icuSources/i18n/repattrn.cpp

   1 //
   2 //  file:  repattrn.cpp
   3 //
   4 /*
   5 ***************************************************************************
   6 *   Copyright (C) 2002-2013 International Business Machines Corporation   *
   7 *   and others. All rights reserved.                                      *
   8 ***************************************************************************
   9 */
  10
  11 #include "unicode/utypes.h"
  12
  13 #if !UCONFIG_NO_REGULAR_EXPRESSIONS
  14
  15 #include "unicode/regex.h"
  16 #include "unicode/uclean.h"
  17 #include "uassert.h"
  18 #include "uvector.h"
  19 #include "uvectr32.h"
  20 #include "uvectr64.h"
  21 #include "regexcmp.h"
  22 #include "regeximp.h"
  23 #include "regexst.h"
  24
  25 U_NAMESPACE_BEGIN
  26
  27 //--------------------------------------------------------------------------
  28 //
  29 //    RegexPattern    Default Constructor
  30 //
  31 //--------------------------------------------------------------------------
  32 RegexPattern::RegexPattern() {
  33     // Init all of this instances data.
  34     init();
  35 }
  36
  37
  38 //--------------------------------------------------------------------------
  39 //
  40 //   Copy Constructor        Note:  This is a rather inefficient implementation,
  41 //                                  but it probably doesn't matter.
  42 //
  43 //--------------------------------------------------------------------------
  44 RegexPattern::RegexPattern(const RegexPattern &other) :  UObject(other) {
  45     init();
  46     *this = other;
  47 }
  48
  49
  50
  51 //--------------------------------------------------------------------------
  52 //
  53 //    Assignment Operator
  54 //
  55 //--------------------------------------------------------------------------
  56 RegexPattern &RegexPattern::operator = (const RegexPattern &other) {
  57     if (this == &other) {
  58         // Source and destination are the same.  Don't do anything.
  59         return *this;
  60     }
  61
  62     // Clean out any previous contents of object being assigned to.
  63     zap();
  64
  65     // Give target object a default initialization
  66     init();
  67
  68     // Copy simple fields
  69     if ( other.fPatternString == NULL ) {
  70         fPatternString = NULL;
  71         fPattern      = utext_clone(fPattern, other.fPattern, FALSE, TRUE, &fDeferredStatus);
  72     } else {
  73         fPatternString = new UnicodeString(*(other.fPatternString));
  74         UErrorCode status = U_ZERO_ERROR;
  75         fPattern      = utext_openConstUnicodeString(NULL, fPatternString, &status);
  76         if (U_FAILURE(status)) {
  77             fDeferredStatus = U_MEMORY_ALLOCATION_ERROR;
  78             return *this;
  79         }
  80     }
  81     fFlags            = other.fFlags;
  82     fLiteralText      = other.fLiteralText;
  83     fDeferredStatus   = other.fDeferredStatus;
  84     fMinMatchLen      = other.fMinMatchLen;
  85     fFrameSize        = other.fFrameSize;
  86     fDataSize         = other.fDataSize;
  87     fMaxCaptureDigits = other.fMaxCaptureDigits;
  88     fStaticSets       = other.fStaticSets;
  89     fStaticSets8      = other.fStaticSets8;
  90
  91     fStartType        = other.fStartType;
  92     fInitialStringIdx = other.fInitialStringIdx;
  93     fInitialStringLen = other.fInitialStringLen;
  94     *fInitialChars    = *other.fInitialChars;
  95     fInitialChar      = other.fInitialChar;
  96     *fInitialChars8   = *other.fInitialChars8;
  97     fNeedsAltInput    = other.fNeedsAltInput;
  98
  99     //  Copy the pattern.  It's just values, nothing deep to copy.
 100     fCompiledPat->assign(*other.fCompiledPat, fDeferredStatus);
 101     fGroupMap->assign(*other.fGroupMap, fDeferredStatus);
 102
 103     //  Copy the Unicode Sets.
 104     //    Could be made more efficient if the sets were reference counted and shared,
 105     //    but I doubt that pattern copying will be particularly common.
