icuSources/test/intltest/rbbiapts.cpp

   1 /********************************************************************
   2  * COPYRIGHT:
   3  * Copyright (c) 1999-2004, International Business Machines Corporation and
   4  * others. All Rights Reserved.
   5  ********************************************************************/
   6 /************************************************************************
   7 *   Date        Name        Description
   8 *   12/14/99    Madhu        Creation.
   9 *   01/12/2000  Madhu        updated for changed API
  10 ************************************************************************/
  11
  12 #include "unicode/utypes.h"
  13
  14 #if !UCONFIG_NO_BREAK_ITERATION
  15
  16 #include "unicode/uchar.h"
  17 #include "intltest.h"
  18 #include "unicode/rbbi.h"
  19 #include "unicode/schriter.h"
  20 #include "rbbiapts.h"
  21 #include "rbbidata.h"
  22 #include "cstring.h"
  23 #include "unicode/ustring.h"
  24
  25 /**
  26  * API Test the RuleBasedBreakIterator class
  27  */
  28
  29
  30 #define TEST_ASSERT_SUCCESS(status) {if (U_FAILURE(status)) {\
  31 errln("Failure at file %s, line %d, error = %s", __FILE__, __LINE__, u_errorName(status));}}
  32
  33 #define TEST_ASSERT(expr) {if ((expr)==FALSE) { \
  34 errln("Test Failure at file %s, line %d", __FILE__, __LINE__);}}
  35
  36 void RBBIAPITest::TestCloneEquals()
  37 {
  38
  39     UErrorCode status=U_ZERO_ERROR;
  40     RuleBasedBreakIterator* bi1     = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createCharacterInstance(Locale::getDefault(), status);
  41     RuleBasedBreakIterator* biequal = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createCharacterInstance(Locale::getDefault(), status);
  42     RuleBasedBreakIterator* bi3     = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createCharacterInstance(Locale::getDefault(), status);
  43     RuleBasedBreakIterator* bi2     = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createWordInstance(Locale::getDefault(), status);
  44     if(U_FAILURE(status)){
  45         errln((UnicodeString)"FAIL : in construction");
  46         return;
  47     }
  48
  49
  50     UnicodeString testString="Testing word break iterators's clone() and equals()";
  51     bi1->setText(testString);
  52     bi2->setText(testString);
  53     biequal->setText(testString);
  54
  55     bi3->setText("hello");
  56
  57     logln((UnicodeString)"Testing equals()");
  58
  59     logln((UnicodeString)"Testing == and !=");
  60     UBool b = (*bi1 != *biequal);
  61     b |= *bi1 == *bi2;
  62     b |= *bi1 == *bi3;
  63     if (b) {
  64         errln((UnicodeString)"ERROR:1 RBBI's == and != operator failed.");
  65     }
  66
  67     if(*bi2 == *biequal || *bi2 == *bi1  || *biequal == *bi3)
  68         errln((UnicodeString)"ERROR:2 RBBI's == and != operator  failed.");
  69
  70
  71     // Quick test of RulesBasedBreakIterator assignment -
  72     // Check that
  73     //    two different iterators are !=
  74     //    they are == after assignment
  75     //    source and dest iterator produce the same next() after assignment.
  76     //    deleting one doesn't disable the other.
  77     logln("Testing assignment");
  78     RuleBasedBreakIterator *bix = (RuleBasedBreakIterator *)BreakIterator::createLineInstance(Locale::getEnglish(), status);
  79     if(U_FAILURE(status)){
  80         errln((UnicodeString)"FAIL : in construction");
  81         return;
  82     }
  83
  84     RuleBasedBreakIterator biDefault, biDefault2;
  85     if(U_FAILURE(status)){
  86         errln((UnicodeString)"FAIL : in construction of default iterator");
  87         return;
  88     }
  89     if (biDefault == *bix) {
  90         errln((UnicodeString)"ERROR: iterators should not compare ==");
  91         return;
  92     }
  93     if (biDefault != biDefault2) {
  94         errln((UnicodeString)"ERROR: iterators should compare ==");
  95         return;
  96     }
  97
  98
  99     UnicodeString   HelloString("Hello Kitty");
 100     bix->setText(HelloString);
 101     if (*bix == *bi2) {
 102         errln(UnicodeString("ERROR: strings should not be equal before assignment."));
 103     }
 104     *bix = *bi2;
 105     if (*bix != *bi2) {
 106         errln(UnicodeString("ERROR: strings should be equal before assignment."));
 107     }
 108
 109     int bixnext = bix->next();
 110     int bi2next = bi2->next();
 111     if (! (bixnext == bi2next && bixnext == 7)) {
 112         errln(UnicodeString("ERROR: iterators behaved differently after assignment."));
 113     }
 114     delete bix;
 115     if (bi2->next() != 8) {
 116         errln(UnicodeString("ERROR: iterator.next() failed after deleting copy."));
 117     }
 118
 119
 120
 121     logln((UnicodeString)"Testing clone()");
 122     RuleBasedBreakIterator* bi1clone=(RuleBasedBreakIterator*)bi1->clone();
