(strsrch, *search)) { | |
| 2006 usearch_close(strsrch); | |
| 2007 return; | |
| 2008 } | |
| 2009 search = &(NONOVERLAPCANONICAL[count]); | |
| 2010 usearch_setAttribute(strsrch, USEARCH_OVERLAP, USEARCH_OFF, &status); | |
| 2011 if (U_FAILURE(status) || | |
| 2012 usearch_getAttribute(strsrch, USEARCH_OVERLAP) != USEARCH_OFF) { | |
| 2013 log_err("Error setting overlap option\n"); | |
| 2014 } | |
| 2015 usearch_reset(strsrch); | |
| 2016 if (!assertEqualWithUStringSearch(strsrch, *search)) { | |
| 2017 usearch_close(strsrch); | |
| 2018 log_err("Error at test number %d\n", count); | |
| 2019 } | |
| 2020 | |
| 2021 count ++; | |
| 2022 usearch_close(strsrch); | |
| 2023 } | |
| 2024 close(); | |
| 2025 } | |
| 2026 | |
| 2027 static void TestCollatorCanonical(void) | |
| 2028 { | |
| 2029 /* test collator that thinks "o" and "p" are the same thing */ | |
| 2030 UChar rules[32]; | |
| 2031 UCollator *tailored = NULL; | |
| 2032 UErrorCode status = U_ZERO_ERROR; | |
| 2033 UChar pattern[32]; | |
| 2034 UChar text[128]; | |
| 2035 UStringSearch *strsrch; | |
| 2036 | |
| 2037 open(&status); | |
| 2038 if (U_FAILURE(status)) { | |
| 2039 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 2040 return; | |
| 2041 } | |
| 2042 u_unescape(COLLATORCANONICAL[0].text, text, 128); | |
| 2043 u_unescape(COLLATORCANONICAL[0].pattern, pattern, 32); | |
| 2044 | |
| 2045 strsrch = usearch_openFromCollator(pattern, -1, text, -1, EN_US_, | |
| 2046 NULL, &status); | |
| 2047 if(status == U_FILE_ACCESS_ERROR) { | |
| 2048 log_data_err("Is your data around?\n"); | |
| 2049 return; | |
| 2050 } else if(U_FAILURE(status)) { | |
| 2051 log_err("Error opening searcher\n"); | |
| 2052 return; | |
| 2053 } | |
| 2054 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 2055 &status); | |
| 2056 if (U_FAILURE(status)) { | |
| 2057 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 2058 } | |
| 2059 if (!assertEqualWithUStringSearch(strsrch, COLLATORCANONICAL[0])) { | |
| 2060 goto ENDTESTCOLLATOR; | |
| 2061 } | |
| 2062 | |
| 2063 u_unescape(TESTCOLLATORRULE, rules, 32); | |
| 2064 tailored = ucol_openRules(rules, -1, UCOL_ON, | |
| 2065 COLLATORCANONICAL[1].strength, NULL, &status); | |
| 2066 if (U_FAILURE(status)) { | |
| 2067 log_err("Error opening rule based collator %s\n", u_errorName(status)); | |
| 2068 } | |
| 2069 | |
| 2070 usearch_setCollator(strsrch, tailored, &status); | |
| 2071 if (U_FAILURE(status) || usearch_getCollator(strsrch) != tailored) { | |
| 2072 log_err("Error setting rule based collator\n"); | |
| 2073 } | |
| 2074 usearch_reset(strsrch); | |
| 2075 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 2076 &status); | |
| 2077 if (!assertEqualWithUStringSearch(strsrch, COLLATORCANONICAL[1])) { | |
| 2078 goto ENDTESTCOLLATOR; | |
| 2079 } | |
| 2080 | |
| 2081 usearch_setCollator(strsrch, EN_US_, &status); | |
| 2082 usearch_reset(strsrch); | |
| 2083 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 2084 &status); | |
| 2085 if (U_FAILURE(status) || usearch_getCollator(strsrch) != EN_US_) { | |
| 2086 log_err("Error setting rule based collator\n"); | |
| 2087 } | |
| 2088 if (!assertEqualWithUStringSearch(strsrch, COLLATORCANONICAL[0])) { | |
| 2089 goto ENDTESTCOLLATOR; | |
| 2090 } | |
| 2091 | |
| 2092 ENDTESTCOLLATOR: | |
| 2093 usearch_close(strsrch); | |
| 2094 if (tailored != NULL) { | |
| 2095 ucol_close(tailored); | |
| 2096 } | |
| 2097 close(); | |
| 2098 } | |
| 2099 | |
| 2100 static void TestPatternCanonical(void) | |
| 2101 { | |
| 2102 UStringSearch *strsrch; | |
| 2103 UChar pattern[32]; | |
| 2104 UChar text[128]; | |
| 2105 const UChar *temp; | |
| 2106 int32_t templength; | |
| 2107 UErrorCode status = U_ZERO_ERROR; | |
| 2108 | |
| 2109 open(&status); | |
| 2110 if (U_FAILURE(status)) { | |
| 2111 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 2112 return; | |
| 2113 } | |
| 2114 u_unescape(PATTERNCANONICAL[0].text, text, 128); | |
| 2115 u_unescape(PATTERNCANONICAL[0].pattern, pattern, 32); | |
| 2116 | |
| 2117 ucol_setStrength(EN_US_, PATTERNCANONICAL[0].strength); | |
| 2118 strsrch = usearch_openFromCollator(pattern, -1, text, -1, EN_US_, | |
| 2119 NULL, &status); | |
| 2120 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 2121 &status); | |
| 2122 if (U_FAILURE(status)) { | |
| 2123 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 2124 goto ENDTESTPATTERN; | |
| 2125 } | |
| 2126 temp = usearch_getPattern(strsrch, &templength); | |
| 2127 if (u_strcmp(pattern, temp) != 0) { | |
| 2128 log_err("Error setting pattern\n"); | |
| 2129 } | |
| 2130 if (!assertEqualWithUStringSearch(strsrch, PATTERNCANONICAL[0])) { | |
| 2131 goto ENDTESTPATTERN; | |
| 2132 } | |
| 2133 | |
| 2134 u_unescape(PATTERNCANONICAL[1].pattern, pattern, 32); | |
| 2135 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 2136 temp = usearch_getPattern(strsrch, &templength); | |
| 2137 if (u_strcmp(pattern, temp) != 0) { | |
| 2138 log_err("Error setting pattern\n"); | |
| 2139 goto ENDTESTPATTERN; | |
| 2140 } | |
| 2141 usearch_reset(strsrch); | |
| 2142 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 2143 &status); | |
| 2144 if (U_FAILURE(status)) { | |
| 2145 log_err("Error setting pattern %s\n", u_errorName(status)); | |
| 2146 } | |
| 2147 if (!assertEqualWithUStringSearch(strsrch, PATTERNCANONICAL[1])) { | |
| 2148 goto ENDTESTPATTERN; | |
| 2149 } | |