 106     //    Note:  init() already added an empty element zero to fSets
 107     int32_t i;
 108     int32_t  numSets = other.fSets->size();
 109     fSets8 = new Regex8BitSet[numSets];
 110     if (fSets8 == NULL) {
 111         fDeferredStatus = U_MEMORY_ALLOCATION_ERROR;
 112         return *this;
 113     }
 114     for (i=1; i<numSets; i++) {
 115         if (U_FAILURE(fDeferredStatus)) {
 116             return *this;
 117         }
 118         UnicodeSet *sourceSet = (UnicodeSet *)other.fSets->elementAt(i);
 119         UnicodeSet *newSet    = new UnicodeSet(*sourceSet);
 120         if (newSet == NULL) {
 121             fDeferredStatus = U_MEMORY_ALLOCATION_ERROR;
 122             break;
 123         }
 124         fSets->addElement(newSet, fDeferredStatus);
 125         fSets8[i] = other.fSets8[i];
 126     }
 127
 128     return *this;
 129 }
 130
 131
 132 //--------------------------------------------------------------------------
 133 //
 134 //    init        Shared initialization for use by constructors.
 135 //                Bring an uninitialized RegexPattern up to a default state.
 136 //
 137 //--------------------------------------------------------------------------
 138 void RegexPattern::init() {
 139     fFlags            = 0;
 140     fCompiledPat      = 0;
 141     fLiteralText.remove();
 142     fSets             = NULL;
 143     fSets8            = NULL;
 144     fDeferredStatus   = U_ZERO_ERROR;
 145     fMinMatchLen      = 0;
 146     fFrameSize        = 0;
 147     fDataSize         = 0;
 148     fGroupMap         = NULL;
 149     fMaxCaptureDigits = 1;
 150     fStaticSets       = NULL;
 151     fStaticSets8      = NULL;
 152     fStartType        = START_NO_INFO;
 153     fInitialStringIdx = 0;
 154     fInitialStringLen = 0;
 155     fInitialChars     = NULL;
 156     fInitialChar      = 0;
 157     fInitialChars8    = NULL;
 158     fNeedsAltInput    = FALSE;
 159
 160     fPattern          = NULL; // will be set later
 161     fPatternString    = NULL; // may be set later
 162     fCompiledPat      = new UVector64(fDeferredStatus);
 163     fGroupMap         = new UVector32(fDeferredStatus);
 164     fSets             = new UVector(fDeferredStatus);
 165     fInitialChars     = new UnicodeSet;
 166     fInitialChars8    = new Regex8BitSet;
 167     if (U_FAILURE(fDeferredStatus)) {
 168         return;
 169     }
 170     if (fCompiledPat == NULL  || fGroupMap == NULL || fSets == NULL ||
 171         fInitialChars == NULL || fInitialChars8 == NULL) {
 172         fDeferredStatus = U_MEMORY_ALLOCATION_ERROR;
 173         return;
 174     }
 175
 176     // Slot zero of the vector of sets is reserved.  Fill it here.
 177     fSets->addElement((int32_t)0, fDeferredStatus);
 178 }
 179
 180
 181 //--------------------------------------------------------------------------
 182 //
 183 //   zap            Delete everything owned by this RegexPattern.
