1<html> 2<head> 3<title>pcrepattern specification</title> 4</head> 5<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB"> 6<h1>pcrepattern man page</h1> 7<p> 8Return to the <a href="index.html">PCRE index page</a>. 9</p> 10<p> 11This page is part of the PCRE HTML documentation. It was generated automatically 12from the original man page. If there is any nonsense in it, please consult the 13man page, in case the conversion went wrong. 14<br> 15<ul> 16<li><a name="TOC1" href="#SEC1">PCRE REGULAR EXPRESSION DETAILS</a> 17<li><a name="TOC2" href="#SEC2">NEWLINE CONVENTIONS</a> 18<li><a name="TOC3" href="#SEC3">CHARACTERS AND METACHARACTERS</a> 19<li><a name="TOC4" href="#SEC4">BACKSLASH</a> 20<li><a name="TOC5" href="#SEC5">CIRCUMFLEX AND DOLLAR</a> 21<li><a name="TOC6" href="#SEC6">FULL STOP (PERIOD, DOT)</a> 22<li><a name="TOC7" href="#SEC7">MATCHING A SINGLE BYTE</a> 23<li><a name="TOC8" href="#SEC8">SQUARE BRACKETS AND CHARACTER CLASSES</a> 24<li><a name="TOC9" href="#SEC9">POSIX CHARACTER CLASSES</a> 25<li><a name="TOC10" href="#SEC10">VERTICAL BAR</a> 26<li><a name="TOC11" href="#SEC11">INTERNAL OPTION SETTING</a> 27<li><a name="TOC12" href="#SEC12">SUBPATTERNS</a> 28<li><a name="TOC13" href="#SEC13">DUPLICATE SUBPATTERN NUMBERS</a> 29<li><a name="TOC14" href="#SEC14">NAMED SUBPATTERNS</a> 30<li><a name="TOC15" href="#SEC15">REPETITION</a> 31<li><a name="TOC16" href="#SEC16">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a> 32<li><a name="TOC17" href="#SEC17">BACK REFERENCES</a> 33<li><a name="TOC18" href="#SEC18">ASSERTIONS</a> 34<li><a name="TOC19" href="#SEC19">CONDITIONAL SUBPATTERNS</a> 35<li><a name="TOC20" href="#SEC20">COMMENTS</a> 36<li><a name="TOC21" href="#SEC21">RECURSIVE PATTERNS</a> 37<li><a name="TOC22" href="#SEC22">SUBPATTERNS AS SUBROUTINES</a> 38<li><a name="TOC23" href="#SEC23">ONIGURUMA SUBROUTINE SYNTAX</a> 39<li><a name="TOC24" href="#SEC24">CALLOUTS</a> 40<li><a name="TOC25" href="#SEC25">BACKTRACKING CONTROL</a> 41<li><a name="TOC26" href="#SEC26">SEE ALSO</a> 42<li><a name="TOC27" href="#SEC27">AUTHOR</a> 43<li><a name="TOC28" href="#SEC28">REVISION</a> 44</ul> 45<br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br> 46<P> 47The syntax and semantics of the regular expressions that are supported by PCRE 48are described in detail below. There is a quick-reference syntax summary in the 49<a href="pcresyntax.html"><b>pcresyntax</b></a> 50page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE 51also supports some alternative regular expression syntax (which does not 52conflict with the Perl syntax) in order to provide some compatibility with 53regular expressions in Python, .NET, and Oniguruma. 54</P> 55<P> 56Perl's regular expressions are described in its own documentation, and 57regular expressions in general are covered in a number of books, some of which 58have copious examples. Jeffrey Friedl's "Mastering Regular Expressions", 59published by O'Reilly, covers regular expressions in great detail. This 60description of PCRE's regular expressions is intended as reference material. 61</P> 62<P> 63The original operation of PCRE was on strings of one-byte characters. However, 64there is now also support for UTF-8 character strings. To use this, 65PCRE must be built to include UTF-8 support, and you must call 66<b>pcre_compile()</b> or <b>pcre_compile2()</b> with the PCRE_UTF8 option. There 67is also a special sequence that can be given at the start of a pattern: 68<pre> 69 (*UTF8) 70</pre> 71Starting a pattern with this sequence is equivalent to setting the PCRE_UTF8 72option. This feature is not Perl-compatible. How setting UTF-8 mode affects 73pattern matching is mentioned in several places below. There is also a summary 74of UTF-8 features in the 75<a href="pcre.html#utf8support">section on UTF-8 support</a> 76in the main 77<a href="pcre.html"><b>pcre</b></a> 78page. 79</P> 80<P> 81The remainder of this document discusses the patterns that are supported by 82PCRE when its main matching function, <b>pcre_exec()</b>, is used. 83From release 6.0, PCRE offers a second matching function, 84<b>pcre_dfa_exec()</b>, which matches using a different algorithm that is not 85Perl-compatible. Some of the features discussed below are not available when 86<b>pcre_dfa_exec()</b> is used. The advantages and disadvantages of the 87alternative function, and how it differs from the normal function, are 88discussed in the 89<a href="pcrematching.html"><b>pcrematching</b></a> 90page. 91</P> 92<br><a name="SEC2" href="#TOC1">NEWLINE CONVENTIONS</a><br> 93<P> 94PCRE supports five different conventions for indicating line breaks in 95strings: a single CR (carriage return) character, a single LF (linefeed) 96character, the two-character sequence CRLF, any of the three preceding, or any 97Unicode newline sequence. The 98<a href="pcreapi.html"><b>pcreapi</b></a> 99page has 100<a href="pcreapi.html#newlines">further discussion</a> 101about newlines, and shows how to set the newline convention in the 102<i>options</i> arguments for the compiling and matching functions. 103</P> 104<P> 105It is also possible to specify a newline convention by starting a pattern 106string with one of the following five sequences: 107<pre> 108 (*CR) carriage return 109 (*LF) linefeed 110 (*CRLF) carriage return, followed by linefeed 111 (*ANYCRLF) any of the three above 112 (*ANY) all Unicode newline sequences 113</pre> 114These override the default and the options given to <b>pcre_compile()</b> or 115<b>pcre_compile2()</b>. For example, on a Unix system where LF is the default 116newline sequence, the pattern 117<pre> 118 (*CR)a.b 119</pre> 120changes the convention to CR. That pattern matches "a\nb" because LF is no 121longer a newline. Note that these special settings, which are not 122Perl-compatible, are recognized only at the very start of a pattern, and that 123they must be in upper case. If more than one of them is present, the last one 124is used. 125</P> 126<P> 127The newline convention does not affect what the \R escape sequence matches. By 128default, this is any Unicode newline sequence, for Perl compatibility. However, 129this can be changed; see the description of \R in the section entitled 130<a href="#newlineseq">"Newline sequences"</a> 131below. A change of \R setting can be combined with a change of newline 132convention. 133</P> 134<br><a name="SEC3" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br> 135<P> 136A regular expression is a pattern that is matched against a subject string from 137left to right. Most characters stand for themselves in a pattern, and match the 138corresponding characters in the subject. As a trivial example, the pattern 139<pre> 140 The quick brown fox 141</pre> 142matches a portion of a subject string that is identical to itself. When 143caseless matching is specified (the PCRE_CASELESS option), letters are matched 144independently of case. In UTF-8 mode, PCRE always understands the concept of 145case for characters whose values are less than 128, so caseless matching is 146always possible. For characters with higher values, the concept of case is 147supported if PCRE is compiled with Unicode property support, but not otherwise. 148If you want to use caseless matching for characters 128 and above, you must 149ensure that PCRE is compiled with Unicode property support as well as with 150UTF-8 support. 151</P> 152<P> 153The power of regular expressions comes from the ability to include alternatives 154and repetitions in the pattern. These are encoded in the pattern by the use of 155<i>metacharacters</i>, which do not stand for themselves but instead are 156interpreted in some special way. 157</P> 158<P> 159There are two different sets of metacharacters: those that are recognized 160anywhere in the pattern except within square brackets, and those that are 161recognized within square brackets. Outside square brackets, the metacharacters 162are as follows: 163<pre> 164 \ general escape character with several uses 165 ^ assert start of string (or line, in multiline mode) 166 $ assert end of string (or line, in multiline mode) 167 . match any character except newline (by default) 168 [ start character class definition 169 | start of alternative branch 170 ( start subpattern 171 ) end subpattern 172 ? extends the meaning of ( 173 also 0 or 1 quantifier 174 also quantifier minimizer 175 * 0 or more quantifier 176 + 1 or more quantifier 177 also "possessive quantifier" 178 { start min/max quantifier 179</pre> 180Part of a pattern that is in square brackets is called a "character class". In 181a character class the only metacharacters are: 182<pre> 183 \ general escape character 184 ^ negate the class, but only if the first character 185 - indicates character range 186 [ POSIX character class (only if followed by POSIX syntax) 187 ] terminates the character class 188</pre> 189The following sections describe the use of each of the metacharacters. 190</P> 191<br><a name="SEC4" href="#TOC1">BACKSLASH</a><br> 192<P> 193The backslash character has several uses. Firstly, if it is followed by a 194non-alphanumeric character, it takes away any special meaning that character 195may have. This use of backslash as an escape character applies both inside and 196outside character classes. 197</P> 198<P> 199For example, if you want to match a * character, you write \* in the pattern. 200This escaping action applies whether or not the following character would 201otherwise be interpreted as a metacharacter, so it is always safe to precede a 202non-alphanumeric with backslash to specify that it stands for itself. In 203particular, if you want to match a backslash, you write \\. 204</P> 205<P> 206If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the 207pattern (other than in a character class) and characters between a # outside 208a character class and the next newline are ignored. An escaping backslash can 209be used to include a whitespace or # character as part of the pattern. 210</P> 211<P> 212If you want to remove the special meaning from a sequence of characters, you 213can do so by putting them between \Q and \E. This is different from Perl in 214that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in 215Perl, $ and @ cause variable interpolation. Note the following examples: 216<pre> 217 Pattern PCRE matches Perl matches 218 219 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz 220 \Qabc\$xyz\E abc\$xyz abc\$xyz 221 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz 222</pre> 223The \Q...\E sequence is recognized both inside and outside character classes. 224<a name="digitsafterbackslash"></a></P> 225<br><b> 226Non-printing characters 227</b><br> 228<P> 229A second use of backslash provides a way of encoding non-printing characters 230in patterns in a visible manner. There is no restriction on the appearance of 231non-printing characters, apart from the binary zero that terminates a pattern, 232but when a pattern is being prepared by text editing, it is often easier to use 233one of the following escape sequences than the binary character it represents: 234<pre> 235 \a alarm, that is, the BEL character (hex 07) 236 \cx "control-x", where x is any character 237 \e escape (hex 1B) 238 \f formfeed (hex 0C) 239 \n linefeed (hex 0A) 240 \r carriage return (hex 0D) 241 \t tab (hex 09) 242 \ddd character with octal code ddd, or back reference 243 \xhh character with hex code hh 244 \x{hhh..} character with hex code hhh.. 245</pre> 246The precise effect of \cx is as follows: if x is a lower case letter, it 247is converted to upper case. Then bit 6 of the character (hex 40) is inverted. 248Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c; becomes hex 2497B. 250</P> 251<P> 252After \x, from zero to two hexadecimal digits are read (letters can be in 253upper or lower case). Any number of hexadecimal digits may appear between \x{ 254and }, but the value of the character code must be less than 256 in non-UTF-8 255mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in 256hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code 257point, which is 10FFFF. 258</P> 259<P> 260If characters other than hexadecimal digits appear between \x{ and }, or if 261there is no terminating }, this form of escape is not recognized. Instead, the 262initial \x will be interpreted as a basic hexadecimal escape, with no 263following digits, giving a character whose value is zero. 264</P> 265<P> 266Characters whose value is less than 256 can be defined by either of the two 267syntaxes for \x. There is no difference in the way they are handled. For 268example, \xdc is exactly the same as \x{dc}. 269</P> 270<P> 271After \0 up to two further octal digits are read. If there are fewer than two 272digits, just those that are present are used. Thus the sequence \0\x\07 273specifies two binary zeros followed by a BEL character (code value 7). Make 274sure you supply two digits after the initial zero if the pattern character that 275follows is itself an octal digit. 276</P> 277<P> 278The handling of a backslash followed by a digit other than 0 is complicated. 279Outside a character class, PCRE reads it and any following digits as a decimal 280number. If the number is less than 10, or if there have been at least that many 281previous capturing left parentheses in the expression, the entire sequence is 282taken as a <i>back reference</i>. A description of how this works is given 283<a href="#backreferences">later,</a> 284following the discussion of 285<a href="#subpattern">parenthesized subpatterns.