icuSources/i18n/unicode/uregex.h

   1 /*
   2 **********************************************************************
   3 *   Copyright (C) 2004-2010, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 **********************************************************************
   6 *   file name:  uregex.h
   7 *   encoding:   US-ASCII
   8 *   indentation:4
   9 *
  10 *   created on: 2004mar09
  11 *   created by: Andy Heninger
  12 *
  13 *   ICU Regular Expressions, API for C
  14 */
  15
  16 /**
  17  * \file
  18  * \brief C API: Regular Expressions
  19  *
  20  * <p>This is a C wrapper around the C++ RegexPattern and RegexMatcher classes.</p>
  21  */
  22
  23 #ifndef UREGEX_H
  24 #define UREGEX_H
  25
  26 #include "unicode/utext.h"
  27 #include "unicode/utypes.h"
  28
  29 #if !UCONFIG_NO_REGULAR_EXPRESSIONS
  30
  31 #include "unicode/localpointer.h"
  32 #include "unicode/parseerr.h"
  33
  34 struct URegularExpression;
  35 /**
  36   * Structure representing a compiled regular rexpression, plus the results
  37   *    of a match operation.
  38   * @stable ICU 3.0
  39   */
  40 typedef struct URegularExpression URegularExpression;
  41
  42
  43 /**
  44  * Constants for Regular Expression Match Modes.
  45  * @stable ICU 2.4
  46  */
  47 typedef enum URegexpFlag{
  48
  49 #ifndef U_HIDE_DRAFT_API
  50     /** Forces normalization of pattern and strings.
  51     Not implemented yet, just a placeholder, hence draft.
  52     @draft ICU 2.4 */
  53     UREGEX_CANON_EQ         = 128,
  54 #endif
  55     /**  Enable case insensitive matching.  @stable ICU 2.4 */
  56     UREGEX_CASE_INSENSITIVE = 2,
  57
  58     /**  Allow white space and comments within patterns  @stable ICU 2.4 */
  59     UREGEX_COMMENTS         = 4,
  60
  61     /**  If set, '.' matches line terminators,  otherwise '.' matching stops at line end.
  62       *  @stable ICU 2.4 */
  63     UREGEX_DOTALL           = 32,
  64
  65     /**  If set, treat the entire pattern as a literal string.
  66       *  Metacharacters or escape sequences in the input sequence will be given
  67       *  no special meaning. Not implemented yet as of ICU 4.4.
  68       *
  69       *  The flags CASE_INSENSITIVE and UNICODE_CASE retain their impact
  70       *  on matching when used in conjunction with this flag.
  71       *  The other flags become superfluous.
  72       *  TODO:  say which escapes are still handled; anything Java does
  73       *         early (\\u) we should still do.
  74       * @stable ICU 4.0
  75       */
  76     UREGEX_LITERAL = 16,
  77
  78     /**   Control behavior of "$" and "^"
  79       *    If set, recognize line terminators within string,
  80       *    otherwise, match only at start and end of input string.
  81       *   @stable ICU 2.4 */
  82     UREGEX_MULTILINE        = 8,
  83
  84     /**   Unix-only line endings.
  85       *   When this mode is enabled, only \\u000a is recognized as a line ending
  86       *    in the behavior of ., ^, and $.
  87       *   @stable ICU 4.0
  88       */
  89     UREGEX_UNIX_LINES = 1,
  90
  91     /**  Unicode word boundaries.
  92       *     If set, \b uses the Unicode TR 29 definition of word boundaries.
  93       *     Warning: Unicode word boundaries are quite different from
  94       *     traditional regular expression word boundaries.  See
  95       *     http://unicode.org/reports/tr29/#Word_Boundaries
  96       *     @stable ICU 2.8
  97       */
  98     UREGEX_UWORD            = 256,
  99
 100      /**  Error on Unrecognized backslash escapes.
 101        *     If set, fail with an error on patterns that contain
 102        *     backslash-escaped ASCII letters without a known specail
 103        *     meaning.  If this flag is not set, these
 104        *     escaped letters represent themselves.
 105        *     @stable ICU 4.0
 106        */
 107      UREGEX_ERROR_ON_UNKNOWN_ESCAPES = 512
 108
 109 }  URegexpFlag;
 110
 111 /**
 112   *  Open (compile) an ICU regular expression.  Compiles the regular expression in
 113   *  string form into an internal representation using the specified match mode flags.
 114   *  The resulting regular expression handle can then be used to perform various
 115   *   matching operations.
 116   *
 117   *
 118   * @param pattern        The Regular Expression pattern to be compiled.
 119   * @param patternLength  The length of the pattern, or -1 if the pattern is
 120   *                       NUL termintated.
 121   * @param flags          Flags that alter the default matching behavior for
 122   *                       the regular expression, UREGEX_CASE_INSENSITIVE, for
 123   *                       example.  For default behavior, set this parameter to zero.
 124   *                       See <code>enum URegexpFlag</code>.  All desired flags
 125   *                       are bitwise-ORed together.
 126   * @param pe             Receives the position (line and column nubers) of any syntax
 127   *                       error within the source regular expression string.  If this
 128   *                       information is not wanted, pass NULL for this parameter.
 129   * @param status         Receives error detected by this function.
 130   * @stable ICU 3.0
 131   *
 132   */
 133 U_STABLE URegularExpression * U_EXPORT2
 134 uregex_open( const  UChar          *pattern,
 135                     int32_t         patternLength,
 136                     uint32_t        flags,
 137                     UParseError    *pe,
 138                     UErrorCode     *status);
 139
 140 /**
 141   *  Open (compile) an ICU regular expression.  Compiles the regular expression in
 142   *  string form into an internal representation using the specified match mode flags.
 143   *  The resulting regular expression handle can then be used to perform various
 144   *   matching operations.
 145   *  <p>
 146   *  The contents of the pattern UText will be extracted and saved. Ownership of the
 147   *   UText struct itself remains with the caller. This is to match the behavior of
 148   *   uregex_open().
 149   *
 150   * @param pattern        The Regular Expression pattern to be compiled.
 151   * @param flags          Flags that alter the default matching behavior for
 152   *                       the regular expression, UREGEX_CASE_INSENSITIVE, for
 153   *                       example.  For default behavior, set this parameter to zero.
 154   *                       See <code>enum URegexpFlag</code>.  All desired flags
 155   *                       are bitwise-ORed together.
 156   * @param pe             Receives the position (line and column nubers) of any syntax
 157   *                       error within the source regular expression string.  If this
 158   *                       information is not wanted, pass NULL for this parameter.
 159   * @param status         Receives error detected by this function.
 160   *
 161   * @draft ICU 4.6
 162   */
 163 U_DRAFT URegularExpression *  U_EXPORT2
 164 uregex_openUText(UText          *pattern,
 165                  uint32_t        flags,
 166                  UParseError    *pe,
 167                  UErrorCode     *status);
 168
 169 /**
 170   *  Open (compile) an ICU regular expression.  The resulting regular expression
 171   *   handle can then be used to perform various matching operations.
 172   *  <p>
 173   *   This function is the same as uregex_open, except that the pattern
 174   *   is supplied as an 8 bit char * string in the default code page.
 175   *
 176   * @param pattern        The Regular Expression pattern to be compiled,
 177   *                       NUL termintated.
 178   * @param flags          Flags that alter the default matching behavior for
 179   *                       the regular expression, UREGEX_CASE_INSENSITIVE, for
 180   *                       example.  For default behavior, set this parameter to zero.
 181   *                       See <code>enum URegexpFlag</code>.  All desired flags
 182   *                       are bitwise-ORed together.
 183   * @param pe             Receives the position (line and column nubers) of any syntax
 184   *                       error within the source regular expression string.  If this
 185   *                       information is not wanted, pass NULL for this parameter.
