icuSources/test/intltest/itercoll.cpp

   1 /********************************************************************
   2  * COPYRIGHT:
   3  * Copyright (c) 1997-2009, International Business Machines Corporation and
   4  * others. All Rights Reserved.
   5  ********************************************************************/
   6
   7 #include "unicode/utypes.h"
   8
   9 #if !UCONFIG_NO_COLLATION
  10
  11 #include "unicode/coll.h"
  12 #include "unicode/tblcoll.h"
  13 #include "unicode/unistr.h"
  14 #include "unicode/sortkey.h"
  15 #include "itercoll.h"
  16 #include "unicode/schriter.h"
  17 #include "unicode/chariter.h"
  18 #include "unicode/uchar.h"
  19 #include "cmemory.h"
  20
  21 #define ARRAY_LENGTH(array) (sizeof array / sizeof array[0])
  22
  23 static UErrorCode status = U_ZERO_ERROR;
  24
  25 CollationIteratorTest::CollationIteratorTest()
  26  : test1("What subset of all possible test cases?", ""),
  27    test2("has the highest probability of detecting", "")
  28 {
  29     en_us = (RuleBasedCollator *)Collator::createInstance(Locale::getUS(), status);
  30     if(U_FAILURE(status)) {
  31       delete en_us;
  32       en_us = 0;
  33       errcheckln(status, "Collator creation failed with %s", u_errorName(status));
  34       return;
  35     }
  36
  37 }
  38
  39 CollationIteratorTest::~CollationIteratorTest()
  40 {
  41     delete en_us;
  42 }
  43
  44 /**
  45  * Test for CollationElementIterator previous and next for the whole set of
  46  * unicode characters.
  47  */
  48 void CollationIteratorTest::TestUnicodeChar()
  49 {
  50     CollationElementIterator *iter;
  51     UChar codepoint;
  52     UnicodeString source;
  53
  54     for (codepoint = 1; codepoint < 0xFFFE;)
  55     {
  56       source.remove();
  57
  58       while (codepoint % 0xFF != 0)
  59       {
  60         if (u_isdefined(codepoint))
  61           source += codepoint;
  62         codepoint ++;
  63       }
  64
  65       if (u_isdefined(codepoint))
  66         source += codepoint;
  67
  68       if (codepoint != 0xFFFF)
  69         codepoint ++;
  70
  71       iter = en_us->createCollationElementIterator(source);
  72       /* A basic test to see if it's working at all */
  73       backAndForth(*iter);
  74       delete iter;
  75     }
  76 }
  77
  78 /**
  79  * Test for CollationElementIterator.previous()
  80  *
  81  * @bug 4108758 - Make sure it works with contracting characters
  82  *
  83  */
  84 void CollationIteratorTest::TestPrevious(/* char* par */)
  85 {
  86     UErrorCode status = U_ZERO_ERROR;
  87     CollationElementIterator *iter = en_us->createCollationElementIterator(test1);
  88
  89     // A basic test to see if it's working at all
  90     backAndForth(*iter);
  91     delete iter;
  92
  93     // Test with a contracting character sequence
  94     UnicodeString source;
  95     RuleBasedCollator *c1 = NULL;
  96     c1 = new RuleBasedCollator(
  97         (UnicodeString)"&a,A < b,B < c,C, d,D < z,Z < ch,cH,Ch,CH", status);
  98
  99     if (c1 == NULL || U_FAILURE(status))
 100     {
 101         errln("Couldn't create a RuleBasedCollator with a contracting sequence.");
 102         delete c1;
 103         return;
 104     }
 105
 106     source = "abchdcba";
 107     iter = c1->createCollationElementIterator(source);
 108     backAndForth(*iter);
 109     delete iter;
 110     delete c1;
 111
 112     // Test with an expanding character sequence
 113     RuleBasedCollator *c2 = NULL;
 114     c2 = new RuleBasedCollator((UnicodeString)"&a < b < c/abd < d", status);
 115
 116     if (c2 == NULL || U_FAILURE(status))
 117     {
 118         errln("Couldn't create a RuleBasedCollator with an expanding sequence.");
 119         delete c2;
 120         return;
 121     }
 122
 123     source = "abcd";
 124     iter = c2->createCollationElementIterator(source);
 125     backAndForth(*iter);
 126     delete iter;
 127     delete c2;
 128
 129     // Now try both
 130     RuleBasedCollator *c3 = NULL;
 131     c3 = new RuleBasedCollator((UnicodeString)"&a < b < c/aba < d < z < ch", status);
 132
 133     if (c3 == NULL || U_FAILURE(status))
 134     {
 135         errln("Couldn't create a RuleBasedCollator with both an expanding and a contracting sequence.");
 136         delete c3;
 137         return;
 138     }
 139
 140     source = "abcdbchdc";