| 2150 | |
| 2151 u_unescape(PATTERNCANONICAL[0].pattern, pattern, 32); | |
| 2152 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 2153 temp = usearch_getPattern(strsrch, &templength); | |
| 2154 if (u_strcmp(pattern, temp) != 0) { | |
| 2155 log_err("Error setting pattern\n"); | |
| 2156 goto ENDTESTPATTERN; | |
| 2157 } | |
| 2158 usearch_reset(strsrch); | |
| 2159 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 2160 &status); | |
| 2161 if (U_FAILURE(status)) { | |
| 2162 log_err("Error setting pattern %s\n", u_errorName(status)); | |
| 2163 } | |
| 2164 if (!assertEqualWithUStringSearch(strsrch, PATTERNCANONICAL[0])) { | |
| 2165 goto ENDTESTPATTERN; | |
| 2166 } | |
| 2167 ENDTESTPATTERN: | |
| 2168 ucol_setStrength(EN_US_, UCOL_TERTIARY); | |
| 2169 if (strsrch != NULL) { | |
| 2170 usearch_close(strsrch); | |
| 2171 } | |
| 2172 close(); | |
| 2173 } | |
| 2174 | |
| 2175 static void TestTextCanonical(void) | |
| 2176 { | |
| 2177 UStringSearch *strsrch; | |
| 2178 UChar pattern[32]; | |
| 2179 UChar text[128]; | |
| 2180 const UChar *temp; | |
| 2181 int32_t templength; | |
| 2182 UErrorCode status = U_ZERO_ERROR; | |
| 2183 | |
| 2184 u_unescape(TEXTCANONICAL[0].text, text, 128); | |
| 2185 u_unescape(TEXTCANONICAL[0].pattern, pattern, 32); | |
| 2186 | |
| 2187 open(&status); | |
| 2188 if (U_FAILURE(status)) { | |
| 2189 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 2190 return; | |
| 2191 } | |
| 2192 strsrch = usearch_openFromCollator(pattern, -1, text, -1, EN_US_, | |
| 2193 NULL, &status); | |
| 2194 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 2195 &status); | |
| 2196 | |
| 2197 if (U_FAILURE(status)) { | |
| 2198 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 2199 goto ENDTESTPATTERN; | |
| 2200 } | |
| 2201 temp = usearch_getText(strsrch, &templength); | |
| 2202 if (u_strcmp(text, temp) != 0) { | |
| 2203 log_err("Error setting text\n"); | |
| 2204 } | |
| 2205 if (!assertEqualWithUStringSearch(strsrch, TEXTCANONICAL[0])) { | |
| 2206 goto ENDTESTPATTERN; | |
| 2207 } | |
| 2208 | |
| 2209 u_unescape(TEXTCANONICAL[1].text, text, 32); | |
| 2210 usearch_setText(strsrch, text, -1, &status); | |
| 2211 temp = usearch_getText(strsrch, &templength); | |
| 2212 if (u_strcmp(text, temp) != 0) { | |
| 2213 log_err("Error setting text\n"); | |
| 2214 goto ENDTESTPATTERN; | |
| 2215 } | |
| 2216 if (U_FAILURE(status)) { | |
| 2217 log_err("Error setting text %s\n", u_errorName(status)); | |
| 2218 } | |
| 2219 if (!assertEqualWithUStringSearch(strsrch, TEXTCANONICAL[1])) { | |
| 2220 goto ENDTESTPATTERN; | |
| 2221 } | |
| 2222 | |
| 2223 u_unescape(TEXTCANONICAL[0].text, text, 32); | |
| 2224 usearch_setText(strsrch, text, -1, &status); | |
| 2225 temp = usearch_getText(strsrch, &templength); | |
| 2226 if (u_strcmp(text, temp) != 0) { | |
| 2227 log_err("Error setting text\n"); | |
| 2228 goto ENDTESTPATTERN; | |
| 2229 } | |
| 2230 if (U_FAILURE(status)) { | |
| 2231 log_err("Error setting pattern %s\n", u_errorName(status)); | |
| 2232 } | |
| 2233 if (!assertEqualWithUStringSearch(strsrch, TEXTCANONICAL[0])) { | |
| 2234 goto ENDTESTPATTERN; | |
| 2235 } | |
| 2236 ENDTESTPATTERN: | |
| 2237 if (strsrch != NULL) { | |
| 2238 usearch_close(strsrch); | |
| 2239 } | |
| 2240 close(); | |
| 2241 } | |
| 2242 | |
| 2243 static void TestCompositeBoundariesCanonical(void) | |
| 2244 { | |
| 2245 int count = 0; | |
| 2246 UErrorCode status = U_ZERO_ERROR; | |
| 2247 open(&status); | |
| 2248 if (U_FAILURE(status)) { | |
| 2249 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 2250 return; | |
| 2251 } | |
| 2252 while (COMPOSITEBOUNDARIESCANONICAL[count].text != NULL) { | |
| 2253 log_verbose("composite %d\n", count); | |
| 2254 if (!assertCanonicalEqual(COMPOSITEBOUNDARIESCANONICAL[count])) { | |
| 2255 log_err("Error at test number %d\n", count); | |
| 2256 } | |
| 2257 count ++; | |
| 2258 } | |
| 2259 close(); | |
| 2260 } | |
| 2261 | |
| 2262 static void TestGetSetOffsetCanonical(void) | |
| 2263 { | |
| 2264 int searchDataIndex = 0; | |
| 2265 UChar pattern[32]; | |
| 2266 UChar text[128]; | |
| 2267 UErrorCode status = U_ZERO_ERROR; | |
| 2268 UStringSearch *strsrch; | |
| 2269 UCollator *collator; | |
| 2270 | |
| 2271 memset(pattern, 0, 32*sizeof(UChar)); | |
| 2272 memset(text, 0, 128*sizeof(UChar)); | |
| 2273 | |
| 2274 open(&status); | |
| 2275 if (U_FAILURE(status)) { | |
| 2276 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 2277 return; | |
| 2278 } | |
| 2279 strsrch = usearch_openFromCollator(pattern, 16, text, 32, EN_US_, NULL, | |
| 2280 &status); | |
| 2281 | |
| 2282 collator = usearch_getCollator(strsrch); | |
| 2283 ucol_setAttribute(collator, UCOL_NORMALIZATION_MODE, UCOL_ON, &status); | |
| 2284 | |
| 2285 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 2286 &status); | |
| 2287 | |
| 2288 /* testing out of bounds error */ | |
| 2289 usearch_setOffset(strsrch, -1, &status); | |
| 2290 if (U_SUCCESS(status)) { | |
| 2291 log_err("Error expecting set offset error\n"); | |
| 2292 } | |
| 2293 usearch_setOffset(strsrch, 128, &status); | |
| 2294 if (U_SUCCESS(status)) { | |
| 2295 log_err("Error expecting set offset error\n"); | |
| 2296 } | |
| 2297 while (BASICCANONICAL[searchDataIndex].text != NULL) { | |
| 2298 int count = 0; | |
| 2299 SearchData search = BASICCANONICAL[searchDataIndex ++]; | |