 184 //
 185 //--------------------------------------------------------------------------
 186 void RegexPattern::zap() {
 187     delete fCompiledPat;
 188     fCompiledPat = NULL;
 189     int i;
 190     for (i=1; i<fSets->size(); i++) {
 191         UnicodeSet *s;
 192         s = (UnicodeSet *)fSets->elementAt(i);
 193         if (s != NULL) {
 194             delete s;
 195         }
 196     }
 197     delete fSets;
 198     fSets = NULL;
 199     delete[] fSets8;
 200     fSets8 = NULL;
 201     delete fGroupMap;
 202     fGroupMap = NULL;
 203     delete fInitialChars;
 204     fInitialChars = NULL;
 205     delete fInitialChars8;
 206     fInitialChars8 = NULL;
 207     if (fPattern != NULL) {
 208         utext_close(fPattern);
 209         fPattern = NULL;
 210     }
 211     if (fPatternString != NULL) {
 212         delete fPatternString;
 213         fPatternString = NULL;
 214     }
 215 }
 216
 217
 218 //--------------------------------------------------------------------------
 219 //
 220 //   Destructor
 221 //
 222 //--------------------------------------------------------------------------
 223 RegexPattern::~RegexPattern() {
 224     zap();
 225 }
 226
 227
 228 //--------------------------------------------------------------------------
 229 //
 230 //   Clone
 231 //
 232 //--------------------------------------------------------------------------
 233 RegexPattern  *RegexPattern::clone() const {
 234     RegexPattern  *copy = new RegexPattern(*this);
 235     return copy;
 236 }
 237
 238
 239 //--------------------------------------------------------------------------
 240 //
 241 //   operator ==   (comparison)    Consider to patterns to be == if the
 242 //                                 pattern strings and the flags are the same.
 243 //                                 Note that pattern strings with the same
 244 //                                 characters can still be considered different.
 245 //
 246 //--------------------------------------------------------------------------
 247 UBool   RegexPattern::operator ==(const RegexPattern &other) const {
 248     if (this->fFlags == other.fFlags && this->fDeferredStatus == other.fDeferredStatus) {
 249         if (this->fPatternString != NULL && other.fPatternString != NULL) {
 250             return *(this->fPatternString) == *(other.fPatternString);
 251         } else if (this->fPattern == NULL) {
 252             if (other.fPattern == NULL) {
 253                 return TRUE;
 254             }
 255         } else if (other.fPattern != NULL) {
 256             UTEXT_SETNATIVEINDEX(this->fPattern, 0);
 257             UTEXT_SETNATIVEINDEX(other.fPattern, 0);
 258             return utext_equals(this->fPattern, other.fPattern);
 259         }
 260     }
 261     return FALSE;
 262 }
 263
 264 //---------------------------------------------------------------------
 265 //
 266 //   compile
 267 //
 268 //---------------------------------------------------------------------
 269 RegexPattern * U_EXPORT2
 270 RegexPattern::compile(const UnicodeString &regex,
 271                       uint32_t             flags,
 272                       UParseError          &pe,
 273                       UErrorCode           &status)
 274 {
 275     if (U_FAILURE(status)) {
 276         return NULL;
 277     }
 278
 279     const uint32_t allFlags = UREGEX_CANON_EQ | UREGEX_CASE_INSENSITIVE | UREGEX_COMMENTS |
 280     UREGEX_DOTALL   | UREGEX_MULTILINE        | UREGEX_UWORD |
 281     UREGEX_ERROR_ON_UNKNOWN_ESCAPES           | UREGEX_UNIX_LINES | UREGEX_LITERAL;
 282
 283     if ((flags & ~allFlags) != 0) {
 284         status = U_REGEX_INVALID_FLAG;
 285         return NULL;
 286     }
 287
 288     if ((flags & UREGEX_CANON_EQ) != 0) {
 289         status = U_REGEX_UNIMPLEMENTED;
 290         return NULL;
 291     }
 292
 293     RegexPattern *This = new RegexPattern;
 294     if (This == NULL) {
 295         status = U_MEMORY_ALLOCATION_ERROR;
 296         return NULL;
 297     }
 298     if (U_FAILURE(This->fDeferredStatus)) {
 299         status = This->fDeferredStatus;
 300         delete This;
 301         return NULL;
 302     }
 303     This->fFlags = flags;
 304
 305     RegexCompile     compiler(This, status);
 306     compiler.compile(regex, pe, status);
 307
 308     if (U_FAILURE(status)) {
 309         delete This;
 310         This = NULL;
 311     }
 312
 313     return This;
 314 }
 315
 316
 317 //
 318 //   compile, UText mode
 319 //
 320 RegexPattern * U_EXPORT2
 321 RegexPattern::compile(UText                *regex,