 186   * @param status         Receives errors detected by this function.
 187   * @return               The URegularExpression object representing the compiled
 188   *                       pattern.
 189   *
 190   * @stable ICU 3.0
 191   */
 192 #if !UCONFIG_NO_CONVERSION
 193 U_STABLE URegularExpression * U_EXPORT2
 194 uregex_openC( const char           *pattern,
 195                     uint32_t        flags,
 196                     UParseError    *pe,
 197                     UErrorCode     *status);
 198 #endif
 199
 200
 201
 202 /**
 203   *  Close the regular expression, recovering all resources (memory) it
 204   *   was holding.
 205   *
 206   * @param regexp   The regular expression to be closed.
 207   * @stable ICU 3.0
 208   */
 209 U_STABLE void U_EXPORT2
 210 uregex_close(URegularExpression *regexp);
 211
 212 #if U_SHOW_CPLUSPLUS_API
 213
 214 U_NAMESPACE_BEGIN
 215
 216 /**
 217  * \class LocalURegularExpressionPointer
 218  * "Smart pointer" class, closes a URegularExpression via uregex_close().
 219  * For most methods see the LocalPointerBase base class.
 220  *
 221  * @see LocalPointerBase
 222  * @see LocalPointer
 223  * @stable ICU 4.4
 224  */
 225 U_DEFINE_LOCAL_OPEN_POINTER(LocalURegularExpressionPointer, URegularExpression, uregex_close);
 226
 227 U_NAMESPACE_END
 228
 229 #endif
 230
 231 /**
 232  * Make a copy of a compiled regular expression.  Cloning a regular
 233  * expression is faster than opening a second instance from the source
 234  * form of the expression, and requires less memory.
 235  * <p>
 236  * Note that the current input string and the position of any matched text
 237  *  within it are not cloned; only the pattern itself and and the
 238  *  match mode flags are copied.
 239  * <p>
 240  * Cloning can be particularly useful to threaded applications that perform
 241  * multiple match operations in parallel.  Each concurrent RE
 242  * operation requires its own instance of a URegularExpression.
 243  *
 244  * @param regexp   The compiled regular expression to be cloned.
 245  * @param status   Receives indication of any errors encountered
 246  * @return the cloned copy of the compiled regular expression.
 247  * @stable ICU 3.0
 248  */
 249 U_STABLE URegularExpression * U_EXPORT2
 250 uregex_clone(const URegularExpression *regexp, UErrorCode *status);
 251
 252 /**
 253  *  Returns a pointer to the source form of the pattern for this regular expression.
 254  *  This function will work even if the pattern was originally specified as a UText.
 255  *
 256  * @param regexp     The compiled regular expression.
 257  * @param patLength  This output parameter will be set to the length of the
 258  *                   pattern string.  A NULL pointer may be used here if the
 259  *                   pattern length is not needed, as would be the case if
 260  *                   the pattern is known in advance to be a NUL terminated
 261  *                   string.
 262  * @param status     Receives errors detected by this function.
 263  * @return a pointer to the pattern string.  The storage for the string is
 264  *                   owned by the regular expression object, and must not be
 265  *                   altered or deleted by the application.  The returned string
 266  *                   will remain valid until the regular expression is closed.
 267  * @stable ICU 3.0
 268  */
 269 U_STABLE const UChar * U_EXPORT2
 270 uregex_pattern(const URegularExpression *regexp,
 271                      int32_t            *patLength,
 272                      UErrorCode         *status);
 273
 274 /**
 275  *  Returns the source text of the pattern for this regular expression.
 276  *  This function will work even if the pattern was originally specified as a UChar string.
 277  *
 278  * @param regexp     The compiled regular expression.
 279  * @param status     Receives errors detected by this function.
 280  * @return the pattern text.  The storage for the text is owned by the regular expression
 281  *                   object, and must not be altered or deleted.
 282  *
 283  * @draft ICU 4.6
 284  */
 285 U_DRAFT UText * U_EXPORT2
 286 uregex_patternUText(const URegularExpression *regexp,
 287                           UErrorCode         *status);
 288
 289
 290 /**
 291   * Get the match mode flags that were specified when compiling this regular expression.
 292   * @param status   Receives errors detected by this function.
 293   * @param regexp   The compiled regular expression.
 294   * @return         The match mode flags
 295   * @see URegexpFlag
 296   * @stable ICU 3.0
 297   */
 298 U_STABLE int32_t U_EXPORT2
 299 uregex_flags(const  URegularExpression   *regexp,
 300                     UErrorCode           *status);
 301
 302
 303 /**
 304   *  Set the subject text string upon which the regular expression will look for matches.
 305   *  This function may be called any number of times, allowing the regular
 306   *  expression pattern to be applied to different strings.
 307   *  <p>
 308   *  Regular expression matching operations work directly on the application's
 309   *  string data.  No copy is made.  The subject string data must not be
 310   *  altered after calling this function until after all regular expression
 311   *  operations involving this string data are completed.
 312   *  <p>
 313   *  Zero length strings are permitted.  In this case, no subsequent match
 314   *  operation will dereference the text string pointer.
 315   *
 316   * @param regexp     The compiled regular expression.
 317   * @param text       The subject text string.
 318   * @param textLength The length of the subject text, or -1 if the string
 319   *                   is NUL terminated.
 320   * @param status     Receives errors detected by this function.
 321   * @stable ICU 3.0
 322   */
 323 U_STABLE void U_EXPORT2
 324 uregex_setText(URegularExpression *regexp,
 325                const UChar        *text,
 326                int32_t             textLength,
 327                UErrorCode         *status);
 328
 329
 330 /**
 331   *  Set the subject text string upon which the regular expression will look for matches.
 332   *  This function may be called any number of times, allowing the regular
 333   *  expression pattern to be applied to different strings.
 334   *  <p>
 335   *  Regular expression matching operations work directly on the application's
 336   *  string data; only a shallow clone is made.  The subject string data must not be
 337   *  altered after calling this function until after all regular expression
 338   *  operations involving this string data are completed.
 339   *
 340   * @param regexp     The compiled regular expression.
 341   * @param text       The subject text string.
 342   * @param status     Receives errors detected by this function.
 343   *
 344   * @draft ICU 4.6
 345   */
 346 U_DRAFT void U_EXPORT2
 347 uregex_setUText(URegularExpression *regexp,
 348                 UText              *text,
 349                 UErrorCode         *status);
 350
 351 /**
 352   *  Get the subject text that is currently associated with this
 353   *   regular expression object.  If the input was supplied using uregex_setText(),
 354   *   that pointer will be returned.  Otherwise, the characters in the input will
 355   *   be extracted to a buffer and returned.  In either case, ownership remains
 356   *   with the regular expression object.
 357   *
 358   *  This function will work even if the input was originally specified as a UText.
 359   *
 360   * @param regexp      The compiled regular expression.
 361   * @param textLength  The length of the string is returned in this output parameter.
 362   *                    A NULL pointer may be used here if the
 363   *                    text length is not needed, as would be the case if
 364   *                    the text is known in advance to be a NUL terminated
 365   *                    string.
 366   * @param status      Receives errors detected by this function.
 367   * @return            Pointer to the subject text string currently associated with
 368   *                    this regular expression.
 369   * @stable ICU 3.0
 370   */
 371 U_STABLE const UChar * U_EXPORT2
 372 uregex_getText(URegularExpression *regexp,
 373                int32_t            *textLength,
 374                UErrorCode         *status);
 375
 376
 377 /**
 378   *  Get the subject text that is currently associated with this
 379   *   regular expression object.
 380   *
 381   *  This function will work even if the input was originally specified as a UChar string.
 382   *
 383   * @param regexp      The compiled regular expression.
 384   * @param dest        A mutable UText in which to store the current input.
