1/* 2********************************************************************** 3* Copyright (C) 2004-2013, 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 34struct URegularExpression; 35/** 36 * Structure representing a compiled regular expression, plus the results 37 * of a match operation. 38 * @stable ICU 3.0 39 */ 40typedef struct URegularExpression URegularExpression; 41 42 43/** 44 * Constants for Regular Expression Match Modes. 45 * @stable ICU 2.4 46 */ 47typedef 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 /* U_HIDE_DRAFT_API */ 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. 68 * 69 * The flag UREGEX_CASE_INSENSITIVE retains its impact 70 * on matching when used in conjunction with this flag. 71 * The other flags become superfluous. 72 * 73 * @stable ICU 4.0 74 */ 75 UREGEX_LITERAL = 16, 76 77 /** Control behavior of "$" and "^" 78 * If set, recognize line terminators within string, 79 * otherwise, match only at start and end of input string. 80 * @stable ICU 2.4 */ 81 UREGEX_MULTILINE = 8, 82 83 /** Unix-only line endings. 84 * When this mode is enabled, only \\u000a is recognized as a line ending 85 * in the behavior of ., ^, and $. 86 * @stable ICU 4.0 87 */ 88 UREGEX_UNIX_LINES = 1, 89 90 /** Unicode word boundaries. 91 * If set, \b uses the Unicode TR 29 definition of word boundaries. 92 * Warning: Unicode word boundaries are quite different from 93 * traditional regular expression word boundaries. See 94 * http://unicode.org/reports/tr29/#Word_Boundaries 95 * @stable ICU 2.8 96 */ 97 UREGEX_UWORD = 256, 98 99 /** Error on Unrecognized backslash escapes. 100 * If set, fail with an error on patterns that contain 101 * backslash-escaped ASCII letters without a known special 102 * meaning. If this flag is not set, these 103 * escaped letters represent themselves. 104 * @stable ICU 4.0 105 */ 106 UREGEX_ERROR_ON_UNKNOWN_ESCAPES = 512 107 108} URegexpFlag; 109 110/** 111 * Open (compile) an ICU regular expression. Compiles the regular expression in 112 * string form into an internal representation using the specified match mode flags. 113 * The resulting regular expression handle can then be used to perform various 114 * matching operations. 115 * 116 * 117 * @param pattern The Regular Expression pattern to be compiled. 118 * @param patternLength The length of the pattern, or -1 if the pattern is 119 * NUL terminated. 120 * @param flags Flags that alter the default matching behavior for 121 * the regular expression, UREGEX_CASE_INSENSITIVE, for 122 * example. For default behavior, set this parameter to zero. 123 * See <code>enum URegexpFlag</code>. All desired flags 124 * are bitwise-ORed together. 125 * @param pe Receives the position (line and column numbers) of any syntax 126 * error within the source regular expression string. If this 127 * information is not wanted, pass NULL for this parameter. 128 * @param status Receives error detected by this function. 129 * @stable ICU 3.0 130 * 131 */ 132U_STABLE URegularExpression * U_EXPORT2 133uregex_open( const UChar *pattern, 134 int32_t patternLength, 135 uint32_t flags, 136 UParseError *pe, 137 UErrorCode *status); 138 139/** 140 * Open (compile) an ICU regular expression. Compiles the regular expression in 141 * string form into an internal representation using the specified match mode flags. 142 * The resulting regular expression handle can then be used to perform various 143 * matching operations. 144 * <p> 145 * The contents of the pattern UText will be extracted and saved. Ownership of the 146 * UText struct itself remains with the caller. This is to match the behavior of 147 * uregex_open(). 148 * 149 * @param pattern The Regular Expression pattern to be compiled. 150 * @param flags Flags that alter the default matching behavior for 151 * the regular expression, UREGEX_CASE_INSENSITIVE, for 152 * example. For default behavior, set this parameter to zero. 153 * See <code>enum URegexpFlag</code>. All desired flags 154 * are bitwise-ORed together. 155 * @param pe Receives the position (line and column numbers) of any syntax 156 * error within the source regular expression string. If this 157 * information is not wanted, pass NULL for this parameter. 158 * @param status Receives error detected by this function. 159 * 160 * @stable ICU 4.6 161 */ 162U_STABLE URegularExpression * U_EXPORT2 163uregex_openUText(UText *pattern, 164 uint32_t flags, 165 UParseError *pe, 166 UErrorCode *status); 167 168/** 169 * Open (compile) an ICU regular expression. The resulting regular expression 170 * handle can then be used to perform various matching operations. 171 * <p> 172 * This function is the same as uregex_open, except that the pattern 173 * is supplied as an 8 bit char * string in the default code page. 174 * 175 * @param pattern The Regular Expression pattern to be compiled, 176 * NUL terminated. 177 * @param flags Flags that alter the default matching behavior for 178 * the regular expression, UREGEX_CASE_INSENSITIVE, for 179 * example. For default behavior, set this parameter to zero. 180 * See <code>enum URegexpFlag</code>. All desired flags 181 * are bitwise-ORed together. 182 * @param pe Receives the position (line and column numbers) of any syntax 183 * error within the source regular expression string. If this 184 * information is not wanted, pass NULL for this parameter. 185 * @param status Receives errors detected by this function. 186 * @return The URegularExpression object representing the compiled 187 * pattern. 188 * 189 * @stable ICU 3.0 190 */ 191#if !UCONFIG_NO_CONVERSION 192U_STABLE URegularExpression * U_EXPORT2 193uregex_openC( const char *pattern, 194 uint32_t flags, 195 UParseError *pe, 196 UErrorCode *status); 197#endif 198 199 200 201/** 202 * Close the regular expression, recovering all resources (memory) it 203 * was holding. 204 * 205 * @param regexp The regular expression to be closed. 206 * @stable ICU 3.0 207 */ 208U_STABLE void U_EXPORT2 209uregex_close(URegularExpression *regexp); 210 211#if U_SHOW_CPLUSPLUS_API 212 213U_NAMESPACE_BEGIN 214 215/** 216 * \class LocalURegularExpressionPointer 217 * "Smart pointer" class, closes a URegularExpression via uregex_close(). 218 * For most methods see the LocalPointerBase base class. 219 * 220 * @see LocalPointerBase 221 * @see LocalPointer 222 * @stable ICU 4.4 223 */ 224U_DEFINE_LOCAL_OPEN_POINTER(LocalURegularExpressionPointer, URegularExpression, uregex_close); 225 226U_NAMESPACE_END 227 228#endif 229 230/** 231 * Make a copy of a compiled regular expression. Cloning a regular 232 * expression is faster than opening a second instance from the source 233 * form of the expression, and requires less memory. 234 * <p> 235 * Note that the current input string and the position of any matched text 236 * within it are not cloned; only the pattern itself and the 237 * match mode flags are copied. 238 * <p> 239 * Cloning can be particularly useful to threaded applications that perform 240 * multiple match operations in parallel. Each concurrent RE 241 * operation requires its own instance of a URegularExpression. 242 * 243 * @param regexp The compiled regular expression to be cloned. 244 * @param status Receives indication of any errors encountered 245 * @return the cloned copy of the compiled regular expression. 246 * @stable ICU 3.0 247 */ 248U_STABLE URegularExpression * U_EXPORT2 249uregex_clone(const URegularExpression *regexp, UErrorCode *status); 250 251/** 252 * Returns a pointer to the source form of the pattern for this regular expression. 253 * This function will work even if the pattern was originally specified as a UText. 254 * 255 * @param regexp The compiled regular expression. 256 * @param patLength This output parameter will be set to the length of the 257 * pattern string. A NULL pointer may be used here if the 258 * pattern length is not needed, as would be the case if 259 * the pattern is known in advance to be a NUL terminated 260 * string. 261 * @param status Receives errors detected by this function. 262 * @return a pointer to the pattern string. The storage for the string is 263 * owned by the regular expression object, and must not be 264 * altered or deleted by the application. The returned string 265 * will remain valid until the regular expression is closed. 