icuSources/test/perf/collperf2/collperf2.cpp

   1 /*
   2 ***********************************************************************
   3 * © 2016 and later: Unicode, Inc. and others.
   4 * License & terms of use: http://www.unicode.org/copyright.html#License
   5 ***********************************************************************
   6 ***********************************************************************
   7 * Copyright (c) 2013-2014, International Business Machines
   8 * Corporation and others.  All Rights Reserved.
   9 ***********************************************************************
  10 */
  11
  12 #include <string.h>
  13 #include "unicode/localpointer.h"
  14 #include "unicode/uperf.h"
  15 #include "unicode/ucol.h"
  16 #include "unicode/coll.h"
  17 #include "unicode/uiter.h"
  18 #include "unicode/ustring.h"
  19 #include "unicode/sortkey.h"
  20 #include "uarrsort.h"
  21 #include "uoptions.h"
  22 #include "ustr_imp.h"
  23
  24 #define COMPACT_ARRAY(CompactArrays, UNIT) \
  25 struct CompactArrays{\
  26     CompactArrays(const CompactArrays & );\
  27     CompactArrays & operator=(const CompactArrays & );\
  28     int32_t   count;/*total number of the strings*/ \
  29     int32_t * index;/*relative offset in data*/ \
  30     UNIT    * data; /*the real space to hold strings*/ \
  31     \
  32     ~CompactArrays(){free(index);free(data);} \
  33     CompactArrays() : count(0), index(NULL), data(NULL) { \
  34         index = (int32_t *) realloc(index, sizeof(int32_t)); \
  35         index[0] = 0; \
  36     } \
  37     void append_one(int32_t theLen){ /*include terminal NULL*/ \
  38         count++; \
  39         index = (int32_t *) realloc(index, sizeof(int32_t) * (count + 1)); \
  40         index[count] = index[count - 1] + theLen; \
  41         data = (UNIT *) realloc(data, sizeof(UNIT) * index[count]); \
  42     } \
  43     UNIT * last(){return data + index[count - 1];} \
  44     const UNIT * dataOf(int32_t i) const {return data + index[i];} \
  45     int32_t lengthOf(int i) const {return index[i+1] - index[i] - 1; } /*exclude terminating NULL*/  \
  46 };
  47
  48 COMPACT_ARRAY(CA_uchar, UChar)
  49 COMPACT_ARRAY(CA_char, char)
  50
  51 #define MAX_TEST_STRINGS_FOR_PERMUTING 1000
  52
  53 // C API test cases
  54
  55 //
  56 // Test case taking a single test data array, calling ucol_strcoll by permuting the test data
  57 //
  58 class Strcoll : public UPerfFunction
  59 {
  60 public:
  61     Strcoll(const UCollator* coll, const CA_uchar* source, UBool useLen);
  62     ~Strcoll();
  63     virtual void call(UErrorCode* status);
  64     virtual long getOperationsPerIteration();
  65
  66 private:
  67     const UCollator *coll;
  68     const CA_uchar *source;
  69     UBool useLen;
  70     int32_t maxTestStrings;
  71 };
  72
  73 Strcoll::Strcoll(const UCollator* coll, const CA_uchar* source, UBool useLen)
  74     :   coll(coll),
  75         source(source),
  76         useLen(useLen)
  77 {
  78     maxTestStrings = source->count > MAX_TEST_STRINGS_FOR_PERMUTING ? MAX_TEST_STRINGS_FOR_PERMUTING : source->count;
  79 }
  80
  81 Strcoll::~Strcoll()
  82 {
  83 }
  84
  85 void Strcoll::call(UErrorCode* status)
  86 {
  87     if (U_FAILURE(*status)) return;
  88
  89     // call strcoll for permutation
  90     int32_t divisor = source->count / maxTestStrings;
  91     int32_t srcLen, tgtLen;
  92     int32_t cmp = 0;
  93     for (int32_t i = 0, numTestStringsI = 0; i < source->count && numTestStringsI < maxTestStrings; i++) {
  94         if (i % divisor) continue;
  95         numTestStringsI++;
  96         srcLen = useLen ? source->lengthOf(i) : -1;
  97         for (int32_t j = 0, numTestStringsJ = 0; j < source->count && numTestStringsJ < maxTestStrings; j++) {
  98             if (j % divisor) continue;
  99             numTestStringsJ++;
 100             tgtLen = useLen ? source->lengthOf(j) : -1;
 101             cmp += ucol_strcoll(coll, source->dataOf(i), srcLen, source->dataOf(j), tgtLen);
 102         }
 103     }
 104     // At the end, cmp must be 0
 105     if (cmp != 0) {
 106         *status = U_INTERNAL_PROGRAM_ERROR;
 107     }
 108 }
 109
 110 long Strcoll::getOperationsPerIteration()
 111 {
 112     return maxTestStrings * maxTestStrings;
 113 }
 114
 115 //
 116 // Test case taking two test data arrays, calling ucol_strcoll for strings at a same index
 117 //
 118 class Strcoll_2 : public UPerfFunction
 119 {
 120 public:
 121     Strcoll_2(const UCollator* coll, const CA_uchar* source, const CA_uchar* target, UBool useLen);
 122     ~Strcoll_2();
 123     virtual void call(UErrorCode* status);
 124     virtual long getOperationsPerIteration();
 125
 126 private:
 127     const UCollator *coll;
 128     const CA_uchar *source;
 129     const CA_uchar *target;
 130     UBool useLen;
 131 };
 132
 133 Strcoll_2::Strcoll_2(const UCollator* coll, const CA_uchar* source, const CA_uchar* target, UBool useLen)
 134     :   coll(coll),
 135         source(source),
 136         target(target),
 137         useLen(useLen)
 138 {
 139 }
 140
 141 Strcoll_2::~Strcoll_2()
 142 {
 143 }
 144
 145 void Strcoll_2::call(UErrorCode* status)
 146 {
 147     if (U_FAILURE(*status)) return;
 148
 149     // call strcoll for two strings at the same index
 150     if (source->count < target->count) {
 151         *status = U_ILLEGAL_ARGUMENT_ERROR;
 152     } else {
 153         for (int32_t i = 0; i < source->count; i++) {
 154             int32_t srcLen = useLen ? source->lengthOf(i) : -1;
 155             int32_t tgtLen = useLen ? target->lengthOf(i) : -1;
 156             ucol_strcoll(coll, source->dataOf(i), srcLen, target->dataOf(i), tgtLen);
 157         }
 158     }
 159 }
 160
 161 long Strcoll_2::getOperationsPerIteration()
 162 {
 163     return source->count;
 164 }
 165
 166
 167 //
 168 // Test case taking a single test data array, calling ucol_strcollUTF8 by permuting the test data
 169 //
 170 class StrcollUTF8 : public UPerfFunction
 171 {
 172 public:
 173     StrcollUTF8(const UCollator* coll, const CA_char* source, UBool useLen);
 174     ~StrcollUTF8();
 175     virtual void call(UErrorCode* status);
 176     virtual long getOperationsPerIteration();
 177
 178 private:
 179     const UCollator *coll;
 180     const CA_char *source;
 181     UBool useLen;
 182     int32_t maxTestStrings;
 183 };
 184
 185 StrcollUTF8::StrcollUTF8(const UCollator* coll, const CA_char* source, UBool useLen)
 186     :   coll(coll),
 187         source(source),
 188         useLen(useLen)
 189 {
 190     maxTestStrings = source->count > MAX_TEST_STRINGS_FOR_PERMUTING ? MAX_TEST_STRINGS_FOR_PERMUTING : source->count;
 191 }
 192
 193 StrcollUTF8::~StrcollUTF8()
 194 {
 195 }
 196
 197 void StrcollUTF8::call(UErrorCode* status)
 198 {
 199     if (U_FAILURE(*status)) return;
 200
 201     // call strcollUTF8 for permutation
 202     int32_t divisor = source->count / maxTestStrings;
 203     int32_t srcLen, tgtLen;
 204     int32_t cmp = 0;
 205     for (int32_t i = 0, numTestStringsI = 0; U_SUCCESS(*status) && i < source->count && numTestStringsI < maxTestStrings; i++) {
 206         if (i % divisor) continue;
 207         numTestStringsI++;
 208         srcLen = useLen ? source->lengthOf(i) : -1;
 209         for (int32_t j = 0, numTestStringsJ = 0; U_SUCCESS(*status) && j < source->count && numTestStringsJ < maxTestStrings; j++) {
 210             if (j % divisor) continue;
 211             numTestStringsJ++;
 212             tgtLen = useLen ? source->lengthOf(j) : -1;
 213             cmp += ucol_strcollUTF8(coll, source->dataOf(i), srcLen, source->dataOf(j), tgtLen, status);
 214         }
 215     }
 216     // At the end, cmp must be 0