 385   *                    If NULL, a new UText will be created as an immutable shallow clone
 386   *                    of the actual input string.
 387   * @param status      Receives errors detected by this function.
 388   * @return            The subject text currently associated with this regular expression.
 389   *                    If a pre-allocated UText was provided, it will always be used and returned.
 390   *
 391   * @draft ICU 4.6
 392   */
 393 U_DRAFT UText * U_EXPORT2
 394 uregex_getUText(URegularExpression *regexp,
 395                 UText              *dest,
 396                 UErrorCode         *status);
 397
 398 /**
 399   *   Attempts to match the input string against the pattern.
 400   *   To succeed, the match must extend to the end of the string,
 401   *   or cover the complete match region.
 402   *
 403   *   If startIndex >= zero the match operation starts at the specified
 404   *   index and must extend to the end of the input string.  Any region
 405   *   that has been specified is reset.
 406   *
 407   *   If startIndex == -1 the match must cover the input region, or the entire
 408   *   input string if no region has been set.  This directly corresponds to
 409   *   Matcher.matches() in Java
 410   *
 411   *    @param  regexp      The compiled regular expression.
 412   *    @param  startIndex  The input string (native) index at which to begin matching, or -1
 413   *                        to match the input Region.
 414   *    @param  status      Receives errors detected by this function.
 415   *    @return             TRUE if there is a match
 416   *    @stable ICU 3.0
 417   */
 418 U_STABLE UBool U_EXPORT2
 419 uregex_matches(URegularExpression *regexp,
 420                 int32_t            startIndex,
 421                 UErrorCode        *status);
 422
 423 /**
 424   *   64bit version of uregex_matches.
 425   *   @draft ICU 4.6
 426   */
 427 U_DRAFT UBool U_EXPORT2
 428 uregex_matches64(URegularExpression *regexp,
 429                  int64_t            startIndex,
 430                  UErrorCode        *status);
 431
 432 /**
 433   *   Attempts to match the input string, starting from the specified index, against the pattern.
 434   *   The match may be of any length, and is not required to extend to the end
 435   *   of the input string.  Contrast with uregex_matches().
 436   *
 437   *   <p>If startIndex is >= 0 any input region that was set for this
 438   *   URegularExpression is reset before the operation begins.
 439   *
 440   *   <p>If the specified starting index == -1 the match begins at the start of the input
 441   *   region, or at the start of the full string if no region has been specified.
 442   *   This corresponds directly with Matcher.lookingAt() in Java.
 443   *
 444   *   <p>If the match succeeds then more information can be obtained via the
 445   *    <code>uregexp_start()</code>, <code>uregexp_end()</code>,
 446   *    and <code>uregexp_group()</code> functions.</p>
 447   *
 448   *    @param   regexp      The compiled regular expression.
 449   *    @param   startIndex  The input string (native) index at which to begin matching, or
 450   *                         -1 to match the Input Region
 451   *    @param   status      A reference to a UErrorCode to receive any errors.
 452   *    @return  TRUE if there is a match.
 453   *    @stable ICU 3.0
 454   */
 455 U_STABLE UBool U_EXPORT2
 456 uregex_lookingAt(URegularExpression *regexp,
 457                  int32_t             startIndex,
 458                  UErrorCode         *status);
 459
 460 /**
 461   *   64bit version of uregex_lookingAt.
 462   *   @draft ICU 4.6
 463   */
 464 U_DRAFT UBool U_EXPORT2
 465 uregex_lookingAt64(URegularExpression *regexp,
 466                    int64_t             startIndex,
 467                    UErrorCode         *status);
 468
 469 /**
 470   *   Find the first matching substring of the input string that matches the pattern.
 471   *   If startIndex is >= zero the search for a match begins at the specified index,
 472   *          and any match region is reset.  This corresponds directly with
 473   *          Matcher.find(startIndex) in Java.
 474   *
 475   *   If startIndex == -1 the search begins at the start of the input region,
 476   *           or at the start of the full string if no region has been specified.
 477   *
 478   *   If a match is found, <code>uregex_start(), uregex_end()</code>, and
 479   *   <code>uregex_group()</code> will provide more information regarding the match.
 480   *
 481   *   @param   regexp      The compiled regular expression.
 482   *   @param   startIndex  The position (native) in the input string to begin the search, or
 483   *                        -1 to search within the Input Region.
 484   *   @param   status      A reference to a UErrorCode to receive any errors.
 485   *   @return              TRUE if a match is found.
 486   *   @stable ICU 3.0
 487   */
 488 U_STABLE UBool U_EXPORT2
 489 uregex_find(URegularExpression *regexp,
 490             int32_t             startIndex,
 491             UErrorCode         *status);
 492
 493 /**
 494   *   64bit version of uregex_find.
 495   *   @draft ICU 4.6
 496   */
 497 U_DRAFT UBool U_EXPORT2
 498 uregex_find64(URegularExpression *regexp,
 499               int64_t             startIndex,
 500               UErrorCode         *status);
 501
 502 /**
 503   *  Find the next pattern match in the input string.  Begin searching
 504   *  the input at the location following the end of he previous match,
 505   *  or at the start of the string (or region) if there is no
 506   *  previous match.  If a match is found, <code>uregex_start(), uregex_end()</code>, and
 507   *  <code>uregex_group()</code> will provide more information regarding the match.
 508   *
 509   *  @param   regexp      The compiled regular expression.
 510   *  @param   status      A reference to a UErrorCode to receive any errors.
 511   *  @return              TRUE if a match is found.
 512   *  @see uregex_reset
 513   *  @stable ICU 3.0
 514   */
 515 U_STABLE UBool U_EXPORT2
 516 uregex_findNext(URegularExpression *regexp,
 517                 UErrorCode         *status);
 518
 519 /**
 520   *   Get the number of capturing groups in this regular expression's pattern.
 521   *   @param   regexp      The compiled regular expression.
 522   *   @param   status      A reference to a UErrorCode to receive any errors.
 523   *   @return the number of capture groups
 524   *   @stable ICU 3.0
 525   */
 526 U_STABLE int32_t U_EXPORT2
 527 uregex_groupCount(URegularExpression *regexp,
 528                   UErrorCode         *status);
 529
 530 /** Extract the string for the specified matching expression or subexpression.
 531   * Group #0 is the complete string of matched text.
 532   * Group #1 is the text matched by the first set of capturing parentheses.
 533   *
 534   *   @param   regexp       The compiled regular expression.
 535   *   @param   groupNum     The capture group to extract.  Group 0 is the complete
 536   *                         match.  The value of this parameter must be
 537   *                         less than or equal to the number of capture groups in
 538   *                         the pattern.
 539   *   @param   dest         Buffer to receive the matching string data
 540   *   @param   destCapacity Capacity of the dest buffer.
 541   *   @param   status       A reference to a UErrorCode to receive any errors.
 542   *   @return               Length of matching data,
 543   *                         or -1 if no applicable match.
 544   *   @stable ICU 3.0
 545   */
 546 U_STABLE int32_t U_EXPORT2
 547 uregex_group(URegularExpression *regexp,
 548              int32_t             groupNum,
 549              UChar              *dest,
 550              int32_t             destCapacity,
 551              UErrorCode          *status);
 552
 553
 554 /** Returns a shallow immutable clone of the entire input string.  The returned UText current native index
 555   *   is set to the beginning of the requested capture group.  The capture group length is also
 556   *   returned via groupLength.
 557   * Group #0 is the complete string of matched text.
 558   * Group #1 is the text matched by the first set of capturing parentheses.
 559   *
 560   *   @param   regexp       The compiled regular expression.
 561   *   @param   groupNum     The capture group to extract.  Group 0 is the complete
 562   *                         match.  The value of this parameter must be
 563   *                         less than or equal to the number of capture groups in
 564   *                         the pattern.