 123     RuleBasedBreakIterator* bi2clone=(RuleBasedBreakIterator*)bi2->clone();
 124
 125     if(*bi1clone != *bi1 || *bi1clone  != *biequal  ||
 126       *bi1clone == *bi3 || *bi1clone == *bi2)
 127         errln((UnicodeString)"ERROR:1 RBBI's clone() method failed");
 128
 129     if(*bi2clone == *bi1 || *bi2clone == *biequal ||
 130        *bi2clone == *bi3 || *bi2clone != *bi2)
 131         errln((UnicodeString)"ERROR:2 RBBI's clone() method failed");
 132
 133     if(bi1->getText() != bi1clone->getText()   ||
 134        bi2clone->getText() != bi2->getText()   ||
 135        *bi2clone == *bi1clone )
 136         errln((UnicodeString)"ERROR: RBBI's clone() method failed");
 137
 138     delete bi1clone;
 139     delete bi2clone;
 140     delete bi1;
 141     delete bi3;
 142     delete bi2;
 143     delete biequal;
 144 }
 145
 146 void RBBIAPITest::TestBoilerPlate()
 147 {
 148     UErrorCode status = U_ZERO_ERROR;
 149     BreakIterator* a = BreakIterator::createLineInstance(Locale("hi"), status);
 150     BreakIterator* b = BreakIterator::createLineInstance(Locale("hi_IN"),status);
 151     if (U_FAILURE(status)) {
 152         errln("Creation of break iterator failed %s", u_errorName(status));
 153         return;
 154     }
 155     if(*a!=*b){
 156         errln("Failed: boilerplate method operator!= does not return correct results");
 157     }
 158     BreakIterator* c = BreakIterator::createLineInstance(Locale("th"),status);
 159     if(a && c){
 160         if(*c==*a){
 161             errln("Failed: boilerplate method opertator== does not return correct results");
 162         }
 163     }else{
 164         errln("creation of break iterator failed");
 165     }
 166     delete a;
 167     delete b;
 168     delete c;
 169 }
 170
 171 void RBBIAPITest::TestgetRules()
 172 {
 173     UErrorCode status=U_ZERO_ERROR;
 174
 175     RuleBasedBreakIterator* bi1=(RuleBasedBreakIterator*)RuleBasedBreakIterator::createCharacterInstance(Locale::getDefault(), status);
 176     RuleBasedBreakIterator* bi2=(RuleBasedBreakIterator*)RuleBasedBreakIterator::createWordInstance(Locale::getDefault(), status);
 177     if(U_FAILURE(status)){
 178         errln((UnicodeString)"FAIL: in construction");
 179         delete bi1;
 180         delete bi2;
 181         return;
 182     }
 183
 184
 185
 186     logln((UnicodeString)"Testing toString()");
 187
 188     bi1->setText((UnicodeString)"Hello there");
 189
 190     RuleBasedBreakIterator* bi3 =(RuleBasedBreakIterator*)bi1->clone();
 191
 192     UnicodeString temp=bi1->getRules();
 193     UnicodeString temp2=bi2->getRules();
 194     UnicodeString temp3=bi3->getRules();
 195     if( temp2.compare(temp3) ==0 || temp.compare(temp2) == 0 || temp.compare(temp3) != 0)
 196         errln((UnicodeString)"ERROR: error in getRules() method");
 197
 198     delete bi1;
 199     delete bi2;
 200     delete bi3;
 201 }
 202 void RBBIAPITest::TestHashCode()
 203 {
 204     UErrorCode status=U_ZERO_ERROR;
 205     RuleBasedBreakIterator* bi1     = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createCharacterInstance(Locale::getDefault(), status);
 206     RuleBasedBreakIterator* bi3     = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createCharacterInstance(Locale::getDefault(), status);
 207     RuleBasedBreakIterator* bi2     = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createWordInstance(Locale::getDefault(), status);
 208     if(U_FAILURE(status)){
 209         errln((UnicodeString)"FAIL : in construction");
 210         delete bi1;
 211         delete bi2;
 212         delete bi3;
 213         return;
 214     }
 215
 216
 217     logln((UnicodeString)"Testing hashCode()");
 218
 219     bi1->setText((UnicodeString)"Hash code");
 220     bi2->setText((UnicodeString)"Hash code");
 221     bi3->setText((UnicodeString)"Hash code");
 222
 223     RuleBasedBreakIterator* bi1clone= (RuleBasedBreakIterator*)bi1->clone();
 224     RuleBasedBreakIterator* bi2clone= (RuleBasedBreakIterator*)bi2->clone();
 225
 226     if(bi1->hashCode() != bi1clone->hashCode() ||  bi1->hashCode() != bi3->hashCode() ||
 227         bi1clone->hashCode() != bi3->hashCode() || bi2->hashCode() != bi2clone->hashCode())
 228         errln((UnicodeString)"ERROR: identical objects have different hashcodes");
 229
 230     if(bi1->hashCode() == bi2->hashCode() ||  bi2->hashCode() == bi3->hashCode() ||
 231         bi1clone->hashCode() == bi2clone->hashCode() || bi1clone->hashCode() == bi2->hashCode())
 232         errln((UnicodeString)"ERROR: different objects have same hashcodes");
 233
 234     delete bi1clone;
 235     delete bi2clone;
 236     delete bi1;
 237     delete bi2;
 238     delete bi3;
 239
 240 }
 241 void RBBIAPITest::TestGetSetAdoptText()
 242 {
 243     logln((UnicodeString)"Testing getText setText ");