 217     if (cmp != 0) {
 218         *status = U_INTERNAL_PROGRAM_ERROR;
 219     }
 220 }
 221
 222 long StrcollUTF8::getOperationsPerIteration()
 223 {
 224     return maxTestStrings * maxTestStrings;
 225 }
 226
 227 //
 228 // Test case taking two test data arrays, calling ucol_strcoll for strings at a same index
 229 //
 230 class StrcollUTF8_2 : public UPerfFunction
 231 {
 232 public:
 233     StrcollUTF8_2(const UCollator* coll, const CA_char* source, const CA_char* target, UBool useLen);
 234     ~StrcollUTF8_2();
 235     virtual void call(UErrorCode* status);
 236     virtual long getOperationsPerIteration();
 237
 238 private:
 239     const UCollator *coll;
 240     const CA_char *source;
 241     const CA_char *target;
 242     UBool useLen;
 243 };
 244
 245 StrcollUTF8_2::StrcollUTF8_2(const UCollator* coll, const CA_char* source, const CA_char* target, UBool useLen)
 246     :   coll(coll),
 247         source(source),
 248         target(target),
 249         useLen(useLen)
 250 {
 251 }
 252
 253 StrcollUTF8_2::~StrcollUTF8_2()
 254 {
 255 }
 256
 257 void StrcollUTF8_2::call(UErrorCode* status)
 258 {
 259     if (U_FAILURE(*status)) return;
 260
 261     // call strcoll for two strings at the same index
 262     if (source->count < target->count) {
 263         *status = U_ILLEGAL_ARGUMENT_ERROR;
 264     } else {
 265         for (int32_t i = 0; U_SUCCESS(*status) && i < source->count; i++) {
 266             int32_t srcLen = useLen ? source->lengthOf(i) : -1;
 267             int32_t tgtLen = useLen ? target->lengthOf(i) : -1;
 268             ucol_strcollUTF8(coll, source->dataOf(i), srcLen, target->dataOf(i), tgtLen, status);
 269         }
 270     }
 271 }
 272
 273 long StrcollUTF8_2::getOperationsPerIteration()
 274 {
 275     return source->count;
 276 }
 277
 278 //
 279 // Test case taking a single test data array, calling ucol_getSortKey for each
 280 //
 281 class GetSortKey : public UPerfFunction
 282 {
 283 public:
 284     GetSortKey(const UCollator* coll, const CA_uchar* source, UBool useLen);
 285     ~GetSortKey();
 286     virtual void call(UErrorCode* status);
 287     virtual long getOperationsPerIteration();
 288
 289 private:
 290     const UCollator *coll;
 291     const CA_uchar *source;
 292     UBool useLen;
 293 };
 294
 295 GetSortKey::GetSortKey(const UCollator* coll, const CA_uchar* source, UBool useLen)
 296     :   coll(coll),
 297         source(source),
 298         useLen(useLen)
 299 {
 300 }
 301
 302 GetSortKey::~GetSortKey()
 303 {
 304 }
 305
 306 #define KEY_BUF_SIZE 512
 307
 308 void GetSortKey::call(UErrorCode* status)
 309 {
 310     if (U_FAILURE(*status)) return;
 311
 312     uint8_t key[KEY_BUF_SIZE];
 313     int32_t len;
 314
 315     if (useLen) {
 316         for (int32_t i = 0; i < source->count; i++) {
 317             len = ucol_getSortKey(coll, source->dataOf(i), source->lengthOf(i), key, KEY_BUF_SIZE);
 318         }
 319     } else {
 320         for (int32_t i = 0; i < source->count; i++) {
 321             len = ucol_getSortKey(coll, source->dataOf(i), -1, key, KEY_BUF_SIZE);
 322         }
 323     }
 324 }
 325
 326 long GetSortKey::getOperationsPerIteration()
 327 {
 328     return source->count;
 329 }
 330
 331 //
 332 // Test case taking a single test data array in UTF-16, calling ucol_nextSortKeyPart for each for the
 333 // given buffer size
 334 //
 335 class NextSortKeyPart : public UPerfFunction
 336 {
 337 public:
 338     NextSortKeyPart(const UCollator* coll, const CA_uchar* source, int32_t bufSize, int32_t maxIteration = -1);
 339     ~NextSortKeyPart();
 340     virtual void call(UErrorCode* status);
 341     virtual long getOperationsPerIteration();
 342     virtual long getEventsPerIteration();
 343
 344 private:
 345     const UCollator *coll;
 346     const CA_uchar *source;
 347     int32_t bufSize;
 348     int32_t maxIteration;
 349     long events;
 350 };
 351
 352 // Note: maxIteration = -1 -> repeat until the end of collation key
 353 NextSortKeyPart::NextSortKeyPart(const UCollator* coll, const CA_uchar* source, int32_t bufSize, int32_t maxIteration /* = -1 */)
 354     :   coll(coll),
 355         source(source),
 356         bufSize(bufSize),
 357         maxIteration(maxIteration),
 358         events(0)
 359 {
 360 }
 361
 362 NextSortKeyPart::~NextSortKeyPart()
 363 {
 364 }
 365
 366 void NextSortKeyPart::call(UErrorCode* status)
 367 {
 368     if (U_FAILURE(*status)) return;
 369
 370     uint8_t *part = (uint8_t *)malloc(bufSize);
 371     uint32_t state[2];
 372     UCharIterator iter;
 373
 374     events = 0;
 375     for (int i = 0; i < source->count && U_SUCCESS(*status); i++) {
 376         uiter_setString(&iter, source->dataOf(i), source->lengthOf(i));
 377         state[0] = 0;
 378         state[1] = 0;
 379         int32_t partLen = bufSize;
 380         for (int32_t n = 0; U_SUCCESS(*status) && partLen == bufSize && (maxIteration < 0 || n < maxIteration); n++) {
 381             partLen = ucol_nextSortKeyPart(coll, &iter, state, part, bufSize, status);
 382             events++;
 383         }
 384     }
 385     free(part);
 386 }
 387
 388 long NextSortKeyPart::getOperationsPerIteration()
 389 {
 390     return source->count;
 391 }
 392
 393 long NextSortKeyPart::getEventsPerIteration()
 394 {
 395     return events;
 396 }
 397
 398 //
 399 // Test case taking a single test data array in UTF-8, calling ucol_nextSortKeyPart for each for the
 400 // given buffer size
 401 //
 402 class NextSortKeyPartUTF8 : public UPerfFunction
 403 {
 404 public:
 405     NextSortKeyPartUTF8(const UCollator* coll, const CA_char* source, int32_t bufSize, int32_t maxIteration = -1);
 406     ~NextSortKeyPartUTF8();
 407     virtual void call(UErrorCode* status);
 408     virtual long getOperationsPerIteration();
 409     virtual long getEventsPerIteration();
 410
 411 private:
 412     const UCollator *coll;
 413     const CA_char *source;
 414     int32_t bufSize;
 415     int32_t maxIteration;
 416     long events;
 417 };
 418
 419 // Note: maxIteration = -1 -> repeat until the end of collation key
 420 NextSortKeyPartUTF8::NextSortKeyPartUTF8(const UCollator* coll, const CA_char* source, int32_t bufSize, int32_t maxIteration /* = -1 */)
 421     :   coll(coll),
 422         source(source),
 423         bufSize(bufSize),
 424         maxIteration(maxIteration),
 425         events(0)
 426 {
 427 }
 428
 429 NextSortKeyPartUTF8::~NextSortKeyPartUTF8()
 430 {
 431 }
 432
 433 void NextSortKeyPartUTF8::call(UErrorCode* status)
 434 {
 435     if (U_FAILURE(*status)) return;
 436
 437     uint8_t *part = (uint8_t *)malloc(bufSize);
 438     uint32_t state[2];
 439     UCharIterator iter;
 440
 441     events = 0;
 442     for (int i = 0; i < source->count && U_SUCCESS(*status); i++) {
 443         uiter_setUTF8(&iter, source->dataOf(i), source->lengthOf(i));
 444         state[0] = 0;
 445         state[1] = 0;
 446         int32_t partLen = bufSize;
 447         for (int32_t n = 0; U_SUCCESS(*status) && partLen == bufSize && (maxIteration < 0 || n < maxIteration); n++) {
 448             partLen = ucol_nextSortKeyPart(coll, &iter, state, part, bufSize, status);
 449             events++;
 450         }
 451     }
 452     free(part);
 453 }
 454
 455 long NextSortKeyPartUTF8::getOperationsPerIteration()
 456 {
 457     return source->count;
 458 }
 459
 460 long NextSortKeyPartUTF8::getEventsPerIteration()
 461 {
 462     return events;
 463 }
 464
 465 // CPP API test cases
 466
 467 //
 468 // Test case taking a single test data array, calling Collator::compare by permuting the test data
 469 //
 470 class CppCompare : public UPerfFunction
 471 {
 472 public:
 473     CppCompare(const Collator* coll, const CA_uchar* source, UBool useLen);
 474     ~CppCompare();
 475     virtual void call(UErrorCode* status);
 476     virtual long getOperationsPerIteration();
 477
 478 private:
 479     const Collator *coll;
 480     const CA_uchar *source;
 481     UBool useLen;
 482     int32_t maxTestStrings;
 483 };
 484
 485 CppCompare::CppCompare(const Collator* coll, const CA_uchar* source, UBool useLen)
 486     :   coll(coll),
 487         source(source),
 488         useLen(useLen)
 489 {
 490     maxTestStrings = source->count > MAX_TEST_STRINGS_FOR_PERMUTING ? MAX_TEST_STRINGS_FOR_PERMUTING : source->count;
 491 }
 492
 493 CppCompare::~CppCompare()
 494 {
 495 }
 496
 497 void CppCompare::call(UErrorCode* status) {
 498     if (U_FAILURE(*status)) return;
 499
 500     // call compare for permutation of test data
 501     int32_t divisor = source->count / maxTestStrings;
 502     int32_t srcLen, tgtLen;
 503     int32_t cmp = 0;
 504     for (int32_t i = 0, numTestStringsI = 0; i < source->count && numTestStringsI < maxTestStrings; i++) {
 505         if (i % divisor) continue;
 506         numTestStringsI++;
 507         srcLen = useLen ? source->lengthOf(i) : -1;
 508         for (int32_t j = 0, numTestStringsJ = 0; j < source->count && numTestStringsJ < maxTestStrings; j++) {
 509             if (j % divisor) continue;
 510             numTestStringsJ++;
 511             tgtLen = useLen ? source->lengthOf(j) : -1;
 512             cmp += coll->compare(source->dataOf(i), srcLen, source->dataOf(j), tgtLen);
 513         }
 514     }
 515     // At the end, cmp must be 0
 516     if (cmp != 0) {
 517         *status = U_INTERNAL_PROGRAM_ERROR;
 518     }
 519 }
 520
 521 long CppCompare::getOperationsPerIteration()
 522 {
 523     return maxTestStrings * maxTestStrings;
 524 }
 525
 526 //
 527 // Test case taking two test data arrays, calling Collator::compare for strings at a same index
 528 //
 529 class CppCompare_2 : public UPerfFunction
 530 {
 531 public:
 532     CppCompare_2(const Collator* coll, const CA_uchar* source, const CA_uchar* target, UBool useLen);
 533     ~CppCompare_2();
 534     virtual void call(UErrorCode* status);
 535     virtual long getOperationsPerIteration();
 536
 537 private:
 538     const Collator *coll;
 539     const CA_uchar *source;
 540     const CA_uchar *target;
 541     UBool useLen;
 542 };
 543
 544 CppCompare_2::CppCompare_2(const Collator* coll, const CA_uchar* source, const CA_uchar* target, UBool useLen)
 545     :   coll(coll),
 546         source(source),
 547         target(target),
 548         useLen(useLen)
 549 {
 550 }
 551
 552 CppCompare_2::~CppCompare_2()
 553 {
 554 }
 555
 556 void CppCompare_2::call(UErrorCode* status) {
 557     if (U_FAILURE(*status)) return;
 558
 559     // call strcoll for two strings at the same index
 560     if (source->count < target->count) {
 561         *status = U_ILLEGAL_ARGUMENT_ERROR;
 562     } else {
 563         for (int32_t i = 0; i < source->count; i++) {
 564             int32_t srcLen = useLen ? source->lengthOf(i) : -1;
 565             int32_t tgtLen = useLen ? target->lengthOf(i) : -1;
 566             coll->compare(source->dataOf(i), srcLen, target->dataOf(i), tgtLen);
 567         }
 568     }
 569 }
 570
 571 long CppCompare_2::getOperationsPerIteration()
 572 {
 573     return source->count;
 574 }
 575
 576
 577 //
 578 // Test case taking a single test data array, calling Collator::compareUTF8 by permuting the test data
 579 //
 580 class CppCompareUTF8 : public UPerfFunction
 581 {
 582 public:
 583     CppCompareUTF8(const Collator* coll, const CA_char* source, UBool useLen);
 584     ~CppCompareUTF8();
 585     virtual void call(UErrorCode* status);
 586     virtual long getOperationsPerIteration();
 587
 588 private:
 589     const Collator *coll;
 590     const CA_char *source;
 591     UBool useLen;
 592     int32_t maxTestStrings;
 593 };
 594
 595 CppCompareUTF8::CppCompareUTF8(const Collator* coll, const CA_char* source, UBool useLen)
 596     :   coll(coll),
 597         source(source),
 598         useLen(useLen)
 599 {
 600     maxTestStrings = source->count > MAX_TEST_STRINGS_FOR_PERMUTING ? MAX_TEST_STRINGS_FOR_PERMUTING : source->count;
 601 }
 602
 603 CppCompareUTF8::~CppCompareUTF8()
 604 {
 605 }
 606
 607 void CppCompareUTF8::call(UErrorCode* status) {
 608     if (U_FAILURE(*status)) return;
 609
 610     // call compareUTF8 for all permutations
 611     int32_t divisor = source->count / maxTestStrings;
 612     StringPiece src, tgt;
 613     int32_t cmp = 0;
 614     for (int32_t i = 0, numTestStringsI = 0; U_SUCCESS(*status) && i < source->count && numTestStringsI < maxTestStrings; i++) {
 615         if (i % divisor) continue;
 616         numTestStringsI++;
 617
 618         if (useLen) {
 619             src.set(source->dataOf(i), source->lengthOf(i));
 620         } else {
 621             src.set(source->dataOf(i));
 622         }
 623         for (int32_t j = 0, numTestStringsJ = 0; U_SUCCESS(*status) && j < source->count && numTestStringsJ < maxTestStrings; j++) {
 624             if (j % divisor) continue;
 625             numTestStringsJ++;
 626
 627             if (useLen) {
 628                 tgt.set(source->dataOf(i), source->lengthOf(i));
 629             } else {
 630                 tgt.set(source->dataOf(i));
 631             }
 632             cmp += coll->compareUTF8(src, tgt, *status);
 633         }
 634     }
 635     // At the end, cmp must be 0
 636     if (cmp != 0) {
 637         *status = U_INTERNAL_PROGRAM_ERROR;
 638     }
 639 }
 640
 641 long CppCompareUTF8::getOperationsPerIteration()
 642 {
 643     return maxTestStrings * maxTestStrings;
 644 }
 645
 646
 647 //
 648 // Test case taking two test data arrays, calling Collator::compareUTF8 for strings at a same index
 649 //
 650 class CppCompareUTF8_2 : public UPerfFunction
 651 {
 652 public:
 653     CppCompareUTF8_2(const Collator* coll, const CA_char* source, const CA_char* target, UBool useLen);
 654     ~CppCompareUTF8_2();
 655     virtual void call(UErrorCode* status);
 656     virtual long getOperationsPerIteration();
 657
 658 private:
 659     const Collator *coll;
 660     const CA_char *source;
 661     const CA_char *target;
 662     UBool useLen;
 663 };
 664
 665 CppCompareUTF8_2::CppCompareUTF8_2(const Collator* coll, const CA_char* source, const CA_char* target, UBool useLen)
 666     :   coll(coll),
 667         source(source),
 668         target(target),
 669         useLen(useLen)
 670 {
 671 }
 672
 673 CppCompareUTF8_2::~CppCompareUTF8_2()
 674 {
 675 }
 676
 677 void CppCompareUTF8_2::call(UErrorCode* status) {
 678     if (U_FAILURE(*status)) return;
 679
 680     // call strcoll for two strings at the same index
 681     StringPiece src, tgt;
 682     if (source->count < target->count) {
 683         *status = U_ILLEGAL_ARGUMENT_ERROR;
 684     } else {
 685         for (int32_t i = 0; U_SUCCESS(*status) && i < source->count; i++) {
 686             if (useLen) {
 687                 src.set(source->dataOf(i), source->lengthOf(i));
 688                 tgt.set(target->dataOf(i), target->lengthOf(i));
 689             } else {
 690                 src.set(source->dataOf(i));
 691                 tgt.set(target->dataOf(i));
 692             }
 693             coll->compareUTF8(src, tgt, *status);
 694         }
 695     }
 696 }
 697
 698 long CppCompareUTF8_2::getOperationsPerIteration()
 699 {
 700     return source->count;
 701 }
 702
 703
 704 //
 705 // Test case taking a single test data array, calling Collator::getCollationKey for each
 706 //
 707 class CppGetCollationKey : public UPerfFunction
 708 {
 709 public:
 710     CppGetCollationKey(const Collator* coll, const CA_uchar* source, UBool useLen);
 711     ~CppGetCollationKey();
 712     virtual void call(UErrorCode* status);
 713     virtual long getOperationsPerIteration();
 714
 715 private:
 716     const Collator *coll;
 717     const CA_uchar *source;
 718     UBool useLen;
 719 };
 720
 721 CppGetCollationKey::CppGetCollationKey(const Collator* coll, const CA_uchar* source, UBool useLen)
 722     :   coll(coll),
 723         source(source),
 724         useLen(useLen)
 725 {
 726 }
 727
 728 CppGetCollationKey::~CppGetCollationKey()
 729 {
 730 }
 731
 732 void CppGetCollationKey::call(UErrorCode* status)
 733 {
 734     if (U_FAILURE(*status)) return;
 735
 736     CollationKey key;
 737     for (int32_t i = 0; U_SUCCESS(*status) && i < source->count; i++) {
 738         coll->getCollationKey(source->dataOf(i), source->lengthOf(i), key, *status);
 739     }
 740 }
 741
 742 long CppGetCollationKey::getOperationsPerIteration() {
 743     return source->count;
 744 }
 745
 746 namespace {
 747
 748 struct CollatorAndCounter {
 749     CollatorAndCounter(const Collator& coll) : coll(coll), ucoll(NULL), counter(0) {}
 750     CollatorAndCounter(const Collator& coll, const UCollator *ucoll)
 751             : coll(coll), ucoll(ucoll), counter(0) {}
 752     const Collator& coll;
 753     const UCollator *ucoll;
 754     int32_t counter;
 755 };
 756
 757 int32_t U_CALLCONV
 758 UniStrCollatorComparator(const void* context, const void* left, const void* right) {
 759     CollatorAndCounter& cc = *(CollatorAndCounter*)context;
 760     const UnicodeString& leftString = **(const UnicodeString**)left;
 761     const UnicodeString& rightString = **(const UnicodeString**)right;
 762     UErrorCode errorCode = U_ZERO_ERROR;
 763     ++cc.counter;
 764     return cc.coll.compare(leftString, rightString, errorCode);
 765 }
 766
 767 }  // namespace
 768
 769 class CollPerfFunction : public UPerfFunction {
 770 public:
 771     CollPerfFunction(const Collator& coll, const UCollator *ucoll)
 772             : coll(coll), ucoll(ucoll), ops(0) {}
 773     virtual ~CollPerfFunction();
 774     /** Calls call() to set the ops field, and returns that. */
 775     virtual long getOperationsPerIteration();
 776
 777 protected:
 778     const Collator& coll;
 779     const UCollator *ucoll;
 780     int32_t ops;
 781 };
 782
 783 CollPerfFunction::~CollPerfFunction() {}
 784
 785 long CollPerfFunction::getOperationsPerIteration() {
 786     UErrorCode errorCode = U_ZERO_ERROR;
 787     call(&errorCode);
 788     return U_SUCCESS(errorCode) ? ops : 0;
 789 }
 790
 791 class UniStrCollPerfFunction : public CollPerfFunction {
 792 public:
 793     UniStrCollPerfFunction(const Collator& coll, const UCollator *ucoll, const CA_uchar* data16)
 794             : CollPerfFunction(coll, ucoll), d16(data16),
 795               source(new UnicodeString*[d16->count]) {
 796         for (int32_t i = 0; i < d16->count; ++i) {
 797             source[i] = new UnicodeString(TRUE, d16->dataOf(i), d16->lengthOf(i));
 798         }
 799     }
 800     virtual ~UniStrCollPerfFunction();
 801
 802 protected:
 803     const CA_uchar* d16;
 804     UnicodeString** source;
 805 };
 806
 807 UniStrCollPerfFunction::~UniStrCollPerfFunction() {
 808     for (int32_t i = 0; i < d16->count; ++i) {
 809         delete source[i];
 810     }
 811     delete[] source;
 812 }
 813
 814 //
 815 // Test case sorting an array of UnicodeString pointers.
 816 //
 817 class UniStrSort : public UniStrCollPerfFunction {
 818 public:
 819     UniStrSort(const Collator& coll, const UCollator *ucoll, const CA_uchar* data16)
 820             : UniStrCollPerfFunction(coll, ucoll, data16),
 821               dest(new UnicodeString*[d16->count]) {}
 822     virtual ~UniStrSort();
 823     virtual void call(UErrorCode* status);
 824
 825 private:
 826     UnicodeString** dest;  // aliases only
 827 };
 828
 829 UniStrSort::~UniStrSort() {
 830     delete[] dest;
 831 }
 832
 833 void UniStrSort::call(UErrorCode* status) {
 834     if (U_FAILURE(*status)) return;
 835
 836     CollatorAndCounter cc(coll);
 837     int32_t count = d16->count;
 838     memcpy(dest, source, count * sizeof(UnicodeString *));
 839     uprv_sortArray(dest, count, (int32_t)sizeof(UnicodeString *),
 840                    UniStrCollatorComparator, &cc, TRUE, status);
 841     ops = cc.counter;
 842 }
 843
 844 namespace {
 845
 846 int32_t U_CALLCONV
 847 StringPieceCollatorComparator(const void* context, const void* left, const void* right) {
 848     CollatorAndCounter& cc = *(CollatorAndCounter*)context;
 849     const StringPiece& leftString = *(const StringPiece*)left;
 850     const StringPiece& rightString = *(const StringPiece*)right;
 851     UErrorCode errorCode = U_ZERO_ERROR;
 852     ++cc.counter;
 853     return cc.coll.compareUTF8(leftString, rightString, errorCode);
 854 }
 855
 856 int32_t U_CALLCONV
 857 StringPieceUCollatorComparator(const void* context, const void* left, const void* right) {
 858     CollatorAndCounter& cc = *(CollatorAndCounter*)context;
 859     const StringPiece& leftString = *(const StringPiece*)left;
 860     const StringPiece& rightString = *(const StringPiece*)right;
 861     UErrorCode errorCode = U_ZERO_ERROR;
 862     ++cc.counter;
 863     return ucol_strcollUTF8(cc.ucoll,
 864                             leftString.data(), leftString.length(),
 865                             rightString.data(), rightString.length(), &errorCode);
 866 }
 867
 868 }  // namespace
 869
 870 class StringPieceCollPerfFunction : public CollPerfFunction {
 871 public:
 872     StringPieceCollPerfFunction(const Collator& coll, const UCollator *ucoll, const CA_char* data8)
 873             : CollPerfFunction(coll, ucoll), d8(data8),
 874               source(new StringPiece[d8->count]) {
 875         for (int32_t i = 0; i < d8->count; ++i) {
 876             source[i].set(d8->dataOf(i), d8->lengthOf(i));
 877         }
 878     }
 879     virtual ~StringPieceCollPerfFunction();
 880
 881 protected:
 882     const CA_char* d8;
 883     StringPiece* source;
 884 };
 885
 886 StringPieceCollPerfFunction::~StringPieceCollPerfFunction() {
 887     delete[] source;
 888 }
 889
 890 class StringPieceSort : public StringPieceCollPerfFunction {
 891 public:
 892     StringPieceSort(const Collator& coll, const UCollator *ucoll, const CA_char* data8)
 893             : StringPieceCollPerfFunction(coll, ucoll, data8),
 894               dest(new StringPiece[d8->count]) {}
 895     virtual ~StringPieceSort();
 896
 897 protected:
 898     StringPiece* dest;
 899 };
 900
 901 StringPieceSort::~StringPieceSort() {
 902     delete[] dest;
 903 }
 904
 905 //
 906 // Test case sorting an array of UTF-8 StringPiece's with Collator::compareUTF8().
 907 //
 908 class StringPieceSortCpp : public StringPieceSort {
 909 public:
 910     StringPieceSortCpp(const Collator& coll, const UCollator *ucoll, const CA_char* data8)
 911             : StringPieceSort(coll, ucoll, data8) {}
 912     virtual ~StringPieceSortCpp();
 913     virtual void call(UErrorCode* status);
 914 };
 915
 916 StringPieceSortCpp::~StringPieceSortCpp() {}
 917
 918 void StringPieceSortCpp::call(UErrorCode* status) {
 919     if (U_FAILURE(*status)) return;
 920
 921     CollatorAndCounter cc(coll);
 922     int32_t count = d8->count;
 923     memcpy(dest, source, count * sizeof(StringPiece));
 924     uprv_sortArray(dest, count, (int32_t)sizeof(StringPiece),
 925                    StringPieceCollatorComparator, &cc, TRUE, status);
 926     ops = cc.counter;
 927 }
 928
 929 //
 930 // Test case sorting an array of UTF-8 StringPiece's with ucol_strcollUTF8().