| 2300 int32_t matchindex = search.offset[count]; | |
| 2301 int32_t textlength; | |
| 2302 | |
| 2303 if (BASICCANONICAL[searchDataIndex].text == NULL) { | |
| 2304 /* skip the last one */ | |
| 2305 break; | |
| 2306 } | |
| 2307 | |
| 2308 u_unescape(search.text, text, 128); | |
| 2309 u_unescape(search.pattern, pattern, 32); | |
| 2310 status = U_ZERO_ERROR; | |
| 2311 usearch_setText(strsrch, text, -1, &status); | |
| 2312 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 2313 while (U_SUCCESS(status) && matchindex >= 0) { | |
| 2314 uint32_t matchlength = search.size[count]; | |
| 2315 usearch_next(strsrch, &status); | |
| 2316 if (matchindex != usearch_getMatchedStart(strsrch) || | |
| 2317 matchlength != (uint32_t)usearch_getMatchedLength(strsrch)) { | |
| 2318 char *str = toCharString(usearch_getText(strsrch, | |
| 2319 &textlength)); | |
| 2320 log_err("Text: %s\n", str); | |
| 2321 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 2322 log_err("Pattern: %s\n", str); | |
| 2323 log_err("Error match found at %d %d\n", | |
| 2324 usearch_getMatchedStart(strsrch), | |
| 2325 usearch_getMatchedLength(strsrch)); | |
| 2326 goto bail; | |
| 2327 } | |
| 2328 matchindex = search.offset[count + 1] == -1 ? -1 : | |
| 2329 search.offset[count + 2]; | |
| 2330 if (search.offset[count + 1] != -1) { | |
| 2331 usearch_setOffset(strsrch, search.offset[count + 1] + 1, | |
| 2332 &status); | |
| 2333 if (usearch_getOffset(strsrch) != search.offset[count + 1] + 1)
{ | |
| 2334 log_err("Error setting offset\n"); | |
| 2335 goto bail; | |
| 2336 } | |
| 2337 } | |
| 2338 | |
| 2339 count += 2; | |
| 2340 } | |
| 2341 usearch_next(strsrch, &status); | |
| 2342 if (usearch_getMatchedStart(strsrch) != USEARCH_DONE) { | |
| 2343 char *str = toCharString(usearch_getText(strsrch, &textlength)); | |
| 2344 log_err("Text: %s\n", str); | |
| 2345 str = toCharString(usearch_getPattern(strsrch, &textlength)); | |
| 2346 log_err("Pattern: %s\n", str); | |
| 2347 log_err("Error match found at %d %d\n", | |
| 2348 usearch_getMatchedStart(strsrch), | |
| 2349 usearch_getMatchedLength(strsrch)); | |
| 2350 goto bail; | |
| 2351 } | |
| 2352 } | |
| 2353 | |
| 2354 bail: | |
| 2355 ucol_setAttribute(collator, UCOL_NORMALIZATION_MODE, UCOL_OFF, &status); | |
| 2356 usearch_close(strsrch); | |
| 2357 close(); | |
| 2358 } | |
| 2359 | |
| 2360 static void TestSupplementaryCanonical(void) | |
| 2361 { | |
| 2362 int count = 0; | |
| 2363 UErrorCode status = U_ZERO_ERROR; | |
| 2364 open(&status); | |
| 2365 if (U_FAILURE(status)) { | |
| 2366 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 2367 return; | |
| 2368 } | |
| 2369 while (SUPPLEMENTARYCANONICAL[count].text != NULL) { | |
| 2370 if (!assertCanonicalEqual(SUPPLEMENTARYCANONICAL[count])) { | |
| 2371 log_err("Error at test number %d\n", count); | |
| 2372 } | |
| 2373 count ++; | |
| 2374 } | |
| 2375 close(); | |
| 2376 } | |
| 2377 | |
| 2378 static void TestContractionCanonical(void) | |
| 2379 { | |
| 2380 UChar rules[128]; | |
| 2381 UChar pattern[128]; | |
| 2382 UChar text[128]; | |
| 2383 UCollator *collator = NULL; | |
| 2384 UErrorCode status = U_ZERO_ERROR; | |
| 2385 int count = 0; | |
| 2386 UStringSearch *strsrch = NULL; | |
| 2387 memset(rules, 0, 128*sizeof(UChar)); | |
| 2388 memset(pattern, 0, 128*sizeof(UChar)); | |
| 2389 memset(text, 0, 128*sizeof(UChar)); | |
| 2390 | |
| 2391 u_unescape(CONTRACTIONRULE, rules, 128); | |
| 2392 collator = ucol_openRules(rules, u_strlen(rules), UCOL_ON, | |
| 2393 UCOL_TERTIARY, NULL, &status); | |
| 2394 if(status == U_FILE_ACCESS_ERROR) { | |
| 2395 log_data_err("Is your data around?\n"); | |
| 2396 return; | |
| 2397 } else if(U_FAILURE(status)) { | |
| 2398 log_err("Error opening collator %s\n", u_errorName(status)); | |
| 2399 return; | |
| 2400 } | |
| 2401 strsrch = usearch_openFromCollator(pattern, 1, text, 1, collator, NULL, | |
| 2402 &status); | |
| 2403 usearch_setAttribute(strsrch, USEARCH_CANONICAL_MATCH, USEARCH_ON, | |
| 2404 &status); | |
| 2405 if (U_FAILURE(status)) { | |
| 2406 log_err("Error opening string search %s\n", u_errorName(status)); | |
| 2407 } | |
| 2408 | |
| 2409 while (CONTRACTIONCANONICAL[count].text != NULL) { | |
| 2410 u_unescape(CONTRACTIONCANONICAL[count].text, text, 128); | |
| 2411 u_unescape(CONTRACTIONCANONICAL[count].pattern, pattern, 128); | |
| 2412 usearch_setText(strsrch, text, -1, &status); | |
| 2413 usearch_setPattern(strsrch, pattern, -1, &status); | |
| 2414 if (!assertEqualWithUStringSearch(strsrch, | |
| 2415 CONTRACTIONCANONICAL[count])) { | |
| 2416 log_err("Error at test number %d\n", count); | |
| 2417 } | |
| 2418 count ++; | |
| 2419 } | |
| 2420 usearch_close(strsrch); | |
| 2421 ucol_close(collator); | |
| 2422 } | |
| 2423 | |
| 2424 static void TestNumeric(void) { | |
| 2425 UCollator *coll = NULL; | |
| 2426 UStringSearch *strsrch = NULL; | |
| 2427 UErrorCode status = U_ZERO_ERROR; | |
| 2428 | |
| 2429 UChar pattern[128]; | |
| 2430 UChar text[128]; | |
| 2431 memset(pattern, 0, 128*sizeof(UChar)); | |
| 2432 memset(text, 0, 128*sizeof(UChar)); | |
| 2433 | |
| 2434 coll = ucol_open("", &status); | |
| 2435 if(U_FAILURE(status)) { | |
| 2436 log_data_err("Could not open UCA. Is your data around?\n"); | |
| 2437 return; | |
| 2438 } | |
| 2439 | |
| 2440 ucol_setAttribute(coll, UCOL_NUMERIC_COLLATION, UCOL_ON, &status); | |
| 2441 | |