 244     UErrorCode status=U_ZERO_ERROR;
 245     UnicodeString str1="first string.";
 246     UnicodeString str2="Second string.";
 247     RuleBasedBreakIterator* charIter1 = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createCharacterInstance(Locale::getDefault(), status);
 248     RuleBasedBreakIterator* wordIter1 = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createWordInstance(Locale::getDefault(), status);
 249     if(U_FAILURE(status)){
 250         errln((UnicodeString)"FAIL : in construction");
 251             return;
 252     }
 253
 254
 255     CharacterIterator* text1= new StringCharacterIterator(str1);
 256     CharacterIterator* text1Clone = text1->clone();
 257     CharacterIterator* text2= new StringCharacterIterator(str2);
 258     CharacterIterator* text3= new StringCharacterIterator(str2, 3, 10, 3); //  "ond str"
 259
 260     wordIter1->setText(str1);
 261     if(wordIter1->getText() != *text1)
 262        errln((UnicodeString)"ERROR:1 error in setText or getText ");
 263     if(wordIter1->current() != 0)
 264         errln((UnicodeString)"ERROR:1 setText did not set the iteration position to the beginning of the text, it is" + wordIter1->current() + (UnicodeString)"\n");
 265
 266     wordIter1->next(2);
 267
 268     wordIter1->setText(str2);
 269     if(wordIter1->current() != 0)
 270         errln((UnicodeString)"ERROR:2 setText did not reset the iteration position to the beginning of the text, it is" + wordIter1->current() + (UnicodeString)"\n");
 271
 272
 273     charIter1->adoptText(text1Clone);
 274     if( wordIter1->getText() == charIter1->getText() ||
 275         wordIter1->getText() != *text2 ||  charIter1->getText() != *text1 )
 276         errln((UnicodeString)"ERROR:2 error is getText or setText()");
 277
 278     RuleBasedBreakIterator* rb=(RuleBasedBreakIterator*)wordIter1->clone();
 279     rb->adoptText(text1);
 280     if(rb->getText() != *text1)
 281         errln((UnicodeString)"ERROR:1 error in adoptText ");
 282     rb->adoptText(text2);
 283     if(rb->getText() != *text2)
 284         errln((UnicodeString)"ERROR:2 error in adoptText ");
 285
 286     // Adopt where iterator range is less than the entire orignal source string.
 287     rb->adoptText(text3);
 288     if(rb->preceding(2) != 3) {
 289         errln((UnicodeString)"ERROR:3 error in adoptText ");
 290     }
 291     if(rb->following(11) != BreakIterator::DONE) {
 292         errln((UnicodeString)"ERROR:4 error in adoptText ");
 293     }
 294
 295     delete wordIter1;
 296     delete charIter1;
 297     delete rb;
 298
 299  }
 300
 301
 302 void RBBIAPITest::TestIteration()
 303 {
 304     // This test just verifies that the API is present.
 305     // Testing for correct operation of the break rules happens elsewhere.
 306
 307     UErrorCode status=U_ZERO_ERROR;
 308     RuleBasedBreakIterator* bi  = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createCharacterInstance(Locale::getDefault(), status);
 309     if (U_FAILURE(status) || bi == NULL)  {
 310         errln("Failure creating character break iterator.  Status = %s", u_errorName(status));
 311     }
 312     delete bi;
 313
 314     status=U_ZERO_ERROR;
 315     bi  = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createWordInstance(Locale::getDefault(), status);
 316     if (U_FAILURE(status) || bi == NULL)  {
 317         errln("Failure creating Word break iterator.  Status = %s", u_errorName(status));
 318     }
 319     delete bi;
 320
 321     status=U_ZERO_ERROR;
 322     bi  = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createLineInstance(Locale::getDefault(), status);
 323     if (U_FAILURE(status) || bi == NULL)  {
 324         errln("Failure creating Line break iterator.  Status = %s", u_errorName(status));
 325     }
 326     delete bi;
 327
 328     status=U_ZERO_ERROR;
 329     bi  = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createSentenceInstance(Locale::getDefault(), status);
 330     if (U_FAILURE(status) || bi == NULL)  {
 331         errln("Failure creating Sentence break iterator.  Status = %s", u_errorName(status));
 332     }
 333     delete bi;
 334
 335     status=U_ZERO_ERROR;
 336     bi  = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createTitleInstance(Locale::getDefault(), status);
 337     if (U_FAILURE(status) || bi == NULL)  {
 338         errln("Failure creating Title break iterator.  Status = %s", u_errorName(status));
 339     }
 340     delete bi;
 341
 342     status=U_ZERO_ERROR;
 343     bi  = (RuleBasedBreakIterator*)RuleBasedBreakIterator::createCharacterInstance(Locale::getDefault(), status);
 344     if (U_FAILURE(status) || bi == NULL)  {
 345         errln("Failure creating character break iterator.  Status = %s", u_errorName(status));
 346         return;   // Skip the rest of these tests.