 565   *   @param   dest         A mutable UText in which to store the current input.
 566   *                         If NULL, a new UText will be created as an immutable shallow clone
 567   *                         of the entire input string.
 568   *   @param   groupLength  The group length of the desired capture group.
 569   *   @param   status       A reference to a UErrorCode to receive any errors.
 570   *   @return               The subject text currently associated with this regular expression.
 571   *                         If a pre-allocated UText was provided, it will always be used and returned.
 572
 573   *
 574   *   @draft ICU 4.6
 575   */
 576 U_DRAFT UText * U_EXPORT2
 577 uregex_groupUText(URegularExpression *regexp,
 578                   int32_t             groupNum,
 579                   UText              *dest,
 580                   int64_t            *groupLength,
 581                   UErrorCode         *status);
 582
 583
 584 /** Extract the string for the specified matching expression or subexpression.
 585   * Group #0 is the complete string of matched text.
 586   * Group #1 is the text matched by the first set of capturing parentheses.
 587   *
 588   *   @param   regexp       The compiled regular expression.
 589   *   @param   groupNum     The capture group to extract.  Group 0 is the complete
 590   *                         match.  The value of this parameter must be
 591   *                         less than or equal to the number of capture groups in
 592   *                         the pattern.
 593   *   @param   dest         Mutable UText to receive the matching string data.
 594   *                         If NULL, a new UText will be created (which may not be mutable).
 595   *   @param   status       A reference to a UErrorCode to receive any errors.
 596   *   @return               The matching string data. If a pre-allocated UText was provided,
 597   *                          it will always be used and returned.
 598   *
 599   *   @internal ICU 4.4 technology preview
 600   */
 601 U_INTERNAL UText * U_EXPORT2
 602 uregex_groupUTextDeep(URegularExpression *regexp,
 603                   int32_t             groupNum,
 604                   UText              *dest,
 605                   UErrorCode         *status);
 606
 607 /**
 608   *   Returns the index in the input string of the start of the text matched by the
 609   *   specified capture group during the previous match operation.  Return -1 if
 610   *   the capture group was not part of the last match.
 611   *   Group #0 refers to the complete range of matched text.
 612   *   Group #1 refers to the text matched by the first set of capturing parentheses.
 613   *
 614   *    @param   regexp      The compiled regular expression.
 615   *    @param   groupNum    The capture group number
 616   *    @param   status      A reference to a UErrorCode to receive any errors.
 617   *    @return              the starting (native) position in the input of the text matched
 618   *                         by the specified group.
 619   *    @stable ICU 3.0
 620   */
 621 U_STABLE int32_t U_EXPORT2
 622 uregex_start(URegularExpression *regexp,
 623              int32_t             groupNum,
 624              UErrorCode          *status);
 625
 626 /**
 627   *   64bit version of uregex_start.
 628   *   @draft ICU 4.6
 629   */
 630 U_DRAFT int64_t U_EXPORT2
 631 uregex_start64(URegularExpression *regexp,
 632                int32_t             groupNum,
 633                UErrorCode          *status);
 634
 635 /**
 636   *   Returns the index in the input string of the position following the end
 637   *   of the text matched by the specified capture group.
 638   *   Return -1 if the capture group was not part of the last match.
 639   *   Group #0 refers to the complete range of matched text.
 640   *   Group #1 refers to the text matched by the first set of capturing parentheses.
 641   *
 642   *    @param   regexp      The compiled regular expression.
 643   *    @param   groupNum    The capture group number
 644   *    @param   status      A reference to a UErrorCode to receive any errors.
 645   *    @return              the (native) index of the position following the last matched character.
 646   *    @stable ICU 3.0
 647   */
 648 U_STABLE int32_t U_EXPORT2
 649 uregex_end(URegularExpression   *regexp,
 650            int32_t               groupNum,
 651            UErrorCode           *status);
 652
 653 /**
 654   *   64bit version of uregex_end.
 655   *   @draft ICU 4.6
 656   */
 657 U_DRAFT int64_t U_EXPORT2
 658 uregex_end64(URegularExpression *regexp,
 659              int32_t               groupNum,
 660              UErrorCode           *status);
 661
 662 /**
 663   *  Reset any saved state from the previous match.  Has the effect of
 664   *  causing uregex_findNext to begin at the specified index, and causing
 665   *  uregex_start(), uregex_end() and uregex_group() to return an error
 666   *  indicating that there is no match information available.  Clears any
 667   *  match region that may have been set.
 668   *
 669   *    @param   regexp      The compiled regular expression.
 670   *    @param   index       The position (native) in the text at which a
 671   *                         uregex_findNext() should begin searching.
 672   *    @param   status      A reference to a UErrorCode to receive any errors.
 673   *    @stable ICU 3.0
 674   */
 675 U_STABLE void U_EXPORT2
 676 uregex_reset(URegularExpression    *regexp,
 677              int32_t               index,
 678              UErrorCode            *status);
 679
 680 /**
 681   *   64bit version of uregex_reset.
 682   *   @draft ICU 4.6
 683   */
 684 U_DRAFT void U_EXPORT2
 685 uregex_reset64(URegularExpression  *regexp,
 686                int64_t               index,
 687                UErrorCode            *status);
 688
 689 /** Sets the limits of the matching region for this URegularExpression.
 690   * The region is the part of the input string that will be considered when matching.
 691   * Invoking this method resets any saved state from the previous match,
 692   * then sets the region to start at the index specified by the start parameter
 693   * and end at the index specified by the end parameter.
 694   *
 695   * Depending on the transparency and anchoring being used (see useTransparentBounds
 696   * and useAnchoringBounds), certain constructs such as anchors may behave differently
 697   * at or around the boundaries of the region
 698   *
 699   * The function will fail if start is greater than limit, or if either index
 700   *  is less than zero or greater than the length of the string being matched.
 701   *
 702   * @param regexp The compiled regular expression.
 703   * @param regionStart  The (native) index to begin searches at.
 704   * @param regionLimit  The (native) index to end searches at (exclusive).
 705   * @param status A pointer to a UErrorCode to receive any errors.
 706   * @stable ICU 4.0
 707   */
 708 U_STABLE void U_EXPORT2
 709 uregex_setRegion(URegularExpression   *regexp,
 710                  int32_t               regionStart,
 711                  int32_t               regionLimit,
 712                  UErrorCode           *status);
 713
 714 /**
 715   *   64bit version of uregex_setRegion.
 716   *   @draft ICU 4.6
 717   */
 718 U_DRAFT void U_EXPORT2
 719 uregex_setRegion64(URegularExpression *regexp,
 720                  int64_t               regionStart,
 721                  int64_t               regionLimit,
 722                  UErrorCode           *status);
 723
 724 /**
 725   *   Variation on uregex_setRegion to set the region without resetting the start index
 726   *     without resetting the position for subsequent matches.
 727   *   @draft ICU 4.6
 728   */
 729 U_DRAFT void U_EXPORT2
 730 uregex_setRegionAndStart(URegularExpression *regexp,
 731                  int64_t               regionStart,
 732                  int64_t               regionLimit,
 733                  int64_t               startIndex,
 734                  UErrorCode           *status);
 735
 736 /**
 737   * Reports the start index of the matching region. Any matches found are limited to
 738   * to the region bounded by regionStart (inclusive) and regionEnd (exclusive).
 739   *
 740   * @param regexp The compiled regular expression.
 741   * @param status A pointer to a UErrorCode to receive any errors.
 742   * @return The starting (native) index of this matcher's region.
 743   * @stable ICU 4.0
 744   */
 745 U_STABLE int32_t U_EXPORT2
 746 uregex_regionStart(const  URegularExpression   *regexp,
 747                           UErrorCode           *status);
 748
 749 /**
 750   *   64bit version of uregex_regionStart.