| 2442 strsrch = usearch_openFromCollator(pattern, 1, text, 1, coll, NULL, &status)
; | |
| 2443 | |
| 2444 if(status != U_UNSUPPORTED_ERROR || U_SUCCESS(status)) { | |
| 2445 log_err("Expected U_UNSUPPORTED_ERROR when trying to instantiate a searc
h object from a CODAN collator, got %s instead\n", u_errorName(status)); | |
| 2446 if(strsrch) { | |
| 2447 usearch_close(strsrch); | |
| 2448 } | |
| 2449 } | |
| 2450 | |
| 2451 ucol_close(coll); | |
| 2452 | |
| 2453 } | |
| 2454 | |
| 2455 /* This test is for ticket 4038 due to incorrect backward searching when certain
patterns have a length > 1 */ | |
| 2456 static void TestForwardBackward(void) { | |
| 2457 UErrorCode status = U_ZERO_ERROR; | |
| 2458 UCollator *coll = NULL; | |
| 2459 UStringSearch *search = NULL; | |
| 2460 UChar usrcstr[32], value[4]; | |
| 2461 int32_t pos= -1; | |
| 2462 int32_t expectedPos = 9; | |
| 2463 | |
| 2464 coll = ucol_open("en_GB", &status); | |
| 2465 if (U_FAILURE(status)) { | |
| 2466 log_err_status(status, "ucol_open failed: %s\n", u_errorName(status)); | |
| 2467 goto exitTestForwardBackward; | |
| 2468 } | |
| 2469 ucol_setAttribute(coll, UCOL_STRENGTH, UCOL_PRIMARY, &status); | |
| 2470 ucol_setAttribute(coll, UCOL_CASE_LEVEL, UCOL_ON, &status); | |
| 2471 ucol_setAttribute(coll, UCOL_ALTERNATE_HANDLING, UCOL_NON_IGNORABLE, &status
); | |
| 2472 | |
| 2473 u_uastrcpy(usrcstr, "QBitArray::bitarr_data"); /* text */ | |
| 2474 u_uastrcpy(value, "::"); /* pattern */ | |
| 2475 | |
| 2476 search = usearch_openFromCollator(value, 2, usrcstr, 22, coll, NULL, &status
); | |
| 2477 if (U_FAILURE(status)) { | |
| 2478 log_err("usearch_openFromCollator failed: %s\n", u_errorName(status)); | |
| 2479 goto exitTestForwardBackward; | |
| 2480 } | |
| 2481 | |
| 2482 usearch_reset(search); | |
| 2483 /* forward search */ | |
| 2484 pos = usearch_first(search, &status); | |
| 2485 if (pos != expectedPos) { | |
| 2486 log_err("Expected search result: %d; Got instead: %d\n", expectedPos, po
s); | |
| 2487 goto exitTestForwardBackward; | |
| 2488 } | |
| 2489 | |
| 2490 pos = -1; | |
| 2491 usearch_reset(search); | |
| 2492 /* backward search */ | |
| 2493 pos = usearch_last(search, &status); | |
| 2494 if (pos != expectedPos) { | |
| 2495 log_err("Expected search result: %d; Got instead: %d\n", expectedPos, po
s); | |
| 2496 } | |
| 2497 | |
| 2498 exitTestForwardBackward : | |
| 2499 if (coll != NULL) { | |
| 2500 ucol_close(coll); | |
| 2501 } | |
| 2502 if (search != NULL) { | |
| 2503 usearch_close(search); | |
| 2504 } | |
| 2505 } | |
| 2506 | |
| 2507 #define TEST_ASSERT(x) \ | |
| 2508 {if (U_FAILURE(x)) {log_err_status(x, "%s:%d: FAIL: test assertion failure \n
", __FILE__, __LINE__);\ | |
| 2509 }} | |
| 2510 | |
| 2511 static void TestSearchForNull(void) { | |
| 2512 UCollator *coll; | |
| 2513 UErrorCode ec; | |
| 2514 UStringSearch *search; | |
| 2515 int pos; | |
| 2516 int len; | |
| 2517 int expectedPos; | |
| 2518 int expectedLen; | |
| 2519 int expectedNum; | |
| 2520 int count = 0; | |
| 2521 const UChar zerodigit = 0x0030; /* 0 */ | |
| 2522 const UChar nulldigit = 0x0000; /* null */ | |
| 2523 | |
| 2524 /* static const UChar var[(length)+1]=U_DECLARE_UTF16(cs) */ | |
| 2525 #define PATTERN_LEN 4 | |
| 2526 #define TEXT_LEN 10 | |
| 2527 | |
| 2528 U_STRING_DECL(_pattern, "IS 0", PATTERN_LEN); | |
| 2529 U_STRING_DECL(_text, "_0IS 0 OK?", TEXT_LEN); | |
| 2530 UChar pattern[PATTERN_LEN + 1], text[TEXT_LEN + 1]; | |
| 2531 | |
| 2532 U_STRING_INIT(_pattern, "IS 0", PATTERN_LEN); | |
| 2533 U_STRING_INIT(_text, "_0IS 0 OK?", TEXT_LEN); | |
| 2534 expectedPos = 2; | |
| 2535 expectedLen = 4; | |
| 2536 expectedNum = 1; | |
| 2537 | |
| 2538 for (pos = 0; pos < PATTERN_LEN; pos++) { | |
| 2539 if (_pattern[pos] == zerodigit) { | |
| 2540 pattern[pos] = nulldigit; | |
| 2541 } else { | |
| 2542 pattern[pos] = _pattern[pos]; | |
| 2543 } | |
| 2544 } | |
| 2545 pattern[PATTERN_LEN] = 0x0000; | |
| 2546 | |
| 2547 for (pos = 0; pos < TEXT_LEN; pos++) { | |
| 2548 if (_text[pos] == zerodigit) { | |
| 2549 text[pos] = nulldigit; | |
| 2550 } else { | |
| 2551 text[pos] = _text[pos]; | |
| 2552 } | |
| 2553 } | |
| 2554 text[TEXT_LEN] = 0x0000; | |
| 2555 | |
| 2556 ec = U_ZERO_ERROR; | |
| 2557 | |
| 2558 /* create a US-English collator */ | |
| 2559 coll = ucol_open("en_US", &ec); | |
| 2560 | |
| 2561 /* make sure we didn't fail. */ | |
| 2562 TEST_ASSERT (ec); | |
| 2563 | |
| 2564 ucol_setStrength(coll, UCOL_IDENTICAL); | |
| 2565 | |
| 2566 /* open a search looking for 0 */ | |
| 2567 search = usearch_openFromCollator(pattern, PATTERN_LEN, text, | |
| 2568 TEXT_LEN, coll, NULL, &ec); | |
| 2569 TEST_ASSERT (ec); | |
| 2570 | |
| 2571 if (coll != NULL && search != NULL) { | |
| 2572 pos = usearch_first(search, &ec); | |
| 2573 len = usearch_getMatchedLength(search); | |
| 2574 if (pos != expectedPos) { | |
| 2575 log_err("Expected search result: %d; Got instead: %d\n", expectedPos
, | |
| 2576 pos); | |
| 2577 } | |
| 2578 | |
| 2579 if (len != expectedLen) { | |
| 2580 log_err("Expected search result length: %d; Got instead: %d\n", | |
| 2581 expectedLen, len); | |
| 2582 } | |
| 2583 | |
| 2584 for (pos = usearch_first(search, &ec); pos != USEARCH_DONE; pos | |
| 2585 = usearch_next(search, &ec)) { | |
| 2586 log_verbose("Match at %d\n", pos); | |
| 2587 count += 1; | |
| 2588 } | |
| 2589 | |
| 2590 if (count != expectedNum) { | |