 347     }
 348
 349
 350     UnicodeString testString="0123456789";
 351     bi->setText(testString);
 352
 353     int32_t i;
 354     i = bi->first();
 355     if (i != 0) {
 356         errln("Incorrect value from bi->first().  Expected 0, got %d.", i);
 357     }
 358
 359     i = bi->last();
 360     if (i != 10) {
 361         errln("Incorrect value from bi->last().  Expected 10, got %d", i);
 362     }
 363
 364     //
 365     // Previous
 366     //
 367     bi->last();
 368     i = bi->previous();
 369     if (i != 9) {
 370         errln("Incorrect value from bi->last() at line %d.  Expected 9, got %d", __LINE__, i);
 371     }
 372
 373
 374     bi->first();
 375     i = bi->previous();
 376     if (i != BreakIterator::DONE) {
 377         errln("Incorrect value from bi->previous() at line %d.  Expected DONE, got %d", __LINE__, i);
 378     }
 379
 380     //
 381     // next()
 382     //
 383     bi->first();
 384     i = bi->next();
 385     if (i != 1) {
 386         errln("Incorrect value from bi->next() at line %d.  Expected 1, got %d", __LINE__, i);
 387     }
 388
 389     bi->last();
 390     i = bi->next();
 391     if (i != BreakIterator::DONE) {
 392         errln("Incorrect value from bi->next() at line %d.  Expected DONE, got %d", __LINE__, i);
 393     }
 394
 395
 396     //
 397     //  current()
 398     //
 399     bi->first();
 400     i = bi->current();
 401     if (i != 0) {
 402         errln("Incorrect value from bi->previous() at line %d.  Expected 0, got %d", __LINE__, i);
 403     }
 404
 405     bi->next();
 406     i = bi->current();
 407     if (i != 1) {
 408         errln("Incorrect value from bi->previous() at line %d.  Expected 1, got %d", __LINE__, i);
 409     }
 410
 411     bi->last();
 412     bi->next();
 413     i = bi->current();
 414     if (i != 10) {
 415         errln("Incorrect value from bi->previous() at line %d.  Expected 10, got %d", __LINE__, i);
 416     }
 417
 418     bi->first();
 419     bi->previous();
 420     i = bi->current();
 421     if (i != 0) {
 422         errln("Incorrect value from bi->previous() at line %d.  Expected 0, got %d", __LINE__, i);
 423     }
 424
 425
 426     //
 427     // Following()
 428     //
 429     i = bi->following(4);
 430     if (i != 5) {
 431         errln("Incorrect value from bi->following() at line %d.  Expected 5, got %d", __LINE__, i);
 432     }
 433
 434     i = bi->following(9);
 435     if (i != 10) {
 436         errln("Incorrect value from bi->following() at line %d.  Expected 10, got %d", __LINE__, i);
 437     }
 438
 439     i = bi->following(10);
 440     if (i != BreakIterator::DONE) {
 441         errln("Incorrect value from bi->following() at line %d.  Expected DONE, got %d", __LINE__, i);
 442     }
 443
 444
 445     //
 446     // Preceding
 447     //
 448     i = bi->preceding(4);
 449     if (i != 3) {
 450         errln("Incorrect value from bi->preceding() at line %d.  Expected 3, got %d", __LINE__, i);
 451     }
 452
 453     i = bi->preceding(10);
 454     if (i != 9) {
 455         errln("Incorrect value from bi->preceding() at line %d.  Expected 9, got %d", __LINE__, i);
 456     }
 457
 458     i = bi->preceding(1);
 459     if (i != 0) {
 460         errln("Incorrect value from bi->preceding() at line %d.  Expected 0, got %d", __LINE__, i);
 461     }
 462
 463     i = bi->preceding(0);
 464     if (i != BreakIterator::DONE) {
 465         errln("Incorrect value from bi->preceding() at line %d.  Expected DONE, got %d", __LINE__, i);
 466     }
 467
 468
 469     //
 470     // isBoundary()
 471     //
 472     bi->first();
 473     if (bi->isBoundary(3) != TRUE) {
 474         errln("Incorrect value from bi->isBoudary() at line %d.  Expected TRUE, got FALSE", __LINE__, i);
 475     }
 476     i = bi->current();
 477     if (i != 3) {
 478         errln("Incorrect value from bi->current() at line %d.  Expected 3, got %d", __LINE__, i);
 479     }
 480
 481
 482     if (bi->isBoundary(11) != FALSE) {
 483         errln("Incorrect value from bi->isBoudary() at line %d.  Expected FALSE, got TRUE", __LINE__, i);
 484     }
 485     i = bi->current();
 486     if (i != 10) {
 487         errln("Incorrect value from bi->current() at line %d.  Expected 10, got %d", __LINE__, i);
 488     }
 489
 490     //
 491     // next(n)
 492     //
 493     bi->first();
 494     i = bi->next(4);
 495     if (i != 4) {
 496         errln("Incorrect value from bi->next() at line %d.  Expected 4, got %d", __LINE__, i);
 497     }
 498
 499     i = bi->next(6);
 500     if (i != 10) {
 501         errln("Incorrect value from bi->next() at line %d.  Expected 10, got %d", __LINE__, i);
 502     }
 503
 504     bi->first();
 505     i = bi->next(11);
 506     if (i != BreakIterator::DONE) {
 507         errln("Incorrect value from bi->next() at line %d.  Expected BreakIterator::DONE, got %d", __LINE__, i);
 508     }
 509
 510     delete bi;
 511
 512 }
 513
 514
 515
 516
 517
 518
 519 void RBBIAPITest::TestBuilder() {
 520      UnicodeString rulesString1 = "$Letters = [:L:];\n"
 521                                   "$Numbers = [:N:];\n"
 522                                   "$Letters+;\n"
 523                                   "$Numbers+;\n"
 524                                   "[^$Letters $Numbers];\n"
 525                                   "!.*;\n";
 526      UnicodeString testString1  = "abc123..abc";
 527                                 // 01234567890
 528      int32_t bounds1[] = {0, 3, 6, 7, 8, 11};
 529      UErrorCode status=U_ZERO_ERROR;
 530      UParseError    parseError;
 531
 532      RuleBasedBreakIterator *bi = new RuleBasedBreakIterator(rulesString1, parseError, status);
 533      if(U_FAILURE(status)) {
 534          errln("FAIL : in construction");
 535      } else {
 536          bi->setText(testString1);
 537          doBoundaryTest(*bi, testString1, bounds1);
 538      }
 539      delete bi;
 540 }
 541
 542
 543 //
 544 //  TestQuoteGrouping
 545 //       Single quotes within rules imply a grouping, so that a modifier
 546 //       following the quoted text (* or +) applies to all of the quoted chars.