</a> 286</P> 287<P> 288Inside a character class, or if the decimal number is greater than 9 and there 289have not been that many capturing subpatterns, PCRE re-reads up to three octal 290digits following the backslash, and uses them to generate a data character. Any 291subsequent digits stand for themselves. In non-UTF-8 mode, the value of a 292character specified in octal must be less than \400. In UTF-8 mode, values up 293to \777 are permitted. For example: 294<pre> 295 \040 is another way of writing a space 296 \40 is the same, provided there are fewer than 40 previous capturing subpatterns 297 \7 is always a back reference 298 \11 might be a back reference, or another way of writing a tab 299 \011 is always a tab 300 \0113 is a tab followed by the character "3" 301 \113 might be a back reference, otherwise the character with octal code 113 302 \377 might be a back reference, otherwise the byte consisting entirely of 1 bits 303 \81 is either a back reference, or a binary zero followed by the two characters "8" and "1" 304</pre> 305Note that octal values of 100 or greater must not be introduced by a leading 306zero, because no more than three octal digits are ever read. 307</P> 308<P> 309All the sequences that define a single character value can be used both inside 310and outside character classes. In addition, inside a character class, the 311sequence \b is interpreted as the backspace character (hex 08), and the 312sequences \R and \X are interpreted as the characters "R" and "X", 313respectively. Outside a character class, these sequences have different 314meanings 315<a href="#uniextseq">(see below).</a> 316</P> 317<br><b> 318Absolute and relative back references 319</b><br> 320<P> 321The sequence \g followed by an unsigned or a negative number, optionally 322enclosed in braces, is an absolute or relative back reference. A named back 323reference can be coded as \g{name}. Back references are discussed 324<a href="#backreferences">later,</a> 325following the discussion of 326<a href="#subpattern">parenthesized subpatterns.</a> 327</P> 328<br><b> 329Absolute and relative subroutine calls 330</b><br> 331<P> 332For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or 333a number enclosed either in angle brackets or single quotes, is an alternative 334syntax for referencing a subpattern as a "subroutine". Details are discussed 335<a href="#onigurumasubroutines">later.</a> 336Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are <i>not</i> 337synonymous. The former is a back reference; the latter is a 338<a href="#subpatternsassubroutines">subroutine</a> 339call. 340</P> 341<br><b> 342Generic character types 343</b><br> 344<P> 345Another use of backslash is for specifying generic character types. The 346following are always recognized: 347<pre> 348 \d any decimal digit 349 \D any character that is not a decimal digit 350 \h any horizontal whitespace character 351 \H any character that is not a horizontal whitespace character 352 \s any whitespace character 353 \S any character that is not a whitespace character 354 \v any vertical whitespace character 355 \V any character that is not a vertical whitespace character 356 \w any "word" character 357 \W any "non-word" character 358</pre> 359Each pair of escape sequences partitions the complete set of characters into 360two disjoint sets. Any given character matches one, and only one, of each pair. 361</P> 362<P> 363These character type sequences can appear both inside and outside character 364classes. They each match one character of the appropriate type. If the current 365matching point is at the end of the subject string, all of them fail, since 366there is no character to match. 367</P> 368<P> 369For compatibility with Perl, \s does not match the VT character (code 11). 370This makes it different from the the POSIX "space" class. The \s characters 371are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is 372included in a Perl script, \s may match the VT character. In PCRE, it never 373does. 374</P> 375<P> 376In UTF-8 mode, characters with values greater than 128 never match \d, \s, or 377\w, and always match \D, \S, and \W. This is true even when Unicode 378character property support is available. These sequences retain their original 379meanings from before UTF-8 support was available, mainly for efficiency 380reasons. Note that this also affects \b, because it is defined in terms of \w 381and \W. 382</P> 383<P> 384The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to the 385other sequences, these do match certain high-valued codepoints in UTF-8 mode. 386The horizontal space characters are: 387<pre> 388 U+0009 Horizontal tab 389 U+0020 Space 390 U+00A0 Non-break space 391 U+1680 Ogham space mark 392 U+180E Mongolian vowel separator 393 U+2000 En quad 394 U+2001 Em quad 395 U+2002 En space 396 U+2003 Em space 397 U+2004 Three-per-em space 398 U+2005 Four-per-em space 399 U+2006 Six-per-em space 400 U+2007 Figure space 401 U+2008 Punctuation space 402 U+2009 Thin space 403 U+200A Hair space 404 U+202F Narrow no-break space 405 U+205F Medium mathematical space 406 U+3000 Ideographic space 407</pre> 408The vertical space characters are: 409<pre> 410 U+000A Linefeed 411 U+000B Vertical tab 412 U+000C Formfeed 413 U+000D Carriage return 414 U+0085 Next line 415 U+2028 Line separator 416 U+2029 Paragraph separator 417</PRE> 418</P> 419<P> 420A "word" character is an underscore or any character less than 256 that is a 421letter or digit. The definition of letters and digits is controlled by PCRE's 422low-valued character tables, and may vary if locale-specific matching is taking 423place (see 424<a href="pcreapi.html#localesupport">"Locale support"</a> 425in the 426<a href="pcreapi.html"><b>pcreapi</b></a> 427page). For example, in a French locale such as "fr_FR" in Unix-like systems, 428or "french" in Windows, some character codes greater than 128 are used for 429accented letters, and these are matched by \w. The use of locales with Unicode 430is discouraged. 431<a name="newlineseq"></a></P> 432<br><b> 433Newline sequences 434</b><br> 435<P> 436Outside a character class, by default, the escape sequence \R matches any 437Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \R is 438equivalent to the following: 439<pre> 440 (?>\r\n|\n|\x0b|\f|\r|\x85) 441</pre> 442This is an example of an "atomic group", details of which are given 443<a href="#atomicgroup">below.</a> 444This particular group matches either the two-character sequence CR followed by 445LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab, 446U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next 447line, U+0085). The two-character sequence is treated as a single unit that 448cannot be split. 449</P> 450<P> 451In UTF-8 mode, two additional characters whose codepoints are greater than 255 452are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029). 453Unicode character property support is not needed for these characters to be 454recognized. 455</P> 456<P> 457It is possible to restrict \R to match only CR, LF, or CRLF (instead of the 458complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF 459either at compile time or when the pattern is matched. (BSR is an abbrevation 460for "backslash R".) This can be made the default when PCRE is built; if this is 461the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option. 462It is also possible to specify these settings by starting a pattern string with 463one of the following sequences: 464<pre> 465 (*BSR_ANYCRLF) CR, LF, or CRLF only 466 (*BSR_UNICODE) any Unicode newline sequence 467</pre> 468These override the default and the options given to <b>pcre_compile()</b> or 469<b>pcre_compile2()</b>, but they can be overridden by options given to 470<b>pcre_exec()</b> or <b>pcre_dfa_exec()</b>. Note that these special settings, 471which are not Perl-compatible, are recognized only at the very start of a 472pattern, and that they must be in upper case. If more than one of them is 473present, the last one is used. They can be combined with a change of newline 474convention, for example, a pattern can start with: 475<pre> 476 (*ANY)(*BSR_ANYCRLF) 477</pre> 478Inside a character class, \R matches the letter "R". 479<a name="uniextseq"></a></P> 480<br><b> 481Unicode character properties 482</b><br> 483<P> 484When PCRE is built with Unicode character property support, three additional 485escape sequences that match characters with specific properties are available. 486When not in UTF-8 mode, these sequences are of course limited to testing 487characters whose codepoints are less than 256, but they do work in this mode. 488The extra escape sequences are: 489<pre> 490 \p{<i>xx</i>} a character with the <i>xx</i> property 491 \P{<i>xx</i>} a character without the <i>xx</i> property 492 \X an extended Unicode sequence 493</pre> 494The property names represented by <i>xx</i> above are limited to the Unicode 495script names, the general category properties, and "Any", which matches any 496character (including newline). Other properties such as "InMusicalSymbols" are 497not currently supported by PCRE. Note that \P{Any} does not match any 498characters, so always causes a match failure. 499</P> 500<P> 501Sets of Unicode characters are defined as belonging to certain scripts. A 502character from one of these sets can be matched using a script name. For 503example: 504<pre> 505 \p{Greek} 506 \P{Han} 507</pre> 508Those that are not part of an identified script are lumped together as 509"Common". The current list of scripts is: 510</P> 511<P> 512Arabic, 513Armenian, 514Avestan, 515Balinese, 516Bamum, 517Bengali, 518Bopomofo, 519Braille, 520Buginese, 521Buhid, 522Canadian_Aboriginal, 523Carian, 524Cham, 525Cherokee, 526Common, 527Coptic, 528Cuneiform, 529Cypriot, 530Cyrillic, 531Deseret, 532Devanagari, 533Egyptian_Hieroglyphs, 534Ethiopic, 535Georgian, 536Glagolitic, 537Gothic, 538Greek, 539Gujarati, 540Gurmukhi, 541Han, 542Hangul, 543Hanunoo, 544Hebrew, 545Hiragana, 546Imperial_Aramaic, 547Inherited, 548Inscriptional_Pahlavi, 549Inscriptional_Parthian, 550Javanese, 551Kaithi, 552Kannada, 553Katakana, 554Kayah_Li, 555Kharoshthi, 556Khmer, 557Lao, 558Latin, 559Lepcha, 560Limbu, 561Linear_B, 562Lisu, 563Lycian, 564Lydian, 565Malayalam, 566Meetei_Mayek, 567Mongolian, 568Myanmar, 569New_Tai_Lue, 570Nko, 571Ogham, 572Old_Italic, 573Old_Persian, 574Old_South_Arabian, 575Old_Turkic, 576Ol_Chiki, 577Oriya, 578Osmanya, 579Phags_Pa, 580Phoenician, 581Rejang, 582Runic, 583Samaritan, 584Saurashtra, 585Shavian, 586Sinhala, 587Sundanese, 588Syloti_Nagri, 589Syriac, 590Tagalog, 591Tagbanwa, 592Tai_Le, 593Tai_Tham, 594Tai_Viet, 595Tamil, 596Telugu, 597Thaana, 598Thai, 599Tibetan, 600Tifinagh, 601Ugaritic, 602Vai, 603Yi. 604</P> 605<P> 606Each character has exactly one general category property, specified by a 607two-letter abbreviation. For compatibility with Perl, negation can be specified 608by including a circumflex between the opening brace and the property name. For 609example, \p{^Lu} is the same as \P{Lu}. 610</P> 611<P> 612If only one letter is specified with \p or \P, it includes all the general 613category properties that start with that letter. In this case, in the absence 614of negation, the curly brackets in the escape sequence are optional; these two 615examples have the same effect: 616<pre> 617 \p{L} 618 \pL 619</pre> 620The following general category property codes are supported: 621<pre> 622 C Other 623 Cc Control 624 Cf Format 625 Cn Unassigned 626 Co Private use 627 Cs Surrogate 628 629 L Letter 630 Ll Lower case letter 631 Lm Modifier letter 632 Lo Other letter 633 Lt Title case letter 634 Lu Upper case letter 635 636 M Mark 637 Mc Spacing mark 638 Me Enclosing mark 639 Mn Non-spacing mark 640 641 N Number 642 Nd Decimal number 643 Nl Letter number 644 No Other number 645 646 P Punctuation 647 Pc Connector punctuation 648 Pd Dash punctuation 649 Pe Close punctuation 650 Pf Final punctuation 651 Pi Initial punctuation 652 Po Other punctuation 653 Ps Open punctuation 654 655 S Symbol 656 Sc Currency symbol 657 Sk Modifier symbol 658 Sm Mathematical symbol 659 So Other symbol 660 661 Z Separator 662 Zl Line separator 663 Zp Paragraph separator 664 Zs Space separator 665</pre> 666The special property L& is also supported: it matches a character that has 667the Lu, Ll, or Lt property, in other words, a letter that is not classified as 668a modifier or "other". 669</P> 670<P> 671The Cs (Surrogate) property applies only to characters in the range U+D800 to 672U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so 673cannot be tested by PCRE, unless UTF-8 validity checking has been turned off 674(see the discussion of PCRE_NO_UTF8_CHECK in the 675<a href="pcreapi.html"><b>pcreapi</b></a> 676page). Perl does not support the Cs property. 677</P> 678<P> 679The long synonyms for property names that Perl supports (such as \p{Letter}) 680are not supported by PCRE, nor is it permitted to prefix any of these 681properties with "Is". 682</P> 683<P> 684No character that is in the Unicode table has the Cn (unassigned) property. 685Instead, this property is assumed for any code point that is not in the 686Unicode table. 687</P> 688<P> 689Specifying caseless matching does not affect these escape sequences. For 690example, \p{Lu} always matches only upper case letters. 691</P> 692<P> 693The \X escape matches any number of Unicode characters that form an extended 694Unicode sequence. \X is equivalent to 695<pre> 696 (?>\PM\pM*) 697</pre> 698That is, it matches a character without the "mark" property, followed by zero 699or more characters with the "mark" property, and treats the sequence as an 700atomic group 701<a href="#atomicgroup">(see below).</a> 702Characters with the "mark" property are typically accents that affect the 703preceding character. None of them have codepoints less than 256, so in 704non-UTF-8 mode \X matches any one character. 