 141     iter = c3->createCollationElementIterator(source);
 142     backAndForth(*iter);
 143     delete iter;
 144     delete c3;
 145
 146     status=U_ZERO_ERROR;
 147     source= CharsToUnicodeString("\\u0e41\\u0e02\\u0e41\\u0e02\\u0e27abc");
 148
 149     Collator *c4 = Collator::createInstance(Locale("th", "TH", ""), status);
 150     if(U_FAILURE(status)){
 151         errln("Couldn't create a collator");
 152     }
 153     iter = ((RuleBasedCollator*)c4)->createCollationElementIterator(source);
 154     backAndForth(*iter);
 155     delete iter;
 156     delete c4;
 157
 158     source= CharsToUnicodeString("\\u0061\\u30CF\\u3099\\u30FC");
 159     Collator *c5 = Collator::createInstance(Locale("ja", "JP", ""), status);
 160
 161     iter = ((RuleBasedCollator*)c5)->createCollationElementIterator(source);
 162     if(U_FAILURE(status)){
 163         errln("Couldn't create Japanese collator\n");
 164     }
 165     backAndForth(*iter);
 166     delete iter;
 167     delete c5;
 168 }
 169
 170 /**
 171  * Test for getOffset() and setOffset()
 172  */
 173 void CollationIteratorTest::TestOffset(/* char* par */)
 174 {
 175     CollationElementIterator *iter = en_us->createCollationElementIterator(test1);
 176     UErrorCode status = U_ZERO_ERROR;
 177     // testing boundaries
 178     iter->setOffset(0, status);
 179     if (U_FAILURE(status) || iter->previous(status) != UCOL_NULLORDER) {
 180         errln("Error: After setting offset to 0, we should be at the end "
 181                 "of the backwards iteration");
 182     }
 183     iter->setOffset(test1.length(), status);
 184     if (U_FAILURE(status) || iter->next(status) != UCOL_NULLORDER) {
 185         errln("Error: After setting offset to end of the string, we should "
 186                 "be at the end of the backwards iteration");
 187     }
 188
 189     // Run all the way through the iterator, then get the offset
 190     int32_t orderLength = 0;
 191     Order *orders = getOrders(*iter, orderLength);
 192
 193     int32_t offset = iter->getOffset();
 194
 195     if (offset != test1.length())
 196     {
 197         UnicodeString msg1("offset at end != length: ");
 198         UnicodeString msg2(" vs ");
 199
 200         errln(msg1 + offset + msg2 + test1.length());
 201     }
 202
 203     // Now set the offset back to the beginning and see if it works
 204     CollationElementIterator *pristine = en_us->createCollationElementIterator(test1);
 205
 206     iter->setOffset(0, status);
 207
 208     if (U_FAILURE(status))
 209     {
 210         errln("setOffset failed.");
 211     }
 212     else
 213     {
 214         assertEqual(*iter, *pristine);
 215     }
 216
 217     // TODO: try iterating halfway through a messy string.
 218
 219     delete pristine;
 220     delete[] orders;
 221     delete iter;
 222 }
 223
 224 /**
 225  * Test for setText()
 226  */
 227 void CollationIteratorTest::TestSetText(/* char* par */)
 228 {
 229     CollationElementIterator *iter1 = en_us->createCollationElementIterator(test1);
 230     CollationElementIterator *iter2 = en_us->createCollationElementIterator(test2);
 231     UErrorCode status = U_ZERO_ERROR;
 232
 233     // Run through the second iterator just to exercise it
 234     int32_t c = iter2->next(status);
 235     int32_t i = 0;
 236
 237     while ( ++i < 10 && c != CollationElementIterator::NULLORDER)
 238     {
 239         if (U_FAILURE(status))
 240         {
 241             errln("iter2->next() returned an error.");
 242             delete iter2;
 243             delete iter1;
 244         }
 245
 246         c = iter2->next(status);
 247     }
 248
 249     // Now set it to point to the same string as the first iterator
 250     iter2->setText(test1, status);
 251
 252     if (U_FAILURE(status))
 253     {
 254         errln("call to iter2->setText(test1) failed.");
 255     }
 256     else
 257     {
 258         assertEqual(*iter1, *iter2);
 259     }
 260     iter1->reset();
 261     //now use the overloaded setText(ChracterIterator&, UErrorCode) function to set the text
 262     CharacterIterator* chariter = new StringCharacterIterator(test1);
 263     iter2->setText(*chariter, status);
 264     if (U_FAILURE(status))
 265     {
 266         errln("call to iter2->setText(chariter(test1)) failed.");
 267     }
 268     else
 269     {
 270         assertEqual(*iter1, *iter2);