 751   *   @draft ICU 4.6
 752   */
 753 U_DRAFT int64_t U_EXPORT2
 754 uregex_regionStart64(const  URegularExpression   *regexp,
 755                             UErrorCode           *status);
 756
 757 /**
 758   * Reports the end index (exclusive) of the matching region for this URegularExpression.
 759   * Any matches found are limited to to the region bounded by regionStart (inclusive)
 760   * and regionEnd (exclusive).
 761   *
 762   * @param regexp The compiled regular expression.
 763   * @param status A pointer to a UErrorCode to receive any errors.
 764   * @return The ending point (native) of this matcher's region.
 765   * @stable ICU 4.0
 766   */
 767 U_STABLE int32_t U_EXPORT2
 768 uregex_regionEnd(const  URegularExpression   *regexp,
 769                         UErrorCode           *status);
 770
 771 /**
 772   *   64bit version of uregex_regionEnd.
 773   *   @draft ICU 4.6
 774   */
 775 U_DRAFT int64_t U_EXPORT2
 776 uregex_regionEnd64(const  URegularExpression   *regexp,
 777                           UErrorCode           *status);
 778
 779 /**
 780   * Queries the transparency of region bounds for this URegularExpression.
 781   * See useTransparentBounds for a description of transparent and opaque bounds.
 782   * By default, matching boundaries are opaque.
 783   *
 784   * @param regexp The compiled regular expression.
 785   * @param status A pointer to a UErrorCode to receive any errors.
 786   * @return TRUE if this matcher is using opaque bounds, false if it is not.
 787   * @stable ICU 4.0
 788   */
 789 U_STABLE UBool U_EXPORT2
 790 uregex_hasTransparentBounds(const  URegularExpression   *regexp,
 791                                    UErrorCode           *status);
 792
 793
 794 /**
 795   * Sets the transparency of region bounds for this URegularExpression.
 796   * Invoking this function with an argument of TRUE will set matches to use transparent bounds.
 797   * If the boolean argument is FALSE, then opaque bounds will be used.
 798   *
 799   * Using transparent bounds, the boundaries of the matching region are transparent
 800   * to lookahead, lookbehind, and boundary matching constructs. Those constructs can
 801   * see text beyond the boundaries of the region while checking for a match.
 802   *
 803   * With opaque bounds, no text outside of the matching region is visible to lookahead,
 804   * lookbehind, and boundary matching constructs.
 805   *
 806   * By default, opaque bounds are used.
 807   *
 808   * @param   regexp The compiled regular expression.
 809   * @param   b      TRUE for transparent bounds; FALSE for opaque bounds
 810   * @param   status A pointer to a UErrorCode to receive any errors.
 811   * @stable ICU 4.0
 812   **/
 813 U_STABLE void U_EXPORT2
 814 uregex_useTransparentBounds(URegularExpression   *regexp,
 815                             UBool                b,
 816                             UErrorCode           *status);
 817
 818
 819 /**
 820   * Return true if this URegularExpression is using anchoring bounds.
 821   * By default, anchoring region bounds are used.
 822   *
 823   * @param  regexp The compiled regular expression.
 824   * @param  status A pointer to a UErrorCode to receive any errors.
 825   * @return TRUE if this matcher is using anchoring bounds.
 826   * @stable ICU 4.0
 827   */
 828 U_STABLE UBool U_EXPORT2
 829 uregex_hasAnchoringBounds(const  URegularExpression   *regexp,
 830                                  UErrorCode           *status);
 831
 832
 833 /**
 834   * Set whether this URegularExpression is using Anchoring Bounds for its region.
 835   * With anchoring bounds, pattern anchors such as ^ and $ will match at the start
 836   * and end of the region.  Without Anchoring Bounds, anchors will only match at
 837   * the positions they would in the complete text.
 838   *
 839   * Anchoring Bounds are the default for regions.
 840   *
 841   * @param regexp The compiled regular expression.
 842   * @param b      TRUE if to enable anchoring bounds; FALSE to disable them.
 843   * @param status A pointer to a UErrorCode to receive any errors.
 844   * @stable ICU 4.0
 845   */
 846 U_STABLE void U_EXPORT2
 847 uregex_useAnchoringBounds(URegularExpression   *regexp,
 848                           UBool                 b,
 849                           UErrorCode           *status);
 850
 851 /**
 852   * Return TRUE if the most recent matching operation touched the
 853   *  end of the text being processed.  In this case, additional input text could
 854   *  change the results of that match.
 855   *
 856   *  @param regexp The compiled regular expression.
 857   *  @param status A pointer to a UErrorCode to receive any errors.
 858   *  @return  TRUE if the most recent match hit the end of input
 859   *  @stable ICU 4.0
 860   */
 861 U_STABLE UBool U_EXPORT2
 862 uregex_hitEnd(const  URegularExpression   *regexp,
 863                      UErrorCode           *status);
 864
 865 /**
 866   * Return TRUE the most recent match succeeded and additional input could cause
 867   * it to fail. If this function returns false and a match was found, then more input
 868   * might change the match but the match won't be lost. If a match was not found,
 869   * then requireEnd has no meaning.
 870   *
 871   * @param regexp The compiled regular expression.
 872   * @param status A pointer to a UErrorCode to receive any errors.
 873   * @return TRUE  if more input could cause the most recent match to no longer match.
 874   * @stable ICU 4.0
 875   */
 876 U_STABLE UBool U_EXPORT2
 877 uregex_requireEnd(const  URegularExpression   *regexp,
 878                          UErrorCode           *status);
 879
 880
 881
 882
 883
 884 /**
 885   *    Replaces every substring of the input that matches the pattern
 886   *    with the given replacement string.  This is a convenience function that
 887   *    provides a complete find-and-replace-all operation.
 888   *
 889   *    This method scans the input string looking for matches of the pattern.
 890   *    Input that is not part of any match is copied unchanged to the
 891   *    destination buffer.  Matched regions are replaced in the output
 892   *    buffer by the replacement string.   The replacement string may contain
 893   *    references to capture groups; these take the form of $1, $2, etc.
 894   *
 895   *    @param   regexp             The compiled regular expression.
 896   *    @param   replacementText    A string containing the replacement text.
 897   *    @param   replacementLength  The length of the replacement string, or
 898   *                                -1 if it is NUL terminated.
 899   *    @param   destBuf            A (UChar *) buffer that will receive the result.
 900   *    @param   destCapacity       The capacity of the desitnation buffer.
 901   *    @param   status             A reference to a UErrorCode to receive any errors.
 902   *    @return                     The length of the string resulting from the find
 903   *                                and replace operation.  In the event that the
 904   *                                destination capacity is inadequate, the return value
 905   *                                is still the full length of the untruncated string.
 906   *    @stable ICU 3.0
 907   */
 908 U_STABLE int32_t U_EXPORT2
 909 uregex_replaceAll(URegularExpression    *regexp,
 910                   const UChar           *replacementText,
 911                   int32_t                replacementLength,
 912                   UChar                 *destBuf,
 913                   int32_t                destCapacity,
 914                   UErrorCode            *status);
 915
 916 /**
 917   *    Replaces every substring of the input that matches the pattern
 918   *    with the given replacement string.  This is a convenience function that
 919   *    provides a complete find-and-replace-all operation.
 920   *
 921   *    This method scans the input string looking for matches of the pattern.
 922   *    Input that is not part of any match is copied unchanged to the
 923   *    destination buffer.  Matched regions are replaced in the output
 924   *    buffer by the replacement string.   The replacement string may contain
 925   *    references to capture groups; these take the form of $1, $2, etc.
 926   *
 927   *    @param   regexp         The compiled regular expression.