| 2591 log_err("Expected %d search hits, found %d\n", expectedNum, count); | |
| 2592 } | |
| 2593 } | |
| 2594 | |
| 2595 ucol_close(coll); | |
| 2596 usearch_close(search); | |
| 2597 } | |
| 2598 | |
| 2599 static void TestStrengthIdentical(void) | |
| 2600 { | |
| 2601 UCollator *coll; | |
| 2602 UErrorCode ec = U_ZERO_ERROR; | |
| 2603 UStringSearch *search; | |
| 2604 | |
| 2605 UChar pattern[] = {0x05E9, 0x0591, 0x05E9}; | |
| 2606 UChar text[] = {0x05E9, 0x0592, 0x05E9}; | |
| 2607 int32_t pLen = sizeof (pattern) / sizeof(pattern[0]); | |
| 2608 int32_t tLen = sizeof(text) / sizeof (text[0]); | |
| 2609 int32_t expectedPos = 0; | |
| 2610 int32_t expectedLen = 3; | |
| 2611 | |
| 2612 int32_t pos; | |
| 2613 int32_t len; | |
| 2614 | |
| 2615 /* create a US-English collator */ | |
| 2616 coll = ucol_open ("en_US", &ec); | |
| 2617 | |
| 2618 /* make sure we didn't fail. */ | |
| 2619 TEST_ASSERT (ec); | |
| 2620 | |
| 2621 ucol_setStrength( coll, UCOL_TERTIARY); | |
| 2622 | |
| 2623 /* open a search looking for 0 */ | |
| 2624 search = usearch_openFromCollator (pattern, pLen, text, tLen, coll, NULL, &e
c); | |
| 2625 TEST_ASSERT (ec); | |
| 2626 | |
| 2627 if (coll != NULL && search != NULL) { | |
| 2628 pos = usearch_first(search, &ec); | |
| 2629 len = usearch_getMatchedLength(search); | |
| 2630 | |
| 2631 if(pos != expectedPos) { | |
| 2632 log_err("Expected search result: %d; Got instead: %d\n", expectedPos
, pos); | |
| 2633 } | |
| 2634 | |
| 2635 if(len != expectedLen) { | |
| 2636 log_err("Expected search result length: %d; Got instead: %d\n", expe
ctedLen, len); | |
| 2637 } | |
| 2638 | |
| 2639 /* Now try it at strength == UCOL_IDENTICAL */ | |
| 2640 ucol_setStrength(coll, UCOL_IDENTICAL); | |
| 2641 usearch_reset(search); | |
| 2642 | |
| 2643 pos = usearch_first(search, &ec); | |
| 2644 len = usearch_getMatchedLength(search); | |
| 2645 | |
| 2646 if(pos != -1) { | |
| 2647 log_err("Expected failure for strentgh = UCOL_IDENTICAL: got %d inst
ead.\n", pos); | |
| 2648 } | |
| 2649 } | |
| 2650 | |
| 2651 usearch_close(search); | |
| 2652 ucol_close(coll); | |
| 2653 } | |
| 2654 | |
| 2655 /** | |
| 2656 * TestUsingSearchCollator | |
| 2657 */ | |
| 2658 | |
| 2659 #define ARRAY_LENGTH(array) (sizeof(array)/sizeof(array[0])) | |
| 2660 | |
| 2661 typedef struct { | |
| 2662 const UChar * pattern; | |
| 2663 const int32_t * offsets; | |
| 2664 int32_t offsetsLen; | |
| 2665 } PatternAndOffsets; | |
| 2666 | |
| 2667 static const UChar scKoText[] = { | |
| 2668 0x0020, | |
| 2669 /*01*/ 0xAC00, 0x0020, /* simple LV Hangul */ | |
| 2670 /*03*/ 0xAC01, 0x0020, /* simple LVT Hangul */ | |
| 2671 /*05*/ 0xAC0F, 0x0020, /* LVTT, last jamo expands for se
arch */ | |
| 2672 /*07*/ 0xAFFF, 0x0020, /* LLVVVTT, every jamo expands fo
r search */ | |
| 2673 /*09*/ 0x1100, 0x1161, 0x11A8, 0x0020, /* 0xAC01 as conjoining jamo */ | |
| 2674 /*13*/ 0x1100, 0x1161, 0x1100, 0x0020, /* 0xAC01 as basic conjoining jam
o (per search rules) */ | |
| 2675 /*17*/ 0x3131, 0x314F, 0x3131, 0x0020, /* 0xAC01 as compatibility jamo *
/ | |
| 2676 /*21*/ 0x1100, 0x1161, 0x11B6, 0x0020, /* 0xAC0F as conjoining jamo; las
t expands for search */ | |
| 2677 /*25*/ 0x1100, 0x1161, 0x1105, 0x1112, 0x0020, /* 0xAC0F as basic conjoining jam
o; last expands for search */ | |
| 2678 /*30*/ 0x1101, 0x1170, 0x11B6, 0x0020, /* 0xAFFF as conjoining jamo; all
expand for search */ | |
| 2679 /*34*/ 0x00E6, 0x0020, /* small letter ae, expands */ | |
| 2680 /*36*/ 0x1E4D, 0x0020, /* small letter o with tilde and
acute, decomposes */ | |
| 2681 0 | |
| 2682 }; | |
| 2683 | |
| 2684 static const UChar scKoPat0[] = { 0xAC01, 0 }; | |
| 2685 static const UChar scKoPat1[] = { 0x1100, 0x1161, 0x11A8, 0 }; /* 0xAC01 as conj
oining jamo */ | |
| 2686 static const UChar scKoPat2[] = { 0xAC0F, 0 }; | |
| 2687 static const UChar scKoPat3[] = { 0x1100, 0x1161, 0x1105, 0x1112, 0 }; /* 0xAC0F
as basic conjoining jamo */ | |
| 2688 static const UChar scKoPat4[] = { 0xAFFF, 0 }; | |
| 2689 static const UChar scKoPat5[] = { 0x1101, 0x1170, 0x11B6, 0 }; /* 0xAFFF as conj
oining jamo */ | |
| 2690 | |
| 2691 static const int32_t scKoSrchOff01[] = { 3, 9, 13 }; | |
| 2692 static const int32_t scKoSrchOff23[] = { 5, 21, 25 }; | |
| 2693 static const int32_t scKoSrchOff45[] = { 7, 30 }; | |
| 2694 | |
| 2695 static const PatternAndOffsets scKoSrchPatternsOffsets[] = { | |
| 2696 { scKoPat0, scKoSrchOff01, ARRAY_LENGTH(scKoSrchOff01) }, | |
| 2697 { scKoPat1, scKoSrchOff01, ARRAY_LENGTH(scKoSrchOff01) }, | |
| 2698 { scKoPat2, scKoSrchOff23, ARRAY_LENGTH(scKoSrchOff23) }, | |
| 2699 { scKoPat3, scKoSrchOff23, ARRAY_LENGTH(scKoSrchOff23) }, | |
| 2700 { scKoPat4, scKoSrchOff45, ARRAY_LENGTH(scKoSrchOff45) }, | |
| 2701 { scKoPat5, scKoSrchOff45, ARRAY_LENGTH(scKoSrchOff45) }, | |
| 2702 { NULL, NULL, 0 } | |
| 2703 }; | |
| 2704 | |
| 2705 static const int32_t scKoStndOff01[] = { 3, 9 }; | |
| 2706 static const int32_t scKoStndOff2[] = { 5, 21 }; | |
| 2707 static const int32_t scKoStndOff3[] = { 25 }; | |
| 2708 static const int32_t scKoStndOff45[] = { 7, 30 }; | |
| 2709 | |
| 2710 static const PatternAndOffsets scKoStndPatternsOffsets[] = { | |
| 2711 { scKoPat0, scKoStndOff01, ARRAY_LENGTH(scKoStndOff01) }, | |
| 2712 { scKoPat1, scKoStndOff01, ARRAY_LENGTH(scKoStndOff01) }, | |
| 2713 { scKoPat2, scKoStndOff2, ARRAY_LENGTH(scKoStndOff2) }, | |