 931 //
 932 class StringPieceSortC : public StringPieceSort {
 933 public:
 934     StringPieceSortC(const Collator& coll, const UCollator *ucoll, const CA_char* data8)
 935             : StringPieceSort(coll, ucoll, data8) {}
 936     virtual ~StringPieceSortC();
 937     virtual void call(UErrorCode* status);
 938 };
 939
 940 StringPieceSortC::~StringPieceSortC() {}
 941
 942 void StringPieceSortC::call(UErrorCode* status) {
 943     if (U_FAILURE(*status)) return;
 944
 945     CollatorAndCounter cc(coll, ucoll);
 946     int32_t count = d8->count;
 947     memcpy(dest, source, count * sizeof(StringPiece));
 948     uprv_sortArray(dest, count, (int32_t)sizeof(StringPiece),
 949                    StringPieceUCollatorComparator, &cc, TRUE, status);
 950     ops = cc.counter;
 951 }
 952
 953 //
 954 // Test case performing binary searches in a sorted array of UnicodeString pointers.
 955 //
 956 class UniStrBinSearch : public UniStrCollPerfFunction {
 957 public:
 958     UniStrBinSearch(const Collator& coll, const UCollator *ucoll, const CA_uchar* data16)
 959             : UniStrCollPerfFunction(coll, ucoll, data16) {}
 960     virtual ~UniStrBinSearch();
 961     virtual void call(UErrorCode* status);
 962 };
 963
 964 UniStrBinSearch::~UniStrBinSearch() {}
 965
 966 void UniStrBinSearch::call(UErrorCode* status) {
 967     if (U_FAILURE(*status)) return;
 968
 969     CollatorAndCounter cc(coll);
 970     int32_t count = d16->count;
 971     for (int32_t i = 0; i < count; ++i) {
 972         (void)uprv_stableBinarySearch((char *)source, count,
 973                                       source + i, (int32_t)sizeof(UnicodeString *),
 974                                       UniStrCollatorComparator, &cc);
 975     }
 976     ops = cc.counter;
 977 }
 978
 979 class StringPieceBinSearch : public StringPieceCollPerfFunction {
 980 public:
 981     StringPieceBinSearch(const Collator& coll, const UCollator *ucoll, const CA_char* data8)
 982             : StringPieceCollPerfFunction(coll, ucoll, data8) {}
 983     virtual ~StringPieceBinSearch();
 984 };
 985
 986 StringPieceBinSearch::~StringPieceBinSearch() {}
 987
 988 //
 989 // Test case performing binary searches in a sorted array of UTF-8 StringPiece's
 990 // with Collator::compareUTF8().
 991 //
 992 class StringPieceBinSearchCpp : public StringPieceBinSearch {
 993 public:
 994     StringPieceBinSearchCpp(const Collator& coll, const UCollator *ucoll, const CA_char* data8)
 995             : StringPieceBinSearch(coll, ucoll, data8) {}
 996     virtual ~StringPieceBinSearchCpp();
 997     virtual void call(UErrorCode* status);
 998 };
 999
1000 StringPieceBinSearchCpp::~StringPieceBinSearchCpp() {}
1001
1002 void StringPieceBinSearchCpp::call(UErrorCode* status) {
1003     if (U_FAILURE(*status)) return;
1004
1005     CollatorAndCounter cc(coll);
1006     int32_t count = d8->count;
1007     for (int32_t i = 0; i < count; ++i) {
1008         (void)uprv_stableBinarySearch((char *)source, count,
1009                                       source + i, (int32_t)sizeof(StringPiece),
1010                                       StringPieceCollatorComparator, &cc);
1011     }
1012     ops = cc.counter;
1013 }
1014
1015 //
1016 // Test case performing binary searches in a sorted array of UTF-8 StringPiece's
1017 // with ucol_strcollUTF8().
1018 //
1019 class StringPieceBinSearchC : public StringPieceBinSearch {
1020 public:
1021     StringPieceBinSearchC(const Collator& coll, const UCollator *ucoll, const CA_char* data8)
1022             : StringPieceBinSearch(coll, ucoll, data8) {}
1023     virtual ~StringPieceBinSearchC();
1024     virtual void call(UErrorCode* status);
1025 };
1026
1027 StringPieceBinSearchC::~StringPieceBinSearchC() {}
1028
1029 void StringPieceBinSearchC::call(UErrorCode* status) {
1030     if (U_FAILURE(*status)) return;
1031
1032     CollatorAndCounter cc(coll, ucoll);
1033     int32_t count = d8->count;
1034     for (int32_t i = 0; i < count; ++i) {
1035         (void)uprv_stableBinarySearch((char *)source, count,
1036                                       source + i, (int32_t)sizeof(StringPiece),
1037                                       StringPieceUCollatorComparator, &cc);
1038     }
1039     ops = cc.counter;
1040 }
1041
1042
1043 class CollPerf2Test : public UPerfTest
1044 {
1045 public:
1046     CollPerf2Test(int32_t argc, const char *argv[], UErrorCode &status);
1047     ~CollPerf2Test();
1048     virtual UPerfFunction* runIndexedTest(
1049         int32_t index, UBool exec, const char *&name, char *par = NULL);
1050
1051 private:
1052     UCollator* coll;
1053     Collator* collObj;
1054
1055     int32_t count;
1056     CA_uchar* data16;
1057     CA_char* data8;
1058
1059     CA_uchar* modData16;
1060     CA_char* modData8;
1061
1062     CA_uchar* sortedData16;
1063     CA_char* sortedData8;
1064
1065     CA_uchar* randomData16;
1066     CA_char* randomData8;
1067
1068     const CA_uchar* getData16(UErrorCode &status);
1069     const CA_char* getData8(UErrorCode &status);
1070
1071     const CA_uchar* getModData16(UErrorCode &status);
1072     const CA_char* getModData8(UErrorCode &status);
1073
1074     const CA_uchar* getSortedData16(UErrorCode &status);
1075     const CA_char* getSortedData8(UErrorCode &status);
1076
1077     const CA_uchar* getRandomData16(UErrorCode &status);
1078     const CA_char* getRandomData8(UErrorCode &status);
1079
1080     static CA_uchar* sortData16(
1081             const CA_uchar* d16,
1082             UComparator *cmp, const void *context,
1083             UErrorCode &status);
1084     static CA_char* getData8FromData16(const CA_uchar* d16, UErrorCode &status);
1085
1086     UPerfFunction* TestStrcoll();
1087     UPerfFunction* TestStrcollNull();
1088     UPerfFunction* TestStrcollSimilar();
1089
1090     UPerfFunction* TestStrcollUTF8();
1091     UPerfFunction* TestStrcollUTF8Null();
1092     UPerfFunction* TestStrcollUTF8Similar();
1093
1094     UPerfFunction* TestGetSortKey();
1095     UPerfFunction* TestGetSortKeyNull();
1096
1097     UPerfFunction* TestNextSortKeyPart_4All();
1098     UPerfFunction* TestNextSortKeyPart_4x2();
1099     UPerfFunction* TestNextSortKeyPart_4x4();
1100     UPerfFunction* TestNextSortKeyPart_4x8();
1101     UPerfFunction* TestNextSortKeyPart_32All();
1102     UPerfFunction* TestNextSortKeyPart_32x2();
1103
1104     UPerfFunction* TestNextSortKeyPartUTF8_4All();
1105     UPerfFunction* TestNextSortKeyPartUTF8_4x2();
1106     UPerfFunction* TestNextSortKeyPartUTF8_4x4();
1107     UPerfFunction* TestNextSortKeyPartUTF8_4x8();
1108     UPerfFunction* TestNextSortKeyPartUTF8_32All();
1109     UPerfFunction* TestNextSortKeyPartUTF8_32x2();
1110
1111     UPerfFunction* TestCppCompare();
1112     UPerfFunction* TestCppCompareNull();
1113     UPerfFunction* TestCppCompareSimilar();
1114
1115     UPerfFunction* TestCppCompareUTF8();
1116     UPerfFunction* TestCppCompareUTF8Null();
1117     UPerfFunction* TestCppCompareUTF8Similar();
1118
1119     UPerfFunction* TestCppGetCollationKey();
1120     UPerfFunction* TestCppGetCollationKeyNull();
1121
1122     UPerfFunction* TestUniStrSort();
1123     UPerfFunction* TestStringPieceSortCpp();
1124     UPerfFunction* TestStringPieceSortC();
1125
1126     UPerfFunction* TestUniStrBinSearch();
1127     UPerfFunction* TestStringPieceBinSearchCpp();
1128     UPerfFunction* TestStringPieceBinSearchC();
1129 };
1130
1131 CollPerf2Test::CollPerf2Test(int32_t argc, const char *argv[], UErrorCode &status) :
1132     UPerfTest(argc, argv, status),
1133     coll(NULL),
1134     collObj(NULL),
1135     count(0),
1136     data16(NULL),
1137     data8(NULL),
1138     modData16(NULL),
1139     modData8(NULL),
1140     sortedData16(NULL),
1141     sortedData8(NULL),
1142     randomData16(NULL),
1143     randomData8(NULL)
1144 {
1145     if (U_FAILURE(status)) {
1146         return;
1147     }
1148
1149     if (locale == NULL){
1150         locale = "root";
1151     }
1152
1153     // Set up an ICU collator.