 271     }
 272
 273     // test for an empty string
 274     UnicodeString empty("");
 275     iter1->setText(empty, status);
 276     if (U_FAILURE(status)
 277         || iter1->next(status) != (int32_t)UCOL_NULLORDER) {
 278         errln("Empty string should have no CEs.");
 279     }
 280     ((StringCharacterIterator *)chariter)->setText(empty);
 281     iter1->setText(*chariter, status);
 282     if (U_FAILURE(status)
 283         || iter1->next(status) != (int32_t)UCOL_NULLORDER) {
 284         errln("Empty string should have no CEs.");
 285     }
 286     delete chariter;
 287     delete iter2;
 288     delete iter1;
 289 }
 290
 291 /** @bug 4108762
 292  * Test for getMaxExpansion()
 293  */
 294 void CollationIteratorTest::TestMaxExpansion(/* char* par */)
 295 {
 296     UErrorCode          status = U_ZERO_ERROR;
 297     UnicodeString rule("&a < ab < c/aba < d < z < ch");
 298     RuleBasedCollator  *coll   = new RuleBasedCollator(rule, status);
 299     UChar               ch     = 0;
 300     UnicodeString       str(ch);
 301
 302     CollationElementIterator *iter   = coll->createCollationElementIterator(str);
 303
 304     while (ch < 0xFFFF && U_SUCCESS(status)) {
 305         int      count = 1;
 306         uint32_t order;
 307         ch ++;
 308         UnicodeString str(ch);
 309         iter->setText(str, status);
 310         order = iter->previous(status);
 311
 312         /* thai management */
 313         if (CollationElementIterator::isIgnorable(order))
 314             order = iter->previous(status);
 315
 316         while (U_SUCCESS(status)
 317             && iter->previous(status) != (int32_t)UCOL_NULLORDER)
 318         {
 319             count ++;
 320         }
 321
 322         if (U_FAILURE(status) && iter->getMaxExpansion(order) < count) {
 323             errln("Failure at codepoint %d, maximum expansion count < %d\n",
 324                 ch, count);
 325         }
 326     }
 327
 328     delete iter;
 329     delete coll;
 330 }
 331
 332 /*
 333  * @bug 4157299
 334  */
 335 void CollationIteratorTest::TestClearBuffers(/* char* par */)
 336 {
 337     UErrorCode status = U_ZERO_ERROR;
 338     RuleBasedCollator *c = new RuleBasedCollator((UnicodeString)"&a < b < c & ab = d", status);
 339
 340     if (c == NULL || U_FAILURE(status))
 341     {
 342         errln("Couldn't create a RuleBasedCollator.");
 343         delete c;
 344         return;
 345     }
 346
 347     UnicodeString source("abcd");
 348     CollationElementIterator *i = c->createCollationElementIterator(source);
 349     int32_t e0 = i->next(status);    // save the first collation element
 350
 351     if (U_FAILURE(status))
 352     {
 353         errln("call to i->next() failed. err=%s", u_errorName(status));
 354     }
 355     else
 356     {
 357         i->setOffset(3, status);        // go to the expanding character
 358
 359         if (U_FAILURE(status))
 360         {
 361             errln("call to i->setOffset(3) failed. err=%s", u_errorName(status));
 362         }
 363         else
 364         {
 365             i->next(status);                // but only use up half of it
 366
 367             if (U_FAILURE(status))
 368             {
 369                 errln("call to i->next() failed. err=%s", u_errorName(status));
 370             }
 371             else
 372             {
 373                 i->setOffset(0, status);        // go back to the beginning
 374
 375                 if (U_FAILURE(status))
 376                 {
 377                     errln("call to i->setOffset(0) failed. err=%s", u_errorName(status));
 378                 }
 379                 else
 380                 {
 381                     int32_t e = i->next(status);    // and get this one again
 382
 383                     if (U_FAILURE(status))
 384                     {
 385                         errln("call to i->next() failed. err=%s", u_errorName(status));
 386                     }
 387                     else if (e != e0)
 388                     {
 389                         errln("got 0x%X, expected 0x%X", e, e0);
 390                     }
 391                 }
 392             }
 393         }
 394     }
 395
 396     delete i;
 397     delete c;
 398 }
 399
 400 /**
 401  * Testing the assignment operator
 402  */
 403 void CollationIteratorTest::TestAssignment()