705</P> 706<P> 707Matching characters by Unicode property is not fast, because PCRE has to search 708a structure that contains data for over fifteen thousand characters. That is 709why the traditional escape sequences such as \d and \w do not use Unicode 710properties in PCRE. 711<a name="resetmatchstart"></a></P> 712<br><b> 713Resetting the match start 714</b><br> 715<P> 716The escape sequence \K, which is a Perl 5.10 feature, causes any previously 717matched characters not to be included in the final matched sequence. For 718example, the pattern: 719<pre> 720 foo\Kbar 721</pre> 722matches "foobar", but reports that it has matched "bar". This feature is 723similar to a lookbehind assertion 724<a href="#lookbehind">(described below).</a> 725However, in this case, the part of the subject before the real match does not 726have to be of fixed length, as lookbehind assertions do. The use of \K does 727not interfere with the setting of 728<a href="#subpattern">captured substrings.</a> 729For example, when the pattern 730<pre> 731 (foo)\Kbar 732</pre> 733matches "foobar", the first substring is still set to "foo". 734</P> 735<P> 736Perl documents that the use of \K within assertions is "not well defined". In 737PCRE, \K is acted upon when it occurs inside positive assertions, but is 738ignored in negative assertions. 739<a name="smallassertions"></a></P> 740<br><b> 741Simple assertions 742</b><br> 743<P> 744The final use of backslash is for certain simple assertions. An assertion 745specifies a condition that has to be met at a particular point in a match, 746without consuming any characters from the subject string. The use of 747subpatterns for more complicated assertions is described 748<a href="#bigassertions">below.</a> 749The backslashed assertions are: 750<pre> 751 \b matches at a word boundary 752 \B matches when not at a word boundary 753 \A matches at the start of the subject 754 \Z matches at the end of the subject 755 also matches before a newline at the end of the subject 756 \z matches only at the end of the subject 757 \G matches at the first matching position in the subject 758</pre> 759These assertions may not appear in character classes (but note that \b has a 760different meaning, namely the backspace character, inside a character class). 761</P> 762<P> 763A word boundary is a position in the subject string where the current character 764and the previous character do not both match \w or \W (i.e. one matches 765\w and the other matches \W), or the start or end of the string if the 766first or last character matches \w, respectively. Neither PCRE nor Perl has a 767separte "start of word" or "end of word" metasequence. However, whatever 768follows \b normally determines which it is. For example, the fragment 769\ba matches "a" at the start of a word. 770</P> 771<P> 772The \A, \Z, and \z assertions differ from the traditional circumflex and 773dollar (described in the next section) in that they only ever match at the very 774start and end of the subject string, whatever options are set. Thus, they are 775independent of multiline mode. These three assertions are not affected by the 776PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the 777circumflex and dollar metacharacters. However, if the <i>startoffset</i> 778argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start 779at a point other than the beginning of the subject, \A can never match. The 780difference between \Z and \z is that \Z matches before a newline at the end 781of the string as well as at the very end, whereas \z matches only at the end. 782</P> 783<P> 784The \G assertion is true only when the current matching position is at the 785start point of the match, as specified by the <i>startoffset</i> argument of 786<b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is 787non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate 788arguments, you can mimic Perl's /g option, and it is in this kind of 789implementation where \G can be useful. 790</P> 791<P> 792Note, however, that PCRE's interpretation of \G, as the start of the current 793match, is subtly different from Perl's, which defines it as the end of the 794previous match. In Perl, these can be different when the previously matched 795string was empty. Because PCRE does just one match at a time, it cannot 796reproduce this behaviour. 797</P> 798<P> 799If all the alternatives of a pattern begin with \G, the expression is anchored 800to the starting match position, and the "anchored" flag is set in the compiled 801regular expression. 802</P> 803<br><a name="SEC5" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br> 804<P> 805Outside a character class, in the default matching mode, the circumflex 806character is an assertion that is true only if the current matching point is 807at the start of the subject string. If the <i>startoffset</i> argument of 808<b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE 809option is unset. Inside a character class, circumflex has an entirely different 810meaning 811<a href="#characterclass">(see below).</a> 812</P> 813<P> 814Circumflex need not be the first character of the pattern if a number of 815alternatives are involved, but it should be the first thing in each alternative 816in which it appears if the pattern is ever to match that branch. If all 817possible alternatives start with a circumflex, that is, if the pattern is 818constrained to match only at the start of the subject, it is said to be an 819"anchored" pattern. (There are also other constructs that can cause a pattern 820to be anchored.) 821</P> 822<P> 823A dollar character is an assertion that is true only if the current matching 824point is at the end of the subject string, or immediately before a newline 825at the end of the string (by default). Dollar need not be the last character of 826the pattern if a number of alternatives are involved, but it should be the last 827item in any branch in which it appears. Dollar has no special meaning in a 828character class. 829</P> 830<P> 831The meaning of dollar can be changed so that it matches only at the very end of 832the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This 833does not affect the \Z assertion. 834</P> 835<P> 836The meanings of the circumflex and dollar characters are changed if the 837PCRE_MULTILINE option is set. When this is the case, a circumflex matches 838immediately after internal newlines as well as at the start of the subject 839string. It does not match after a newline that ends the string. A dollar 840matches before any newlines in the string, as well as at the very end, when 841PCRE_MULTILINE is set. When newline is specified as the two-character 842sequence CRLF, isolated CR and LF characters do not indicate newlines. 843</P> 844<P> 845For example, the pattern /^abc$/ matches the subject string "def\nabc" (where 846\n represents a newline) in multiline mode, but not otherwise. Consequently, 847patterns that are anchored in single line mode because all branches start with 848^ are not anchored in multiline mode, and a match for circumflex is possible 849when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The 850PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set. 851</P> 852<P> 853Note that the sequences \A, \Z, and \z can be used to match the start and 854end of the subject in both modes, and if all branches of a pattern start with 855\A it is always anchored, whether or not PCRE_MULTILINE is set. 856</P> 857<br><a name="SEC6" href="#TOC1">FULL STOP (PERIOD, DOT)</a><br> 858<P> 859Outside a character class, a dot in the pattern matches any one character in 860the subject string except (by default) a character that signifies the end of a 861line. In UTF-8 mode, the matched character may be more than one byte long. 862</P> 863<P> 864When a line ending is defined as a single character, dot never matches that 865character; when the two-character sequence CRLF is used, dot does not match CR 866if it is immediately followed by LF, but otherwise it matches all characters 867(including isolated CRs and LFs). When any Unicode line endings are being 868recognized, dot does not match CR or LF or any of the other line ending 869characters. 870</P> 871<P> 872The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL 873option is set, a dot matches any one character, without exception. If the 874two-character sequence CRLF is present in the subject string, it takes two dots 875to match it. 876</P> 877<P> 878The handling of dot is entirely independent of the handling of circumflex and 879dollar, the only relationship being that they both involve newlines. Dot has no 880special meaning in a character class. 881</P> 882<br><a name="SEC7" href="#TOC1">MATCHING A SINGLE BYTE</a><br> 883<P> 884Outside a character class, the escape sequence \C matches any one byte, both 885in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending 886characters. The feature is provided in Perl in order to match individual bytes 887in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes, 888what remains in the string may be a malformed UTF-8 string. For this reason, 889the \C escape sequence is best avoided. 890</P> 891<P> 892PCRE does not allow \C to appear in lookbehind assertions 893<a href="#lookbehind">(described below),</a> 894because in UTF-8 mode this would make it impossible to calculate the length of 895the lookbehind. 896<a name="characterclass"></a></P> 897<br><a name="SEC8" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br> 898<P> 899An opening square bracket introduces a character class, terminated by a closing 900square bracket. A closing square bracket on its own is not special by default. 901However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square 902bracket causes a compile-time error. If a closing square bracket is required as 903a member of the class, it should be the first data character in the class 904(after an initial circumflex, if present) or escaped with a backslash. 905</P> 906<P> 907A character class matches a single character in the subject. In UTF-8 mode, the 908character may be more than one byte long. A matched character must be in the 909set of characters defined by the class, unless the first character in the class 910definition is a circumflex, in which case the subject character must not be in 911the set defined by the class. If a circumflex is actually required as a member 912of the class, ensure it is not the first character, or escape it with a 913backslash. 914</P> 915<P> 916For example, the character class [aeiou] matches any lower case vowel, while 917[^aeiou] matches any character that is not a lower case vowel. Note that a 918circumflex is just a convenient notation for specifying the characters that 919are in the class by enumerating those that are not. A class that starts with a 920circumflex is not an assertion; it still consumes a character from the subject 921string, and therefore it fails if the current pointer is at the end of the 922string. 923</P> 924<P> 925In UTF-8 mode, characters with values greater than 255 can be included in a 926class as a literal string of bytes, or by using the \x{ escaping mechanism. 927</P> 928<P> 929When caseless matching is set, any letters in a class represent both their 930upper case and lower case versions, so for example, a caseless [aeiou] matches 931"A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a 932caseful version would. In UTF-8 mode, PCRE always understands the concept of 933case for characters whose values are less than 128, so caseless matching is 934always possible. For characters with higher values, the concept of case is 935supported if PCRE is compiled with Unicode property support, but not otherwise. 936If you want to use caseless matching in UTF8-mode for characters 128 and above, 937you must ensure that PCRE is compiled with Unicode property support as well as 938with UTF-8 support. 939</P> 940<P> 941Characters that might indicate line breaks are never treated in any special way 942when matching character classes, whatever line-ending sequence is in use, and 943whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class 944such as [^a] always matches one of these characters. 945</P> 946<P> 947The minus (hyphen) character can be used to specify a range of characters in a 948character class. For example, [d-m] matches any letter between d and m, 949inclusive. If a minus character is required in a class, it must be escaped with 950a backslash or appear in a position where it cannot be interpreted as 951indicating a range, typically as the first or last character in the class. 952</P> 953<P> 954It is not possible to have the literal character "]" as the end character of a 955range. A pattern such as [W-]46] is interpreted as a class of two characters 956("W" and "-") followed by a literal string "46]", so it would match "W46]" or 957"-46]". However, if the "]" is escaped with a backslash it is interpreted as 958the end of range, so [W-\]46] is interpreted as a class containing a range 959followed by two other characters. The octal or hexadecimal representation of 960"]" can also be used to end a range. 961</P> 962<P> 963Ranges operate in the collating sequence of character values. They can also be 964used for characters specified numerically, for example [\000-\037]. In UTF-8 965mode, ranges can include characters whose values are greater than 255, for 966example [\x{100}-\x{2ff}]. 967</P> 968<P> 969If a range that includes letters is used when caseless matching is set, it 970matches the letters in either case. For example, [W-c] is equivalent to 971[][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character 972tables for a French locale are in use, [\xc8-\xcb] matches accented E 973characters in both cases. In UTF-8 mode, PCRE supports the concept of case for 974characters with values greater than 128 only when it is compiled with Unicode 975property support. 