266 * @stable ICU 3.0 267 */ 268U_STABLE const UChar * U_EXPORT2 269uregex_pattern(const URegularExpression *regexp, 270 int32_t *patLength, 271 UErrorCode *status); 272 273/** 274 * Returns the source text of the pattern for this regular expression. 275 * This function will work even if the pattern was originally specified as a UChar string. 276 * 277 * @param regexp The compiled regular expression. 278 * @param status Receives errors detected by this function. 279 * @return the pattern text. The storage for the text is owned by the regular expression 280 * object, and must not be altered or deleted. 281 * 282 * @stable ICU 4.6 283 */ 284U_STABLE UText * U_EXPORT2 285uregex_patternUText(const URegularExpression *regexp, 286 UErrorCode *status); 287 288/** 289 * Get the match mode flags that were specified when compiling this regular expression. 290 * @param status Receives errors detected by this function. 291 * @param regexp The compiled regular expression. 292 * @return The match mode flags 293 * @see URegexpFlag 294 * @stable ICU 3.0 295 */ 296U_STABLE int32_t U_EXPORT2 297uregex_flags(const URegularExpression *regexp, 298 UErrorCode *status); 299 300 301/** 302 * Set the subject text string upon which the regular expression will look for matches. 303 * This function may be called any number of times, allowing the regular 304 * expression pattern to be applied to different strings. 305 * <p> 306 * Regular expression matching operations work directly on the application's 307 * string data. No copy is made. The subject string data must not be 308 * altered after calling this function until after all regular expression 309 * operations involving this string data are completed. 310 * <p> 311 * Zero length strings are permitted. In this case, no subsequent match 312 * operation will dereference the text string pointer. 313 * 314 * @param regexp The compiled regular expression. 315 * @param text The subject text string. 316 * @param textLength The length of the subject text, or -1 if the string 317 * is NUL terminated. 318 * @param status Receives errors detected by this function. 319 * @stable ICU 3.0 320 */ 321U_STABLE void U_EXPORT2 322uregex_setText(URegularExpression *regexp, 323 const UChar *text, 324 int32_t textLength, 325 UErrorCode *status); 326 327 328/** 329 * Set the subject text string upon which the regular expression will look for matches. 330 * This function may be called any number of times, allowing the regular 331 * expression pattern to be applied to different strings. 332 * <p> 333 * Regular expression matching operations work directly on the application's 334 * string data; only a shallow clone is made. The subject string data must not be 335 * altered after calling this function until after all regular expression 336 * operations involving this string data are completed. 337 * 338 * @param regexp The compiled regular expression. 339 * @param text The subject text string. 340 * @param status Receives errors detected by this function. 341 * 342 * @stable ICU 4.6 343 */ 344U_STABLE void U_EXPORT2 345uregex_setUText(URegularExpression *regexp, 346 UText *text, 347 UErrorCode *status); 348 349/** 350 * Get the subject text that is currently associated with this 351 * regular expression object. If the input was supplied using uregex_setText(), 352 * that pointer will be returned. Otherwise, the characters in the input will 353 * be extracted to a buffer and returned. In either case, ownership remains 354 * with the regular expression object. 355 * 356 * This function will work even if the input was originally specified as a UText. 357 * 358 * @param regexp The compiled regular expression. 359 * @param textLength The length of the string is returned in this output parameter. 360 * A NULL pointer may be used here if the 361 * text length is not needed, as would be the case if 362 * the text is known in advance to be a NUL terminated 363 * string. 364 * @param status Receives errors detected by this function. 365 * @return Pointer to the subject text string currently associated with 366 * this regular expression. 367 * @stable ICU 3.0 368 */ 369U_STABLE const UChar * U_EXPORT2 370uregex_getText(URegularExpression *regexp, 371 int32_t *textLength, 372 UErrorCode *status); 373 374/** 375 * Get the subject text that is currently associated with this 376 * regular expression object. 377 * 378 * This function will work even if the input was originally specified as a UChar string. 379 * 380 * @param regexp The compiled regular expression. 381 * @param dest A mutable UText in which to store the current input. 382 * If NULL, a new UText will be created as an immutable shallow clone 383 * of the actual input string. 384 * @param status Receives errors detected by this function. 385 * @return The subject text currently associated with this regular expression. 386 * If a pre-allocated UText was provided, it will always be used and returned. 387 * 388 * @stable ICU 4.6 389 */ 390U_STABLE UText * U_EXPORT2 391uregex_getUText(URegularExpression *regexp, 392 UText *dest, 393 UErrorCode *status); 394 395/** 396 * Set the subject text string upon which the regular expression is looking for matches 397 * without changing any other aspect of the matching state. 398 * The new and previous text strings must have the same content. 399 * 400 * This function is intended for use in environments where ICU is operating on 401 * strings that may move around in memory. It provides a mechanism for notifying 402 * ICU that the string has been relocated, and providing a new UText to access the 403 * string in its new position. 404 * 405 * Note that the regular expression implementation never copies the underlying text 406 * of a string being matched, but always operates directly on the original text 407 * provided by the user. Refreshing simply drops the references to the old text 408 * and replaces them with references to the new. 409 * 410 * Caution: this function is normally used only by very specialized 411 * system-level code. One example use case is with garbage collection 412 * that moves the text in memory. 413 * 414 * @param regexp The compiled regular expression. 415 * @param text The new (moved) text string. 416 * @param status Receives errors detected by this function. 417 * 418 * @stable ICU 4.8 419 */ 420U_STABLE void U_EXPORT2 421uregex_refreshUText(URegularExpression *regexp, 422 UText *text, 423 UErrorCode *status); 424 425/** 426 * Attempts to match the input string against the pattern. 427 * To succeed, the match must extend to the end of the string, 428 * or cover the complete match region. 429 * 430 * If startIndex >= zero the match operation starts at the specified 431 * index and must extend to the end of the input string. Any region 432 * that has been specified is reset. 433 * 434 * If startIndex == -1 the match must cover the input region, or the entire 435 * input string if no region has been set. This directly corresponds to 436 * Matcher.matches() in Java 437 * 438 * @param regexp The compiled regular expression. 439 * @param startIndex The input string (native) index at which to begin matching, or -1 440 * to match the input Region. 441 * @param status Receives errors detected by this function. 442 * @return TRUE if there is a match 443 * @stable ICU 3.0 444 */ 445U_STABLE UBool U_EXPORT2 446uregex_matches(URegularExpression *regexp, 447 int32_t startIndex, 448 UErrorCode *status); 449 450/** 451 * 64bit version of uregex_matches. 452 * Attempts to match the input string against the pattern. 453 * To succeed, the match must extend to the end of the string, 454 * or cover the complete match region. 455 * 456 * If startIndex >= zero the match operation starts at the specified 457 * index and must extend to the end of the input string. Any region 458 * that has been specified is reset. 459 * 460 * If startIndex == -1 the match must cover the input region, or the entire 461 * input string if no region has been set. This directly corresponds to 462 * Matcher.matches() in Java 463 * 464 * @param regexp The compiled regular expression. 