 404 {
 405     UErrorCode status = U_ZERO_ERROR;
 406     RuleBasedCollator *coll =
 407         (RuleBasedCollator *)Collator::createInstance(status);
 408
 409     if (coll == NULL || U_FAILURE(status))
 410     {
 411         errln("Couldn't create a default collator.");
 412         return;
 413     }
 414
 415     UnicodeString source("abcd");
 416     CollationElementIterator *iter1 =
 417         coll->createCollationElementIterator(source);
 418
 419     CollationElementIterator iter2 = *iter1;
 420
 421     if (*iter1 != iter2) {
 422         errln("Fail collation iterator assignment does not produce the same elements");
 423     }
 424
 425     CollationElementIterator iter3(*iter1);
 426
 427     if (*iter1 != iter3) {
 428         errln("Fail collation iterator copy constructor does not produce the same elements");
 429     }
 430
 431     source = CharsToUnicodeString("a\\u0300\\u0325");
 432     coll->setAttribute(UCOL_NORMALIZATION_MODE, UCOL_ON, status);
 433     CollationElementIterator *iter4
 434                         = coll->createCollationElementIterator(source);
 435     CollationElementIterator iter5(*iter4);
 436     if (*iter4 != iter5) {
 437         errln("collation iterator assignment does not produce the same elements");
 438     }
 439     iter4->next(status);
 440     if (U_FAILURE(status) || *iter4 == iter5) {
 441         errln("collation iterator not equal");
 442     }
 443     iter5.next(status);
 444     if (U_FAILURE(status) || *iter4 != iter5) {
 445         errln("collation iterator equal");
 446     }
 447     iter4->next(status);
 448     if (U_FAILURE(status) || *iter4 == iter5) {
 449         errln("collation iterator not equal");
 450     }
 451     iter5.next(status);
 452     if (U_FAILURE(status) || *iter4 != iter5) {
 453         errln("collation iterator equal");
 454     }
 455     CollationElementIterator iter6(*iter4);
 456     if (*iter4 != iter6) {
 457         errln("collation iterator equal");
 458     }
 459     iter4->next(status);
 460     if (U_FAILURE(status) || *iter4 == iter5) {
 461         errln("collation iterator not equal");
 462     }
 463     iter5.next(status);
 464     if (U_FAILURE(status) || *iter4 != iter5) {
 465         errln("collation iterator equal");
 466     }
 467     iter4->next(status);
 468     if (U_FAILURE(status) || *iter4 == iter5) {
 469         errln("collation iterator not equal");
 470     }
 471     iter5.next(status);
 472     if (U_FAILURE(status) || *iter4 != iter5) {
 473         errln("collation iterator equal");
 474     }
 475     delete iter1;
 476     delete iter4;
 477     delete coll;
 478 }
 479
 480 /**
 481  * Testing the constructors
 482  */
 483 void CollationIteratorTest::TestConstructors()
 484 {
 485     UErrorCode status = U_ZERO_ERROR;
 486     RuleBasedCollator *coll =
 487         (RuleBasedCollator *)Collator::createInstance(status);
 488     if (coll == NULL || U_FAILURE(status))
 489     {
 490         errln("Couldn't create a default collator.");
 491         return;
 492     }
 493
 494     // testing protected constructor with character iterator as argument
 495     StringCharacterIterator chariter(test1);
 496     CollationElementIterator *iter1 =
 497         coll->createCollationElementIterator(chariter);
 498     if (U_FAILURE(status)) {
 499         errln("Couldn't create collation element iterator with character iterator.");
 500         return;
 501     }
 502     CollationElementIterator *iter2 =
 503         coll->createCollationElementIterator(test1);
 504
 505     // initially the 2 collation element iterators should be the same
 506     if (*iter1 != *iter1 || *iter2 != *iter2 || *iter1 != *iter2
 507         || *iter2 != *iter1) {
 508         errln("CollationElementIterators constructed with the same string data should be the same at the start");
 509     }
 510     assertEqual(*iter1, *iter2);
 511
 512     delete iter1;
 513     delete iter2;
 514
 515     // tests empty strings
 516     UnicodeString empty("");
 517     iter1 = coll->createCollationElementIterator(empty);
 518     chariter.setText(empty);
 519     iter2 = coll->createCollationElementIterator(chariter);
 520     if (*iter1 != *iter1 || *iter2 != *iter2 || *iter1 != *iter2
 521         || *iter2 != *iter1) {
 522         errln("CollationElementIterators constructed with the same string data should be the same at the start");