 322                       uint32_t             flags,
 323                       UParseError          &pe,
 324                       UErrorCode           &status)
 325 {
 326     if (U_FAILURE(status)) {
 327         return NULL;
 328     }
 329
 330     const uint32_t allFlags = UREGEX_CANON_EQ | UREGEX_CASE_INSENSITIVE | UREGEX_COMMENTS |
 331                               UREGEX_DOTALL   | UREGEX_MULTILINE        | UREGEX_UWORD |
 332                               UREGEX_ERROR_ON_UNKNOWN_ESCAPES           | UREGEX_UNIX_LINES | UREGEX_LITERAL;
 333
 334     if ((flags & ~allFlags) != 0) {
 335         status = U_REGEX_INVALID_FLAG;
 336         return NULL;
 337     }
 338
 339     if ((flags & UREGEX_CANON_EQ) != 0) {
 340         status = U_REGEX_UNIMPLEMENTED;
 341         return NULL;
 342     }
 343
 344     RegexPattern *This = new RegexPattern;
 345     if (This == NULL) {
 346         status = U_MEMORY_ALLOCATION_ERROR;
 347         return NULL;
 348     }
 349     if (U_FAILURE(This->fDeferredStatus)) {
 350         status = This->fDeferredStatus;
 351         delete This;
 352         return NULL;
 353     }
 354     This->fFlags = flags;
 355
 356     RegexCompile     compiler(This, status);
 357     compiler.compile(regex, pe, status);
 358
 359     if (U_FAILURE(status)) {
 360         delete This;
 361         This = NULL;
 362     }
 363
 364     return This;
 365 }
 366
 367 //
 368 //   compile with default flags.
 369 //
 370 RegexPattern * U_EXPORT2
 371 RegexPattern::compile(const UnicodeString &regex,
 372                       UParseError         &pe,
 373                       UErrorCode          &err)
 374 {
 375     return compile(regex, 0, pe, err);
 376 }
 377
 378
 379 //
 380 //   compile with default flags, UText mode
 381 //
 382 RegexPattern * U_EXPORT2
 383 RegexPattern::compile(UText               *regex,
 384                       UParseError         &pe,
 385                       UErrorCode          &err)
 386 {
 387     return compile(regex, 0, pe, err);
 388 }
 389
 390
 391 //
 392 //   compile with no UParseErr parameter.
 393 //
 394 RegexPattern * U_EXPORT2
 395 RegexPattern::compile(const UnicodeString &regex,
 396                       uint32_t             flags,
 397                       UErrorCode          &err)
 398 {
 399     UParseError pe;
 400     return compile(regex, flags, pe, err);
 401 }
 402
 403
 404 //
 405 //   compile with no UParseErr parameter, UText mode
 406 //
 407 RegexPattern * U_EXPORT2
 408 RegexPattern::compile(UText                *regex,
 409                       uint32_t             flags,
 410                       UErrorCode           &err)
 411 {
 412     UParseError pe;
 413     return compile(regex, flags, pe, err);
 414 }
 415
 416
 417 //---------------------------------------------------------------------
 418 //
 419 //   flags
 420 //
 421 //---------------------------------------------------------------------
 422 uint32_t RegexPattern::flags() const {
 423     return fFlags;
 424 }
 425
 426
 427 //---------------------------------------------------------------------
 428 //
 429 //   matcher(UnicodeString, err)
 430 //
 431 //---------------------------------------------------------------------
 432 RegexMatcher *RegexPattern::matcher(const UnicodeString &input,
 433                                     UErrorCode          &status)  const {
 434     RegexMatcher    *retMatcher = matcher(status);
 435     if (retMatcher != NULL) {
 436         retMatcher->fDeferredStatus = status;
 437         retMatcher->reset(input);
 438     }
 439     return retMatcher;
 440 }
 441
 442
 443 //---------------------------------------------------------------------
 444 //
 445 //   matcher(status)
 446 //
 447 //---------------------------------------------------------------------
 448 RegexMatcher *RegexPattern::matcher(UErrorCode &status)  const {
 449     RegexMatcher    *retMatcher = NULL;
 450
 451     if (U_FAILURE(status)) {
 452         return NULL;
 453     }
 454     if (U_FAILURE(fDeferredStatus)) {
 455         status = fDeferredStatus;
 456         return NULL;
 457     }
 458
 459     retMatcher = new RegexMatcher(this);
 460     if (retMatcher == NULL) {
 461         status = U_MEMORY_ALLOCATION_ERROR;
 462         return NULL;
 463     }
 464     return retMatcher;
 465 }
 466
 467
 468
 469 //---------------------------------------------------------------------
 470 //
 471 //   matches        Convenience function to test for a match, starting
 472 //                  with a pattern string and a data string.