465 * @param startIndex The input string (native) index at which to begin matching, or -1 466 * to match the input Region. 467 * @param status Receives errors detected by this function. 468 * @return TRUE if there is a match 469 * @stable ICU 4.6 470 */ 471U_STABLE UBool U_EXPORT2 472uregex_matches64(URegularExpression *regexp, 473 int64_t startIndex, 474 UErrorCode *status); 475 476/** 477 * Attempts to match the input string, starting from the specified index, against the pattern. 478 * The match may be of any length, and is not required to extend to the end 479 * of the input string. Contrast with uregex_matches(). 480 * 481 * <p>If startIndex is >= 0 any input region that was set for this 482 * URegularExpression is reset before the operation begins. 483 * 484 * <p>If the specified starting index == -1 the match begins at the start of the input 485 * region, or at the start of the full string if no region has been specified. 486 * This corresponds directly with Matcher.lookingAt() in Java. 487 * 488 * <p>If the match succeeds then more information can be obtained via the 489 * <code>uregexp_start()</code>, <code>uregexp_end()</code>, 490 * and <code>uregexp_group()</code> functions.</p> 491 * 492 * @param regexp The compiled regular expression. 493 * @param startIndex The input string (native) index at which to begin matching, or 494 * -1 to match the Input Region 495 * @param status A reference to a UErrorCode to receive any errors. 496 * @return TRUE if there is a match. 497 * @stable ICU 3.0 498 */ 499U_STABLE UBool U_EXPORT2 500uregex_lookingAt(URegularExpression *regexp, 501 int32_t startIndex, 502 UErrorCode *status); 503 504/** 505 * 64bit version of uregex_lookingAt. 506 * Attempts to match the input string, starting from the specified index, against the pattern. 507 * The match may be of any length, and is not required to extend to the end 508 * of the input string. Contrast with uregex_matches(). 509 * 510 * <p>If startIndex is >= 0 any input region that was set for this 511 * URegularExpression is reset before the operation begins. 512 * 513 * <p>If the specified starting index == -1 the match begins at the start of the input 514 * region, or at the start of the full string if no region has been specified. 515 * This corresponds directly with Matcher.lookingAt() in Java. 516 * 517 * <p>If the match succeeds then more information can be obtained via the 518 * <code>uregexp_start()</code>, <code>uregexp_end()</code>, 519 * and <code>uregexp_group()</code> functions.</p> 520 * 521 * @param regexp The compiled regular expression. 522 * @param startIndex The input string (native) index at which to begin matching, or 523 * -1 to match the Input Region 524 * @param status A reference to a UErrorCode to receive any errors. 525 * @return TRUE if there is a match. 526 * @stable ICU 4.6 527 */ 528U_STABLE UBool U_EXPORT2 529uregex_lookingAt64(URegularExpression *regexp, 530 int64_t startIndex, 531 UErrorCode *status); 532 533/** 534 * Find the first matching substring of the input string that matches the pattern. 535 * If startIndex is >= zero the search for a match begins at the specified index, 536 * and any match region is reset. This corresponds directly with 537 * Matcher.find(startIndex) in Java. 538 * 539 * If startIndex == -1 the search begins at the start of the input region, 540 * or at the start of the full string if no region has been specified. 541 * 542 * If a match is found, <code>uregex_start(), uregex_end()</code>, and 543 * <code>uregex_group()</code> will provide more information regarding the match. 544 * 545 * @param regexp The compiled regular expression. 546 * @param startIndex The position (native) in the input string to begin the search, or 547 * -1 to search within the Input Region. 548 * @param status A reference to a UErrorCode to receive any errors. 549 * @return TRUE if a match is found. 550 * @stable ICU 3.0 551 */ 552U_STABLE UBool U_EXPORT2 553uregex_find(URegularExpression *regexp, 554 int32_t startIndex, 555 UErrorCode *status); 556 557/** 558 * 64bit version of uregex_find. 559 * Find the first matching substring of the input string that matches the pattern. 560 * If startIndex is >= zero the search for a match begins at the specified index, 561 * and any match region is reset. This corresponds directly with 562 * Matcher.find(startIndex) in Java. 563 * 564 * If startIndex == -1 the search begins at the start of the input region, 565 * or at the start of the full string if no region has been specified. 566 * 567 * If a match is found, <code>uregex_start(), uregex_end()</code>, and 568 * <code>uregex_group()</code> will provide more information regarding the match. 569 * 570 * @param regexp The compiled regular expression. 571 * @param startIndex The position (native) in the input string to begin the search, or 572 * -1 to search within the Input Region. 573 * @param status A reference to a UErrorCode to receive any errors. 574 * @return TRUE if a match is found. 575 * @stable ICU 4.6 576 */ 577U_STABLE UBool U_EXPORT2 578uregex_find64(URegularExpression *regexp, 579 int64_t startIndex, 580 UErrorCode *status); 581 582/** 583 * Find the next pattern match in the input string. Begin searching 584 * the input at the location following the end of he previous match, 585 * or at the start of the string (or region) if there is no 586 * previous match. If a match is found, <code>uregex_start(), uregex_end()</code>, and 587 * <code>uregex_group()</code> will provide more information regarding the match. 588 * 589 * @param regexp The compiled regular expression. 590 * @param status A reference to a UErrorCode to receive any errors. 591 * @return TRUE if a match is found. 592 * @see uregex_reset 593 * @stable ICU 3.0 594 */ 595U_STABLE UBool U_EXPORT2 596uregex_findNext(URegularExpression *regexp, 597 UErrorCode *status); 598 599/** 600 * Get the number of capturing groups in this regular expression's pattern. 601 * @param regexp The compiled regular expression. 602 * @param status A reference to a UErrorCode to receive any errors. 603 * @return the number of capture groups 604 * @stable ICU 3.0 605 */ 606U_STABLE int32_t U_EXPORT2 607uregex_groupCount(URegularExpression *regexp, 608 UErrorCode *status); 609 610/** Extract the string for the specified matching expression or subexpression. 611 * Group #0 is the complete string of matched text. 612 * Group #1 is the text matched by the first set of capturing parentheses. 613 * 614 * @param regexp The compiled regular expression. 615 * @param groupNum The capture group to extract. Group 0 is the complete 616 * match. The value of this parameter must be 617 * less than or equal to the number of capture groups in 618 * the pattern. 619 * @param dest Buffer to receive the matching string data 620 * @param destCapacity Capacity of the dest buffer. 621 * @param status A reference to a UErrorCode to receive any errors. 622 * @return Length of matching data, 623 * or -1 if no applicable match. 624 * @stable ICU 3.0 625 */ 626U_STABLE int32_t U_EXPORT2 627uregex_group(URegularExpression *regexp, 628 int32_t groupNum, 629 UChar *dest, 630 int32_t destCapacity, 631 UErrorCode *status); 632 633/** Returns a shallow immutable clone of the entire input string. The returned UText current native index 634 * is set to the beginning of the requested capture group. The capture group length is also 635 * returned via groupLength. 636 * Group #0 is the complete string of matched text. 637 * Group #1 is the text matched by the first set of capturing parentheses. 638 * 639 * @param regexp The compiled regular expression. 640 * @param groupNum The capture group to extract. Group 0 is the complete 641 * match. The value of this parameter must be 642 * less than or equal to the number of capture groups in 643 * the pattern. 644 * @param dest A mutable UText in which to store the current input. 645 * If NULL, a new UText will be created as an immutable shallow clone 646 * of the entire input string. 647 * @param groupLength The group length of the desired capture group. 648 * @param status A reference to a UErrorCode to receive any errors. 649 * @return The subject text currently associated with this regular expression. 