1154     // Starting with ICU 54 (ticket #8260), this supports standard collation locale keywords.
1155     coll = ucol_open(locale, &status);
1156     collObj = Collator::createInstance(locale, status);
1157 }
1158
1159 CollPerf2Test::~CollPerf2Test()
1160 {
1161     ucol_close(coll);
1162     delete collObj;
1163
1164     delete data16;
1165     delete data8;
1166     delete modData16;
1167     delete modData8;
1168     delete sortedData16;
1169     delete sortedData8;
1170     delete randomData16;
1171     delete randomData8;
1172 }
1173
1174 #define MAX_NUM_DATA 10000
1175
1176 const CA_uchar* CollPerf2Test::getData16(UErrorCode &status)
1177 {
1178     if (U_FAILURE(status)) return NULL;
1179     if (data16) return data16;
1180
1181     CA_uchar* d16 = new CA_uchar();
1182     const UChar *line = NULL;
1183     int32_t len = 0;
1184     int32_t numData = 0;
1185
1186     for (;;) {
1187         line = ucbuf_readline(ucharBuf, &len, &status);
1188         if (line == NULL || U_FAILURE(status)) break;
1189
1190         // Refer to the source code of ucbuf_readline()
1191         // 1. 'len' includes the line terminal symbols
1192         // 2. The length of the line terminal symbols is only one character
1193         // 3. The Windows CR LF line terminal symbols will be converted to CR
1194
1195         if (len == 1 || line[0] == 0x23 /* '#' */) {
1196             continue; // skip empty/comment line
1197         } else {
1198             d16->append_one(len);
1199             UChar *p = d16->last();
1200             u_memcpy(p, line, len - 1);  // exclude the CR
1201             p[len - 1] = 0;  // NUL-terminate
1202
1203             numData++;
1204             if (numData >= MAX_NUM_DATA) break;
1205         }
1206     }
1207
1208     if (U_SUCCESS(status)) {
1209         data16 = d16;
1210     } else {
1211         delete d16;
1212     }
1213
1214     return data16;
1215 }
1216
1217 const CA_char* CollPerf2Test::getData8(UErrorCode &status)
1218 {
1219     if (U_FAILURE(status)) return NULL;
1220     if (data8) return data8;
1221     return data8 = getData8FromData16(getData16(status), status);
1222 }
1223
1224 const CA_uchar* CollPerf2Test::getModData16(UErrorCode &status)
1225 {
1226     if (U_FAILURE(status)) return NULL;
1227     if (modData16) return modData16;
1228
1229     const CA_uchar* d16 = getData16(status);
1230     if (U_FAILURE(status)) return NULL;
1231
1232     CA_uchar* modData16 = new CA_uchar();
1233
1234     for (int32_t i = 0; i < d16->count; i++) {
1235         const UChar *s = d16->dataOf(i);
1236         int32_t len = d16->lengthOf(i) + 1; // including NULL terminator
1237
1238         modData16->append_one(len);
1239         u_memcpy(modData16->last(), s, len);
1240
1241         // replacing the last character with a different character
1242         UChar *lastChar = &modData16->last()[len -2];
1243         for (int32_t j = i + 1; j != i; j++) {
1244             if (j >= d16->count) {
1245                 j = 0;
1246             }
1247             const UChar *s1 = d16->dataOf(j);
1248             UChar lastChar1 = s1[d16->lengthOf(j) - 1];
1249             if (*lastChar != lastChar1) {
1250                 *lastChar = lastChar1;
1251                 break;
1252             }
1253         }
1254     }
1255
1256     return modData16;
1257 }
1258
1259 const CA_char* CollPerf2Test::getModData8(UErrorCode &status)
1260 {
1261     if (U_FAILURE(status)) return NULL;
1262     if (modData8) return modData8;
1263     return modData8 = getData8FromData16(getModData16(status), status);
1264 }
1265
1266 namespace {
1267
1268 struct ArrayAndColl {
1269     ArrayAndColl(const CA_uchar* a, const Collator& c) : d16(a), coll(c) {}
1270     const CA_uchar* d16;
1271     const Collator& coll;
1272 };
1273
1274 int32_t U_CALLCONV
1275 U16CollatorComparator(const void* context, const void* left, const void* right) {
1276     const ArrayAndColl& ac = *(const ArrayAndColl*)context;
1277     const CA_uchar* d16 = ac.d16;
1278     int32_t leftIndex = *(const int32_t*)left;
1279     int32_t rightIndex = *(const int32_t*)right;
1280     UErrorCode errorCode = U_ZERO_ERROR;
1281     return ac.coll.compare(d16->dataOf(leftIndex), d16->lengthOf(leftIndex),
1282                            d16->dataOf(rightIndex), d16->lengthOf(rightIndex),
1283                            errorCode);
1284 }
1285
1286 int32_t U_CALLCONV
1287 U16HashComparator(const void* context, const void* left, const void* right) {
1288     const CA_uchar* d16 = (const CA_uchar*)context;
1289     int32_t leftIndex = *(const int32_t*)left;
1290     int32_t rightIndex = *(const int32_t*)right;
1291     int32_t leftHash = ustr_hashUCharsN(d16->dataOf(leftIndex), d16->lengthOf(leftIndex));
1292     int32_t rightHash = ustr_hashUCharsN(d16->dataOf(rightIndex), d16->lengthOf(rightIndex));
1293     return leftHash < rightHash ? -1 : leftHash == rightHash ? 0 : 1;
1294 }
1295
1296 }  // namespace
1297
1298 const CA_uchar* CollPerf2Test::getSortedData16(UErrorCode &status) {
1299     if (U_FAILURE(status)) return NULL;
1300     if (sortedData16) return sortedData16;
1301
1302     ArrayAndColl ac(getData16(status), *collObj);
1303     return sortedData16 = sortData16(ac.d16, U16CollatorComparator, &ac, status);
1304 }
1305
1306 const CA_char* CollPerf2Test::getSortedData8(UErrorCode &status) {
1307     if (U_FAILURE(status)) return NULL;
1308     if (sortedData8) return sortedData8;
1309     return sortedData8 = getData8FromData16(getSortedData16(status), status);
1310 }
1311
1312 const CA_uchar* CollPerf2Test::getRandomData16(UErrorCode &status) {
1313     if (U_FAILURE(status)) return NULL;
1314     if (randomData16) return randomData16;
1315
1316     // Sort the strings by their hash codes, which should be a reasonably pseudo-random order.
1317     const CA_uchar* d16 = getData16(status);
1318     return randomData16 = sortData16(d16, U16HashComparator, d16, status);
1319 }
1320
1321 const CA_char* CollPerf2Test::getRandomData8(UErrorCode &status) {
1322     if (U_FAILURE(status)) return NULL;
1323     if (randomData8) return randomData8;
1324     return randomData8 = getData8FromData16(getRandomData16(status), status);
1325 }
1326
1327 CA_uchar* CollPerf2Test::sortData16(const CA_uchar* d16,
1328                                     UComparator *cmp, const void *context,
1329                                     UErrorCode &status) {
1330     if (U_FAILURE(status)) return NULL;
1331
1332     LocalArray<int32_t> indexes(new int32_t[d16->count]);
1333     for (int32_t i = 0; i < d16->count; ++i) {
1334         indexes[i] = i;
1335     }
1336     uprv_sortArray(indexes.getAlias(), d16->count, 4, cmp, context, TRUE, &status);
1337     if (U_FAILURE(status)) return NULL;
1338
1339     // Copy the strings in sorted order into a new array.