976</P> 977<P> 978The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear 979in a character class, and add the characters that they match to the class. For 980example, [\dABCDEF] matches any hexadecimal digit. A circumflex can 981conveniently be used with the upper case character types to specify a more 982restricted set of characters than the matching lower case type. For example, 983the class [^\W_] matches any letter or digit, but not underscore. 984</P> 985<P> 986The only metacharacters that are recognized in character classes are backslash, 987hyphen (only where it can be interpreted as specifying a range), circumflex 988(only at the start), opening square bracket (only when it can be interpreted as 989introducing a POSIX class name - see the next section), and the terminating 990closing square bracket. However, escaping other non-alphanumeric characters 991does no harm. 992</P> 993<br><a name="SEC9" href="#TOC1">POSIX CHARACTER CLASSES</a><br> 994<P> 995Perl supports the POSIX notation for character classes. This uses names 996enclosed by [: and :] within the enclosing square brackets. PCRE also supports 997this notation. For example, 998<pre> 999 [01[:alpha:]%] 1000</pre> 1001matches "0", "1", any alphabetic character, or "%". The supported class names 1002are 1003<pre> 1004 alnum letters and digits 1005 alpha letters 1006 ascii character codes 0 - 127 1007 blank space or tab only 1008 cntrl control characters 1009 digit decimal digits (same as \d) 1010 graph printing characters, excluding space 1011 lower lower case letters 1012 print printing characters, including space 1013 punct printing characters, excluding letters and digits 1014 space white space (not quite the same as \s) 1015 upper upper case letters 1016 word "word" characters (same as \w) 1017 xdigit hexadecimal digits 1018</pre> 1019The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and 1020space (32). Notice that this list includes the VT character (code 11). This 1021makes "space" different to \s, which does not include VT (for Perl 1022compatibility). 1023</P> 1024<P> 1025The name "word" is a Perl extension, and "blank" is a GNU extension from Perl 10265.8. Another Perl extension is negation, which is indicated by a ^ character 1027after the colon. For example, 1028<pre> 1029 [12[:^digit:]] 1030</pre> 1031matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX 1032syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not 1033supported, and an error is given if they are encountered. 1034</P> 1035<P> 1036In UTF-8 mode, characters with values greater than 128 do not match any of 1037the POSIX character classes. 1038</P> 1039<br><a name="SEC10" href="#TOC1">VERTICAL BAR</a><br> 1040<P> 1041Vertical bar characters are used to separate alternative patterns. For example, 1042the pattern 1043<pre> 1044 gilbert|sullivan 1045</pre> 1046matches either "gilbert" or "sullivan". Any number of alternatives may appear, 1047and an empty alternative is permitted (matching the empty string). The matching 1048process tries each alternative in turn, from left to right, and the first one 1049that succeeds is used. If the alternatives are within a subpattern 1050<a href="#subpattern">(defined below),</a> 1051"succeeds" means matching the rest of the main pattern as well as the 1052alternative in the subpattern. 1053</P> 1054<br><a name="SEC11" href="#TOC1">INTERNAL OPTION SETTING</a><br> 1055<P> 1056The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and 1057PCRE_EXTENDED options (which are Perl-compatible) can be changed from within 1058the pattern by a sequence of Perl option letters enclosed between "(?" and ")". 1059The option letters are 1060<pre> 1061 i for PCRE_CASELESS 1062 m for PCRE_MULTILINE 1063 s for PCRE_DOTALL 1064 x for PCRE_EXTENDED 1065</pre> 1066For example, (?im) sets caseless, multiline matching. It is also possible to 1067unset these options by preceding the letter with a hyphen, and a combined 1068setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and 1069PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also 1070permitted. If a letter appears both before and after the hyphen, the option is 1071unset. 1072</P> 1073<P> 1074The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be 1075changed in the same way as the Perl-compatible options by using the characters 1076J, U and X respectively. 1077</P> 1078<P> 1079When one of these option changes occurs at top level (that is, not inside 1080subpattern parentheses), the change applies to the remainder of the pattern 1081that follows. If the change is placed right at the start of a pattern, PCRE 1082extracts it into the global options (and it will therefore show up in data 1083extracted by the <b>pcre_fullinfo()</b> function). 1084</P> 1085<P> 1086An option change within a subpattern (see below for a description of 1087subpatterns) affects only that part of the current pattern that follows it, so 1088<pre> 1089 (a(?i)b)c 1090</pre> 1091matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used). 1092By this means, options can be made to have different settings in different 1093parts of the pattern. Any changes made in one alternative do carry on 1094into subsequent branches within the same subpattern. For example, 1095<pre> 1096 (a(?i)b|c) 1097</pre> 1098matches "ab", "aB", "c", and "C", even though when matching "C" the first 1099branch is abandoned before the option setting. This is because the effects of 1100option settings happen at compile time. There would be some very weird 1101behaviour otherwise. 1102</P> 1103<P> 1104<b>Note:</b> There are other PCRE-specific options that can be set by the 1105application when the compile or match functions are called. In some cases the 1106pattern can contain special leading sequences such as (*CRLF) to override what 1107the application has set or what has been defaulted. Details are given in the 1108section entitled 1109<a href="#newlineseq">"Newline sequences"</a> 1110above. There is also the (*UTF8) leading sequence that can be used to set UTF-8 1111mode; this is equivalent to setting the PCRE_UTF8 option. 1112<a name="subpattern"></a></P> 1113<br><a name="SEC12" href="#TOC1">SUBPATTERNS</a><br> 1114<P> 1115Subpatterns are delimited by parentheses (round brackets), which can be nested. 1116Turning part of a pattern into a subpattern does two things: 1117<br> 1118<br> 11191. It localizes a set of alternatives. For example, the pattern 1120<pre> 1121 cat(aract|erpillar|) 1122</pre> 1123matches one of the words "cat", "cataract", or "caterpillar". Without the 1124parentheses, it would match "cataract", "erpillar" or an empty string. 1125<br> 1126<br> 11272. It sets up the subpattern as a capturing subpattern. This means that, when 1128the whole pattern matches, that portion of the subject string that matched the 1129subpattern is passed back to the caller via the <i>ovector</i> argument of 1130<b>pcre_exec()</b>. Opening parentheses are counted from left to right (starting 1131from 1) to obtain numbers for the capturing subpatterns. 1132</P> 1133<P> 1134For example, if the string "the red king" is matched against the pattern 1135<pre> 1136 the ((red|white) (king|queen)) 1137</pre> 1138the captured substrings are "red king", "red", and "king", and are numbered 1, 11392, and 3, respectively. 1140</P> 1141<P> 1142The fact that plain parentheses fulfil two functions is not always helpful. 1143There are often times when a grouping subpattern is required without a 1144capturing requirement. If an opening parenthesis is followed by a question mark 1145and a colon, the subpattern does not do any capturing, and is not counted when 1146computing the number of any subsequent capturing subpatterns. For example, if 1147the string "the white queen" is matched against the pattern 1148<pre> 1149 the ((?:red|white) (king|queen)) 1150</pre> 1151the captured substrings are "white queen" and "queen", and are numbered 1 and 11522. The maximum number of capturing subpatterns is 65535. 1153</P> 1154<P> 1155As a convenient shorthand, if any option settings are required at the start of 1156a non-capturing subpattern, the option letters may appear between the "?" and 1157the ":". Thus the two patterns 1158<pre> 1159 (?i:saturday|sunday) 1160 (?:(?i)saturday|sunday) 1161</pre> 1162match exactly the same set of strings. Because alternative branches are tried 1163from left to right, and options are not reset until the end of the subpattern 1164is reached, an option setting in one branch does affect subsequent branches, so 1165the above patterns match "SUNDAY" as well as "Saturday". 1166<a name="dupsubpatternnumber"></a></P> 1167<br><a name="SEC13" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br> 1168<P> 1169Perl 5.10 introduced a feature whereby each alternative in a subpattern uses 1170the same numbers for its capturing parentheses. Such a subpattern starts with 1171(?| and is itself a non-capturing subpattern. For example, consider this 1172pattern: 1173<pre> 1174 (?|(Sat)ur|(Sun))day 1175</pre> 1176Because the two alternatives are inside a (?| group, both sets of capturing 1177parentheses are numbered one. Thus, when the pattern matches, you can look 1178at captured substring number one, whichever alternative matched. This construct 1179is useful when you want to capture part, but not all, of one of a number of 1180alternatives. Inside a (?| group, parentheses are numbered as usual, but the 1181number is reset at the start of each branch. The numbers of any capturing 1182buffers that follow the subpattern start after the highest number used in any 1183branch. The following example is taken from the Perl documentation. 1184The numbers underneath show in which buffer the captured content will be 1185stored. 1186<pre> 1187 # before ---------------branch-reset----------- after 1188 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x 1189 # 1 2 2 3 2 3 4 1190</pre> 1191A back reference to a numbered subpattern uses the most recent value that is 1192set for that number by any subpattern. The following pattern matches "abcabc" 1193or "defdef": 1194<pre> 1195 /(?|(abc)|(def))\1/ 1196</pre> 1197In contrast, a recursive or "subroutine" call to a numbered subpattern always 1198refers to the first one in the pattern with the given number. The following 1199pattern matches "abcabc" or "defabc": 1200<pre> 1201 /(?|(abc)|(def))(?1)/ 1202</pre> 1203If a 1204<a href="#conditions">condition test</a> 1205for a subpattern's having matched refers to a non-unique number, the test is 1206true if any of the subpatterns of that number have matched. 1207</P> 1208<P> 1209An alternative approach to using this "branch reset" feature is to use 1210duplicate named subpatterns, as described in the next section. 1211</P> 1212<br><a name="SEC14" href="#TOC1">NAMED SUBPATTERNS</a><br> 1213<P> 1214Identifying capturing parentheses by number is simple, but it can be very hard 1215to keep track of the numbers in complicated regular expressions. Furthermore, 1216if an expression is modified, the numbers may change. To help with this 1217difficulty, PCRE supports the naming of subpatterns. This feature was not 1218added to Perl until release 5.10. Python had the feature earlier, and PCRE 1219introduced it at release 4.0, using the Python syntax. PCRE now supports both 1220the Perl and the Python syntax. Perl allows identically numbered subpatterns to 1221have different names, but PCRE does not. 1222</P> 1223<P> 1224In PCRE, a subpattern can be named in one of three ways: (?<name>...) or 1225(?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing 1226parentheses from other parts of the pattern, such as 1227<a href="#backreferences">back references,</a> 1228<a href="#recursion">recursion,</a> 1229and 1230<a href="#conditions">conditions,</a> 1231can be made by name as well as by number. 1232</P> 1233<P> 1234Names consist of up to 32 alphanumeric characters and underscores. Named 1235capturing parentheses are still allocated numbers as well as names, exactly as 1236if the names were not present. The PCRE API provides function calls for 1237extracting the name-to-number translation table from a compiled pattern. There 1238is also a convenience function for extracting a captured substring by name. 1239</P> 1240<P> 1241By default, a name must be unique within a pattern, but it is possible to relax 1242this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate 1243names are also always permitted for subpatterns with the same number, set up as 1244described in the previous section.) Duplicate names can be useful for patterns 1245where only one instance of the named parentheses can match. Suppose you want to 1246match the name of a weekday, either as a 3-letter abbreviation or as the full 1247name, and in both cases you want to extract the abbreviation. This pattern 1248(ignoring the line breaks) does the job: 1249<pre> 1250 (?<DN>Mon|Fri|Sun)(?:day)?| 1251 (?<DN>Tue)(?:sday)?| 1252 (?<DN>Wed)(?:nesday)?| 1253 (?<DN>Thu)(?:rsday)?| 1254 (?<DN>Sat)(?:urday)? 1255</pre> 1256There are five capturing substrings, but only one is ever set after a match. 1257(An alternative way of solving this problem is to use a "branch reset" 1258subpattern, as described in the previous section.) 