 473 //
 474 //---------------------------------------------------------------------
 475 UBool U_EXPORT2 RegexPattern::matches(const UnicodeString   &regex,
 476               const UnicodeString   &input,
 477                     UParseError     &pe,
 478                     UErrorCode      &status) {
 479
 480     if (U_FAILURE(status)) {return FALSE;}
 481
 482     UBool         retVal;
 483     RegexPattern *pat     = NULL;
 484     RegexMatcher *matcher = NULL;
 485
 486     pat     = RegexPattern::compile(regex, 0, pe, status);
 487     matcher = pat->matcher(input, status);
 488     retVal  = matcher->matches(status);
 489
 490     delete matcher;
 491     delete pat;
 492     return retVal;
 493 }
 494
 495
 496 //
 497 //   matches, UText mode
 498 //
 499 UBool U_EXPORT2 RegexPattern::matches(UText                *regex,
 500                     UText           *input,
 501                     UParseError     &pe,
 502                     UErrorCode      &status) {
 503
 504     if (U_FAILURE(status)) {return FALSE;}
 505
 506     UBool         retVal  = FALSE;
 507     RegexPattern *pat     = NULL;
 508     RegexMatcher *matcher = NULL;
 509
 510     pat     = RegexPattern::compile(regex, 0, pe, status);
 511     matcher = pat->matcher(status);
 512     if (U_SUCCESS(status)) {
 513         matcher->reset(input);
 514         retVal  = matcher->matches(status);
 515     }
 516
 517     delete matcher;
 518     delete pat;
 519     return retVal;
 520 }
 521
 522
 523
 524
 525
 526 //---------------------------------------------------------------------
 527 //
 528 //   pattern
 529 //
 530 //---------------------------------------------------------------------
 531 UnicodeString RegexPattern::pattern() const {
 532     if (fPatternString != NULL) {
 533         return *fPatternString;
 534     } else if (fPattern == NULL) {
 535         return UnicodeString();
 536     } else {
 537         UErrorCode status = U_ZERO_ERROR;
 538         int64_t nativeLen = utext_nativeLength(fPattern);
 539         int32_t len16 = utext_extract(fPattern, 0, nativeLen, NULL, 0, &status); // buffer overflow error
 540         UnicodeString result;
 541
 542         status = U_ZERO_ERROR;
 543         UChar *resultChars = result.getBuffer(len16);
 544         utext_extract(fPattern, 0, nativeLen, resultChars, len16, &status); // unterminated warning
 545         result.releaseBuffer(len16);
 546
 547         return result;
 548     }
 549 }
 550
 551
 552
 553
 554 //---------------------------------------------------------------------
 555 //
 556 //   patternText
 557 //
 558 //---------------------------------------------------------------------
 559 UText *RegexPattern::patternText(UErrorCode      &status) const {
 560     if (U_FAILURE(status)) {return NULL;}
 561     status = U_ZERO_ERROR;
 562
 563     if (fPattern != NULL) {
 564         return fPattern;
 565     } else {
 566         RegexStaticSets::initGlobals(&status);
 567         return RegexStaticSets::gStaticSets->fEmptyText;
 568     }
 569 }
 570
 571
 572
 573 //---------------------------------------------------------------------
 574 //
 575 //   split
 576 //
 577 //---------------------------------------------------------------------
 578 int32_t  RegexPattern::split(const UnicodeString &input,
 579         UnicodeString    dest[],
 580         int32_t          destCapacity,
 581         UErrorCode      &status) const
 582 {
 583     if (U_FAILURE(status)) {
 584         return 0;
 585     };
 586
 587     RegexMatcher  m(this);
 588     int32_t r = 0;
 589     // Check m's status to make sure all is ok.