 547 //
 548 void RBBIAPITest::TestQuoteGrouping() {
 549      UnicodeString rulesString1 = "#Here comes the rule...\n"
 550                                   "'$@!'*;\n"   //  (\$\@\!)*
 551                                   ".;\n";
 552
 553      UnicodeString testString1  = "$@!$@!X$@!!X";
 554                                 // 0123456789012
 555      int32_t bounds1[] = {0, 6, 7, 10, 11, 12};
 556      UErrorCode status=U_ZERO_ERROR;
 557      UParseError    parseError;
 558
 559      RuleBasedBreakIterator *bi = new RuleBasedBreakIterator(rulesString1, parseError, status);
 560      if(U_FAILURE(status)) {
 561          errln("FAIL : in construction");
 562      } else {
 563          bi->setText(testString1);
 564          doBoundaryTest(*bi, testString1, bounds1);
 565      }
 566      delete bi;
 567 }
 568
 569 //
 570 //  TestRuleStatus
 571 //      Test word break rule status constants.
 572 //
 573 void RBBIAPITest::TestRuleStatus() {
 574      UChar str[30];
 575      u_unescape("plain word 123.45 \\u9160\\u9161 \\u30a1\\u30a2 \\u3041\\u3094",
 576               // 012345678901234567  8      9    0  1      2    3  4      5    6
 577               //                    Ideographic    Katakana       Hiragana
 578                 str, 30);
 579      UnicodeString testString1(str);
 580      int32_t bounds1[] = {0, 5, 6, 10, 11, 17, 18, 19, 20, 21, 23, 24, 25, 26};
 581      int32_t tag_lo[]  = {UBRK_WORD_NONE,     UBRK_WORD_LETTER, UBRK_WORD_NONE,    UBRK_WORD_LETTER,
 582                           UBRK_WORD_NONE,     UBRK_WORD_NUMBER, UBRK_WORD_NONE,
 583                           UBRK_WORD_IDEO,     UBRK_WORD_IDEO,   UBRK_WORD_NONE,
 584                           UBRK_WORD_KANA,     UBRK_WORD_NONE,   UBRK_WORD_KANA,    UBRK_WORD_KANA};
 585
 586      int32_t tag_hi[]  = {UBRK_WORD_NONE_LIMIT, UBRK_WORD_LETTER_LIMIT, UBRK_WORD_NONE_LIMIT, UBRK_WORD_LETTER_LIMIT,
 587                           UBRK_WORD_NONE_LIMIT, UBRK_WORD_NUMBER_LIMIT, UBRK_WORD_NONE_LIMIT,
 588                           UBRK_WORD_IDEO_LIMIT, UBRK_WORD_IDEO_LIMIT,   UBRK_WORD_NONE_LIMIT,
 589                           UBRK_WORD_KANA_LIMIT, UBRK_WORD_NONE_LIMIT,   UBRK_WORD_KANA_LIMIT, UBRK_WORD_KANA_LIMIT};
 590
 591      UErrorCode status=U_ZERO_ERROR;
 592
 593      RuleBasedBreakIterator *bi = (RuleBasedBreakIterator *)BreakIterator::createWordInstance(Locale::getDefault(), status);
 594      if(U_FAILURE(status)) {
 595          errln("FAIL : in construction");
 596      } else {
 597          bi->setText(testString1);
 598          // First test that the breaks are in the right spots.
 599          doBoundaryTest(*bi, testString1, bounds1);
 600
 601          // Then go back and check tag values
 602          int32_t i = 0;
 603          int32_t pos, tag;
 604          for (pos = bi->first(); pos != BreakIterator::DONE; pos = bi->next(), i++) {
 605              if (pos != bounds1[i]) {
 606                  errln("FAIL: unexpected word break at postion %d", pos);
 607                  break;
 608              }
 609              tag = bi->getRuleStatus();
 610              if (tag < tag_lo[i] || tag >= tag_hi[i]) {
 611                  errln("FAIL: incorrect tag value %d at position %d", tag, pos);
 612                  break;
 613              }
 614
 615              // Check that we get the same tag values from getRuleStatusVec()
 616              int32_t vec[10];
 617              int t = bi->getRuleStatusVec(vec, 10, status);
 618              TEST_ASSERT_SUCCESS(status);
 619              TEST_ASSERT(t==1);
 620              TEST_ASSERT(vec[0] == tag);
 621          }
 622      }
 623      delete bi;
 624
 625      // Now test line break status.  This test mostly is to confirm that the status constants
 626      //                              are correctly declared in the header.