650 * If a pre-allocated UText was provided, it will always be used and returned. 651 652 * 653 * @stable ICU 4.6 654 */ 655U_STABLE UText * U_EXPORT2 656uregex_groupUText(URegularExpression *regexp, 657 int32_t groupNum, 658 UText *dest, 659 int64_t *groupLength, 660 UErrorCode *status); 661 662#ifndef U_HIDE_INTERNAL_API 663/** Extract the string for the specified matching expression or subexpression. 664 * Group #0 is the complete string of matched text. 665 * Group #1 is the text matched by the first set of capturing parentheses. 666 * 667 * @param regexp The compiled regular expression. 668 * @param groupNum The capture group to extract. Group 0 is the complete 669 * match. The value of this parameter must be 670 * less than or equal to the number of capture groups in 671 * the pattern. 672 * @param dest Mutable UText to receive the matching string data. 673 * If NULL, a new UText will be created (which may not be mutable). 674 * @param status A reference to a UErrorCode to receive any errors. 675 * @return The matching string data. If a pre-allocated UText was provided, 676 * it will always be used and returned. 677 * 678 * @internal ICU 4.4 technology preview 679 */ 680U_INTERNAL UText * U_EXPORT2 681uregex_groupUTextDeep(URegularExpression *regexp, 682 int32_t groupNum, 683 UText *dest, 684 UErrorCode *status); 685#endif /* U_HIDE_INTERNAL_API */ 686 687/** 688 * Returns the index in the input string of the start of the text matched by the 689 * specified capture group during the previous match operation. Return -1 if 690 * the capture group was not part of the last match. 691 * Group #0 refers to the complete range of matched text. 692 * Group #1 refers to the text matched by the first set of capturing parentheses. 693 * 694 * @param regexp The compiled regular expression. 695 * @param groupNum The capture group number 696 * @param status A reference to a UErrorCode to receive any errors. 697 * @return the starting (native) position in the input of the text matched 698 * by the specified group. 699 * @stable ICU 3.0 700 */ 701U_STABLE int32_t U_EXPORT2 702uregex_start(URegularExpression *regexp, 703 int32_t groupNum, 704 UErrorCode *status); 705 706/** 707 * 64bit version of uregex_start. 708 * Returns the index in the input string of the start of the text matched by the 709 * specified capture group during the previous match operation. Return -1 if 710 * the capture group was not part of the last match. 711 * Group #0 refers to the complete range of matched text. 712 * Group #1 refers to the text matched by the first set of capturing parentheses. 713 * 714 * @param regexp The compiled regular expression. 715 * @param groupNum The capture group number 716 * @param status A reference to a UErrorCode to receive any errors. 717 * @return the starting (native) position in the input of the text matched 718 * by the specified group. 719 * @stable ICU 4.6 720 */ 721U_STABLE int64_t U_EXPORT2 722uregex_start64(URegularExpression *regexp, 723 int32_t groupNum, 724 UErrorCode *status); 725 726/** 727 * Returns the index in the input string of the position following the end 728 * of the text matched by the specified capture group. 729 * Return -1 if the capture group was not part of the last match. 730 * Group #0 refers to the complete range of matched text. 731 * Group #1 refers to the text matched by the first set of capturing parentheses. 732 * 733 * @param regexp The compiled regular expression. 734 * @param groupNum The capture group number 735 * @param status A reference to a UErrorCode to receive any errors. 736 * @return the (native) index of the position following the last matched character. 737 * @stable ICU 3.0 738 */ 739U_STABLE int32_t U_EXPORT2 740uregex_end(URegularExpression *regexp, 741 int32_t groupNum, 742 UErrorCode *status); 743 744/** 745 * 64bit version of uregex_end. 746 * Returns the index in the input string of the position following the end 747 * of the text matched by the specified capture group. 748 * Return -1 if the capture group was not part of the last match. 749 * Group #0 refers to the complete range of matched text. 750 * Group #1 refers to the text matched by the first set of capturing parentheses. 751 * 752 * @param regexp The compiled regular expression. 753 * @param groupNum The capture group number 754 * @param status A reference to a UErrorCode to receive any errors. 755 * @return the (native) index of the position following the last matched character. 756 * @stable ICU 4.6 757 */ 758U_STABLE int64_t U_EXPORT2 759uregex_end64(URegularExpression *regexp, 760 int32_t groupNum, 761 UErrorCode *status); 762 763/** 764 * Reset any saved state from the previous match. Has the effect of 765 * causing uregex_findNext to begin at the specified index, and causing 766 * uregex_start(), uregex_end() and uregex_group() to return an error 767 * indicating that there is no match information available. Clears any 768 * match region that may have been set. 769 * 770 * @param regexp The compiled regular expression. 771 * @param index The position (native) in the text at which a 772 * uregex_findNext() should begin searching. 773 * @param status A reference to a UErrorCode to receive any errors. 774 * @stable ICU 3.0 775 */ 776U_STABLE void U_EXPORT2 777uregex_reset(URegularExpression *regexp, 778 int32_t index, 779 UErrorCode *status); 780 781/** 782 * 64bit version of uregex_reset. 783 * Reset any saved state from the previous match. Has the effect of 784 * causing uregex_findNext to begin at the specified index, and causing 785 * uregex_start(), uregex_end() and uregex_group() to return an error 786 * indicating that there is no match information available. Clears any 787 * match region that may have been set. 788 * 789 * @param regexp The compiled regular expression. 790 * @param index The position (native) in the text at which a 791 * uregex_findNext() should begin searching. 792 * @param status A reference to a UErrorCode to receive any errors. 793 * @stable ICU 4.6 794 */ 795U_STABLE void U_EXPORT2 796uregex_reset64(URegularExpression *regexp, 797 int64_t index, 798 UErrorCode *status); 799 800/** 801 * Sets the limits of the matching region for this URegularExpression. 802 * The region is the part of the input string that will be considered when matching. 803 * Invoking this method resets any saved state from the previous match, 804 * then sets the region to start at the index specified by the start parameter 805 * and end at the index specified by the end parameter. 806 * 807 * Depending on the transparency and anchoring being used (see useTransparentBounds 808 * and useAnchoringBounds), certain constructs such as anchors may behave differently 809 * at or around the boundaries of the region 810 * 811 * The function will fail if start is greater than limit, or if either index 812 * is less than zero or greater than the length of the string being matched. 813 * 814 * @param regexp The compiled regular expression. 815 * @param regionStart The (native) index to begin searches at. 816 * @param regionLimit The (native) index to end searches at (exclusive). 817 * @param status A pointer to a UErrorCode to receive any errors. 818 * @stable ICU 4.0 819 */ 820U_STABLE void U_EXPORT2 821uregex_setRegion(URegularExpression *regexp, 822 int32_t regionStart, 823 int32_t regionLimit, 824 UErrorCode *status); 825 826/** 827 * 64bit version of uregex_setRegion. 828 * Sets the limits of the matching region for this URegularExpression. 829 * The region is the part of the input string that will be considered when matching. 830 * Invoking this method resets any saved state from the previous match, 831 * then sets the region to start at the index specified by the start parameter 832 * and end at the index specified by the end parameter. 833 * 834 * Depending on the transparency and anchoring being used (see useTransparentBounds 835 * and useAnchoringBounds), certain constructs such as anchors may behave differently 836 * at or around the boundaries of the region 837 * 838 * The function will fail if start is greater than limit, or if either index 839 * is less than zero or greater than the length of the string being matched. 840 * 841 * @param regexp The compiled regular expression. 