 928   *    @param   replacement    A string containing the replacement text.
 929   *    @param   dest           A mutable UText that will receive the result.
 930   *                             If NULL, a new UText will be created (which may not be mutable).
 931   *    @param   status         A reference to a UErrorCode to receive any errors.
 932   *    @return                 A UText containing the results of the find and replace.
 933   *                             If a pre-allocated UText was provided, it will always be used and returned.
 934   *
 935   *    @draft ICU 4.6
 936   */
 937 U_DRAFT UText * U_EXPORT2
 938 uregex_replaceAllUText(URegularExpression *regexp,
 939                        UText              *replacement,
 940                        UText              *dest,
 941                        UErrorCode         *status);
 942
 943 /**
 944   *    Replaces the first substring of the input that matches the pattern
 945   *    with the given replacement string.  This is a convenience function that
 946   *    provides a complete find-and-replace operation.
 947   *
 948   *    This method scans the input string looking for a match of the pattern.
 949   *    All input that is not part of the match is copied unchanged to the
 950   *    destination buffer.  The matched region is replaced in the output
 951   *    buffer by the replacement string.   The replacement string may contain
 952   *    references to capture groups; these take the form of $1, $2, etc.
 953   *
 954   *    @param   regexp             The compiled regular expression.
 955   *    @param   replacementText    A string containing the replacement text.
 956   *    @param   replacementLength  The length of the replacement string, or
 957   *                                -1 if it is NUL terminated.
 958   *    @param   destBuf            A (UChar *) buffer that will receive the result.
 959   *    @param   destCapacity       The capacity of the desitnation buffer.
 960   *    @param   status             a reference to a UErrorCode to receive any errors.
 961   *    @return                     The length of the string resulting from the find
 962   *                                and replace operation.  In the event that the
 963   *                                destination capacity is inadequate, the return value
 964   *                                is still the full length of the untruncated string.
 965   *    @stable ICU 3.0
 966   */
 967 U_STABLE int32_t U_EXPORT2
 968 uregex_replaceFirst(URegularExpression  *regexp,
 969                     const UChar         *replacementText,
 970                     int32_t              replacementLength,
 971                     UChar               *destBuf,
 972                     int32_t              destCapacity,
 973                     UErrorCode          *status);
 974
 975 /**
 976   *    Replaces the first substring of the input that matches the pattern
 977   *    with the given replacement string.  This is a convenience function that
 978   *    provides a complete find-and-replace operation.
 979   *
 980   *    This method scans the input string looking for a match of the pattern.
 981   *    All input that is not part of the match is copied unchanged to the
 982   *    destination buffer.  The matched region is replaced in the output
 983   *    buffer by the replacement string.   The replacement string may contain
 984   *    references to capture groups; these take the form of $1, $2, etc.
 985   *
 986   *    @param   regexp         The compiled regular expression.
 987   *    @param   replacement    A string containing the replacement text.
 988   *    @param   dest           A mutable UText that will receive the result.
 989   *                             If NULL, a new UText will be created (which may not be mutable).
 990   *    @param   status         A reference to a UErrorCode to receive any errors.
 991   *    @return                 A UText containing the results of the find and replace.
 992   *                             If a pre-allocated UText was provided, it will always be used and returned.
 993   *
 994   *    @draft ICU 4.6
 995   */
 996 U_DRAFT UText * U_EXPORT2
 997 uregex_replaceFirstUText(URegularExpression *regexp,
 998                          UText              *replacement,
 999                          UText              *dest,
1000                          UErrorCode         *status);
1001
1002
1003 /**
1004   *   Implements a replace operation intended to be used as part of an
1005   *   incremental find-and-replace.
1006   *
1007   *   <p>The input string, starting from the end of the previous match and ending at
1008   *   the start of the current match, is appended to the destination string.  Then the
1009   *   replacement string is appended to the output string,
1010   *   including handling any substitutions of captured text.</p>
1011   *
1012   *   <p>A note on preflight computation of buffersize and error handling:
1013   *   Calls to uregex_appendReplacement() and uregex_appendTail() are
1014   *   designed to be chained, one after another, with the destination
1015   *   buffer pointer and buffer capacity updated after each in preparation
1016   *   to for the next.  If the destination buffer is exhausted partway through such a
1017   *   sequence, a U_BUFFER_OVERFLOW_ERROR status will be returned.  Normal
1018   *   ICU conventions are for a function to perform no action if it is
1019   *   called with an error status, but for this one case, uregex_appendRepacement()
1020   *   will operate normally so that buffer size computations will complete
1021   *   correctly.
1022   *
1023   *   <p>For simple, prepackaged, non-incremental find-and-replace
1024   *      operations, see replaceFirst() or replaceAll().</p>
1025   *
1026   *   @param   regexp      The regular expression object.
1027   *   @param   replacementText The string that will replace the matched portion of the
1028   *                        input string as it is copied to the destination buffer.
1029   *                        The replacement text may contain references ($1, for
1030   *                        example) to capture groups from the match.
1031   *   @param   replacementLength  The length of the replacement text string,
1032   *                        or -1 if the string is NUL terminated.
1033   *   @param   destBuf     The buffer into which the results of the
1034   *                        find-and-replace are placed.  On return, this pointer
1035   *                        will be updated to refer to the beginning of the
1036   *                        unused portion of buffer, leaving it in position for
1037   *                        a subsequent call to this function.
1038   *   @param   destCapacity The size of the output buffer,  On return, this
1039   *                        parameter will be updated to reflect the space remaining
1040   *                        unused in the output buffer.
1041   *   @param   status      A reference to a UErrorCode to receive any errors.
1042   *   @return              The length of the result string.  In the event that
1043   *                        destCapacity is inadequate, the full length of the
1044   *                        untruncated output string is returned.
1045   *
1046   *   @stable ICU 3.0
1047   *
1048   */
1049 U_STABLE int32_t U_EXPORT2
1050 uregex_appendReplacement(URegularExpression    *regexp,
1051                          const UChar           *replacementText,
1052                          int32_t                replacementLength,
1053                          UChar                **destBuf,
1054                          int32_t               *destCapacity,
1055                          UErrorCode            *status);
1056
1057
1058 /**
1059   *   Implements a replace operation intended to be used as part of an
1060   *   incremental find-and-replace.
1061   *
1062   *   <p>The input string, starting from the end of the previous match and ending at
1063   *   the start of the current match, is appended to the destination string.  Then the
1064   *   replacement string is appended to the output string,
1065   *   including handling any substitutions of captured text.</p>
1066   *
1067   *   <p>For simple, prepackaged, non-incremental find-and-replace
1068   *      operations, see replaceFirst() or replaceAll().</p>
1069   *
1070   *   @param   regexp      The regular expression object.
1071   *   @param   replacementText The string that will replace the matched portion of the
1072   *                        input string as it is copied to the destination buffer.
1073   *                        The replacement text may contain references ($1, for
1074   *                        example) to capture groups from the match.
1075   *   @param   dest        A mutable UText that will receive the result. Must not be NULL.
1076   *   @param   status      A reference to a UErrorCode to receive any errors.
1077   *
1078   *   @draft ICU 4.6
1079   */
1080 U_DRAFT void U_EXPORT2
1081 uregex_appendReplacementUText(URegularExpression    *regexp,
1082                               UText                 *replacementText,
1083                               UText                 *dest,
1084                               UErrorCode            *status);
1085
1086
1087 /**
1088   * As the final step in a find-and-replace operation, append the remainder
1089   * of the input string, starting at the position following the last match,
1090   * to the destination string. <code>uregex_appendTail()</code> is intended
1091   *  to be invoked after one or more invocations of the
1092   *  <code>uregex_appendReplacement()</code> function.