 627      testString1 =   "test line. \n";
 628      // break type    s    s     h
 629
 630      bi = (RuleBasedBreakIterator *)
 631          BreakIterator::createLineInstance(Locale::getEnglish(), status);
 632      if(U_FAILURE(status)) {
 633          errln("failed to create word break iterator.");
 634      } else {
 635          int32_t i = 0;
 636          int32_t pos, tag;
 637          UBool   success;
 638
 639          bi->setText(testString1);
 640          pos = bi->current();
 641          tag = bi->getRuleStatus();
 642          for (i=0; i<3; i++) {
 643              switch (i) {
 644              case 0:
 645                  success = pos==0  && tag==UBRK_LINE_SOFT; break;
 646              case 1:
 647                  success = pos==5  && tag==UBRK_LINE_SOFT; break;
 648              case 2:
 649                  success = pos==12 && tag==UBRK_LINE_HARD; break;
 650              default:
 651                  success = FALSE; break;
 652              }
 653              if (success == FALSE) {
 654                  errln("Fail: incorrect word break status or position.  i=%d, pos=%d, tag=%d",
 655                      i, pos, tag);
 656                  break;
 657              }
 658              pos = bi->next();
 659              tag = bi->getRuleStatus();
 660          }
 661          if (UBRK_LINE_SOFT >= UBRK_LINE_SOFT_LIMIT ||
 662              UBRK_LINE_HARD >= UBRK_LINE_HARD_LIMIT ||
 663              UBRK_LINE_HARD > UBRK_LINE_SOFT && UBRK_LINE_HARD < UBRK_LINE_SOFT_LIMIT ) {
 664              errln("UBRK_LINE_* constants from header are inconsistent.");
 665          }
 666      }
 667      delete bi;
 668
 669 }
 670
 671
 672 //
 673 //  TestRuleStatusVec
 674 //      Test the vector form of  break rule status.
 675 //
 676 void RBBIAPITest::TestRuleStatusVec() {
 677     UnicodeString rulesString  = "[A-N]{100}; \n"
 678                                  "[a-w]{200}; \n"
 679                                  "[\\p{L}]{300}; \n"
 680                                  "[\\p{N}]{400}; \n"
 681                                  "[0-5]{500}; \n"
 682                                   "!.*;\n";
 683      UnicodeString testString1  = "Aapz5?";
 684      int32_t  statusVals[10];
 685      int32_t  numStatuses;
 686      int32_t  pos;
 687
 688      UErrorCode status=U_ZERO_ERROR;
 689      UParseError    parseError;
 690
 691      RuleBasedBreakIterator *bi = new RuleBasedBreakIterator(rulesString, parseError, status);
 692      TEST_ASSERT_SUCCESS(status);
 693      if (U_SUCCESS(status)) {
 694          bi->setText(testString1);
 695
 696          // A
 697          pos = bi->next();
 698          TEST_ASSERT(pos==1);
 699          numStatuses = bi->getRuleStatusVec(statusVals, 10, status);
 700          TEST_ASSERT_SUCCESS(status);
 701          TEST_ASSERT(numStatuses == 2);
 702          TEST_ASSERT(statusVals[0] == 100);
 703          TEST_ASSERT(statusVals[1] == 300);
 704
 705          // a
 706          pos = bi->next();
 707          TEST_ASSERT(pos==2);
 708          numStatuses = bi->getRuleStatusVec(statusVals, 10, status);
 709          TEST_ASSERT_SUCCESS(status);
 710          TEST_ASSERT(numStatuses == 2);
 711          TEST_ASSERT(statusVals[0] == 200);
 712          TEST_ASSERT(statusVals[1] == 300);
 713
 714          // p
 715          pos = bi->next();
 716          TEST_ASSERT(pos==3);
 717          numStatuses = bi->getRuleStatusVec(statusVals, 10, status);
 718          TEST_ASSERT_SUCCESS(status);
 719          TEST_ASSERT(numStatuses == 2);
 720          TEST_ASSERT(statusVals[0] == 200);
 721          TEST_ASSERT(statusVals[1] == 300);
 722
 723          // z
 724          pos = bi->next();
 725          TEST_ASSERT(pos==4);
 726          numStatuses = bi->getRuleStatusVec(statusVals, 10, status);
 727          TEST_ASSERT_SUCCESS(status);
 728          TEST_ASSERT(numStatuses == 1);
 729          TEST_ASSERT(statusVals[0] == 300);
 730
 731          // 5
 732          pos = bi->next();
 733          TEST_ASSERT(pos==5);
 734          numStatuses = bi->getRuleStatusVec(statusVals, 10, status);
 735          TEST_ASSERT_SUCCESS(status);
 736          TEST_ASSERT(numStatuses == 2);
 737          TEST_ASSERT(statusVals[0] == 400);
 738          TEST_ASSERT(statusVals[1] == 500);
 739
 740          // ?