1259</P> 1260<P> 1261The convenience function for extracting the data by name returns the substring 1262for the first (and in this example, the only) subpattern of that name that 1263matched. This saves searching to find which numbered subpattern it was. 1264</P> 1265<P> 1266If you make a back reference to a non-unique named subpattern from elsewhere in 1267the pattern, the one that corresponds to the first occurrence of the name is 1268used. In the absence of duplicate numbers (see the previous section) this is 1269the one with the lowest number. If you use a named reference in a condition 1270test (see the 1271<a href="#conditions">section about conditions</a> 1272below), either to check whether a subpattern has matched, or to check for 1273recursion, all subpatterns with the same name are tested. If the condition is 1274true for any one of them, the overall condition is true. This is the same 1275behaviour as testing by number. For further details of the interfaces for 1276handling named subpatterns, see the 1277<a href="pcreapi.html"><b>pcreapi</b></a> 1278documentation. 1279</P> 1280<P> 1281<b>Warning:</b> You cannot use different names to distinguish between two 1282subpatterns with the same number because PCRE uses only the numbers when 1283matching. For this reason, an error is given at compile time if different names 1284are given to subpatterns with the same number. However, you can give the same 1285name to subpatterns with the same number, even when PCRE_DUPNAMES is not set. 1286</P> 1287<br><a name="SEC15" href="#TOC1">REPETITION</a><br> 1288<P> 1289Repetition is specified by quantifiers, which can follow any of the following 1290items: 1291<pre> 1292 a literal data character 1293 the dot metacharacter 1294 the \C escape sequence 1295 the \X escape sequence (in UTF-8 mode with Unicode properties) 1296 the \R escape sequence 1297 an escape such as \d that matches a single character 1298 a character class 1299 a back reference (see next section) 1300 a parenthesized subpattern (unless it is an assertion) 1301 a recursive or "subroutine" call to a subpattern 1302</pre> 1303The general repetition quantifier specifies a minimum and maximum number of 1304permitted matches, by giving the two numbers in curly brackets (braces), 1305separated by a comma. The numbers must be less than 65536, and the first must 1306be less than or equal to the second. For example: 1307<pre> 1308 z{2,4} 1309</pre> 1310matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special 1311character. If the second number is omitted, but the comma is present, there is 1312no upper limit; if the second number and the comma are both omitted, the 1313quantifier specifies an exact number of required matches. Thus 1314<pre> 1315 [aeiou]{3,} 1316</pre> 1317matches at least 3 successive vowels, but may match many more, while 1318<pre> 1319 \d{8} 1320</pre> 1321matches exactly 8 digits. An opening curly bracket that appears in a position 1322where a quantifier is not allowed, or one that does not match the syntax of a 1323quantifier, is taken as a literal character. For example, {,6} is not a 1324quantifier, but a literal string of four characters. 1325</P> 1326<P> 1327In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual 1328bytes. Thus, for example, \x{100}{2} matches two UTF-8 characters, each of 1329which is represented by a two-byte sequence. Similarly, when Unicode property 1330support is available, \X{3} matches three Unicode extended sequences, each of 1331which may be several bytes long (and they may be of different lengths). 1332</P> 1333<P> 1334The quantifier {0} is permitted, causing the expression to behave as if the 1335previous item and the quantifier were not present. This may be useful for 1336subpatterns that are referenced as 1337<a href="#subpatternsassubroutines">subroutines</a> 1338from elsewhere in the pattern. Items other than subpatterns that have a {0} 1339quantifier are omitted from the compiled pattern. 1340</P> 1341<P> 1342For convenience, the three most common quantifiers have single-character 1343abbreviations: 1344<pre> 1345 * is equivalent to {0,} 1346 + is equivalent to {1,} 1347 ? is equivalent to {0,1} 1348</pre> 1349It is possible to construct infinite loops by following a subpattern that can 1350match no characters with a quantifier that has no upper limit, for example: 1351<pre> 1352 (a?)* 1353</pre> 1354Earlier versions of Perl and PCRE used to give an error at compile time for 1355such patterns. However, because there are cases where this can be useful, such 1356patterns are now accepted, but if any repetition of the subpattern does in fact 1357match no characters, the loop is forcibly broken. 1358</P> 1359<P> 1360By default, the quantifiers are "greedy", that is, they match as much as 1361possible (up to the maximum number of permitted times), without causing the 1362rest of the pattern to fail. The classic example of where this gives problems 1363is in trying to match comments in C programs. These appear between /* and */ 1364and within the comment, individual * and / characters may appear. An attempt to 1365match C comments by applying the pattern 1366<pre> 1367 /\*.*\*/ 1368</pre> 1369to the string 1370<pre> 1371 /* first comment */ not comment /* second comment */ 1372</pre> 1373fails, because it matches the entire string owing to the greediness of the .* 1374item. 1375</P> 1376<P> 1377However, if a quantifier is followed by a question mark, it ceases to be 1378greedy, and instead matches the minimum number of times possible, so the 1379pattern 1380<pre> 1381 /\*.*?\*/ 1382</pre> 1383does the right thing with the C comments. The meaning of the various 1384quantifiers is not otherwise changed, just the preferred number of matches. 1385Do not confuse this use of question mark with its use as a quantifier in its 1386own right. Because it has two uses, it can sometimes appear doubled, as in 1387<pre> 1388 \d??\d 1389</pre> 1390which matches one digit by preference, but can match two if that is the only 1391way the rest of the pattern matches. 1392</P> 1393<P> 1394If the PCRE_UNGREEDY option is set (an option that is not available in Perl), 1395the quantifiers are not greedy by default, but individual ones can be made 1396greedy by following them with a question mark. In other words, it inverts the 1397default behaviour. 1398</P> 1399<P> 1400When a parenthesized subpattern is quantified with a minimum repeat count that 1401is greater than 1 or with a limited maximum, more memory is required for the 1402compiled pattern, in proportion to the size of the minimum or maximum. 1403</P> 1404<P> 1405If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent 1406to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is 1407implicitly anchored, because whatever follows will be tried against every 1408character position in the subject string, so there is no point in retrying the 1409overall match at any position after the first. PCRE normally treats such a 1410pattern as though it were preceded by \A. 1411</P> 1412<P> 1413In cases where it is known that the subject string contains no newlines, it is 1414worth setting PCRE_DOTALL in order to obtain this optimization, or 1415alternatively using ^ to indicate anchoring explicitly. 1416</P> 1417<P> 1418However, there is one situation where the optimization cannot be used. When .* 1419is inside capturing parentheses that are the subject of a back reference 1420elsewhere in the pattern, a match at the start may fail where a later one 1421succeeds. Consider, for example: 1422<pre> 1423 (.*)abc\1 1424</pre> 1425If the subject is "xyz123abc123" the match point is the fourth character. For 1426this reason, such a pattern is not implicitly anchored. 1427</P> 1428<P> 1429When a capturing subpattern is repeated, the value captured is the substring 1430that matched the final iteration. For example, after 1431<pre> 1432 (tweedle[dume]{3}\s*)+ 1433</pre> 1434has matched "tweedledum tweedledee" the value of the captured substring is 1435"tweedledee". However, if there are nested capturing subpatterns, the 1436corresponding captured values may have been set in previous iterations. For 1437example, after 1438<pre> 1439 /(a|(b))+/ 1440</pre> 1441matches "aba" the value of the second captured substring is "b". 1442<a name="atomicgroup"></a></P> 1443<br><a name="SEC16" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br> 1444<P> 1445With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy") 1446repetition, failure of what follows normally causes the repeated item to be 1447re-evaluated to see if a different number of repeats allows the rest of the 1448pattern to match. Sometimes it is useful to prevent this, either to change the 1449nature of the match, or to cause it fail earlier than it otherwise might, when 1450the author of the pattern knows there is no point in carrying on. 1451</P> 1452<P> 1453Consider, for example, the pattern \d+foo when applied to the subject line 1454<pre> 1455 123456bar 1456</pre> 1457After matching all 6 digits and then failing to match "foo", the normal 1458action of the matcher is to try again with only 5 digits matching the \d+ 1459item, and then with 4, and so on, before ultimately failing. "Atomic grouping" 1460(a term taken from Jeffrey Friedl's book) provides the means for specifying 1461that once a subpattern has matched, it is not to be re-evaluated in this way. 1462</P> 1463<P> 1464If we use atomic grouping for the previous example, the matcher gives up 1465immediately on failing to match "foo" the first time. The notation is a kind of 1466special parenthesis, starting with (?> as in this example: 1467<pre> 1468 (?>\d+)foo 1469</pre> 1470This kind of parenthesis "locks up" the part of the pattern it contains once 1471it has matched, and a failure further into the pattern is prevented from 1472backtracking into it. Backtracking past it to previous items, however, works as 1473normal. 1474</P> 1475<P> 1476An alternative description is that a subpattern of this type matches the string 1477of characters that an identical standalone pattern would match, if anchored at 1478the current point in the subject string. 1479</P> 1480<P> 1481Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as 1482the above example can be thought of as a maximizing repeat that must swallow 1483everything it can. So, while both \d+ and \d+? are prepared to adjust the 1484number of digits they match in order to make the rest of the pattern match, 1485(?>\d+) can only match an entire sequence of digits. 1486</P> 1487<P> 1488Atomic groups in general can of course contain arbitrarily complicated 1489subpatterns, and can be nested. However, when the subpattern for an atomic 1490group is just a single repeated item, as in the example above, a simpler 1491notation, called a "possessive quantifier" can be used. This consists of an 1492additional + character following a quantifier. Using this notation, the 1493previous example can be rewritten as 1494<pre> 1495 \d++foo 1496</pre> 1497Note that a possessive quantifier can be used with an entire group, for 1498example: 1499<pre> 1500 (abc|xyz){2,3}+ 1501</pre> 1502Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY 1503option is ignored. They are a convenient notation for the simpler forms of 1504atomic group. However, there is no difference in the meaning of a possessive 1505quantifier and the equivalent atomic group, though there may be a performance 1506difference; possessive quantifiers should be slightly faster. 1507</P> 1508<P> 1509The possessive quantifier syntax is an extension to the Perl 5.8 syntax. 1510Jeffrey Friedl originated the idea (and the name) in the first edition of his 1511book. Mike McCloskey liked it, so implemented it when he built Sun's Java 1512package, and PCRE copied it from there. It ultimately found its way into Perl 1513at release 5.10. 1514</P> 1515<P> 1516PCRE has an optimization that automatically "possessifies" certain simple 1517pattern constructs. For example, the sequence A+B is treated as A++B because 1518there is no point in backtracking into a sequence of A's when B must follow. 1519</P> 1520<P> 1521When a pattern contains an unlimited repeat inside a subpattern that can itself 1522be repeated an unlimited number of times, the use of an atomic group is the 1523only way to avoid some failing matches taking a very long time indeed. The 1524pattern 1525<pre> 1526 (\D+|<\d+>)*[!?] 1527</pre> 1528matches an unlimited number of substrings that either consist of non-digits, or 1529digits enclosed in <>, followed by either ! or ?. When it matches, it runs 1530quickly. However, if it is applied to 1531<pre> 1532 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa 1533</pre> 1534it takes a long time before reporting failure. This is because the string can 1535be divided between the internal \D+ repeat and the external * repeat in a 1536large number of ways, and all have to be tried. (The example uses [!?] rather 1537than a single character at the end, because both PCRE and Perl have an 1538optimization that allows for fast failure when a single character is used. They 1539remember the last single character that is required for a match, and fail early 1540if it is not present in the string.) If the pattern is changed so that it uses 1541an atomic group, like this: 1542<pre> 1543 ((?>\D+)|<\d+>)*[!?] 1544</pre> 1545sequences of non-digits cannot be broken, and failure happens quickly. 1546<a name="backreferences"></a></P> 1547<br><a name="SEC17" href="#TOC1">BACK REFERENCES</a><br> 1548<P> 1549Outside a character class, a backslash followed by a digit greater than 0 (and 1550possibly further digits) is a back reference to a capturing subpattern earlier 1551(that is, to its left) in the pattern, provided there have been that many 1552previous capturing left parentheses. 