842 * @param regionStart The (native) index to begin searches at. 843 * @param regionLimit The (native) index to end searches at (exclusive). 844 * @param status A pointer to a UErrorCode to receive any errors. 845 * @stable ICU 4.6 846 */ 847U_STABLE void U_EXPORT2 848uregex_setRegion64(URegularExpression *regexp, 849 int64_t regionStart, 850 int64_t regionLimit, 851 UErrorCode *status); 852 853/** 854 * Set the matching region and the starting index for subsequent matches 855 * in a single operation. 856 * This is useful because the usual function for setting the starting 857 * index, urgex_reset(), also resets any region limits. 858 * 859 * @param regexp The compiled regular expression. 860 * @param regionStart The (native) index to begin searches at. 861 * @param regionLimit The (native) index to end searches at (exclusive). 862 * @param startIndex The index in the input text at which the next 863 * match operation should begin. 864 * @param status A pointer to a UErrorCode to receive any errors. 865 * @stable ICU 4.6 866 */ 867U_STABLE void U_EXPORT2 868uregex_setRegionAndStart(URegularExpression *regexp, 869 int64_t regionStart, 870 int64_t regionLimit, 871 int64_t startIndex, 872 UErrorCode *status); 873 874/** 875 * Reports the start index of the matching region. Any matches found are limited to 876 * to the region bounded by regionStart (inclusive) and regionEnd (exclusive). 877 * 878 * @param regexp The compiled regular expression. 879 * @param status A pointer to a UErrorCode to receive any errors. 880 * @return The starting (native) index of this matcher's region. 881 * @stable ICU 4.0 882 */ 883U_STABLE int32_t U_EXPORT2 884uregex_regionStart(const URegularExpression *regexp, 885 UErrorCode *status); 886 887/** 888 * 64bit version of uregex_regionStart. 889 * Reports the start index of the matching region. Any matches found are limited to 890 * to the region bounded by regionStart (inclusive) and regionEnd (exclusive). 891 * 892 * @param regexp The compiled regular expression. 893 * @param status A pointer to a UErrorCode to receive any errors. 894 * @return The starting (native) index of this matcher's region. 895 * @stable ICU 4.6 896 */ 897U_STABLE int64_t U_EXPORT2 898uregex_regionStart64(const URegularExpression *regexp, 899 UErrorCode *status); 900 901/** 902 * Reports the end index (exclusive) of the matching region for this URegularExpression. 903 * Any matches found are limited to to the region bounded by regionStart (inclusive) 904 * and regionEnd (exclusive). 905 * 906 * @param regexp The compiled regular expression. 907 * @param status A pointer to a UErrorCode to receive any errors. 908 * @return The ending point (native) of this matcher's region. 909 * @stable ICU 4.0 910 */ 911U_STABLE int32_t U_EXPORT2 912uregex_regionEnd(const URegularExpression *regexp, 913 UErrorCode *status); 914 915/** 916 * 64bit version of uregex_regionEnd. 917 * Reports the end index (exclusive) of the matching region for this URegularExpression. 918 * Any matches found are limited to to the region bounded by regionStart (inclusive) 919 * and regionEnd (exclusive). 920 * 921 * @param regexp The compiled regular expression. 922 * @param status A pointer to a UErrorCode to receive any errors. 923 * @return The ending point (native) of this matcher's region. 924 * @stable ICU 4.6 925 */ 926U_STABLE int64_t U_EXPORT2 927uregex_regionEnd64(const URegularExpression *regexp, 928 UErrorCode *status); 929 930/** 931 * Queries the transparency of region bounds for this URegularExpression. 932 * See useTransparentBounds for a description of transparent and opaque bounds. 933 * By default, matching boundaries are opaque. 934 * 935 * @param regexp The compiled regular expression. 936 * @param status A pointer to a UErrorCode to receive any errors. 937 * @return TRUE if this matcher is using opaque bounds, false if it is not. 938 * @stable ICU 4.0 939 */ 940U_STABLE UBool U_EXPORT2 941uregex_hasTransparentBounds(const URegularExpression *regexp, 942 UErrorCode *status); 943 944 945/** 946 * Sets the transparency of region bounds for this URegularExpression. 947 * Invoking this function with an argument of TRUE will set matches to use transparent bounds. 948 * If the boolean argument is FALSE, then opaque bounds will be used. 949 * 950 * Using transparent bounds, the boundaries of the matching region are transparent 951 * to lookahead, lookbehind, and boundary matching constructs. Those constructs can 952 * see text beyond the boundaries of the region while checking for a match. 953 * 954 * With opaque bounds, no text outside of the matching region is visible to lookahead, 955 * lookbehind, and boundary matching constructs. 956 * 957 * By default, opaque bounds are used. 958 * 959 * @param regexp The compiled regular expression. 960 * @param b TRUE for transparent bounds; FALSE for opaque bounds 961 * @param status A pointer to a UErrorCode to receive any errors. 962 * @stable ICU 4.0 963 **/ 964U_STABLE void U_EXPORT2 965uregex_useTransparentBounds(URegularExpression *regexp, 966 UBool b, 967 UErrorCode *status); 968 969 970/** 971 * Return true if this URegularExpression is using anchoring bounds. 972 * By default, anchoring region bounds are used. 973 * 974 * @param regexp The compiled regular expression. 975 * @param status A pointer to a UErrorCode to receive any errors. 976 * @return TRUE if this matcher is using anchoring bounds. 977 * @stable ICU 4.0 978 */ 979U_STABLE UBool U_EXPORT2 980uregex_hasAnchoringBounds(const URegularExpression *regexp, 981 UErrorCode *status); 982 983 984/** 985 * Set whether this URegularExpression is using Anchoring Bounds for its region. 986 * With anchoring bounds, pattern anchors such as ^ and $ will match at the start 987 * and end of the region. Without Anchoring Bounds, anchors will only match at 988 * the positions they would in the complete text. 989 * 990 * Anchoring Bounds are the default for regions. 991 * 992 * @param regexp The compiled regular expression. 993 * @param b TRUE if to enable anchoring bounds; FALSE to disable them. 994 * @param status A pointer to a UErrorCode to receive any errors. 995 * @stable ICU 4.0 996 */ 997U_STABLE void U_EXPORT2 998uregex_useAnchoringBounds(URegularExpression *regexp, 999 UBool b, 1000 UErrorCode *status); 1001 1002/** 1003 * Return TRUE if the most recent matching operation touched the 1004 * end of the text being processed. In this case, additional input text could 1005 * change the results of that match. 1006 * 1007 * @param regexp The compiled regular expression. 1008 * @param status A pointer to a UErrorCode to receive any errors. 1009 * @return TRUE if the most recent match hit the end of input 1010 * @stable ICU 4.0 1011 */ 1012U_STABLE UBool U_EXPORT2 1013uregex_hitEnd(const URegularExpression *regexp, 1014 UErrorCode *status); 1015 1016/** 1017 * Return TRUE the most recent match succeeded and additional input could cause 1018 * it to fail. If this function returns false and a match was found, then more input 1019 * might change the match but the match won't be lost. If a match was not found, 1020 * then requireEnd has no meaning. 1021 * 1022 * @param regexp The compiled regular expression. 1023 * @param status A pointer to a UErrorCode to receive any errors. 1024 * @return TRUE if more input could cause the most recent match to no longer match. 1025 * @stable ICU 4.0 1026 */ 1027U_STABLE UBool U_EXPORT2 1028uregex_requireEnd(const URegularExpression *regexp, 1029 UErrorCode *status); 1030 1031 1032 1033 1034 1035/** 1036 * Replaces every substring of the input that matches the pattern 1037 * with the given replacement string. This is a convenience function that 1038 * provides a complete find-and-replace-all operation. 1039 * 1040 * This method scans the input string looking for matches of the pattern. 1041 * Input that is not part of any match is copied unchanged to the 1042 * destination buffer. Matched regions are replaced in the output 1043 * buffer by the replacement string. The replacement string may contain 1044 * references to capture groups; these take the form of $1, $2, etc. 1045 * 1046 * @param regexp The compiled regular expression. 