 590     if (U_SUCCESS(m.fDeferredStatus)) {
 591         r = m.split(input, dest, destCapacity, status);
 592     }
 593     return r;
 594 }
 595
 596 //
 597 //   split, UText mode
 598 //
 599 int32_t  RegexPattern::split(UText *input,
 600         UText           *dest[],
 601         int32_t          destCapacity,
 602         UErrorCode      &status) const
 603 {
 604     if (U_FAILURE(status)) {
 605         return 0;
 606     };
 607
 608     RegexMatcher  m(this);
 609     int32_t r = 0;
 610     // Check m's status to make sure all is ok.
 611     if (U_SUCCESS(m.fDeferredStatus)) {
 612         r = m.split(input, dest, destCapacity, status);
 613     }
 614     return r;
 615 }
 616
 617
 618
 619 //---------------------------------------------------------------------
 620 //
 621 //   dump    Output the compiled form of the pattern.
 622 //           Debugging function only.
 623 //
 624 //---------------------------------------------------------------------
 625 void   RegexPattern::dumpOp(int32_t index) const {
 626     (void)index;  // Suppress warnings in non-debug build.
 627 #if defined(REGEX_DEBUG)
 628     static const char * const opNames[] = {URX_OPCODE_NAMES};
 629     int32_t op          = fCompiledPat->elementAti(index);
 630     int32_t val         = URX_VAL(op);
 631     int32_t type        = URX_TYPE(op);
 632     int32_t pinnedType  = type;
 633     if ((uint32_t)pinnedType >= sizeof(opNames)/sizeof(char *)) {
 634         pinnedType = 0;
 635     }
 636
 637     printf("%4d   %08x    %-15s  ", index, op, opNames[pinnedType]);
 638     switch (type) {
 639     case URX_NOP:
 640     case URX_DOTANY:
 641     case URX_DOTANY_ALL:
 642     case URX_FAIL:
 643     case URX_CARET:
 644     case URX_DOLLAR:
 645     case URX_BACKSLASH_G:
 646     case URX_BACKSLASH_X:
 647     case URX_END:
 648     case URX_DOLLAR_M:
 649     case URX_CARET_M:
 650         // Types with no operand field of interest.
 651         break;
 652
 653     case URX_RESERVED_OP:
 654     case URX_START_CAPTURE:
 655     case URX_END_CAPTURE:
 656     case URX_STATE_SAVE:
 657     case URX_JMP:
 658     case URX_JMP_SAV:
 659     case URX_JMP_SAV_X:
 660     case URX_BACKSLASH_B:
 661     case URX_BACKSLASH_BU:
 662     case URX_BACKSLASH_D:
 663     case URX_BACKSLASH_Z:
 664     case URX_STRING_LEN:
 665     case URX_CTR_INIT:
 666     case URX_CTR_INIT_NG:
 667     case URX_CTR_LOOP:
 668     case URX_CTR_LOOP_NG:
 669     case URX_RELOC_OPRND:
 670     case URX_STO_SP:
 671     case URX_LD_SP:
 672     case URX_BACKREF:
 673     case URX_STO_INP_LOC:
 674     case URX_JMPX:
 675     case URX_LA_START:
 676     case URX_LA_END:
 677     case URX_BACKREF_I:
 678     case URX_LB_START:
 679     case URX_LB_CONT:
 680     case URX_LB_END:
 681     case URX_LBN_CONT:
 682     case URX_LBN_END:
 683     case URX_LOOP_C:
 684     case URX_LOOP_DOT_I:
 685         // types with an integer operand field.