1553</P> 1554<P> 1555However, if the decimal number following the backslash is less than 10, it is 1556always taken as a back reference, and causes an error only if there are not 1557that many capturing left parentheses in the entire pattern. In other words, the 1558parentheses that are referenced need not be to the left of the reference for 1559numbers less than 10. A "forward back reference" of this type can make sense 1560when a repetition is involved and the subpattern to the right has participated 1561in an earlier iteration. 1562</P> 1563<P> 1564It is not possible to have a numerical "forward back reference" to a subpattern 1565whose number is 10 or more using this syntax because a sequence such as \50 is 1566interpreted as a character defined in octal. See the subsection entitled 1567"Non-printing characters" 1568<a href="#digitsafterbackslash">above</a> 1569for further details of the handling of digits following a backslash. There is 1570no such problem when named parentheses are used. A back reference to any 1571subpattern is possible using named parentheses (see below). 1572</P> 1573<P> 1574Another way of avoiding the ambiguity inherent in the use of digits following a 1575backslash is to use the \g escape sequence, which is a feature introduced in 1576Perl 5.10. This escape must be followed by an unsigned number or a negative 1577number, optionally enclosed in braces. These examples are all identical: 1578<pre> 1579 (ring), \1 1580 (ring), \g1 1581 (ring), \g{1} 1582</pre> 1583An unsigned number specifies an absolute reference without the ambiguity that 1584is present in the older syntax. It is also useful when literal digits follow 1585the reference. A negative number is a relative reference. Consider this 1586example: 1587<pre> 1588 (abc(def)ghi)\g{-1} 1589</pre> 1590The sequence \g{-1} is a reference to the most recently started capturing 1591subpattern before \g, that is, is it equivalent to \2. Similarly, \g{-2} 1592would be equivalent to \1. The use of relative references can be helpful in 1593long patterns, and also in patterns that are created by joining together 1594fragments that contain references within themselves. 1595</P> 1596<P> 1597A back reference matches whatever actually matched the capturing subpattern in 1598the current subject string, rather than anything matching the subpattern 1599itself (see 1600<a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a> 1601below for a way of doing that). So the pattern 1602<pre> 1603 (sens|respons)e and \1ibility 1604</pre> 1605matches "sense and sensibility" and "response and responsibility", but not 1606"sense and responsibility". If caseful matching is in force at the time of the 1607back reference, the case of letters is relevant. For example, 1608<pre> 1609 ((?i)rah)\s+\1 1610</pre> 1611matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original 1612capturing subpattern is matched caselessly. 1613</P> 1614<P> 1615There are several different ways of writing back references to named 1616subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or 1617\k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified 1618back reference syntax, in which \g can be used for both numeric and named 1619references, is also supported. We could rewrite the above example in any of 1620the following ways: 1621<pre> 1622 (?<p1>(?i)rah)\s+\k<p1> 1623 (?'p1'(?i)rah)\s+\k{p1} 1624 (?P<p1>(?i)rah)\s+(?P=p1) 1625 (?<p1>(?i)rah)\s+\g{p1} 1626</pre> 1627A subpattern that is referenced by name may appear in the pattern before or 1628after the reference. 1629</P> 1630<P> 1631There may be more than one back reference to the same subpattern. If a 1632subpattern has not actually been used in a particular match, any back 1633references to it always fail by default. For example, the pattern 1634<pre> 1635 (a|(bc))\2 1636</pre> 1637always fails if it starts to match "a" rather than "bc". However, if the 1638PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an 1639unset value matches an empty string. 1640</P> 1641<P> 1642Because there may be many capturing parentheses in a pattern, all digits 1643following a backslash are taken as part of a potential back reference number. 1644If the pattern continues with a digit character, some delimiter must be used to 1645terminate the back reference. If the PCRE_EXTENDED option is set, this can be 1646whitespace. Otherwise, the \g{ syntax or an empty comment (see 1647<a href="#comments">"Comments"</a> 1648below) can be used. 1649</P> 1650<br><b> 1651Recursive back references 1652</b><br> 1653<P> 1654A back reference that occurs inside the parentheses to which it refers fails 1655when the subpattern is first used, so, for example, (a\1) never matches. 1656However, such references can be useful inside repeated subpatterns. For 1657example, the pattern 1658<pre> 1659 (a|b\1)+ 1660</pre> 1661matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of 1662the subpattern, the back reference matches the character string corresponding 1663to the previous iteration. In order for this to work, the pattern must be such 1664that the first iteration does not need to match the back reference. This can be 1665done using alternation, as in the example above, or by a quantifier with a 1666minimum of zero. 1667</P> 1668<P> 1669Back references of this type cause the group that they reference to be treated 1670as an 1671<a href="#atomicgroup">atomic group.</a> 1672Once the whole group has been matched, a subsequent matching failure cannot 1673cause backtracking into the middle of the group. 1674<a name="bigassertions"></a></P> 1675<br><a name="SEC18" href="#TOC1">ASSERTIONS</a><br> 1676<P> 1677An assertion is a test on the characters following or preceding the current 1678matching point that does not actually consume any characters. The simple 1679assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described 1680<a href="#smallassertions">above.</a> 1681</P> 1682<P> 1683More complicated assertions are coded as subpatterns. There are two kinds: 1684those that look ahead of the current position in the subject string, and those 1685that look behind it. An assertion subpattern is matched in the normal way, 1686except that it does not cause the current matching position to be changed. 1687</P> 1688<P> 1689Assertion subpatterns are not capturing subpatterns, and may not be repeated, 1690because it makes no sense to assert the same thing several times. If any kind 1691of assertion contains capturing subpatterns within it, these are counted for 1692the purposes of numbering the capturing subpatterns in the whole pattern. 1693However, substring capturing is carried out only for positive assertions, 1694because it does not make sense for negative assertions. 1695</P> 1696<br><b> 1697Lookahead assertions 1698</b><br> 1699<P> 1700Lookahead assertions start with (?= for positive assertions and (?! for 1701negative assertions. For example, 1702<pre> 1703 \w+(?=;) 1704</pre> 1705matches a word followed by a semicolon, but does not include the semicolon in 1706the match, and 1707<pre> 1708 foo(?!bar) 1709</pre> 1710matches any occurrence of "foo" that is not followed by "bar". Note that the 1711apparently similar pattern 1712<pre> 1713 (?!foo)bar 1714</pre> 1715does not find an occurrence of "bar" that is preceded by something other than 1716"foo"; it finds any occurrence of "bar" whatsoever, because the assertion 1717(?!foo) is always true when the next three characters are "bar". A 1718lookbehind assertion is needed to achieve the other effect. 1719</P> 1720<P> 1721If you want to force a matching failure at some point in a pattern, the most 1722convenient way to do it is with (?!) because an empty string always matches, so 1723an assertion that requires there not to be an empty string must always fail. 1724The Perl 5.10 backtracking control verb (*FAIL) or (*F) is essentially a 1725synonym for (?!). 1726<a name="lookbehind"></a></P> 1727<br><b> 1728Lookbehind assertions 1729</b><br> 1730<P> 1731Lookbehind assertions start with (?<= for positive assertions and (?<! for 1732negative assertions. For example, 1733<pre> 1734 (?<!foo)bar 1735</pre> 1736does find an occurrence of "bar" that is not preceded by "foo". The contents of 1737a lookbehind assertion are restricted such that all the strings it matches must 1738have a fixed length. However, if there are several top-level alternatives, they 1739do not all have to have the same fixed length. Thus 1740<pre> 1741 (?<=bullock|donkey) 1742</pre> 1743is permitted, but 1744<pre> 1745 (?<!dogs?|cats?) 1746</pre> 1747causes an error at compile time. Branches that match different length strings 1748are permitted only at the top level of a lookbehind assertion. This is an 1749extension compared with Perl (5.8 and 5.10), which requires all branches to 1750match the same length of string. An assertion such as 1751<pre> 1752 (?<=ab(c|de)) 1753</pre> 1754is not permitted, because its single top-level branch can match two different 1755lengths, but it is acceptable to PCRE if rewritten to use two top-level 1756branches: 1757<pre> 1758 (?<=abc|abde) 1759</pre> 1760In some cases, the Perl 5.10 escape sequence \K 1761<a href="#resetmatchstart">(see above)</a> 1762can be used instead of a lookbehind assertion to get round the fixed-length 1763restriction. 1764</P> 1765<P> 1766The implementation of lookbehind assertions is, for each alternative, to 1767temporarily move the current position back by the fixed length and then try to 1768match. If there are insufficient characters before the current position, the 1769assertion fails. 1770</P> 1771<P> 1772PCRE does not allow the \C escape (which matches a single byte in UTF-8 mode) 1773to appear in lookbehind assertions, because it makes it impossible to calculate 1774the length of the lookbehind. The \X and \R escapes, which can match 1775different numbers of bytes, are also not permitted. 1776</P> 1777<P> 1778<a href="#subpatternsassubroutines">"Subroutine"</a> 1779calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long 1780as the subpattern matches a fixed-length string. 1781<a href="#recursion">Recursion,</a> 1782however, is not supported. 1783</P> 1784<P> 1785Possessive quantifiers can be used in conjunction with lookbehind assertions to 1786specify efficient matching of fixed-length strings at the end of subject 1787strings. Consider a simple pattern such as 1788<pre> 1789 abcd$ 1790</pre> 1791when applied to a long string that does not match. Because matching proceeds 1792from left to right, PCRE will look for each "a" in the subject and then see if 1793what follows matches the rest of the pattern. If the pattern is specified as 1794<pre> 1795 ^.*abcd$ 1796</pre> 1797the initial .* matches the entire string at first, but when this fails (because 1798there is no following "a"), it backtracks to match all but the last character, 1799then all but the last two characters, and so on. Once again the search for "a" 1800covers the entire string, from right to left, so we are no better off. However, 1801if the pattern is written as 1802<pre> 1803 ^.*+(?<=abcd) 1804</pre> 1805there can be no backtracking for the .*+ item; it can match only the entire 1806string. The subsequent lookbehind assertion does a single test on the last four 1807characters. If it fails, the match fails immediately. For long strings, this 1808approach makes a significant difference to the processing time. 1809</P> 1810<br><b> 1811Using multiple assertions 1812</b><br> 1813<P> 1814Several assertions (of any sort) may occur in succession. For example, 1815<pre> 1816 (?<=\d{3})(?<!999)foo 1817</pre> 1818matches "foo" preceded by three digits that are not "999". Notice that each of 1819the assertions is applied independently at the same point in the subject 1820string. First there is a check that the previous three characters are all 1821digits, and then there is a check that the same three characters are not "999". 1822This pattern does <i>not</i> match "foo" preceded by six characters, the first 1823of which are digits and the last three of which are not "999". For example, it 1824doesn't match "123abcfoo". A pattern to do that is 1825<pre> 1826 (?<=\d{3}...)(?<!999)foo 1827</pre> 1828This time the first assertion looks at the preceding six characters, checking 1829that the first three are digits, and then the second assertion checks that the 1830preceding three characters are not "999". 1831</P> 1832<P> 1833Assertions can be nested in any combination. For example, 1834<pre> 1835 (?<=(?<!foo)bar)baz 1836</pre> 1837matches an occurrence of "baz" that is preceded by "bar" which in turn is not 1838preceded by "foo", while 1839<pre> 1840 (?<=\d{3}(?!999)...)foo 1841</pre> 1842is another pattern that matches "foo" preceded by three digits and any three 1843characters that are not "999". 1844<a name="conditions"></a></P> 1845<br><a name="SEC19" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br> 1846<P> 1847It is possible to cause the matching process to obey a subpattern 1848conditionally or to choose between two alternative subpatterns, depending on 1849the result of an assertion, or whether a specific capturing subpattern has 1850already been matched. The two possible forms of conditional subpattern are: 1851<pre> 1852 (?(condition)yes-pattern) 1853 (?(condition)yes-pattern|no-pattern) 1854</pre> 1855If the condition is satisfied, the yes-pattern is used; otherwise the 1856no-pattern (if present) is used. If there are more than two alternatives in the 1857subpattern, a compile-time error occurs. 1858</P> 1859<P> 1860There are four kinds of condition: references to subpatterns, references to 1861recursion, a pseudo-condition called DEFINE, and assertions. 