| 2714 { scKoPat3, scKoStndOff3, ARRAY_LENGTH(scKoStndOff3) }, | |
| 2715 { scKoPat4, scKoStndOff45, ARRAY_LENGTH(scKoStndOff45) }, | |
| 2716 { scKoPat5, scKoStndOff45, ARRAY_LENGTH(scKoStndOff45) }, | |
| 2717 { NULL, NULL, 0 } | |
| 2718 }; | |
| 2719 | |
| 2720 typedef struct { | |
| 2721 const char * locale; | |
| 2722 const UChar * text; | |
| 2723 const PatternAndOffsets * patternsAndOffsets; | |
| 2724 } TUSCItem; | |
| 2725 | |
| 2726 static const TUSCItem tuscItems[] = { | |
| 2727 { "root", scKoText, scKoStndPatternsOffsets }, | |
| 2728 { "root@collation=search", scKoText, scKoSrchPatternsOffsets }, | |
| 2729 { "ko@collation=search", scKoText, scKoSrchPatternsOffsets }, | |
| 2730 { NULL, NULL, NULL } | |
| 2731 }; | |
| 2732 | |
| 2733 static const UChar dummyPat[] = { 0x0061, 0 }; | |
| 2734 | |
| 2735 static void TestUsingSearchCollator(void) | |
| 2736 { | |
| 2737 const TUSCItem * tuscItemPtr; | |
| 2738 for (tuscItemPtr = tuscItems; tuscItemPtr->locale != NULL; tuscItemPtr++) { | |
| 2739 UErrorCode status = U_ZERO_ERROR; | |
| 2740 UCollator* ucol = ucol_open(tuscItemPtr->locale, &status); | |
| 2741 if ( U_SUCCESS(status) ) { | |
| 2742 UStringSearch* usrch = usearch_openFromCollator(dummyPat, -1, tuscIt
emPtr->text, -1, ucol, NULL, &status); | |
| 2743 if ( U_SUCCESS(status) ) { | |
| 2744 const PatternAndOffsets * patternsOffsetsPtr; | |
| 2745 for ( patternsOffsetsPtr = tuscItemPtr->patternsAndOffsets; patt
ernsOffsetsPtr->pattern != NULL; patternsOffsetsPtr++) { | |
| 2746 usearch_setPattern(usrch, patternsOffsetsPtr->pattern, -1, &
status); | |
| 2747 if ( U_SUCCESS(status) ) { | |
| 2748 int32_t offset; | |
| 2749 const int32_t * nextOffsetPtr; | |
| 2750 const int32_t * limitOffsetPtr; | |
| 2751 | |
| 2752 usearch_reset(usrch); | |
| 2753 nextOffsetPtr = patternsOffsetsPtr->offsets; | |
| 2754 limitOffsetPtr = patternsOffsetsPtr->offsets + patternsO
ffsetsPtr->offsetsLen; | |
| 2755 while (TRUE) { | |
| 2756 offset = usearch_next(usrch, &status); | |
| 2757 if ( U_FAILURE(status) || offset == USEARCH_DONE ) { | |
| 2758 break; | |
| 2759 } | |
| 2760 if ( nextOffsetPtr < limitOffsetPtr ) { | |
| 2761 if (offset != *nextOffsetPtr) { | |
| 2762 log_err("error, locale %s, expected usearch
_next %d, got %d\n", tuscItemPtr->locale, *nextOffsetPtr, offset); | |
| 2763 nextOffsetPtr = limitOffsetPtr; | |
| 2764 break; | |
| 2765 } | |
| 2766 nextOffsetPtr++; | |
| 2767 } else { | |
| 2768 log_err("error, locale %s, usearch_next returned
more matches than expected\n", tuscItemPtr->locale ); | |
| 2769 } | |
| 2770 } | |
| 2771 if ( U_FAILURE(status) ) { | |
| 2772 log_err("error, locale %s, usearch_next failed: %s\n
", tuscItemPtr->locale, u_errorName(status) ); | |
| 2773 } else if ( nextOffsetPtr < limitOffsetPtr ) { | |
| 2774 log_err("error, locale %s, usearch_next returned few
er matches than expected\n", tuscItemPtr->locale ); | |
| 2775 } | |
| 2776 | |
| 2777 status = U_ZERO_ERROR; | |
| 2778 usearch_reset(usrch); | |
| 2779 nextOffsetPtr = patternsOffsetsPtr->offsets + patternsOf
fsetsPtr->offsetsLen; | |
| 2780 limitOffsetPtr = patternsOffsetsPtr->offsets; | |
| 2781 while (TRUE) { | |
| 2782 offset = usearch_previous(usrch, &status); | |
| 2783 if ( U_FAILURE(status) || offset == USEARCH_DONE ) { | |
| 2784 break; | |
| 2785 } | |
| 2786 if ( nextOffsetPtr > limitOffsetPtr ) { | |
| 2787 nextOffsetPtr--; | |
| 2788 if (offset != *nextOffsetPtr) { | |
| 2789 log_err("error, locale %s, expected usearch
_previous %d, got %d\n", tuscItemPtr->locale, *nextOffsetPtr, offset); | |
| 2790 nextOffsetPtr = limitOffsetPtr; | |
| 2791 break; | |
| 2792 } | |
| 2793 } else { | |
| 2794 log_err("error, locale %s, usearch_previous retu
rned more matches than expected\n", tuscItemPtr->locale ); | |
| 2795 } | |
| 2796 } | |
| 2797 if ( U_FAILURE(status) ) { | |
| 2798 log_err("error, locale %s, usearch_previous failed:
%s\n", tuscItemPtr->locale, u_errorName(status) ); | |
| 2799 } else if ( nextOffsetPtr > limitOffsetPtr ) { | |
| 2800 log_err("error, locale %s, usearch_previous returned
fewer matches than expected\n", tuscItemPtr->locale ); | |
| 2801 } | |
| 2802 | |
| 2803 } else { | |
| 2804 log_err("error, locale %s, usearch_setPattern failed: %s
\n", tuscItemPtr->locale, u_errorName(status) ); | |
| 2805 } | |
| 2806 } | |
| 2807 usearch_close(usrch); | |
| 2808 } else { | |
| 2809 log_err("error, locale %s, usearch_openFromCollator failed: %s\n
", tuscItemPtr->locale, u_errorName(status) ); | |
| 2810 } | |
| 2811 ucol_close(ucol); | |
| 2812 } else { | |
| 2813 log_data_err("error, locale %s, ucol_open failed: %s\n", tuscItemPtr
->locale, u_errorName(status) ); | |
| 2814 } | |
| 2815 } | |
| 2816 } | |
| 2817 | |
| 2818 | |
| 2819 static void TestPCEBuffer_with(const UChar *search, uint32_t searchLen, const UC
har *source, uint32_t sourceLen) { | |
| 2820 UErrorCode icuStatus = U_ZERO_ERROR; | |
| 2821 UCollator *coll; | |
| 2822 const char *locale; | |
| 2823 UBreakIterator *ubrk; | |
| 2824 UStringSearch *usearch; | |
| 2825 int32_t match = 0; | |
| 2826 | |
| 2827 | |
| 2828 coll = ucol_openFromShortString( "LSK_AS_CX_EX_FX_HX_NX_S4", | |
| 2829 FALSE, | |
| 2830 NULL, | |
| 2831 &icuStatus ); | |
| 2832 if ( U_FAILURE(icuStatus) ) | |
| 2833 { | |
| 2834 log_data_err( "ucol_openFromShortString error %s\n" , u_errorName(icuStatus
)); | |
| 2835 goto exit; | |
| 2836 } | |