1047 * @param replacementText A string containing the replacement text. 1048 * @param replacementLength The length of the replacement string, or 1049 * -1 if it is NUL terminated. 1050 * @param destBuf A (UChar *) buffer that will receive the result. 1051 * @param destCapacity The capacity of the destination buffer. 1052 * @param status A reference to a UErrorCode to receive any errors. 1053 * @return The length of the string resulting from the find 1054 * and replace operation. In the event that the 1055 * destination capacity is inadequate, the return value 1056 * is still the full length of the untruncated string. 1057 * @stable ICU 3.0 1058 */ 1059U_STABLE int32_t U_EXPORT2 1060uregex_replaceAll(URegularExpression *regexp, 1061 const UChar *replacementText, 1062 int32_t replacementLength, 1063 UChar *destBuf, 1064 int32_t destCapacity, 1065 UErrorCode *status); 1066 1067/** 1068 * Replaces every substring of the input that matches the pattern 1069 * with the given replacement string. This is a convenience function that 1070 * provides a complete find-and-replace-all operation. 1071 * 1072 * This method scans the input string looking for matches of the pattern. 1073 * Input that is not part of any match is copied unchanged to the 1074 * destination buffer. Matched regions are replaced in the output 1075 * buffer by the replacement string. The replacement string may contain 1076 * references to capture groups; these take the form of $1, $2, etc. 1077 * 1078 * @param regexp The compiled regular expression. 1079 * @param replacement A string containing the replacement text. 1080 * @param dest A mutable UText that will receive the result. 1081 * If NULL, a new UText will be created (which may not be mutable). 1082 * @param status A reference to a UErrorCode to receive any errors. 1083 * @return A UText containing the results of the find and replace. 1084 * If a pre-allocated UText was provided, it will always be used and returned. 1085 * 1086 * @stable ICU 4.6 1087 */ 1088U_STABLE UText * U_EXPORT2 1089uregex_replaceAllUText(URegularExpression *regexp, 1090 UText *replacement, 1091 UText *dest, 1092 UErrorCode *status); 1093 1094/** 1095 * Replaces the first substring of the input that matches the pattern 1096 * with the given replacement string. This is a convenience function that 1097 * provides a complete find-and-replace operation. 1098 * 1099 * This method scans the input string looking for a match of the pattern. 1100 * All input that is not part of the match is copied unchanged to the 1101 * destination buffer. The matched region is replaced in the output 1102 * buffer by the replacement string. The replacement string may contain 1103 * references to capture groups; these take the form of $1, $2, etc. 1104 * 1105 * @param regexp The compiled regular expression. 1106 * @param replacementText A string containing the replacement text. 1107 * @param replacementLength The length of the replacement string, or 1108 * -1 if it is NUL terminated. 1109 * @param destBuf A (UChar *) buffer that will receive the result. 1110 * @param destCapacity The capacity of the destination buffer. 1111 * @param status a reference to a UErrorCode to receive any errors. 1112 * @return The length of the string resulting from the find 1113 * and replace operation. In the event that the 1114 * destination capacity is inadequate, the return value 1115 * is still the full length of the untruncated string. 1116 * @stable ICU 3.0 1117 */ 1118U_STABLE int32_t U_EXPORT2 1119uregex_replaceFirst(URegularExpression *regexp, 1120 const UChar *replacementText, 1121 int32_t replacementLength, 1122 UChar *destBuf, 1123 int32_t destCapacity, 1124 UErrorCode *status); 1125 1126/** 1127 * Replaces the first substring of the input that matches the pattern 1128 * with the given replacement string. This is a convenience function that 1129 * provides a complete find-and-replace operation. 1130 * 1131 * This method scans the input string looking for a match of the pattern. 1132 * All input that is not part of the match is copied unchanged to the 1133 * destination buffer. The matched region is replaced in the output 1134 * buffer by the replacement string. The replacement string may contain 1135 * references to capture groups; these take the form of $1, $2, etc. 1136 * 1137 * @param regexp The compiled regular expression. 1138 * @param replacement A string containing the replacement text. 1139 * @param dest A mutable UText that will receive the result. 1140 * If NULL, a new UText will be created (which may not be mutable). 1141 * @param status A reference to a UErrorCode to receive any errors. 1142 * @return A UText containing the results of the find and replace. 1143 * If a pre-allocated UText was provided, it will always be used and returned. 1144 * 1145 * @stable ICU 4.6 1146 */ 1147U_STABLE UText * U_EXPORT2 1148uregex_replaceFirstUText(URegularExpression *regexp, 1149 UText *replacement, 1150 UText *dest, 1151 UErrorCode *status); 1152 1153/** 1154 * Implements a replace operation intended to be used as part of an 1155 * incremental find-and-replace. 1156 * 1157 * <p>The input string, starting from the end of the previous match and ending at 1158 * the start of the current match, is appended to the destination string. Then the 1159 * replacement string is appended to the output string, 1160 * including handling any substitutions of captured text.</p> 1161 * 1162 * <p>A note on preflight computation of buffersize and error handling: 1163 * Calls to uregex_appendReplacement() and uregex_appendTail() are 1164 * designed to be chained, one after another, with the destination 1165 * buffer pointer and buffer capacity updated after each in preparation 1166 * to for the next. If the destination buffer is exhausted partway through such a 1167 * sequence, a U_BUFFER_OVERFLOW_ERROR status will be returned. Normal 1168 * ICU conventions are for a function to perform no action if it is 1169 * called with an error status, but for this one case, uregex_appendRepacement() 1170 * will operate normally so that buffer size computations will complete 1171 * correctly. 1172 * 1173 * <p>For simple, prepackaged, non-incremental find-and-replace 1174 * operations, see replaceFirst() or replaceAll().</p> 1175 * 1176 * @param regexp The regular expression object. 1177 * @param replacementText The string that will replace the matched portion of the 1178 * input string as it is copied to the destination buffer. 1179 * The replacement text may contain references ($1, for 1180 * example) to capture groups from the match. 1181 * @param replacementLength The length of the replacement text string, 1182 * or -1 if the string is NUL terminated. 1183 * @param destBuf The buffer into which the results of the 1184 * find-and-replace are placed. On return, this pointer 1185 * will be updated to refer to the beginning of the 1186 * unused portion of buffer, leaving it in position for 1187 * a subsequent call to this function. 1188 * @param destCapacity The size of the output buffer, On return, this 1189 * parameter will be updated to reflect the space remaining 1190 * unused in the output buffer. 1191 * @param status A reference to a UErrorCode to receive any errors. 1192 * @return The length of the result string. In the event that 1193 * destCapacity is inadequate, the full length of the 1194 * untruncated output string is returned. 1195 * 1196 * @stable ICU 3.0 1197 * 1198 */ 1199U_STABLE int32_t U_EXPORT2 1200uregex_appendReplacement(URegularExpression *regexp, 1201 const UChar *replacementText, 1202 int32_t replacementLength, 1203 UChar **destBuf, 1204 int32_t *destCapacity, 1205 UErrorCode *status); 1206 1207/** 1208 * Implements a replace operation intended to be used as part of an 1209 * incremental find-and-replace. 1210 * 1211 * <p>The input string, starting from the end of the previous match and ending at 1212 * the start of the current match, is appended to the destination string. Then the 1213 * replacement string is appended to the output string, 1214 * including handling any substitutions of captured text.</p> 1215 * 1216 * <p>For simple, prepackaged, non-incremental find-and-replace 1217 * operations, see replaceFirst() or replaceAll().