1862</P> 1863<br><b> 1864Checking for a used subpattern by number 1865</b><br> 1866<P> 1867If the text between the parentheses consists of a sequence of digits, the 1868condition is true if a capturing subpattern of that number has previously 1869matched. If there is more than one capturing subpattern with the same number 1870(see the earlier 1871<a href="#recursion">section about duplicate subpattern numbers),</a> 1872the condition is true if any of them have been set. An alternative notation is 1873to precede the digits with a plus or minus sign. In this case, the subpattern 1874number is relative rather than absolute. The most recently opened parentheses 1875can be referenced by (?(-1), the next most recent by (?(-2), and so on. In 1876looping constructs it can also make sense to refer to subsequent groups with 1877constructs such as (?(+2). 1878</P> 1879<P> 1880Consider the following pattern, which contains non-significant white space to 1881make it more readable (assume the PCRE_EXTENDED option) and to divide it into 1882three parts for ease of discussion: 1883<pre> 1884 ( \( )? [^()]+ (?(1) \) ) 1885</pre> 1886The first part matches an optional opening parenthesis, and if that 1887character is present, sets it as the first captured substring. The second part 1888matches one or more characters that are not parentheses. The third part is a 1889conditional subpattern that tests whether the first set of parentheses matched 1890or not. If they did, that is, if subject started with an opening parenthesis, 1891the condition is true, and so the yes-pattern is executed and a closing 1892parenthesis is required. Otherwise, since no-pattern is not present, the 1893subpattern matches nothing. In other words, this pattern matches a sequence of 1894non-parentheses, optionally enclosed in parentheses. 1895</P> 1896<P> 1897If you were embedding this pattern in a larger one, you could use a relative 1898reference: 1899<pre> 1900 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ... 1901</pre> 1902This makes the fragment independent of the parentheses in the larger pattern. 1903</P> 1904<br><b> 1905Checking for a used subpattern by name 1906</b><br> 1907<P> 1908Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used 1909subpattern by name. For compatibility with earlier versions of PCRE, which had 1910this facility before Perl, the syntax (?(name)...) is also recognized. However, 1911there is a possible ambiguity with this syntax, because subpattern names may 1912consist entirely of digits. PCRE looks first for a named subpattern; if it 1913cannot find one and the name consists entirely of digits, PCRE looks for a 1914subpattern of that number, which must be greater than zero. Using subpattern 1915names that consist entirely of digits is not recommended. 1916</P> 1917<P> 1918Rewriting the above example to use a named subpattern gives this: 1919<pre> 1920 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) ) 1921</pre> 1922If the name used in a condition of this kind is a duplicate, the test is 1923applied to all subpatterns of the same name, and is true if any one of them has 1924matched. 1925</P> 1926<br><b> 1927Checking for pattern recursion 1928</b><br> 1929<P> 1930If the condition is the string (R), and there is no subpattern with the name R, 1931the condition is true if a recursive call to the whole pattern or any 1932subpattern has been made. If digits or a name preceded by ampersand follow the 1933letter R, for example: 1934<pre> 1935 (?(R3)...) or (?(R&name)...) 1936</pre> 1937the condition is true if the most recent recursion is into a subpattern whose 1938number or name is given. This condition does not check the entire recursion 1939stack. If the name used in a condition of this kind is a duplicate, the test is 1940applied to all subpatterns of the same name, and is true if any one of them is 1941the most recent recursion. 1942</P> 1943<P> 1944At "top level", all these recursion test conditions are false. 1945<a href="#recursion">The syntax for recursive patterns</a> 1946is described below. 1947</P> 1948<br><b> 1949Defining subpatterns for use by reference only 1950</b><br> 1951<P> 1952If the condition is the string (DEFINE), and there is no subpattern with the 1953name DEFINE, the condition is always false. In this case, there may be only one 1954alternative in the subpattern. It is always skipped if control reaches this 1955point in the pattern; the idea of DEFINE is that it can be used to define 1956"subroutines" that can be referenced from elsewhere. (The use of 1957<a href="#subpatternsassubroutines">"subroutines"</a> 1958is described below.) For example, a pattern to match an IPv4 address could be 1959written like this (ignore whitespace and line breaks): 1960<pre> 1961 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) ) 1962 \b (?&byte) (\.(?&byte)){3} \b 1963</pre> 1964The first part of the pattern is a DEFINE group inside which a another group 1965named "byte" is defined. This matches an individual component of an IPv4 1966address (a number less than 256). When matching takes place, this part of the 1967pattern is skipped because DEFINE acts like a false condition. The rest of the 1968pattern uses references to the named group to match the four dot-separated 1969components of an IPv4 address, insisting on a word boundary at each end. 1970</P> 1971<br><b> 1972Assertion conditions 1973</b><br> 1974<P> 1975If the condition is not in any of the above formats, it must be an assertion. 1976This may be a positive or negative lookahead or lookbehind assertion. Consider 1977this pattern, again containing non-significant white space, and with the two 1978alternatives on the second line: 1979<pre> 1980 (?(?=[^a-z]*[a-z]) 1981 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} ) 1982</pre> 1983The condition is a positive lookahead assertion that matches an optional 1984sequence of non-letters followed by a letter. In other words, it tests for the 1985presence of at least one letter in the subject. If a letter is found, the 1986subject is matched against the first alternative; otherwise it is matched 1987against the second. This pattern matches strings in one of the two forms 1988dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits. 1989<a name="comments"></a></P> 1990<br><a name="SEC20" href="#TOC1">COMMENTS</a><br> 1991<P> 1992The sequence (?# marks the start of a comment that continues up to the next 1993closing parenthesis. Nested parentheses are not permitted. The characters 1994that make up a comment play no part in the pattern matching at all. 1995</P> 1996<P> 1997If the PCRE_EXTENDED option is set, an unescaped # character outside a 1998character class introduces a comment that continues to immediately after the 1999next newline in the pattern. 2000<a name="recursion"></a></P> 2001<br><a name="SEC21" href="#TOC1">RECURSIVE PATTERNS</a><br> 2002<P> 2003Consider the problem of matching a string in parentheses, allowing for 2004unlimited nested parentheses. Without the use of recursion, the best that can 2005be done is to use a pattern that matches up to some fixed depth of nesting. It 2006is not possible to handle an arbitrary nesting depth. 2007</P> 2008<P> 2009For some time, Perl has provided a facility that allows regular expressions to 2010recurse (amongst other things). It does this by interpolating Perl code in the 2011expression at run time, and the code can refer to the expression itself. A Perl 2012pattern using code interpolation to solve the parentheses problem can be 2013created like this: 2014<pre> 2015 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x; 2016</pre> 2017The (?p{...}) item interpolates Perl code at run time, and in this case refers 2018recursively to the pattern in which it appears. 2019</P> 2020<P> 2021Obviously, PCRE cannot support the interpolation of Perl code. Instead, it 2022supports special syntax for recursion of the entire pattern, and also for 2023individual subpattern recursion. After its introduction in PCRE and Python, 2024this kind of recursion was subsequently introduced into Perl at release 5.10. 2025</P> 2026<P> 2027A special item that consists of (? followed by a number greater than zero and a 2028closing parenthesis is a recursive call of the subpattern of the given number, 2029provided that it occurs inside that subpattern. (If not, it is a 2030<a href="#subpatternsassubroutines">"subroutine"</a> 2031call, which is described in the next section.) The special item (?R) or (?0) is 2032a recursive call of the entire regular expression. 2033</P> 2034<P> 2035This PCRE pattern solves the nested parentheses problem (assume the 2036PCRE_EXTENDED option is set so that white space is ignored): 2037<pre> 2038 \( ( [^()]++ | (?R) )* \) 2039</pre> 2040First it matches an opening parenthesis. Then it matches any number of 2041substrings which can either be a sequence of non-parentheses, or a recursive 2042match of the pattern itself (that is, a correctly parenthesized substring). 2043Finally there is a closing parenthesis. Note the use of a possessive quantifier 2044to avoid backtracking into sequences of non-parentheses. 2045</P> 2046<P> 2047If this were part of a larger pattern, you would not want to recurse the entire 2048pattern, so instead you could use this: 2049<pre> 2050 ( \( ( [^()]++ | (?1) )* \) ) 2051</pre> 2052We have put the pattern into parentheses, and caused the recursion to refer to 2053them instead of the whole pattern. 2054</P> 2055<P> 2056In a larger pattern, keeping track of parenthesis numbers can be tricky. This 2057is made easier by the use of relative references (a Perl 5.10 feature). 2058Instead of (?1) in the pattern above you can write (?-2) to refer to the second 2059most recently opened parentheses preceding the recursion. In other words, a 2060negative number counts capturing parentheses leftwards from the point at which 2061it is encountered. 2062</P> 2063<P> 2064It is also possible to refer to subsequently opened parentheses, by writing 2065references such as (?+2). However, these cannot be recursive because the 2066reference is not inside the parentheses that are referenced. They are always 2067<a href="#subpatternsassubroutines">"subroutine"</a> 2068calls, as described in the next section. 2069</P> 2070<P> 2071An alternative approach is to use named parentheses instead. The Perl syntax 2072for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We 2073could rewrite the above example as follows: 2074<pre> 2075 (?<pn> \( ( [^()]++ | (?&pn) )* \) ) 2076</pre> 2077If there is more than one subpattern with the same name, the earliest one is 2078used. 2079</P> 2080<P> 2081This particular example pattern that we have been looking at contains nested 2082unlimited repeats, and so the use of a possessive quantifier for matching 2083strings of non-parentheses is important when applying the pattern to strings 2084that do not match. For example, when this pattern is applied to 2085<pre> 2086 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa() 2087</pre> 2088it yields "no match" quickly. However, if a possessive quantifier is not used, 2089the match runs for a very long time indeed because there are so many different 2090ways the + and * repeats can carve up the subject, and all have to be tested 2091before failure can be reported. 2092</P> 2093<P> 2094At the end of a match, the values of capturing parentheses are those from 2095the outermost level. If you want to obtain intermediate values, a callout 2096function can be used (see below and the 2097<a href="pcrecallout.html"><b>pcrecallout</b></a> 2098documentation). If the pattern above is matched against 2099<pre> 2100 (ab(cd)ef) 2101</pre> 2102the value for the inner capturing parentheses (numbered 2) is "ef", which is 2103the last value taken on at the top level. If a capturing subpattern is not 2104matched at the top level, its final value is unset, even if it is (temporarily) 2105set at a deeper level. 2106</P> 2107<P> 2108If there are more than 15 capturing parentheses in a pattern, PCRE has to 2109obtain extra memory to store data during a recursion, which it does by using 2110<b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no memory can 2111be obtained, the match fails with the PCRE_ERROR_NOMEMORY error. 2112</P> 2113<P> 2114Do not confuse the (?R) item with the condition (R), which tests for recursion. 2115Consider this pattern, which matches text in angle brackets, allowing for 2116arbitrary nesting. Only digits are allowed in nested brackets (that is, when 2117recursing), whereas any characters are permitted at the outer level. 2118<pre> 2119 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * > 2120</pre> 2121In this pattern, (?(R) is the start of a conditional subpattern, with two 2122different alternatives for the recursive and non-recursive cases. The (?R) item 2123is the actual recursive call. 2124<a name="recursiondifference"></a></P> 2125<br><b> 2126Recursion difference from Perl 2127</b><br> 2128<P> 2129In PCRE (like Python, but unlike Perl), a recursive subpattern call is always 2130treated as an atomic group. That is, once it has matched some of the subject 2131string, it is never re-entered, even if it contains untried alternatives and 2132there is a subsequent matching failure. This can be illustrated by the 2133following pattern, which purports to match a palindromic string that contains 2134an odd number of characters (for example, "a", "aba", "abcba", "abcdcba"): 2135<pre> 2136 ^(.|(.)(?1)\2)$ 2137</pre> 2138The idea is that it either matches a single character, or two identical 2139characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE 2140it does not if the pattern is longer than three characters. Consider the 2141subject string "abcba": 2142</P> 2143<P> 2144At the top level, the first character is matched, but as it is not at the end 2145of the string, the first alternative fails; the second alternative is taken 2146and the recursion kicks in. The recursive call to subpattern 1 successfully 2147matches the next character ("b"). (Note that the beginning and end of line 2148tests are not part of the recursion). 