 523     }
 524     if (iter1->next(status) != (int32_t)UCOL_NULLORDER) {
 525         errln("Empty string should have no CEs.");
 526     }
 527     if (iter2->next(status) != (int32_t)UCOL_NULLORDER) {
 528         errln("Empty string should have no CEs.");
 529     }
 530     delete iter1;
 531     delete iter2;
 532     delete coll;
 533 }
 534
 535 /**
 536  * Testing the strength order
 537  */
 538 void CollationIteratorTest::TestStrengthOrder()
 539 {
 540     int order = 0x0123ABCD;
 541
 542     UErrorCode status = U_ZERO_ERROR;
 543     RuleBasedCollator *coll =
 544         (RuleBasedCollator *)Collator::createInstance(status);
 545     if (coll == NULL || U_FAILURE(status))
 546     {
 547         errln("Couldn't create a default collator.");
 548         return;
 549     }
 550
 551     coll->setStrength(Collator::PRIMARY);
 552     CollationElementIterator *iter =
 553         coll->createCollationElementIterator(test1);
 554
 555     if (iter == NULL) {
 556         errln("Couldn't create a collation element iterator from default collator");
 557         return;
 558     }
 559
 560     if (iter->strengthOrder(order) != 0x01230000) {
 561         errln("Strength order for a primary strength collator should be the first 2 bytes");
 562         return;
 563     }
 564
 565     coll->setStrength(Collator::SECONDARY);
 566     if (iter->strengthOrder(order) != 0x0123AB00) {
 567         errln("Strength order for a secondary strength collator should be the third byte");
 568         return;
 569     }
 570
 571     coll->setStrength(Collator::TERTIARY);
 572     if (iter->strengthOrder(order) != order) {
 573         errln("Strength order for a tertiary strength collator should be the third byte");
 574         return;
 575     }
 576     delete iter;
 577     delete coll;
 578 }
 579
 580 /**
 581  * Return a string containing all of the collation orders
 582  * returned by calls to next on the specified iterator
 583  */
 584 UnicodeString &CollationIteratorTest::orderString(CollationElementIterator &iter, UnicodeString &target)
 585 {
 586     int32_t order;
 587     UErrorCode status = U_ZERO_ERROR;
 588
 589     while ((order = iter.next(status)) != CollationElementIterator::NULLORDER)
 590     {
 591         target += "0x";
 592         appendHex(order, 8, target);
 593         target += " ";
 594     }
 595
 596     return target;
 597 }
 598
 599 void CollationIteratorTest::assertEqual(CollationElementIterator &i1, CollationElementIterator &i2)
 600 {
 601     int32_t c1, c2, count = 0;
 602     UErrorCode status = U_ZERO_ERROR;
 603
 604     do
 605     {
 606         c1 = i1.next(status);
 607         c2 = i2.next(status);
 608
 609         if (c1 != c2)
 610         {
 611             errln("    %d: strength(0x%X) != strength(0x%X)", count, c1, c2);
 612             break;
 613         }
 614
 615         count += 1;
 616     }
 617     while (c1 != CollationElementIterator::NULLORDER);
 618 }
 619
 620 void CollationIteratorTest::runIndexedTest(int32_t index, UBool exec, const char* &name, char* /*par*/)
 621 {
 622     if (exec)
 623     {
 624         logln("Collation Iteration Tests: ");
 625     }
 626
 627     if(en_us) {
 628       switch (index)
 629       {
 630           case  0: name = "TestPrevious";      if (exec) TestPrevious(/* par */);     break;
 631           case  1: name = "TestOffset";        if (exec) TestOffset(/* par */);       break;
 632           case  2: name = "TestSetText";       if (exec) TestSetText(/* par */);      break;
 633           case  3: name = "TestMaxExpansion";  if (exec) TestMaxExpansion(/* par */); break;
 634           case  4: name = "TestClearBuffers";  if (exec) TestClearBuffers(/* par */); break;
 635           case  5: name = "TestUnicodeChar";   if (exec) TestUnicodeChar(/* par */);  break;
 636           case  6: name = "TestAssignment";    if (exec) TestAssignment(/* par */);    break;
 637           case  7: name = "TestConstructors";  if (exec) TestConstructors(/* par */); break;
 638           case  8: name = "TestStrengthOrder"; if (exec) TestStrengthOrder(/* par */); break;
 639           default: name = ""; break;
 640       }
 641     } else {
 642       dataerrln("Class iterator not instantiated");
 643       name = "";
 644     }
 645 }
 646
 647 #endif /* #if !UCONFIG_NO_COLLATION */