</p> 1218 * 1219 * @param regexp The regular expression object. 1220 * @param replacementText The string that will replace the matched portion of the 1221 * input string as it is copied to the destination buffer. 1222 * The replacement text may contain references ($1, for 1223 * example) to capture groups from the match. 1224 * @param dest A mutable UText that will receive the result. Must not be NULL. 1225 * @param status A reference to a UErrorCode to receive any errors. 1226 * 1227 * @stable ICU 4.6 1228 */ 1229U_STABLE void U_EXPORT2 1230uregex_appendReplacementUText(URegularExpression *regexp, 1231 UText *replacementText, 1232 UText *dest, 1233 UErrorCode *status); 1234 1235/** 1236 * As the final step in a find-and-replace operation, append the remainder 1237 * of the input string, starting at the position following the last match, 1238 * to the destination string. <code>uregex_appendTail()</code> is intended 1239 * to be invoked after one or more invocations of the 1240 * <code>uregex_appendReplacement()</code> function. 1241 * 1242 * @param regexp The regular expression object. This is needed to 1243 * obtain the input string and with the position 1244 * of the last match within it. 1245 * @param destBuf The buffer in which the results of the 1246 * find-and-replace are placed. On return, the pointer 1247 * will be updated to refer to the beginning of the 1248 * unused portion of buffer. 1249 * @param destCapacity The size of the output buffer, On return, this 1250 * value will be updated to reflect the space remaining 1251 * unused in the output buffer. 1252 * @param status A reference to a UErrorCode to receive any errors. 1253 * @return The length of the result string. In the event that 1254 * destCapacity is inadequate, the full length of the 1255 * untruncated output string is returned. 1256 * 1257 * @stable ICU 3.0 1258 */ 1259U_STABLE int32_t U_EXPORT2 1260uregex_appendTail(URegularExpression *regexp, 1261 UChar **destBuf, 1262 int32_t *destCapacity, 1263 UErrorCode *status); 1264 1265/** 1266 * As the final step in a find-and-replace operation, append the remainder 1267 * of the input string, starting at the position following the last match, 1268 * to the destination string. <code>uregex_appendTailUText()</code> is intended 1269 * to be invoked after one or more invocations of the 1270 * <code>uregex_appendReplacementUText()</code> function. 1271 * 1272 * @param regexp The regular expression object. This is needed to 1273 * obtain the input string and with the position 1274 * of the last match within it. 1275 * @param dest A mutable UText that will receive the result. Must not be NULL. 1276 * 1277 * @param status Error code 1278 * 1279 * @return The destination UText. 1280 * 1281 * @stable ICU 4.6 1282 */ 1283U_STABLE UText * U_EXPORT2 1284uregex_appendTailUText(URegularExpression *regexp, 1285 UText *dest, 1286 UErrorCode *status); 1287 1288 /** 1289 * Split a string into fields. Somewhat like split() from Perl. 1290 * The pattern matches identify delimiters that separate the input 1291 * into fields. The input data between the matches becomes the 1292 * fields themselves. 1293 * 1294 * Each of the fields is copied from the input string to the destination 1295 * buffer, and NUL terminated. The position of each field within 1296 * the destination buffer is returned in the destFields array. 1297 * 1298 * If the delimiter pattern includes capture groups, the captured text will 1299 * also appear in the destination array of output strings, interspersed 1300 * with the fields. This is similar to Perl, but differs from Java, 1301 * which ignores the presence of capture groups in the pattern. 1302 * 1303 * Trailing empty fields will always be returned, assuming sufficient 1304 * destination capacity. This differs from the default behavior for Java 1305 * and Perl where trailing empty fields are not returned. 1306 * 1307 * The number of strings produced by the split operation is returned. 1308 * This count includes the strings from capture groups in the delimiter pattern. 1309 * This behavior differs from Java, which ignores capture groups. 1310 * 1311 * @param regexp The compiled regular expression. 1312 * @param destBuf A (UChar *) buffer to receive the fields that 1313 * are extracted from the input string. These 1314 * field pointers will refer to positions within the 1315 * destination buffer supplied by the caller. Any 1316 * extra positions within the destFields array will be 1317 * set to NULL. 1318 * @param destCapacity The capacity of the destBuf. 1319 * @param requiredCapacity The actual capacity required of the destBuf. 1320 * If destCapacity is too small, requiredCapacity will return 1321 * the total capacity required to hold all of the output, and 1322 * a U_BUFFER_OVERFLOW_ERROR will be returned. 1323 * @param destFields An array to be filled with the position of each 1324 * of the extracted fields within destBuf. 1325 * @param destFieldsCapacity The number of elements in the destFields array. 1326 * If the number of fields found is less than destFieldsCapacity, 1327 * the extra destFields elements are set to zero. 1328 * If destFieldsCapacity is too small, the trailing part of the 1329 * input, including any field delimiters, is treated as if it 1330 * were the last field - it is copied to the destBuf, and 1331 * its position is in the destBuf is stored in the last element 1332 * of destFields. This behavior mimics that of Perl. It is not 1333 * an error condition, and no error status is returned when all destField 1334 * positions are used. 1335 * @param status A reference to a UErrorCode to receive any errors. 1336 * @return The number of fields into which the input string was split. 1337 * @stable ICU 3.0 1338 */ 1339U_STABLE int32_t U_EXPORT2 1340uregex_split( URegularExpression *regexp, 1341 UChar *destBuf, 1342 int32_t destCapacity, 1343 int32_t *requiredCapacity, 1344 UChar *destFields[], 1345 int32_t destFieldsCapacity, 1346 UErrorCode *status); 1347 1348 /** 1349 * Split a string into fields. Somewhat like split() from Perl. 1350 * The pattern matches identify delimiters that separate the input 1351 * into fields. The input data between the matches becomes the 1352 * fields themselves. 1353 * <p> 1354 * The behavior of this function is not very closely aligned with uregex_split(); 1355 * instead, it is based on (and implemented directly on top of) the C++ split method. 1356 * 1357 * @param regexp The compiled regular expression. 1358 * @param destFields An array of mutable UText structs to receive the results of the split. 1359 * If a field is NULL, a new UText is allocated to contain the results for 1360 * that field. This new UText is not guaranteed to be mutable. 1361 * @param destFieldsCapacity The number of elements in the destination array. 1362 * If the number of fields found is less than destCapacity, the 1363 * extra strings in the destination array are not altered. 1364 * If the number of destination strings is less than the number 1365 * of fields, the trailing part of the input string, including any 1366 * field delimiters, is placed in the last destination string. 1367 * This behavior mimics that of Perl. It is not an error condition, and no 1368 * error status is returned when all destField positions are used. 1369 * @param status A reference to a UErrorCode to receive any errors. 1370 * @return The number of fields into which the input string was split. 1371 * 1372 * @stable ICU 4.6 1373 */ 1374U_STABLE int32_t U_EXPORT2 1375uregex_splitUText(URegularExpression *regexp, 1376 UText *destFields[], 1377 int32_t destFieldsCapacity, 1378 UErrorCode *status); 1379 1380/** 1381 * Set a processing time limit for match operations with this URegularExpression. 1382 * 1383 * Some patterns, when matching certain strings, can run in exponential time. 1384 * For practical purposes, the match operation may appear to be in an 1385 * infinite loop. 1386 * When a limit is set a match operation will fail with an error if the 1387 * limit is exceeded. 1388 * <p> 1389 * The units of the limit are steps of the match engine. 1390 * Correspondence with actual processor time will depend on the speed 1391 * of the processor and the details of the specific pattern, but will 1392 * typically be on the order of milliseconds. 