1340     LocalPointer<CA_uchar> newD16(new CA_uchar());
1341     for (int32_t i = 0; i < d16->count; i++) {
1342         int32_t j = indexes[i];
1343         const UChar* s = d16->dataOf(j);
1344         int32_t len = d16->lengthOf(j);
1345         int32_t capacity = len + 1;  // including NULL terminator
1346         newD16->append_one(capacity);
1347         u_memcpy(newD16->last(), s, capacity);
1348     }
1349
1350     if (U_SUCCESS(status)) {
1351         return newD16.orphan();
1352     } else {
1353         return NULL;
1354     }
1355 }
1356
1357 CA_char* CollPerf2Test::getData8FromData16(const CA_uchar* d16, UErrorCode &status) {
1358     if (U_FAILURE(status)) return NULL;
1359
1360     // UTF-16 -> UTF-8 conversion
1361     LocalPointer<CA_char> d8(new CA_char());
1362     for (int32_t i = 0; i < d16->count; i++) {
1363         const UChar *s16 = d16->dataOf(i);
1364         int32_t length16 = d16->lengthOf(i);
1365
1366         // get length in UTF-8
1367         int32_t length8;
1368         u_strToUTF8(NULL, 0, &length8, s16, length16, &status);
1369         if (status == U_BUFFER_OVERFLOW_ERROR || status == U_ZERO_ERROR){
1370             status = U_ZERO_ERROR;
1371         } else {
1372             break;
1373         }
1374         int32_t capacity8 = length8 + 1;  // plus terminal NULL
1375         d8->append_one(capacity8);
1376
1377         // convert to UTF-8
1378         u_strToUTF8(d8->last(), capacity8, NULL, s16, length16, &status);
1379         if (U_FAILURE(status)) break;
1380     }
1381
1382     if (U_SUCCESS(status)) {
1383         return d8.orphan();
1384     } else {
1385         return NULL;
1386     }
1387 }
1388
1389 UPerfFunction*
1390 CollPerf2Test::runIndexedTest(int32_t index, UBool exec, const char *&name, char *par /*= NULL*/)
1391 {
1392     (void)par;
1393     TESTCASE_AUTO_BEGIN;
1394
1395     TESTCASE_AUTO(TestStrcoll);
1396     TESTCASE_AUTO(TestStrcollNull);
1397     TESTCASE_AUTO(TestStrcollSimilar);
1398
1399     TESTCASE_AUTO(TestStrcollUTF8);
1400     TESTCASE_AUTO(TestStrcollUTF8Null);
1401     TESTCASE_AUTO(TestStrcollUTF8Similar);
1402
1403     TESTCASE_AUTO(TestGetSortKey);
1404     TESTCASE_AUTO(TestGetSortKeyNull);
1405
1406     TESTCASE_AUTO(TestNextSortKeyPart_4All);
1407     TESTCASE_AUTO(TestNextSortKeyPart_4x4);
1408     TESTCASE_AUTO(TestNextSortKeyPart_4x8);
1409     TESTCASE_AUTO(TestNextSortKeyPart_32All);
1410     TESTCASE_AUTO(TestNextSortKeyPart_32x2);
1411
1412     TESTCASE_AUTO(TestNextSortKeyPartUTF8_4All);
1413     TESTCASE_AUTO(TestNextSortKeyPartUTF8_4x4);
1414     TESTCASE_AUTO(TestNextSortKeyPartUTF8_4x8);
1415     TESTCASE_AUTO(TestNextSortKeyPartUTF8_32All);
1416     TESTCASE_AUTO(TestNextSortKeyPartUTF8_32x2);
1417
1418     TESTCASE_AUTO(TestCppCompare);
1419     TESTCASE_AUTO(TestCppCompareNull);
1420     TESTCASE_AUTO(TestCppCompareSimilar);
1421
1422     TESTCASE_AUTO(TestCppCompareUTF8);
1423     TESTCASE_AUTO(TestCppCompareUTF8Null);
1424     TESTCASE_AUTO(TestCppCompareUTF8Similar);
1425
1426     TESTCASE_AUTO(TestCppGetCollationKey);
1427     TESTCASE_AUTO(TestCppGetCollationKeyNull);
1428
1429     TESTCASE_AUTO(TestUniStrSort);
1430     TESTCASE_AUTO(TestStringPieceSortCpp);
1431     TESTCASE_AUTO(TestStringPieceSortC);
1432
1433     TESTCASE_AUTO(TestUniStrBinSearch);
1434     TESTCASE_AUTO(TestStringPieceBinSearchCpp);
1435     TESTCASE_AUTO(TestStringPieceBinSearchC);
1436
1437     TESTCASE_AUTO_END;
1438     return NULL;
1439 }
1440
1441
1442
1443 UPerfFunction* CollPerf2Test::TestStrcoll()
1444 {
1445     UErrorCode status = U_ZERO_ERROR;
1446     Strcoll *testCase = new Strcoll(coll, getData16(status), TRUE /* useLen */);
1447     if (U_FAILURE(status)) {
1448         delete testCase;
1449         return NULL;
1450     }
1451     return testCase;
1452 }
1453
1454 UPerfFunction* CollPerf2Test::TestStrcollNull()
1455 {
1456     UErrorCode status = U_ZERO_ERROR;
1457     Strcoll *testCase = new Strcoll(coll, getData16(status), FALSE /* useLen */);
1458     if (U_FAILURE(status)) {
1459         delete testCase;
1460         return NULL;
1461     }
1462     return testCase;
1463 }
1464
1465 UPerfFunction* CollPerf2Test::TestStrcollSimilar()
1466 {
1467     UErrorCode status = U_ZERO_ERROR;
1468     Strcoll_2 *testCase = new Strcoll_2(coll, getData16(status), getModData16(status), TRUE /* useLen */);
1469     if (U_FAILURE(status)) {
1470         delete testCase;
1471         return NULL;
1472     }
1473     return testCase;
1474 }
1475
1476 UPerfFunction* CollPerf2Test::TestStrcollUTF8()
1477 {
1478     UErrorCode status = U_ZERO_ERROR;
1479     StrcollUTF8 *testCase = new StrcollUTF8(coll, getData8(status), TRUE /* useLen */);
1480     if (U_FAILURE(status)) {
1481         delete testCase;
1482         return NULL;
1483     }
1484     return testCase;
1485 }
1486
1487 UPerfFunction* CollPerf2Test::TestStrcollUTF8Null()
1488 {
1489     UErrorCode status = U_ZERO_ERROR;
1490     StrcollUTF8 *testCase = new StrcollUTF8(coll, getData8(status),FALSE /* useLen */);
1491     if (U_FAILURE(status)) {
1492         delete testCase;
1493         return NULL;
1494     }
1495     return testCase;
1496 }
1497
1498 UPerfFunction* CollPerf2Test::TestStrcollUTF8Similar()
1499 {
1500     UErrorCode status = U_ZERO_ERROR;
1501     StrcollUTF8_2 *testCase = new StrcollUTF8_2(coll, getData8(status), getModData8(status), TRUE /* useLen */);
1502     if (U_FAILURE(status)) {
1503         delete testCase;
1504         return NULL;
1505     }
1506     return testCase;
1507 }
1508
1509 UPerfFunction* CollPerf2Test::TestGetSortKey()
1510 {
1511     UErrorCode status = U_ZERO_ERROR;
1512     GetSortKey *testCase = new GetSortKey(coll, getData16(status), TRUE /* useLen */);
1513     if (U_FAILURE(status)) {
1514         delete testCase;
1515         return NULL;
1516     }
1517     return testCase;
1518 }
1519
1520 UPerfFunction* CollPerf2Test::TestGetSortKeyNull()
1521 {
1522     UErrorCode status = U_ZERO_ERROR;
1523     GetSortKey *testCase = new GetSortKey(coll, getData16(status), FALSE /* useLen */);
1524     if (U_FAILURE(status)) {
1525         delete testCase;
1526         return NULL;
1527     }
1528     return testCase;
1529 }
1530
1531 UPerfFunction* CollPerf2Test::TestNextSortKeyPart_4All()
1532 {
1533     UErrorCode status = U_ZERO_ERROR;
1534     NextSortKeyPart *testCase = new NextSortKeyPart(coll, getData16(status), 4 /* bufSize */);
1535     if (U_FAILURE(status)) {
1536         delete testCase;
1537         return NULL;
1538     }
1539     return testCase;
1540 }
1541
1542 UPerfFunction* CollPerf2Test::TestNextSortKeyPart_4x4()
1543 {
1544     UErrorCode status = U_ZERO_ERROR;
1545     NextSortKeyPart *testCase = new NextSortKeyPart(coll, getData16(status), 4 /* bufSize */, 4 /* maxIteration */);
1546     if (U_FAILURE(status)) {
1547         delete testCase;
1548         return NULL;
1549     }
1550     return testCase;
1551 }
1552
1553 UPerfFunction* CollPerf2Test::TestNextSortKeyPart_4x8()
1554 {
1555     UErrorCode status = U_ZERO_ERROR;
1556     NextSortKeyPart *testCase = new NextSortKeyPart(coll, getData16(status), 4 /* bufSize */, 8 /* maxIteration */);
1557     if (U_FAILURE(status)) {
1558         delete testCase;
1559         return NULL;
1560     }
1561     return testCase;
1562 }
1563
1564 UPerfFunction* CollPerf2Test::TestNextSortKeyPart_32All()
1565 {
1566     UErrorCode status = U_ZERO_ERROR;
1567     NextSortKeyPart *testCase = new NextSortKeyPart(coll, getData16(status), 32 /* bufSize */);
1568     if (U_FAILURE(status)) {
1569         delete testCase;
1570         return NULL;
1571     }
1572     return testCase;
1573 }
1574