 741          pos = bi->next();
 742          TEST_ASSERT(pos==6);
 743          numStatuses = bi->getRuleStatusVec(statusVals, 10, status);
 744          TEST_ASSERT_SUCCESS(status);
 745          TEST_ASSERT(numStatuses == 1);
 746          TEST_ASSERT(statusVals[0] == 0);
 747
 748          //
 749          //  Check buffer overflow error handling.   Char == A
 750          //
 751          bi->first();
 752          pos = bi->next();
 753          TEST_ASSERT(pos==1);
 754          memset(statusVals, -1, sizeof(statusVals));
 755          numStatuses = bi->getRuleStatusVec(statusVals, 0, status);
 756          TEST_ASSERT(status == U_BUFFER_OVERFLOW_ERROR);
 757          TEST_ASSERT(numStatuses == 2);
 758          TEST_ASSERT(statusVals[0] == -1);
 759
 760          status = U_ZERO_ERROR;
 761          memset(statusVals, -1, sizeof(statusVals));
 762          numStatuses = bi->getRuleStatusVec(statusVals, 1, status);
 763          TEST_ASSERT(status == U_BUFFER_OVERFLOW_ERROR);
 764          TEST_ASSERT(numStatuses == 2);
 765          TEST_ASSERT(statusVals[0] == 100);
 766          TEST_ASSERT(statusVals[1] == -1);
 767
 768          status = U_ZERO_ERROR;
 769          memset(statusVals, -1, sizeof(statusVals));
 770          numStatuses = bi->getRuleStatusVec(statusVals, 2, status);
 771          TEST_ASSERT_SUCCESS(status);
 772          TEST_ASSERT(numStatuses == 2);
 773          TEST_ASSERT(statusVals[0] == 100);
 774          TEST_ASSERT(statusVals[1] == 300);
 775          TEST_ASSERT(statusVals[2] == -1);
 776      }
 777      delete bi;
 778
 779 }
 780
 781 //
 782 //   Bug 2190 Regression test.   Builder crash on rule consisting of only a
 783 //                               $variable reference
 784 void RBBIAPITest::TestBug2190() {
 785      UnicodeString rulesString1 = "$aaa = abcd;\n"
 786                                   "$bbb = $aaa;\n"
 787                                   "$bbb;\n";
 788      UnicodeString testString1  = "abcdabcd";
 789                                 // 01234567890
 790      int32_t bounds1[] = {0, 4, 8};
 791      UErrorCode status=U_ZERO_ERROR;
 792      UParseError    parseError;
 793
 794      RuleBasedBreakIterator *bi = new RuleBasedBreakIterator(rulesString1, parseError, status);
 795      if(U_FAILURE(status)) {
 796          errln("FAIL : in construction");
 797      } else {
 798          bi->setText(testString1);
 799          doBoundaryTest(*bi, testString1, bounds1);
 800      }
 801      delete bi;
 802 }
 803
 804
 805 void RBBIAPITest::TestRegistration() {
 806 #if !UCONFIG_NO_SERVICE
 807     UErrorCode status = U_ZERO_ERROR;
 808     BreakIterator* thai_word = BreakIterator::createWordInstance("th_TH", status);
 809
 810     // ok to not delete these if we exit because of error?
 811     BreakIterator* thai_char = BreakIterator::createCharacterInstance("th_TH", status);
 812     BreakIterator* root_word = BreakIterator::createWordInstance("", status);
 813     BreakIterator* root_char = BreakIterator::createCharacterInstance("", status);
 814
 815     URegistryKey key = BreakIterator::registerInstance(thai_word, "xx", UBRK_WORD, status);
 816     {
 817         if (thai_word && *thai_word == *root_word) {
 818             errln("thai not different from root");
 819         }
 820     }
 821
 822     {
 823         BreakIterator* result = BreakIterator::createWordInstance("xx_XX", status);
 824         UBool fail = TRUE;
 825         if(result){
 826             fail = *result != *thai_word;
 827         }
 828         delete result;
 829         if (fail) {
 830             errln("bad result for xx_XX/word");
 831         }
 832     }
 833
 834     {
 835         BreakIterator* result = BreakIterator::createCharacterInstance("th_TH", status);
 836         UBool fail = TRUE;
 837         if(result){
 838             fail = *result != *thai_char;
 839         }
 840         delete result;
 841         if (fail) {
 842             errln("bad result for th_TH/char");
 843         }
 844     }
 845
 846     {
 847         BreakIterator* result = BreakIterator::createCharacterInstance("xx_XX", status);
 848         UBool fail = TRUE;
 849         if(result){
 850             fail = *result != *root_char;
 851         }
 852         delete result;
 853         if (fail) {
 854             errln("bad result for xx_XX/char");
 855         }
 856     }
 857
 858     {
 859         StringEnumeration* avail = BreakIterator::getAvailableLocales();
 860         UBool found = FALSE;
 861         const UnicodeString* p;
 862         while ((p = avail->snext(status))) {
 863             if (p->compare("xx") == 0) {
 864                 found = TRUE;
 865                 break;
 866             }
 867         }
 868         delete avail;
 869         if (!found) {
 870             errln("did not find test locale");
 871         }
 872     }
 873
 874     {
 875         UBool unreg = BreakIterator::unregister(key, status);
 876         if (!unreg) {
 877             errln("unable to unregister");
 878         }
 879     }
 880
 881     {
 882         BreakIterator* result = BreakIterator::createWordInstance("xx", status);
 883         BreakIterator* root = BreakIterator::createWordInstance("", status);
 884         UBool fail = TRUE;
 885         if(root){
 886           fail = *root != *result;
 887         }
 888         delete root;
 889         delete result;
 890         if (fail) {
 891             errln("did not get root break");
 892         }
 893     }
 894
 895     {
 896         StringEnumeration* avail = BreakIterator::getAvailableLocales();
 897         UBool found = FALSE;
 898         const UnicodeString* p;
 899         while ((p = avail->snext(status))) {
 900             if (p->compare("xx") == 0) {
 901                 found = TRUE;
 902                 break;
 903             }
 904         }
 905         delete avail;
 906         if (found) {
 907             errln("found test locale");
 908         }
 909     }
 910
 911     {
 912         int32_t count;