1093   *
1094   *   @param   regexp      The regular expression object.  This is needed to
1095   *                        obtain the input string and with the position
1096   *                        of the last match within it.
1097   *   @param   destBuf     The buffer in which the results of the
1098   *                        find-and-replace are placed.  On return, the pointer
1099   *                        will be updated to refer to the beginning of the
1100   *                        unused portion of buffer.
1101   *   @param   destCapacity The size of the output buffer,  On return, this
1102   *                        value will be updated to reflect the space remaining
1103   *                        unused in the output buffer.
1104   *   @param   status      A reference to a UErrorCode to receive any errors.
1105   *   @return              The length of the result string.  In the event that
1106   *                        destCapacity is inadequate, the full length of the
1107   *                        untruncated output string is returned.
1108   *
1109   *   @stable ICU 3.0
1110   */
1111 U_STABLE int32_t U_EXPORT2
1112 uregex_appendTail(URegularExpression    *regexp,
1113                   UChar                **destBuf,
1114                   int32_t               *destCapacity,
1115                   UErrorCode            *status);
1116
1117
1118 /**
1119   * As the final step in a find-and-replace operation, append the remainder
1120   * of the input string, starting at the position following the last match,
1121   * to the destination string. <code>uregex_appendTailUText()</code> is intended
1122   *  to be invoked after one or more invocations of the
1123   *  <code>uregex_appendReplacementUText()</code> function.
1124   *
1125   *   @param   regexp      The regular expression object.  This is needed to
1126   *                        obtain the input string and with the position
1127   *                        of the last match within it.
1128   *   @param   dest        A mutable UText that will receive the result. Must not be NULL.
1129   *   @return              The destination UText.
1130   *
1131   *   @draft ICU 4.6
1132   */
1133 U_DRAFT UText * U_EXPORT2
1134 uregex_appendTailUText(URegularExpression    *regexp,
1135                        UText                 *dest,
1136                        UErrorCode            *status);
1137
1138
1139
1140  /**
1141    * Split a string into fields.  Somewhat like split() from Perl.
1142    *  The pattern matches identify delimiters that separate the input
1143    *  into fields.  The input data between the matches becomes the
1144    *  fields themselves.
1145    * <p>
1146    *  Each of the fields is copied from the input string to the destination
1147    *  buffer, and NUL terminated.  The position of each field within
1148    *  the destination buffer is returned in the destFields array.
1149    *
1150    *  Note:  another choice for the design of this function would be to not
1151    *         copy the resulting fields at all, but to return indexes and
1152    *         lengths within the source text.
1153    *           Advantages would be
1154    *             o  Faster.  No Copying.
1155    *             o  Nothing extra needed when field data may contain embedded NUL chars.
1156    *             o  Less memory needed if working on large data.
1157    *           Disadvantages
1158    *             o  Less consistent with C++ split, which copies into an
1159    *                array of UnicodeStrings.
1160    *             o  No NUL termination, extracted fields would be less convenient
1161    *                to use in most cases.
1162    *             o  Possible problems in the future, when support Unicode Normalization
1163    *                could cause the fields to not correspond exactly to
1164    *                a range of the source text.
1165    *
1166    *    @param   regexp      The compiled regular expression.
1167    *    @param   destBuf     A (UChar *) buffer to receive the fields that
1168    *                         are extracted from the input string. These
1169    *                         field pointers will refer to positions within the
1170    *                         destination buffer supplied by the caller.  Any
1171    *                         extra positions within the destFields array will be
1172    *                         set to NULL.
1173    *    @param   destCapacity The capacity of the destBuf.
1174    *    @param   requiredCapacity  The actual capacity required of the destBuf.
1175    *                         If destCapacity is too small, requiredCapacity will return
1176    *                         the total capacity required to hold all of the output, and
1177    *                         a U_BUFFER_OVERFLOW_ERROR will be returned.
1178    *    @param   destFields  An array to be filled with the position of each
1179    *                         of the extracted fields within destBuf.
1180    *    @param   destFieldsCapacity  The number of elements in the destFields array.
1181    *                If the number of fields found is less than destFieldsCapacity,
1182    *                the extra destFields elements are set to zero.
1183    *                If destFieldsCapacity is too small, the trailing part of the
1184    *                input, including any field delimiters, is treated as if it
1185    *                were the last field - it is copied to the destBuf, and
1186    *                its position is in the destBuf is stored in the last element
1187    *                of destFields.  This behavior mimics that of Perl.  It is not
1188    *                an error condition, and no error status is returned when all destField
1189    *                positions are used.
1190    * @param status  A reference to a UErrorCode to receive any errors.
1191    * @return        The number of fields into which the input string was split.
1192    * @stable ICU 3.0
1193    */
1194 U_STABLE int32_t U_EXPORT2
1195 uregex_split(   URegularExpression      *regexp,
1196                   UChar                 *destBuf,
1197                   int32_t                destCapacity,
1198                   int32_t               *requiredCapacity,
1199                   UChar                 *destFields[],
1200                   int32_t                destFieldsCapacity,
1201                   UErrorCode            *status);
1202
1203
1204   /**
1205    * Split a string into fields.  Somewhat like split() from Perl.
1206    * The pattern matches identify delimiters that separate the input
1207    *  into fields.  The input data between the matches becomes the
1208    *  fields themselves.
1209    * <p>
1210    * The behavior of this function is not very closely aligned with uregex_split();
1211    * instead, it is based on (and implemented directly on top of) the C++ split method.
1212    *
1213    * @param regexp  The compiled regular expression.
1214    * @param destFields    An array of mutable UText structs to receive the results of the split.
1215    *                If a field is NULL, a new UText is allocated to contain the results for
1216    *                that field. This new UText is not guaranteed to be mutable.
1217    * @param destFieldsCapacity  The number of elements in the destination array.
1218    *                If the number of fields found is less than destCapacity, the
1219    *                extra strings in the destination array are not altered.
1220    *                If the number of destination strings is less than the number
1221    *                of fields, the trailing part of the input string, including any
1222    *                field delimiters, is placed in the last destination string.
1223    *                This behavior mimics that of Perl.  It is not  an error condition, and no
1224    *                error status is returned when all destField positions are used.
1225    * @param status  A reference to a UErrorCode to receive any errors.
1226    * @return        The number of fields into which the input string was split.
1227    *
1228    * @draft ICU 4.6
1229    */
1230 U_DRAFT int32_t U_EXPORT2
1231 uregex_splitUText(URegularExpression    *regexp,
1232                   UText                 *destFields[],
1233                   int32_t                destFieldsCapacity,
1234                   UErrorCode            *status);
1235
1236
1237
1238
1239 /**
1240  * Set a processing time limit for match operations with this URegularExpression.
1241  *
1242  * Some patterns, when matching certain strings, can run in exponential time.
1243  * For practical purposes, the match operation may appear to be in an
1244  * infinite loop.
1245  * When a limit is set a match operation will fail with an error if the
1246  * limit is exceeded.
1247  * <p>
1248  * The units of the limit are steps of the match engine.
1249  * Correspondence with actual processor time will depend on the speed
1250  * of the processor and the details of the specific pattern, but will
1251  * typically be on the order of milliseconds.
1252  * <p>
1253  * By default, the matching time is not limited.
1254  * <p>
1255  *
1256  * @param   regexp      The compiled regular expression.
1257  * @param   limit       The limit value, or 0 for no limit.
1258  * @param   status      A reference to a UErrorCode to receive any errors.
1259  * @stable ICU 4.0
1260  */
1261 U_STABLE void U_EXPORT2
1262 uregex_setTimeLimit(URegularExpression      *regexp,
1263                     int32_t                  limit,
1264                     UErrorCode              *status);
1265
1266 /**
1267  * Get the time limit for for matches with this URegularExpression.
1268  * A return value of zero indicates that there is no limit.