 686         printf("%d", val);
 687         break;
 688
 689     case URX_ONECHAR:
 690     case URX_ONECHAR_I:
 691         printf("%c", val<256?val:'?');
 692         break;
 693
 694     case URX_STRING:
 695     case URX_STRING_I:
 696         {
 697             int32_t lengthOp       = fCompiledPat->elementAti(index+1);
 698             U_ASSERT(URX_TYPE(lengthOp) == URX_STRING_LEN);
 699             int32_t length = URX_VAL(lengthOp);
 700             int32_t i;
 701             for (i=val; i<val+length; i++) {
 702                 UChar c = fLiteralText[i];
 703                 if (c < 32 || c >= 256) {c = '.';}
 704                 printf("%c", c);
 705             }
 706         }
 707         break;
 708
 709     case URX_SETREF:
 710     case URX_LOOP_SR_I:
 711         {
 712             UnicodeString s;
 713             UnicodeSet *set = (UnicodeSet *)fSets->elementAt(val);
 714             set->toPattern(s, TRUE);
 715             for (int32_t i=0; i<s.length(); i++) {
 716                 printf("%c", s.charAt(i));
 717             }
 718         }
 719         break;
 720
 721     case URX_STATIC_SETREF:
 722     case URX_STAT_SETREF_N:
 723         {
 724             UnicodeString s;
 725             if (val & URX_NEG_SET) {
 726                 printf("NOT ");
 727                 val &= ~URX_NEG_SET;
 728             }
 729             UnicodeSet *set = fStaticSets[val];
 730             set->toPattern(s, TRUE);
 731             for (int32_t i=0; i<s.length(); i++) {
 732                 printf("%c", s.charAt(i));
 733             }
 734         }
 735         break;
 736
 737
 738     default:
 739         printf("??????");
 740         break;
 741     }
 742     printf("\n");
 743 #endif
 744 }
 745
 746
 747 void RegexPattern::dumpPattern() const {
 748 #if defined(REGEX_DEBUG)
 749     int      index;
 750     int      i;
 751
 752     printf("Original Pattern:  ");
 753     UChar32 c = utext_next32From(fPattern, 0);
 754     while (c != U_SENTINEL) {
 755         if (c<32 || c>256) {
 756             c = '.';
 757         }
 758         printf("%c", c);
 759
 760         c = UTEXT_NEXT32(fPattern);
 761     }
 762     printf("\n");
 763     printf("   Min Match Length:  %d\n", fMinMatchLen);
 764     printf("   Match Start Type:  %s\n", START_OF_MATCH_STR(fStartType));
 765     if (fStartType == START_STRING) {
 766         printf("    Initial match string: \"");
 767         for (i=fInitialStringIdx; i<fInitialStringIdx+fInitialStringLen; i++) {
 768             printf("%c", fLiteralText[i]);   // TODO:  non-printables, surrogates.
 769         }
 770         printf("\"\n");
 771
 772     } else if (fStartType == START_SET) {
 773         int32_t numSetChars = fInitialChars->size();
 774         if (numSetChars > 20) {
 775             numSetChars = 20;
 776         }
 777         printf("     Match First Chars : ");
 778         for (i=0; i<numSetChars; i++) {
 779             UChar32 c = fInitialChars->charAt(i);
 780             if (0x20<c && c <0x7e) {
 781                 printf("%c ", c);
 782             } else {
 783                 printf("%#x ", c);
 784             }
 785         }
 786         if (numSetChars < fInitialChars->size()) {
 787             printf(" ...");
 788         }
 789         printf("\n");
 790
 791     } else if (fStartType == START_CHAR) {
 792         printf("    First char of Match : ");
 793         if (0x20 < fInitialChar && fInitialChar<0x7e) {
 794                 printf("%c\n", fInitialChar);
 795             } else {
 796                 printf("%#x\n", fInitialChar);
 797             }
 798     }
 799
 800     printf("\nIndex   Binary     Type             Operand\n" \
 801            "-------------------------------------------\n");
 802     for (index = 0; index<fCompiledPat->size(); index++) {
 803         dumpOp(index);
 804     }
 805     printf("\n\n");
 806 #endif
 807 }
 808
 809
 810
 811 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(RegexPattern)
 812
 813 U_NAMESPACE_END
 814 #endif  // !UCONFIG_NO_REGULAR_EXPRESSIONS