| 2837 | |
| 2838 locale = ucol_getLocaleByType( coll, | |
| 2839 ULOC_VALID_LOCALE, | |
| 2840 &icuStatus ); | |
| 2841 if ( U_FAILURE(icuStatus) ) | |
| 2842 { | |
| 2843 log_err( "ucol_getLocaleByType error %s\n", u_errorName(icuStatus) ); | |
| 2844 goto exit; | |
| 2845 } | |
| 2846 | |
| 2847 log_verbose("locale=%s\n", locale); | |
| 2848 | |
| 2849 ubrk = ubrk_open( UBRK_CHARACTER, | |
| 2850 locale, | |
| 2851 source, | |
| 2852 sourceLen, | |
| 2853 &icuStatus ); | |
| 2854 if ( U_FAILURE(icuStatus) ) | |
| 2855 { | |
| 2856 log_err( "ubrk_open error %s\n", u_errorName(icuStatus) ); | |
| 2857 goto exit; | |
| 2858 } | |
| 2859 | |
| 2860 usearch = usearch_openFromCollator( search, | |
| 2861 searchLen, | |
| 2862 source, | |
| 2863 sourceLen, | |
| 2864 coll, | |
| 2865 ubrk, | |
| 2866 &icuStatus ); | |
| 2867 if ( U_FAILURE(icuStatus) ) | |
| 2868 { | |
| 2869 log_err( "usearch_openFromCollator error %s\n", u_errorName(icuStatus) ); | |
| 2870 goto exit; | |
| 2871 } | |
| 2872 | |
| 2873 match = usearch_first( usearch, | |
| 2874 &icuStatus ); | |
| 2875 if ( U_FAILURE(icuStatus) ) | |
| 2876 { | |
| 2877 log_err( "usearch_first error %s\n", u_errorName(icuStatus) ); | |
| 2878 goto exit; | |
| 2879 } | |
| 2880 | |
| 2881 if(match==0) { | |
| 2882 log_verbose("OK: match=%d\n", match); | |
| 2883 } else { | |
| 2884 log_err("Err: match expected 0 got %d\n", match); | |
| 2885 } | |
| 2886 | |
| 2887 usearch_close(usearch); | |
| 2888 ubrk_close(ubrk); | |
| 2889 ucol_close(coll); | |
| 2890 | |
| 2891 exit: | |
| 2892 return; | |
| 2893 } | |
| 2894 | |
| 2895 | |
| 2896 static void TestPCEBuffer_100df(void) { | |
| 2897 UChar search[] = | |
| 2898 { 0x0020, 0x0020, 0x00df, 0x0020, 0x0041, 0x00df, 0x0020, 0x0061, 0x00df, 0x
0020, 0x00c5, 0x00df, 0x0020, 0x212b, 0x00df, 0x0020, 0x0041, 0x030a, 0x00df, 0x
0020, 0x00e5, 0x00df, 0x0020, 0x0061, 0x02da, 0x00df, 0x0020, 0x0061, 0x030a, 0x
00df, 0x0020, 0xd8fa, 0xdeae, 0x00df, 0x0020, 0x2027, 0x00df }; /* 38 cp, 9 of t
hem unpaired surrogates */ | |
| 2899 UChar source[] = | |
| 2900 { 0x0020, 0x0020, 0x00df, 0x0020, 0x0041, 0x00df, 0x0020, 0x0061, 0x00df, 0x
0020, 0x00c5, 0x00df, 0x0020, 0x212b, 0x00df, 0x0020, 0x0041, 0x030a, 0x00df, 0x
0020, 0x00e5, 0x00df, 0x0020, 0x0061, 0x02da, 0x00df, 0x0020, 0x0061, 0x030a, 0x
00df, 0x0020, 0xd8fa, 0xdeae, 0x00df, 0x0020, 0x2027, 0x00df }; | |
| 2901 uint32_t searchLen = sizeof(search)/sizeof(UChar); | |
| 2902 uint32_t sourceLen = sizeof(source)/sizeof(UChar); | |
| 2903 TestPCEBuffer_with(search,searchLen,source,sourceLen); | |
| 2904 } | |
| 2905 | |
| 2906 | |
| 2907 static void TestPCEBuffer_2surr(void) { | |
| 2908 UChar search[] = | |
| 2909 { 0x0020, 0x0020, 0xdfff, 0x0020, 0x0041, 0xdfff, 0x0020, 0x0061, 0xdfff, 0x
0020, 0x00c5, 0xdfff, 0x0020, 0x212b, 0xdfff, 0x0020, 0x0041, 0x030a, 0xdfff, 0x
0020, 0x00e5, 0xdfff, 0x0020, 0x0061, 0x02da, 0xdfff, 0x0020, 0x0061, 0x030a, 0x
dfff, 0x0020, 0xd8fa, 0xdeae, 0xdfff, 0x0020, 0x2027, 0xdfff }; /* 38 cp, 9 of t
hem unpaired surrogates */ | |
| 2910 UChar source[] = | |
| 2911 { 0x0020, 0x0020, 0xdfff, 0x0020, 0x0041, 0xdfff, 0x0020, 0x0061, 0xdfff, 0x
0020, 0x00c5, 0xdfff, 0x0020, 0x212b, 0xdfff, 0x0020, 0x0041, 0x030a, 0xdfff, 0x
0020, 0x00e5, 0xdfff, 0x0020, 0x0061, 0x02da, 0xdfff, 0x0020, 0x0061, 0x030a, 0x
dfff, 0x0020, 0xd8fa, 0xdeae, 0xdfff, 0x0020, 0x2027, 0xdfff }; | |
| 2912 uint32_t searchLen = sizeof(search)/sizeof(UChar); | |
| 2913 uint32_t sourceLen = sizeof(source)/sizeof(UChar); | |
| 2914 TestPCEBuffer_with(search,searchLen,source,sourceLen); | |
| 2915 } | |
| 2916 | |
| 2917 static void TestMatchFollowedByIgnorables(void) { | |
| 2918 /* test case for ticket#8482 */ | |
| 2919 UChar search[] = { 0x00c9 }; | |
| 2920 UChar source[] = { 0x00c9, 0x0000, 0x0041 }; | |
| 2921 int32_t searchLen; | |
| 2922 int32_t sourceLen; | |
| 2923 UErrorCode icuStatus = U_ZERO_ERROR; | |
| 2924 UCollator *coll; | |
| 2925 const char *locale; | |
| 2926 UBreakIterator *ubrk; | |
| 2927 UStringSearch *usearch; | |
| 2928 int32_t match = 0; | |
| 2929 int32_t matchLength = 0; | |
| 2930 const int32_t expectedMatchLength = 1; | |
| 2931 | |
| 2932 searchLen = sizeof(search)/sizeof(UChar); | |
| 2933 sourceLen = sizeof(source)/sizeof(UChar); | |
| 2934 | |
| 2935 coll = ucol_openFromShortString("LHR_AN_CX_EX_FX_HX_NX_S3", | |
| 2936 FALSE, | |
| 2937 NULL, | |
| 2938 &icuStatus); | |
| 2939 if (U_FAILURE(icuStatus)) { | |
| 2940 log_data_err("ucol_openFromShortString error - %s\n", u_errorName(icuSta
tus)); | |
| 2941 } | |
| 2942 | |
| 2943 locale = ucol_getLocaleByType(coll, | |
| 2944 ULOC_VALID_LOCALE, | |
| 2945 &icuStatus); | |
| 2946 if (U_FAILURE(icuStatus)) { | |
| 2947 log_data_err("ucol_getLocaleByType error - %s\n", u_errorName(icuStatus)
); | |
| 2948 } | |
| 2949 | |
| 2950 ubrk = ubrk_open(UBRK_CHARACTER, | |
| 2951 locale, | |
| 2952 source, | |
| 2953 sourceLen, | |
| 2954 &icuStatus); | |
| 2955 if (U_FAILURE(icuStatus)) { | |
| 2956 log_data_err("ubrk_open error - %s\n", u_errorName(icuStatus)); | |
| 2957 } | |
| 2958 | |
| 2959 usearch = usearch_openFromCollator(search, | |
| 2960 searchLen, | |
| 2961 source, | |
| 2962 sourceLen, | |
| 2963 coll, | |
| 2964 ubrk, | |
| 2965 &icuStatus); | |
| 2966 if (U_FAILURE(icuStatus)) { | |
| 2967 log_data_err("usearch_openFromCollator error - %s\n", u_errorName(icuSta
tus)); | |
| 2968 } | |