1269  *
1270  * @param   regexp      The compiled regular expression.
1271  * @param   status      A reference to a UErrorCode to receive any errors.
1272  * @return the maximum allowed time for a match, in units of processing steps.
1273  * @stable ICU 4.0
1274  */
1275 U_STABLE int32_t U_EXPORT2
1276 uregex_getTimeLimit(const URegularExpression      *regexp,
1277                           UErrorCode              *status);
1278
1279 /**
1280  * Set the amount of heap storage avaliable for use by the match backtracking stack.
1281  * <p>
1282  * ICU uses a backtracking regular expression engine, with the backtrack stack
1283  * maintained on the heap.  This function sets the limit to the amount of memory
1284  * that can be used  for this purpose.  A backtracking stack overflow will
1285  * result in an error from the match operation that caused it.
1286  * <p>
1287  * A limit is desirable because a malicious or poorly designed pattern can use
1288  * excessive memory, potentially crashing the process.  A limit is enabled
1289  * by default.
1290  * <p>
1291  * @param   regexp      The compiled regular expression.
1292  * @param   limit       The maximum size, in bytes, of the matching backtrack stack.
1293  *                      A value of -1 means no limit.
1294  *                      The limit must be greater than zero, or -1.
1295  * @param   status      A reference to a UErrorCode to receive any errors.
1296  *
1297  * @stable ICU 4.0
1298  */
1299 U_STABLE void U_EXPORT2
1300 uregex_setStackLimit(URegularExpression      *regexp,
1301                      int32_t                  limit,
1302                      UErrorCode              *status);
1303
1304 /**
1305  * Get the size of the heap storage available for use by the back tracking stack.
1306  *
1307  * @return  the maximum backtracking stack size, in bytes, or zero if the
1308  *          stack size is unlimited.
1309  * @stable ICU 4.0
1310  */
1311 U_STABLE int32_t U_EXPORT2
1312 uregex_getStackLimit(const URegularExpression      *regexp,
1313                            UErrorCode              *status);
1314
1315
1316 /**
1317  * Function pointer for a regular expression matching callback function.
1318  * When set, a callback function will be called periodically during matching
1319  * operations.  If the call back function returns FALSE, the matching
1320  * operation will be terminated early.
1321  *
1322  * Note:  the callback function must not call other functions on this
1323  *        URegularExpression.
1324  *
1325  * @param context  context pointer.  The callback function will be invoked
1326  *                 with the context specified at the time that
1327  *                 uregex_setMatchCallback() is called.
1328  * @param steps    the accumulated processing time, in match steps,
1329  *                 for this matching operation.
1330  * @return         TRUE to continue the matching operation.
1331  *                 FALSE to terminate the matching operation.
1332  * @stable ICU 4.0
1333  */
1334 U_CDECL_BEGIN
1335 typedef UBool U_CALLCONV URegexMatchCallback (
1336                    const void *context,
1337                    int32_t     steps);
1338 U_CDECL_END
1339
1340 /**
1341  * Set a callback function for this URegularExpression.
1342  * During matching operations the function will be called periodically,
1343  * giving the application the opportunity to terminate a long-running
1344  * match.
1345  *
1346  * @param   regexp      The compiled regular expression.
1347  * @param   callback    A pointer to the user-supplied callback function.
1348  * @param   context     User context pointer.  The value supplied at the
1349  *                      time the callback function is set will be saved
1350  *                      and passed to the callback each time that it is called.
1351  * @param   status      A reference to a UErrorCode to receive any errors.
1352  * @stable ICU 4.0
1353  */
1354 U_STABLE void U_EXPORT2
1355 uregex_setMatchCallback(URegularExpression      *regexp,
1356                         URegexMatchCallback     *callback,
1357                         const void              *context,
1358                         UErrorCode              *status);
1359
1360
1361 /**
1362  *  Get the callback function for this URegularExpression.
1363  *
1364  * @param   regexp      The compiled regular expression.
1365  * @param   callback    Out paramater, receives a pointer to the user-supplied
1366  *                      callback function.
1367  * @param   context     Out parameter, receives the user context pointer that
1368  *                      was set when uregex_setMatchCallback() was called.
1369  * @param   status      A reference to a UErrorCode to receive any errors.
1370  * @stable ICU 4.0
1371  */
1372 U_STABLE void U_EXPORT2
1373 uregex_getMatchCallback(const URegularExpression    *regexp,
1374                         URegexMatchCallback        **callback,
1375                         const void                 **context,
1376                         UErrorCode                  *status);
1377
1378
1379 /**
1380  * Function pointer for a regular expression find callback function.
1381  *
1382  * When set, a callback function will be called during a find operation
1383  * and for operations that depend on find, such as findNext, split and some replace
1384  * operations like replaceFirst.
1385  * The callback will usually be called after each attempt at a match, but this is not a
1386  * guarantee that the callback will be invoked at each character.  For finds where the
1387  * match engine is invoked at each character, this may be close to true, but less likely
1388  * for more optimized loops where the pattern is known to only start, and the match
1389  * engine invoked, at certain characters.
1390  * When invoked, this callback will specify the index at which a match operation is about
1391  * to be attempted, giving the application the opportunity to terminate a long-running
1392  * find operation.
1393  *
1394  * If the call back function returns FALSE, the find operation will be terminated early.
1395  *
1396  * Note:  the callback function must not call other functions on this
1397  *        URegularExpression
1398  *
1399  * @param context  context pointer.  The callback function will be invoked
1400  *                 with the context specified at the time that
1401  *                 uregex_setFindProgressCallback() is called.
1402  * @param matchIndex  the next index at which a match attempt will be attempted for this
1403  *                 find operation.  If this callback interrupts the search, this is the
1404  *                 index at which a find/findNext operation may be re-initiated.
1405  * @return         TRUE to continue the matching operation.
1406  *                 FALSE to terminate the matching operation.
1407  * @draft ICU 4.6
1408  */
1409 U_CDECL_BEGIN
1410 typedef UBool U_CALLCONV URegexFindProgressCallback (
1411                    const void *context,
1412                    int64_t     matchIndex);
1413 U_CDECL_END
1414
1415 /**
1416  *  Set the find progress callback function for this URegularExpression.
1417  *
1418  * @param   regexp      The compiled regular expression.
1419  * @param   callback    A pointer to the user-supplied callback function.
1420  * @param   context     User context pointer.  The value supplied at the
1421  *                      time the callback function is set will be saved
1422  *                      and passed to the callback each time that it is called.
1423  * @param   status      A reference to a UErrorCode to receive any errors.
1424  * @draft ICU 4.6
1425  */
1426 U_DRAFT void U_EXPORT2
1427 uregex_setFindProgressCallback(URegularExpression              *regexp,
1428                                 URegexFindProgressCallback      *callback,
1429                                 const void                      *context,
1430                                 UErrorCode                      *status);
1431
1432
1433 /**
1434  *  Get the find progress callback function for this URegularExpression.
1435  *
1436  * @param   regexp      The compiled regular expression.
1437  * @param   callback    Out paramater, receives a pointer to the user-supplied
1438  *                      callback function.
1439  * @param   context     Out parameter, receives the user context pointer that
1440  *                      was set when uregex_setFindProgressCallback() was called.
1441  * @param   status      A reference to a UErrorCode to receive any errors.
1442  * @draft ICU 4.6
1443  */
1444 U_DRAFT void U_EXPORT2
1445 uregex_getFindProgressCallback(const URegularExpression          *regexp,
1446                                 URegexFindProgressCallback        **callback,
1447                                 const void                        **context,
1448                                 UErrorCode                        *status);
1449
1450 #endif   /*  !UCONFIG_NO_REGULAR_EXPRESSIONS  */
1451 #endif   /*  UREGEX_H  */