2149</P> 2150<P> 2151Back at the top level, the next character ("c") is compared with what 2152subpattern 2 matched, which was "a". This fails. Because the recursion is 2153treated as an atomic group, there are now no backtracking points, and so the 2154entire match fails. (Perl is able, at this point, to re-enter the recursion and 2155try the second alternative.) However, if the pattern is written with the 2156alternatives in the other order, things are different: 2157<pre> 2158 ^((.)(?1)\2|.)$ 2159</pre> 2160This time, the recursing alternative is tried first, and continues to recurse 2161until it runs out of characters, at which point the recursion fails. But this 2162time we do have another alternative to try at the higher level. That is the big 2163difference: in the previous case the remaining alternative is at a deeper 2164recursion level, which PCRE cannot use. 2165</P> 2166<P> 2167To change the pattern so that matches all palindromic strings, not just those 2168with an odd number of characters, it is tempting to change the pattern to this: 2169<pre> 2170 ^((.)(?1)\2|.?)$ 2171</pre> 2172Again, this works in Perl, but not in PCRE, and for the same reason. When a 2173deeper recursion has matched a single character, it cannot be entered again in 2174order to match an empty string. The solution is to separate the two cases, and 2175write out the odd and even cases as alternatives at the higher level: 2176<pre> 2177 ^(?:((.)(?1)\2|)|((.)(?3)\4|.)) 2178</pre> 2179If you want to match typical palindromic phrases, the pattern has to ignore all 2180non-word characters, which can be done like this: 2181<pre> 2182 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$ 2183</pre> 2184If run with the PCRE_CASELESS option, this pattern matches phrases such as "A 2185man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note 2186the use of the possessive quantifier *+ to avoid backtracking into sequences of 2187non-word characters. Without this, PCRE takes a great deal longer (ten times or 2188more) to match typical phrases, and Perl takes so long that you think it has 2189gone into a loop. 2190</P> 2191<P> 2192<b>WARNING</b>: The palindrome-matching patterns above work only if the subject 2193string does not start with a palindrome that is shorter than the entire string. 2194For example, although "abcba" is correctly matched, if the subject is "ababa", 2195PCRE finds the palindrome "aba" at the start, then fails at top level because 2196the end of the string does not follow. Once again, it cannot jump back into the 2197recursion to try other alternatives, so the entire match fails. 2198<a name="subpatternsassubroutines"></a></P> 2199<br><a name="SEC22" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br> 2200<P> 2201If the syntax for a recursive subpattern reference (either by number or by 2202name) is used outside the parentheses to which it refers, it operates like a 2203subroutine in a programming language. The "called" subpattern may be defined 2204before or after the reference. A numbered reference can be absolute or 2205relative, as in these examples: 2206<pre> 2207 (...(absolute)...)...(?2)... 2208 (...(relative)...)...(?-1)... 2209 (...(?+1)...(relative)... 2210</pre> 2211An earlier example pointed out that the pattern 2212<pre> 2213 (sens|respons)e and \1ibility 2214</pre> 2215matches "sense and sensibility" and "response and responsibility", but not 2216"sense and responsibility". If instead the pattern 2217<pre> 2218 (sens|respons)e and (?1)ibility 2219</pre> 2220is used, it does match "sense and responsibility" as well as the other two 2221strings. Another example is given in the discussion of DEFINE above. 2222</P> 2223<P> 2224Like recursive subpatterns, a subroutine call is always treated as an atomic 2225group. That is, once it has matched some of the subject string, it is never 2226re-entered, even if it contains untried alternatives and there is a subsequent 2227matching failure. Any capturing parentheses that are set during the subroutine 2228call revert to their previous values afterwards. 2229</P> 2230<P> 2231When a subpattern is used as a subroutine, processing options such as 2232case-independence are fixed when the subpattern is defined. They cannot be 2233changed for different calls. For example, consider this pattern: 2234<pre> 2235 (abc)(?i:(?-1)) 2236</pre> 2237It matches "abcabc". It does not match "abcABC" because the change of 2238processing option does not affect the called subpattern. 2239<a name="onigurumasubroutines"></a></P> 2240<br><a name="SEC23" href="#TOC1">ONIGURUMA SUBROUTINE SYNTAX</a><br> 2241<P> 2242For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or 2243a number enclosed either in angle brackets or single quotes, is an alternative 2244syntax for referencing a subpattern as a subroutine, possibly recursively. Here 2245are two of the examples used above, rewritten using this syntax: 2246<pre> 2247 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) ) 2248 (sens|respons)e and \g'1'ibility 2249</pre> 2250PCRE supports an extension to Oniguruma: if a number is preceded by a 2251plus or a minus sign it is taken as a relative reference. For example: 2252<pre> 2253 (abc)(?i:\g<-1>) 2254</pre> 2255Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are <i>not</i> 2256synonymous. The former is a back reference; the latter is a subroutine call. 2257</P> 2258<br><a name="SEC24" href="#TOC1">CALLOUTS</a><br> 2259<P> 2260Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl 2261code to be obeyed in the middle of matching a regular expression. This makes it 2262possible, amongst other things, to extract different substrings that match the 2263same pair of parentheses when there is a repetition. 2264</P> 2265<P> 2266PCRE provides a similar feature, but of course it cannot obey arbitrary Perl 2267code. The feature is called "callout". The caller of PCRE provides an external 2268function by putting its entry point in the global variable <i>pcre_callout</i>. 2269By default, this variable contains NULL, which disables all calling out. 2270</P> 2271<P> 2272Within a regular expression, (?C) indicates the points at which the external 2273function is to be called. If you want to identify different callout points, you 2274can put a number less than 256 after the letter C. The default value is zero. 2275For example, this pattern has two callout points: 2276<pre> 2277 (?C1)abc(?C2)def 2278</pre> 2279If the PCRE_AUTO_CALLOUT flag is passed to <b>pcre_compile()</b>, callouts are 2280automatically installed before each item in the pattern. They are all numbered 2281255. 2282</P> 2283<P> 2284During matching, when PCRE reaches a callout point (and <i>pcre_callout</i> is 2285set), the external function is called. It is provided with the number of the 2286callout, the position in the pattern, and, optionally, one item of data 2287originally supplied by the caller of <b>pcre_exec()</b>. The callout function 2288may cause matching to proceed, to backtrack, or to fail altogether. A complete 2289description of the interface to the callout function is given in the 2290<a href="pcrecallout.html"><b>pcrecallout</b></a> 2291documentation. 2292</P> 2293<br><a name="SEC25" href="#TOC1">BACKTRACKING CONTROL</a><br> 2294<P> 2295Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which 2296are described in the Perl documentation as "experimental and subject to change 2297or removal in a future version of Perl". It goes on to say: "Their usage in 2298production code should be noted to avoid problems during upgrades." The same 2299remarks apply to the PCRE features described in this section. 2300</P> 2301<P> 2302Since these verbs are specifically related to backtracking, most of them can be 2303used only when the pattern is to be matched using <b>pcre_exec()</b>, which uses 2304a backtracking algorithm. With the exception of (*FAIL), which behaves like a 2305failing negative assertion, they cause an error if encountered by 2306<b>pcre_dfa_exec()</b>. 2307</P> 2308<P> 2309If any of these verbs are used in an assertion or subroutine subpattern 2310(including recursive subpatterns), their effect is confined to that subpattern; 2311it does not extend to the surrounding pattern. Note that such subpatterns are 2312processed as anchored at the point where they are tested. 2313</P> 2314<P> 2315The new verbs make use of what was previously invalid syntax: an opening 2316parenthesis followed by an asterisk. In Perl, they are generally of the form 2317(*VERB:ARG) but PCRE does not support the use of arguments, so its general 2318form is just (*VERB). Any number of these verbs may occur in a pattern. There 2319are two kinds: 2320</P> 2321<br><b> 2322Verbs that act immediately 2323</b><br> 2324<P> 2325The following verbs act as soon as they are encountered: 2326<pre> 2327 (*ACCEPT) 2328</pre> 2329This verb causes the match to end successfully, skipping the remainder of the 2330pattern. When inside a recursion, only the innermost pattern is ended 2331immediately. If (*ACCEPT) is inside capturing parentheses, the data so far is 2332captured. (This feature was added to PCRE at release 8.00.) For example: 2333<pre> 2334 A((?:A|B(*ACCEPT)|C)D) 2335</pre> 2336This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by 2337the outer parentheses. 2338<pre> 2339 (*FAIL) or (*F) 2340</pre> 2341This verb causes the match to fail, forcing backtracking to occur. It is 2342equivalent to (?!) but easier to read. The Perl documentation notes that it is 2343probably useful only when combined with (?{}) or (??{}). Those are, of course, 2344Perl features that are not present in PCRE. The nearest equivalent is the 2345callout feature, as for example in this pattern: 2346<pre> 2347 a+(?C)(*FAIL) 2348</pre> 2349A match with the string "aaaa" always fails, but the callout is taken before 2350each backtrack happens (in this example, 10 times). 2351</P> 2352<br><b> 2353Verbs that act after backtracking 2354</b><br> 2355<P> 2356The following verbs do nothing when they are encountered. Matching continues 2357with what follows, but if there is no subsequent match, a failure is forced. 2358The verbs differ in exactly what kind of failure occurs. 2359<pre> 2360 (*COMMIT) 2361</pre> 2362This verb causes the whole match to fail outright if the rest of the pattern 2363does not match. Even if the pattern is unanchored, no further attempts to find 2364a match by advancing the starting point take place. Once (*COMMIT) has been 2365passed, <b>pcre_exec()</b> is committed to finding a match at the current 2366starting point, or not at all. For example: 2367<pre> 2368 a+(*COMMIT)b 2369</pre> 2370This matches "xxaab" but not "aacaab". It can be thought of as a kind of 2371dynamic anchor, or "I've started, so I must finish." 2372<pre> 2373 (*PRUNE) 2374</pre> 2375This verb causes the match to fail at the current position if the rest of the 2376pattern does not match. If the pattern is unanchored, the normal "bumpalong" 2377advance to the next starting character then happens. Backtracking can occur as 2378usual to the left of (*PRUNE), or when matching to the right of (*PRUNE), but 2379if there is no match to the right, backtracking cannot cross (*PRUNE). 2380In simple cases, the use of (*PRUNE) is just an alternative to an atomic 2381group or possessive quantifier, but there are some uses of (*PRUNE) that cannot 2382be expressed in any other way. 2383<pre> 2384 (*SKIP) 2385</pre> 2386This verb is like (*PRUNE), except that if the pattern is unanchored, the 2387"bumpalong" advance is not to the next character, but to the position in the 2388subject where (*SKIP) was encountered. (*SKIP) signifies that whatever text 2389was matched leading up to it cannot be part of a successful match. Consider: 2390<pre> 2391 a+(*SKIP)b 2392</pre> 2393If the subject is "aaaac...", after the first match attempt fails (starting at 2394the first character in the string), the starting point skips on to start the 2395next attempt at "c". Note that a possessive quantifer does not have the same 2396effect as this example; although it would suppress backtracking during the 2397first match attempt, the second attempt would start at the second character 2398instead of skipping on to "c". 2399<pre> 2400 (*THEN) 2401</pre> 2402This verb causes a skip to the next alternation if the rest of the pattern does 2403not match. That is, it cancels pending backtracking, but only within the 2404current alternation. Its name comes from the observation that it can be used 2405for a pattern-based if-then-else block: 2406<pre> 2407 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ... 2408</pre> 2409If the COND1 pattern matches, FOO is tried (and possibly further items after 2410the end of the group if FOO succeeds); on failure the matcher skips to the 2411second alternative and tries COND2, without backtracking into COND1. If (*THEN) 2412is used outside of any alternation, it acts exactly like (*PRUNE). 2413</P> 2414<br><a name="SEC26" href="#TOC1">SEE ALSO</a><br> 2415<P> 2416<b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3), 2417<b>pcresyntax</b>(3), <b>pcre</b>(3). 2418</P> 2419<br><a name="SEC27" href="#TOC1">AUTHOR</a><br> 2420<P> 2421Philip Hazel 2422<br> 2423University Computing Service 2424<br> 2425Cambridge CB2 3QH, England. 2426<br> 2427</P> 2428<br><a name="SEC28" href="#TOC1">REVISION</a><br> 2429<P> 2430Last updated: 06 March 2010 2431<br> 2432Copyright © 1997-2010 University of Cambridge. 2433<br> 2434<p> 2435Return to the <a href="index.html">PCRE index page</a>. 2436</p> 2437