| 2969 | |
| 2970 match = usearch_first(usearch, | |
| 2971 &icuStatus); | |
| 2972 if (U_FAILURE(icuStatus)) { | |
| 2973 log_data_err("usearch_first error - %s\n", u_errorName(icuStatus)); | |
| 2974 } else { | |
| 2975 | |
| 2976 log_verbose("match=%d\n", match); | |
| 2977 | |
| 2978 matchLength = usearch_getMatchedLength(usearch); | |
| 2979 | |
| 2980 if (matchLength != expectedMatchLength) { | |
| 2981 log_err("Error: matchLength=%d, expected=%d\n", matchLength, expecte
dMatchLength); | |
| 2982 } | |
| 2983 } | |
| 2984 | |
| 2985 usearch_close(usearch); | |
| 2986 ubrk_close(ubrk); | |
| 2987 ucol_close(coll); | |
| 2988 } | |
| 2989 | |
| 2990 static void TestIndicPrefixMatch(void) | |
| 2991 { | |
| 2992 int count = 0; | |
| 2993 UErrorCode status = U_ZERO_ERROR; | |
| 2994 open(&status); | |
| 2995 if (U_FAILURE(status)) { | |
| 2996 log_err_status(status, "Unable to open static collators %s\n", u_errorNa
me(status)); | |
| 2997 return; | |
| 2998 } | |
| 2999 while (INDICPREFIXMATCH[count].text != NULL) { | |
| 3000 if (!assertEqual(INDICPREFIXMATCH[count])) { | |
| 3001 log_err("Error at test number %d\n", count); | |
| 3002 } | |
| 3003 count ++; | |
| 3004 } | |
| 3005 close(); | |
| 3006 } | |
| 3007 | |
| 3008 /** | |
| 3009 * addSearchTest | |
| 3010 */ | |
| 3011 | |
| 3012 void addSearchTest(TestNode** root) | |
| 3013 { | |
| 3014 addTest(root, &TestStart, "tscoll/usrchtst/TestStart"); | |
| 3015 addTest(root, &TestOpenClose, "tscoll/usrchtst/TestOpenClose"); | |
| 3016 addTest(root, &TestInitialization, "tscoll/usrchtst/TestInitialization"); | |
| 3017 addTest(root, &TestBasic, "tscoll/usrchtst/TestBasic"); | |
| 3018 addTest(root, &TestNormExact, "tscoll/usrchtst/TestNormExact"); | |
| 3019 addTest(root, &TestStrength, "tscoll/usrchtst/TestStrength"); | |
| 3020 addTest(root, &TestBreakIterator, "tscoll/usrchtst/TestBreakIterator"); | |
| 3021 addTest(root, &TestVariable, "tscoll/usrchtst/TestVariable"); | |
| 3022 addTest(root, &TestOverlap, "tscoll/usrchtst/TestOverlap"); | |
| 3023 addTest(root, &TestCollator, "tscoll/usrchtst/TestCollator"); | |
| 3024 addTest(root, &TestPattern, "tscoll/usrchtst/TestPattern"); | |
| 3025 addTest(root, &TestText, "tscoll/usrchtst/TestText"); | |
| 3026 addTest(root, &TestCompositeBoundaries, | |
| 3027 "tscoll/usrchtst/TestCompositeBoundaries"); | |
| 3028 addTest(root, &TestGetSetOffset, "tscoll/usrchtst/TestGetSetOffset"); | |
| 3029 addTest(root, &TestGetSetAttribute, | |
| 3030 "tscoll/usrchtst/TestGetSetAttribute"); | |
| 3031 addTest(root, &TestGetMatch, "tscoll/usrchtst/TestGetMatch"); | |
| 3032 addTest(root, &TestSetMatch, "tscoll/usrchtst/TestSetMatch"); | |
| 3033 addTest(root, &TestReset, "tscoll/usrchtst/TestReset"); | |
| 3034 addTest(root, &TestSupplementary, "tscoll/usrchtst/TestSupplementary"); | |
| 3035 addTest(root, &TestContraction, "tscoll/usrchtst/TestContraction"); | |
| 3036 addTest(root, &TestIgnorable, "tscoll/usrchtst/TestIgnorable"); | |
| 3037 addTest(root, &TestCanonical, "tscoll/usrchtst/TestCanonical"); | |
| 3038 addTest(root, &TestNormCanonical, "tscoll/usrchtst/TestNormCanonical"); | |
| 3039 addTest(root, &TestStrengthCanonical, | |
| 3040 "tscoll/usrchtst/TestStrengthCanonical"); | |
| 3041 addTest(root, &TestBreakIteratorCanonical, | |
| 3042 "tscoll/usrchtst/TestBreakIteratorCanonical"); | |
| 3043 addTest(root, &TestVariableCanonical, | |
| 3044 "tscoll/usrchtst/TestVariableCanonical"); | |
| 3045 addTest(root, &TestOverlapCanonical, | |
| 3046 "tscoll/usrchtst/TestOverlapCanonical"); | |
| 3047 addTest(root, &TestCollatorCanonical, | |
| 3048 "tscoll/usrchtst/TestCollatorCanonical"); | |
| 3049 addTest(root, &TestPatternCanonical, | |
| 3050 "tscoll/usrchtst/TestPatternCanonical"); | |
| 3051 addTest(root, &TestTextCanonical, "tscoll/usrchtst/TestTextCanonical"); | |
| 3052 addTest(root, &TestCompositeBoundariesCanonical, | |
| 3053 "tscoll/usrchtst/TestCompositeBoundariesCanonical"); | |
| 3054 addTest(root, &TestGetSetOffsetCanonical, | |
| 3055 "tscoll/usrchtst/TestGetSetOffsetCanonical"); | |
| 3056 addTest(root, &TestSupplementaryCanonical, | |
| 3057 "tscoll/usrchtst/TestSupplementaryCanonical"); | |
| 3058 addTest(root, &TestContractionCanonical, | |
| 3059 "tscoll/usrchtst/TestContractionCanonical"); | |
| 3060 addTest(root, &TestEnd, "tscoll/usrchtst/TestEnd"); | |
| 3061 addTest(root, &TestNumeric, "tscoll/usrchtst/TestNumeric"); | |
| 3062 addTest(root, &TestDiacriticMatch, "tscoll/usrchtst/TestDiacriticMatch"); | |
| 3063 addTest(root, &TestForwardBackward, "tscoll/usrchtst/TestForwardBackward"); | |
| 3064 addTest(root, &TestSearchForNull, "tscoll/usrchtst/TestSearchForNull"); | |
| 3065 addTest(root, &TestStrengthIdentical, "tscoll/usrchtst/TestStrengthIdentical
"); | |
| 3066 addTest(root, &TestUsingSearchCollator, "tscoll/usrchtst/TestUsingSearchColl
ator"); | |
| 3067 addTest(root, &TestPCEBuffer_100df, "tscoll/usrchtst/TestPCEBuffer/1_00df"); | |
| 3068 addTest(root, &TestPCEBuffer_2surr, "tscoll/usrchtst/TestPCEBuffer/2_dfff"); | |
| 3069 addTest(root, &TestMatchFollowedByIgnorables, "tscoll/usrchtst/TestMatchFoll
owedByIgnorables"); | |
| 3070 addTest(root, &TestIndicPrefixMatch, "tscoll/usrchtst/TestIndicPrefixMatch")
; | |
| 3071 } | |
| 3072 | |
| 3073 #endif /* #if !UCONFIG_NO_COLLATION */ | |
| OLD | NEW |
|---|
Chromium Code Reviews
Issue 2435373002:
Delete source/test (Closed)