1575 UPerfFunction* CollPerf2Test::TestNextSortKeyPart_32x2()
1576 {
1577     UErrorCode status = U_ZERO_ERROR;
1578     NextSortKeyPart *testCase = new NextSortKeyPart(coll, getData16(status), 32 /* bufSize */, 2 /* maxIteration */);
1579     if (U_FAILURE(status)) {
1580         delete testCase;
1581         return NULL;
1582     }
1583     return testCase;
1584 }
1585
1586 UPerfFunction* CollPerf2Test::TestNextSortKeyPartUTF8_4All()
1587 {
1588     UErrorCode status = U_ZERO_ERROR;
1589     NextSortKeyPartUTF8 *testCase = new NextSortKeyPartUTF8(coll, getData8(status), 4 /* bufSize */);
1590     if (U_FAILURE(status)) {
1591         delete testCase;
1592         return NULL;
1593     }
1594     return testCase;
1595 }
1596
1597 UPerfFunction* CollPerf2Test::TestNextSortKeyPartUTF8_4x4()
1598 {
1599     UErrorCode status = U_ZERO_ERROR;
1600     NextSortKeyPartUTF8 *testCase = new NextSortKeyPartUTF8(coll, getData8(status), 4 /* bufSize */, 4 /* maxIteration */);
1601     if (U_FAILURE(status)) {
1602         delete testCase;
1603         return NULL;
1604     }
1605     return testCase;
1606 }
1607
1608 UPerfFunction* CollPerf2Test::TestNextSortKeyPartUTF8_4x8()
1609 {
1610     UErrorCode status = U_ZERO_ERROR;
1611     NextSortKeyPartUTF8 *testCase = new NextSortKeyPartUTF8(coll, getData8(status), 4 /* bufSize */, 8 /* maxIteration */);
1612     if (U_FAILURE(status)) {
1613         delete testCase;
1614         return NULL;
1615     }
1616     return testCase;
1617 }
1618
1619 UPerfFunction* CollPerf2Test::TestNextSortKeyPartUTF8_32All()
1620 {
1621     UErrorCode status = U_ZERO_ERROR;
1622     NextSortKeyPartUTF8 *testCase = new NextSortKeyPartUTF8(coll, getData8(status), 32 /* bufSize */);
1623     if (U_FAILURE(status)) {
1624         delete testCase;
1625         return NULL;
1626     }
1627     return testCase;
1628 }
1629
1630 UPerfFunction* CollPerf2Test::TestNextSortKeyPartUTF8_32x2()
1631 {
1632     UErrorCode status = U_ZERO_ERROR;
1633     NextSortKeyPartUTF8 *testCase = new NextSortKeyPartUTF8(coll, getData8(status), 32 /* bufSize */, 2 /* maxIteration */);
1634     if (U_FAILURE(status)) {
1635         delete testCase;
1636         return NULL;
1637     }
1638     return testCase;
1639 }
1640
1641 UPerfFunction* CollPerf2Test::TestCppCompare()
1642 {
1643     UErrorCode status = U_ZERO_ERROR;
1644     CppCompare *testCase = new CppCompare(collObj, getData16(status), TRUE /* useLen */);
1645     if (U_FAILURE(status)) {
1646         delete testCase;
1647         return NULL;
1648     }
1649     return testCase;
1650 }
1651
1652 UPerfFunction* CollPerf2Test::TestCppCompareNull()
1653 {
1654     UErrorCode status = U_ZERO_ERROR;
1655     CppCompare *testCase = new CppCompare(collObj, getData16(status), FALSE /* useLen */);
1656     if (U_FAILURE(status)) {
1657         delete testCase;
1658         return NULL;
1659     }
1660     return testCase;
1661 }
1662
1663 UPerfFunction* CollPerf2Test::TestCppCompareSimilar()
1664 {
1665     UErrorCode status = U_ZERO_ERROR;
1666     CppCompare_2 *testCase = new CppCompare_2(collObj, getData16(status), getModData16(status), TRUE /* useLen */);
1667     if (U_FAILURE(status)) {
1668         delete testCase;
1669         return NULL;
1670     }
1671     return testCase;
1672 }
1673
1674 UPerfFunction* CollPerf2Test::TestCppCompareUTF8()
1675 {
1676     UErrorCode status = U_ZERO_ERROR;
1677     CppCompareUTF8 *testCase = new CppCompareUTF8(collObj, getData8(status), TRUE /* useLen */);
1678     if (U_FAILURE(status)) {
1679         delete testCase;
1680         return NULL;
1681     }
1682     return testCase;
1683 }
1684
1685 UPerfFunction* CollPerf2Test::TestCppCompareUTF8Null()
1686 {
1687     UErrorCode status = U_ZERO_ERROR;
1688     CppCompareUTF8 *testCase = new CppCompareUTF8(collObj, getData8(status), FALSE /* useLen */);
1689     if (U_FAILURE(status)) {
1690         delete testCase;
1691         return NULL;
1692     }
1693     return testCase;
1694 }
1695
1696 UPerfFunction* CollPerf2Test::TestCppCompareUTF8Similar()
1697 {
1698     UErrorCode status = U_ZERO_ERROR;
1699     CppCompareUTF8_2 *testCase = new CppCompareUTF8_2(collObj, getData8(status), getModData8(status), TRUE /* useLen */);
1700     if (U_FAILURE(status)) {
1701         delete testCase;
1702         return NULL;
1703     }
1704     return testCase;
1705 }
1706
1707 UPerfFunction* CollPerf2Test::TestCppGetCollationKey()
1708 {
1709     UErrorCode status = U_ZERO_ERROR;
1710     CppGetCollationKey *testCase = new CppGetCollationKey(collObj, getData16(status), TRUE /* useLen */);
1711     if (U_FAILURE(status)) {
1712         delete testCase;
1713         return NULL;
1714     }
1715     return testCase;
1716 }
1717
1718 UPerfFunction* CollPerf2Test::TestCppGetCollationKeyNull()
1719 {
1720     UErrorCode status = U_ZERO_ERROR;
1721     CppGetCollationKey *testCase = new CppGetCollationKey(collObj, getData16(status), FALSE /* useLen */);
1722     if (U_FAILURE(status)) {
1723         delete testCase;
1724         return NULL;
1725     }
1726     return testCase;
1727 }
1728
1729 UPerfFunction* CollPerf2Test::TestUniStrSort() {
1730     UErrorCode status = U_ZERO_ERROR;
1731     UPerfFunction *testCase = new UniStrSort(*collObj, coll, getRandomData16(status));
1732     if (U_FAILURE(status)) {
1733         delete testCase;
1734         return NULL;
1735     }
1736     return testCase;
1737 }
1738
1739 UPerfFunction* CollPerf2Test::TestStringPieceSortCpp() {
1740     UErrorCode status = U_ZERO_ERROR;
1741     UPerfFunction *testCase = new StringPieceSortCpp(*collObj, coll, getRandomData8(status));
1742     if (U_FAILURE(status)) {
1743         delete testCase;
1744         return NULL;
1745     }
1746     return testCase;
1747 }
1748
1749 UPerfFunction* CollPerf2Test::TestStringPieceSortC() {
1750     UErrorCode status = U_ZERO_ERROR;
1751     UPerfFunction *testCase = new StringPieceSortC(*collObj, coll, getRandomData8(status));
1752     if (U_FAILURE(status)) {
1753         delete testCase;
1754         return NULL;
1755     }
1756     return testCase;
1757 }
1758
1759 UPerfFunction* CollPerf2Test::TestUniStrBinSearch() {
1760     UErrorCode status = U_ZERO_ERROR;
1761     UPerfFunction *testCase = new UniStrBinSearch(*collObj, coll, getSortedData16(status));
1762     if (U_FAILURE(status)) {
1763         delete testCase;
1764         return NULL;
1765     }
1766     return testCase;
1767 }
1768
1769 UPerfFunction* CollPerf2Test::TestStringPieceBinSearchCpp() {
1770     UErrorCode status = U_ZERO_ERROR;
1771     UPerfFunction *testCase = new StringPieceBinSearchCpp(*collObj, coll, getSortedData8(status));
1772     if (U_FAILURE(status)) {
1773         delete testCase;
1774         return NULL;
1775     }
1776     return testCase;
1777 }
1778
1779 UPerfFunction* CollPerf2Test::TestStringPieceBinSearchC() {
1780     UErrorCode status = U_ZERO_ERROR;
1781     UPerfFunction *testCase = new StringPieceBinSearchC(*collObj, coll, getSortedData8(status));
1782     if (U_FAILURE(status)) {
1783         delete testCase;
1784         return NULL;
1785     }
1786     return testCase;
1787 }
1788
1789
1790 int main(int argc, const char *argv[])
1791 {
1792     UErrorCode status = U_ZERO_ERROR;
1793     CollPerf2Test test(argc, argv, status);
1794
1795     if (U_FAILURE(status)){
1796         printf("The error is %s\n", u_errorName(status));
1797         //TODO: print usage here
1798         return status;
1799     }
1800
1801     if (test.run() == FALSE){
1802         fprintf(stderr, "FAILED: Tests could not be run please check the arguments.\n");
1803         return -1;
1804     }
1805     return 0;
1806 }