1393 * <p> 1394 * By default, the matching time is not limited. 1395 * <p> 1396 * 1397 * @param regexp The compiled regular expression. 1398 * @param limit The limit value, or 0 for no limit. 1399 * @param status A reference to a UErrorCode to receive any errors. 1400 * @stable ICU 4.0 1401 */ 1402U_STABLE void U_EXPORT2 1403uregex_setTimeLimit(URegularExpression *regexp, 1404 int32_t limit, 1405 UErrorCode *status); 1406 1407/** 1408 * Get the time limit for for matches with this URegularExpression. 1409 * A return value of zero indicates that there is no limit. 1410 * 1411 * @param regexp The compiled regular expression. 1412 * @param status A reference to a UErrorCode to receive any errors. 1413 * @return the maximum allowed time for a match, in units of processing steps. 1414 * @stable ICU 4.0 1415 */ 1416U_STABLE int32_t U_EXPORT2 1417uregex_getTimeLimit(const URegularExpression *regexp, 1418 UErrorCode *status); 1419 1420/** 1421 * Set the amount of heap storage available for use by the match backtracking stack. 1422 * <p> 1423 * ICU uses a backtracking regular expression engine, with the backtrack stack 1424 * maintained on the heap. This function sets the limit to the amount of memory 1425 * that can be used for this purpose. A backtracking stack overflow will 1426 * result in an error from the match operation that caused it. 1427 * <p> 1428 * A limit is desirable because a malicious or poorly designed pattern can use 1429 * excessive memory, potentially crashing the process. A limit is enabled 1430 * by default. 1431 * <p> 1432 * @param regexp The compiled regular expression. 1433 * @param limit The maximum size, in bytes, of the matching backtrack stack. 1434 * A value of -1 means no limit. 1435 * The limit must be greater than zero, or -1. 1436 * @param status A reference to a UErrorCode to receive any errors. 1437 * 1438 * @stable ICU 4.0 1439 */ 1440U_STABLE void U_EXPORT2 1441uregex_setStackLimit(URegularExpression *regexp, 1442 int32_t limit, 1443 UErrorCode *status); 1444 1445/** 1446 * Get the size of the heap storage available for use by the back tracking stack. 1447 * 1448 * @return the maximum backtracking stack size, in bytes, or zero if the 1449 * stack size is unlimited. 1450 * @stable ICU 4.0 1451 */ 1452U_STABLE int32_t U_EXPORT2 1453uregex_getStackLimit(const URegularExpression *regexp, 1454 UErrorCode *status); 1455 1456 1457/** 1458 * Function pointer for a regular expression matching callback function. 1459 * When set, a callback function will be called periodically during matching 1460 * operations. If the call back function returns FALSE, the matching 1461 * operation will be terminated early. 1462 * 1463 * Note: the callback function must not call other functions on this 1464 * URegularExpression. 1465 * 1466 * @param context context pointer. The callback function will be invoked 1467 * with the context specified at the time that 1468 * uregex_setMatchCallback() is called. 1469 * @param steps the accumulated processing time, in match steps, 1470 * for this matching operation. 1471 * @return TRUE to continue the matching operation. 1472 * FALSE to terminate the matching operation. 1473 * @stable ICU 4.0 1474 */ 1475U_CDECL_BEGIN 1476typedef UBool U_CALLCONV URegexMatchCallback ( 1477 const void *context, 1478 int32_t steps); 1479U_CDECL_END 1480 1481/** 1482 * Set a callback function for this URegularExpression. 1483 * During matching operations the function will be called periodically, 1484 * giving the application the opportunity to terminate a long-running 1485 * match. 1486 * 1487 * @param regexp The compiled regular expression. 1488 * @param callback A pointer to the user-supplied callback function. 1489 * @param context User context pointer. The value supplied at the 1490 * time the callback function is set will be saved 1491 * and passed to the callback each time that it is called. 1492 * @param status A reference to a UErrorCode to receive any errors. 1493 * @stable ICU 4.0 1494 */ 1495U_STABLE void U_EXPORT2 1496uregex_setMatchCallback(URegularExpression *regexp, 1497 URegexMatchCallback *callback, 1498 const void *context, 1499 UErrorCode *status); 1500 1501 1502/** 1503 * Get the callback function for this URegularExpression. 1504 * 1505 * @param regexp The compiled regular expression. 1506 * @param callback Out parameter, receives a pointer to the user-supplied 1507 * callback function. 1508 * @param context Out parameter, receives the user context pointer that 1509 * was set when uregex_setMatchCallback() was called. 1510 * @param status A reference to a UErrorCode to receive any errors. 1511 * @stable ICU 4.0 1512 */ 1513U_STABLE void U_EXPORT2 1514uregex_getMatchCallback(const URegularExpression *regexp, 1515 URegexMatchCallback **callback, 1516 const void **context, 1517 UErrorCode *status); 1518 1519/** 1520 * Function pointer for a regular expression find callback function. 1521 * 1522 * When set, a callback function will be called during a find operation 1523 * and for operations that depend on find, such as findNext, split and some replace 1524 * operations like replaceFirst. 1525 * The callback will usually be called after each attempt at a match, but this is not a 1526 * guarantee that the callback will be invoked at each character. For finds where the 1527 * match engine is invoked at each character, this may be close to true, but less likely 1528 * for more optimized loops where the pattern is known to only start, and the match 1529 * engine invoked, at certain characters. 1530 * When invoked, this callback will specify the index at which a match operation is about 1531 * to be attempted, giving the application the opportunity to terminate a long-running 1532 * find operation. 1533 * 1534 * If the call back function returns FALSE, the find operation will be terminated early. 1535 * 1536 * Note: the callback function must not call other functions on this 1537 * URegularExpression 1538 * 1539 * @param context context pointer. The callback function will be invoked 1540 * with the context specified at the time that 1541 * uregex_setFindProgressCallback() is called. 1542 * @param matchIndex the next index at which a match attempt will be attempted for this 1543 * find operation. If this callback interrupts the search, this is the 1544 * index at which a find/findNext operation may be re-initiated. 1545 * @return TRUE to continue the matching operation. 1546 * FALSE to terminate the matching operation. 1547 * @stable ICU 4.6 1548 */ 1549U_CDECL_BEGIN 1550typedef UBool U_CALLCONV URegexFindProgressCallback ( 1551 const void *context, 1552 int64_t matchIndex); 1553U_CDECL_END 1554 1555 1556/** 1557 * Set the find progress callback function for this URegularExpression. 1558 * 1559 * @param regexp The compiled regular expression. 1560 * @param callback A pointer to the user-supplied callback function. 1561 * @param context User context pointer. The value supplied at the 1562 * time the callback function is set will be saved 1563 * and passed to the callback each time that it is called. 1564 * @param status A reference to a UErrorCode to receive any errors. 1565 * @stable ICU 4.6 1566 */ 1567U_STABLE void U_EXPORT2 1568uregex_setFindProgressCallback(URegularExpression *regexp, 1569 URegexFindProgressCallback *callback, 1570 const void *context, 1571 UErrorCode *status); 1572 1573/** 1574 * Get the find progress callback function for this URegularExpression. 1575 * 1576 * @param regexp The compiled regular expression. 1577 * @param callback Out parameter, receives a pointer to the user-supplied 1578 * callback function. 1579 * @param context Out parameter, receives the user context pointer that 1580 * was set when uregex_setFindProgressCallback() was called. 1581 * @param status A reference to a UErrorCode to receive any errors. 1582 * @stable ICU 4.6 1583 */ 1584U_STABLE void U_EXPORT2 1585uregex_getFindProgressCallback(const URegularExpression *regexp, 1586 URegexFindProgressCallback **callback, 1587 const void **context, 1588 UErrorCode *status); 1589 1590#endif /* !UCONFIG_NO_REGULAR_EXPRESSIONS */ 1591#endif /* UREGEX_H */ 1592