 913         UBool   foundLocale = FALSE;
 914         const Locale *avail = BreakIterator::getAvailableLocales(count);
 915         for (int i=0; i<count; i++) {
 916             if (avail[i] == Locale::getEnglish()) {
 917                 foundLocale = TRUE;
 918                 break;
 919             }
 920         }
 921         if (foundLocale == FALSE) {
 922             errln("BreakIterator::getAvailableLocales(&count), failed to find EN.");
 923         }
 924     }
 925
 926
 927     // that_word was adopted by factory
 928     delete thai_char;
 929     delete root_word;
 930     delete root_char;
 931 #endif
 932 }
 933
 934 void RBBIAPITest::RoundtripRule(const char *dataFile) {
 935     UErrorCode status = U_ZERO_ERROR;
 936     UParseError parseError;
 937     parseError.line = 0;
 938     parseError.offset = 0;
 939     UDataMemory *data = udata_open(NULL, "brk", dataFile, &status);
 940     uint32_t length;
 941     const UChar *builtSource;
 942     const uint8_t *rbbiRules;
 943     const uint8_t *builtRules;
 944
 945     if (U_FAILURE(status)) {
 946         errln("Can't open \"%s\"", dataFile);
 947         return;
 948     }
 949
 950     builtRules = (const uint8_t *)udata_getMemory(data);
 951     builtSource = (const UChar *)(builtRules + ((RBBIDataHeader*)builtRules)->fRuleSource);
 952     RuleBasedBreakIterator *brkItr = new RuleBasedBreakIterator(builtSource, parseError, status);
 953     if (U_FAILURE(status)) {
 954         errln("createRuleBasedBreakIterator: ICU Error \"%s\"  at line %d, column %d\n",
 955                 u_errorName(status), parseError.line, parseError.offset);
 956         return;
 957     };
 958     rbbiRules = brkItr->getBinaryRules(length);
 959     logln("Comparing \"%s\" len=%d", dataFile, length);
 960     if (memcmp(builtRules, rbbiRules, (int32_t)length) != 0) {
 961         errln("Built rules and rebuilt rules are different %s", dataFile);
 962         return;
 963     }
 964     delete brkItr;
 965     udata_close(data);
 966 }
 967
 968 void RBBIAPITest::TestRoundtripRules() {
 969     RoundtripRule("word");
 970     RoundtripRule("title");
 971     RoundtripRule("sent");
 972     RoundtripRule("line");
 973     RoundtripRule("char");
 974     if (!quick) {
 975         RoundtripRule("word_th");
 976         RoundtripRule("line_th");
 977     }
 978 }
 979
 980 //---------------------------------------------
 981 // runIndexedTest
 982 //---------------------------------------------
 983
 984 void RBBIAPITest::runIndexedTest( int32_t index, UBool exec, const char* &name, char* /*par*/ )
 985 {
 986     if (exec) logln((UnicodeString)"TestSuite RuleBasedBreakIterator API ");
 987     switch (index) {
 988      //   case 0: name = "TestConstruction"; if (exec) TestConstruction(); break;
 989         case  0: name = "TestCloneEquals"; if (exec) TestCloneEquals(); break;
 990         case  1: name = "TestgetRules"; if (exec) TestgetRules(); break;
 991         case  2: name = "TestHashCode"; if (exec) TestHashCode(); break;
 992         case  3: name = "TestGetSetAdoptText"; if (exec) TestGetSetAdoptText(); break;
 993         case  4: name = "TestIteration"; if (exec) TestIteration(); break;
 994         case  5: name = "extra"; break;   // Extra
 995         case  6: name = "extra"; break;   // Extra
 996         case  7: name = "TestBuilder"; if (exec) TestBuilder(); break;
 997         case  8: name = "TestQuoteGrouping"; if (exec) TestQuoteGrouping(); break;
 998         case  9: name = "TestRuleStatus"; if (exec) TestRuleStatus(); break;
 999         case 10: name = "TestRuleStatusVec"; if (exec) TestRuleStatusVec(); break;
1000         case 11: name = "TestBug2190"; if (exec) TestBug2190(); break;
1001         case 12: name = "TestRegistration"; if (exec) TestRegistration(); break;
1002         case 13: name = "TestBoilerPlate"; if (exec) TestBoilerPlate(); break;
1003         case 14: name = "TestRoundtripRules"; if (exec) TestRoundtripRules(); break;
1004
1005         default: name = ""; break; // needed to end loop
1006     }
1007 }
1008
1009 //---------------------------------------------
1010 //Internal subroutines
1011 //---------------------------------------------
1012
1013 void RBBIAPITest::doBoundaryTest(RuleBasedBreakIterator& bi, UnicodeString& text, int32_t *boundaries){
1014      logln((UnicodeString)"testIsBoundary():");
1015         int32_t p = 0;
1016         UBool isB;
1017         for (int32_t i = 0; i < text.length(); i++) {
1018             isB = bi.isBoundary(i);
1019             logln((UnicodeString)"bi.isBoundary(" + i + ") -> " + isB);
1020
1021             if (i == boundaries[p]) {
1022                 if (!isB)
1023                     errln((UnicodeString)"Wrong result from isBoundary() for " + i + (UnicodeString)": expected true, got false");
1024                 p++;
1025             }
1026             else {
1027                 if (isB)
1028                     errln((UnicodeString)"Wrong result from isBoundary() for " + i + (UnicodeString)": expected false, got true");
1029             }
1030         }
1031 }
1032 void RBBIAPITest::doTest(UnicodeString& testString, int32_t start, int32_t gotoffset, int32_t expectedOffset, const char* expectedString){
1033     UnicodeString selected;
1034     UnicodeString expected=CharsToUnicodeString(expectedString);
1035
1036     if(gotoffset != expectedOffset)
1037          errln((UnicodeString)"ERROR:****returned #" + gotoffset + (UnicodeString)" instead of #" + expectedOffset);
1038     if(start <= gotoffset){
1039         testString.extractBetween(start, gotoffset, selected);
1040     }
1041     else{
1042         testString.extractBetween(gotoffset, start, selected);
1043     }
1044     if(selected.compare(expected) != 0)
1045          errln(prettify((UnicodeString)"ERROR:****selected \"" + selected + "\" instead of \"" + expected + "\""));
1046     else
1047         logln(prettify("****selected \"" + selected + "\""));
1048 }
1049
1050 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */