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">SPECIAL START-OF-PATTERN ITEMS</a> 18<li><a name="TOC3" href="#SEC3">EBCDIC CHARACTER CODES</a> 19<li><a name="TOC4" href="#SEC4">CHARACTERS AND METACHARACTERS</a> 20<li><a name="TOC5" href="#SEC5">BACKSLASH</a> 21<li><a name="TOC6" href="#SEC6">CIRCUMFLEX AND DOLLAR</a> 22<li><a name="TOC7" href="#SEC7">FULL STOP (PERIOD, DOT) AND \N</a> 23<li><a name="TOC8" href="#SEC8">MATCHING A SINGLE DATA UNIT</a> 24<li><a name="TOC9" href="#SEC9">SQUARE BRACKETS AND CHARACTER CLASSES</a> 25<li><a name="TOC10" href="#SEC10">POSIX CHARACTER CLASSES</a> 26<li><a name="TOC11" href="#SEC11">COMPATIBILITY FEATURE FOR WORD BOUNDARIES</a> 27<li><a name="TOC12" href="#SEC12">VERTICAL BAR</a> 28<li><a name="TOC13" href="#SEC13">INTERNAL OPTION SETTING</a> 29<li><a name="TOC14" href="#SEC14">SUBPATTERNS</a> 30<li><a name="TOC15" href="#SEC15">DUPLICATE SUBPATTERN NUMBERS</a> 31<li><a name="TOC16" href="#SEC16">NAMED SUBPATTERNS</a> 32<li><a name="TOC17" href="#SEC17">REPETITION</a> 33<li><a name="TOC18" href="#SEC18">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a> 34<li><a name="TOC19" href="#SEC19">BACK REFERENCES</a> 35<li><a name="TOC20" href="#SEC20">ASSERTIONS</a> 36<li><a name="TOC21" href="#SEC21">CONDITIONAL SUBPATTERNS</a> 37<li><a name="TOC22" href="#SEC22">COMMENTS</a> 38<li><a name="TOC23" href="#SEC23">RECURSIVE PATTERNS</a> 39<li><a name="TOC24" href="#SEC24">SUBPATTERNS AS SUBROUTINES</a> 40<li><a name="TOC25" href="#SEC25">ONIGURUMA SUBROUTINE SYNTAX</a> 41<li><a name="TOC26" href="#SEC26">CALLOUTS</a> 42<li><a name="TOC27" href="#SEC27">BACKTRACKING CONTROL</a> 43<li><a name="TOC28" href="#SEC28">SEE ALSO</a> 44<li><a name="TOC29" href="#SEC29">AUTHOR</a> 45<li><a name="TOC30" href="#SEC30">REVISION</a> 46</ul> 47<br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br> 48<P> 49The syntax and semantics of the regular expressions that are supported by PCRE 50are described in detail below. There is a quick-reference syntax summary in the 51<a href="pcresyntax.html"><b>pcresyntax</b></a> 52page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE 53also supports some alternative regular expression syntax (which does not 54conflict with the Perl syntax) in order to provide some compatibility with 55regular expressions in Python, .NET, and Oniguruma. 56</P> 57<P> 58Perl's regular expressions are described in its own documentation, and 59regular expressions in general are covered in a number of books, some of which 60have copious examples. Jeffrey Friedl's "Mastering Regular Expressions", 61published by O'Reilly, covers regular expressions in great detail. This 62description of PCRE's regular expressions is intended as reference material. 63</P> 64<P> 65This document discusses the patterns that are supported by PCRE when one its 66main matching functions, <b>pcre_exec()</b> (8-bit) or <b>pcre[16|32]_exec()</b> 67(16- or 32-bit), is used. PCRE also has alternative matching functions, 68<b>pcre_dfa_exec()</b> and <b>pcre[16|32_dfa_exec()</b>, which match using a 69different algorithm that is not Perl-compatible. Some of the features discussed 70below are not available when DFA matching is used. The advantages and 71disadvantages of the alternative functions, and how they differ from the normal 72functions, are discussed in the 73<a href="pcrematching.html"><b>pcrematching</b></a> 74page. 75</P> 76<br><a name="SEC2" href="#TOC1">SPECIAL START-OF-PATTERN ITEMS</a><br> 77<P> 78A number of options that can be passed to <b>pcre_compile()</b> can also be set 79by special items at the start of a pattern. These are not Perl-compatible, but 80are provided to make these options accessible to pattern writers who are not 81able to change the program that processes the pattern. Any number of these 82items may appear, but they must all be together right at the start of the 83pattern string, and the letters must be in upper case. 84</P> 85<br><b> 86UTF support 87</b><br> 88<P> 89The original operation of PCRE was on strings of one-byte characters. However, 90there is now also support for UTF-8 strings in the original library, an 91extra library that supports 16-bit and UTF-16 character strings, and a 92third library that supports 32-bit and UTF-32 character strings. To use these 93features, PCRE must be built to include appropriate support. When using UTF 94strings you must either call the compiling function with the PCRE_UTF8, 95PCRE_UTF16, or PCRE_UTF32 option, or the pattern must start with one of 96these special sequences: 97<pre> 98 (*UTF8) 99 (*UTF16) 100 (*UTF32) 101 (*UTF) 102</pre> 103(*UTF) is a generic sequence that can be used with any of the libraries. 104Starting a pattern with such a sequence is equivalent to setting the relevant 105option. How setting a UTF mode affects pattern matching is mentioned in several 106places below. There is also a summary of features in the 107<a href="pcreunicode.html"><b>pcreunicode</b></a> 108page. 109</P> 110<P> 111Some applications that allow their users to supply patterns may wish to 112restrict them to non-UTF data for security reasons. If the PCRE_NEVER_UTF 113option is set at compile time, (*UTF) etc. are not allowed, and their 114appearance causes an error. 115</P> 116<br><b> 117Unicode property support 118</b><br> 119<P> 120Another special sequence that may appear at the start of a pattern is (*UCP). 121This has the same effect as setting the PCRE_UCP option: it causes sequences 122such as \d and \w to use Unicode properties to determine character types, 123instead of recognizing only characters with codes less than 128 via a lookup 124table. 125</P> 126<br><b> 127Disabling auto-possessification 128</b><br> 129<P> 130If a pattern starts with (*NO_AUTO_POSSESS), it has the same effect as setting 131the PCRE_NO_AUTO_POSSESS option at compile time. This stops PCRE from making 132quantifiers possessive when what follows cannot match the repeated item. For 133example, by default a+b is treated as a++b. For more details, see the 134<a href="pcreapi.html"><b>pcreapi</b></a> 135documentation. 136</P> 137<br><b> 138Disabling start-up optimizations 139</b><br> 140<P> 141If a pattern starts with (*NO_START_OPT), it has the same effect as setting the 142PCRE_NO_START_OPTIMIZE option either at compile or matching time. This disables 143several optimizations for quickly reaching "no match" results. For more 144details, see the 145<a href="pcreapi.html"><b>pcreapi</b></a> 146documentation. 147<a name="newlines"></a></P> 148<br><b> 149Newline conventions 150</b><br> 151<P> 152PCRE supports five different conventions for indicating line breaks in 153strings: a single CR (carriage return) character, a single LF (linefeed) 154character, the two-character sequence CRLF, any of the three preceding, or any 155Unicode newline sequence. The 156<a href="pcreapi.html"><b>pcreapi</b></a> 157page has 158<a href="pcreapi.html#newlines">further discussion</a> 159about newlines, and shows how to set the newline convention in the 160<i>options</i> arguments for the compiling and matching functions. 161</P> 162<P> 163It is also possible to specify a newline convention by starting a pattern 164string with one of the following five sequences: 165<pre> 166 (*CR) carriage return 167 (*LF) linefeed 168 (*CRLF) carriage return, followed by linefeed 169 (*ANYCRLF) any of the three above 170 (*ANY) all Unicode newline sequences 171</pre> 172These override the default and the options given to the compiling function. For 173example, on a Unix system where LF is the default newline sequence, the pattern 174<pre> 175 (*CR)a.b 176</pre> 177changes the convention to CR. That pattern matches "a\nb" because LF is no 178longer a newline. If more than one of these settings is present, the last one 179is used. 180</P> 181<P> 182The newline convention affects where the circumflex and dollar assertions are 183true. It also affects the interpretation of the dot metacharacter when 184PCRE_DOTALL is not set, and the behaviour of \N. However, it does not affect 185what the \R escape sequence matches. By default, this is any Unicode newline 186sequence, for Perl compatibility. However, this can be changed; see the 187description of \R in the section entitled 188<a href="#newlineseq">"Newline sequences"</a> 189below. A change of \R setting can be combined with a change of newline 190convention. 191</P> 192<br><b> 193Setting match and recursion limits 194</b><br> 195<P> 196The caller of <b>pcre_exec()</b> can set a limit on the number of times the 197internal <b>match()</b> function is called and on the maximum depth of 198recursive calls. These facilities are provided to catch runaway matches that 199are provoked by patterns with huge matching trees (a typical example is a 200pattern with nested unlimited repeats) and to avoid running out of system stack 201by too much recursion. When one of these limits is reached, <b>pcre_exec()</b> 202gives an error return. The limits can also be set by items at the start of the 203pattern of the form 204<pre> 205 (*LIMIT_MATCH=d) 206 (*LIMIT_RECURSION=d) 207</pre> 208where d is any number of decimal digits. However, the value of the setting must 209be less than the value set (or defaulted) by the caller of <b>pcre_exec()</b> 210for it to have any effect. In other words, the pattern writer can lower the 211limits set by the programmer, but not raise them. If there is more than one 212setting of one of these limits, the lower value is used. 213</P> 214<br><a name="SEC3" href="#TOC1">EBCDIC CHARACTER CODES</a><br> 215<P> 216PCRE can be compiled to run in an environment that uses EBCDIC as its character 217code rather than ASCII or Unicode (typically a mainframe system). In the 218sections below, character code values are ASCII or Unicode; in an EBCDIC 219environment these characters may have different code values, and there are no 220code points greater than 255. 221</P> 222<br><a name="SEC4" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br> 223<P> 224A regular expression is a pattern that is matched against a subject string from 225left to right. Most characters stand for themselves in a pattern, and match the 226corresponding characters in the subject. As a trivial example, the pattern 227<pre> 228 The quick brown fox 229</pre> 230matches a portion of a subject string that is identical to itself. When 231caseless matching is specified (the PCRE_CASELESS option), letters are matched 232independently of case. In a UTF mode, PCRE always understands the concept of 233case for characters whose values are less than 128, so caseless matching is 234always possible. For characters with higher values, the concept of case is 235supported if PCRE is compiled with Unicode property support, but not otherwise. 236If you want to use caseless matching for characters 128 and above, you must 237ensure that PCRE is compiled with Unicode property support as well as with 238UTF support. 239</P> 240<P> 241The power of regular expressions comes from the ability to include alternatives 242and repetitions in the pattern. These are encoded in the pattern by the use of 243<i>metacharacters</i>, which do not stand for themselves but instead are 244interpreted in some special way. 245</P> 246<P> 247There are two different sets of metacharacters: those that are recognized 248anywhere in the pattern except within square brackets, and those that are 249recognized within square brackets. Outside square brackets, the metacharacters 250are as follows: 251<pre> 252 \ general escape character with several uses 253 ^ assert start of string (or line, in multiline mode) 254 $ assert end of string (or line, in multiline mode) 255 . match any character except newline (by default) 256 [ start character class definition 257 | start of alternative branch 258 ( start subpattern 259 ) end subpattern 260 ? extends the meaning of ( 261 also 0 or 1 quantifier 262 also quantifier minimizer 263 * 0 or more quantifier 264 + 1 or more quantifier 265 also "possessive quantifier" 266 { start min/max quantifier 267</pre> 268Part of a pattern that is in square brackets is called a "character class". In 269a character class the only metacharacters are: 270<pre> 271 \ general escape character 272 ^ negate the class, but only if the first character 273 - indicates character range 274 [ POSIX character class (only if followed by POSIX syntax) 275 ] terminates the character class 276</pre> 277The following sections describe the use of each of the metacharacters. 278</P> 279<br><a name="SEC5" href="#TOC1">BACKSLASH</a><br> 280<P> 281The backslash character has several uses. Firstly, if it is followed by a 282character that is not a number or a letter, it takes away any special meaning 283that character may have. This use of backslash as an escape character applies 284both inside and outside character classes. 285</P> 286<P> 287For example, if you want to match a * character, you write \* in the pattern. 288This escaping action applies whether or not the following character would 289otherwise be interpreted as a metacharacter, so it is always safe to precede a 290non-alphanumeric with backslash to specify that it stands for itself. In 291particular, if you want to match a backslash, you write \\. 292</P> 293<P> 294In a UTF mode, only ASCII numbers and letters have any special meaning after a 295backslash. All other characters (in particular, those whose codepoints are 296greater than 127) are treated as literals. 297</P> 298<P> 299If a pattern is compiled with the PCRE_EXTENDED option, most white space in the 300pattern (other than in a character class), and characters between a # outside a 301character class and the next newline, inclusive, are ignored. An escaping 302backslash can be used to include a white space or # character as part of the 303pattern. 304</P> 305<P> 306If you want to remove the special meaning from a sequence of characters, you 307can do so by putting them between \Q and \E. This is different from Perl in 308that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in 309Perl, $ and @ cause variable interpolation. Note the following examples: 310<pre> 311 Pattern PCRE matches Perl matches 312 313 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz 314 \Qabc\$xyz\E abc\$xyz abc\$xyz 315 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz 316</pre> 317The \Q...\E sequence is recognized both inside and outside character classes. 318An isolated \E that is not preceded by \Q is ignored. If \Q is not followed 319by \E later in the pattern, the literal interpretation continues to the end of 320the pattern (that is, \E is assumed at the end). If the isolated \Q is inside 321a character class, this causes an error, because the character class is not 322terminated. 323<a name="digitsafterbackslash"></a></P> 324<br><b> 325Non-printing characters 326</b><br> 327<P> 328A second use of backslash provides a way of encoding non-printing characters 329in patterns in a visible manner. There is no restriction on the appearance of 330non-printing characters, apart from the binary zero that terminates a pattern, 331but when a pattern is being prepared by text editing, it is often easier to use 332one of the following escape sequences than the binary character it represents. 333In an ASCII or Unicode environment, these escapes are as follows: 334<pre> 335 \a alarm, that is, the BEL character (hex 07) 336 \cx "control-x", where x is any ASCII character 337 \e escape (hex 1B) 338 \f form feed (hex 0C) 339 \n linefeed (hex 0A) 340 \r carriage return (hex 0D) 341 \t tab (hex 09) 342 \0dd character with octal code 0dd 343 \ddd character with octal code ddd, or back reference 344 \o{ddd..} character with octal code ddd.. 345 \xhh character with hex code hh 346 \x{hhh..} character with hex code hhh.. (non-JavaScript mode) 347 \uhhhh character with hex code hhhh (JavaScript mode only) 348</pre> 349The precise effect of \cx on ASCII characters is as follows: if x is a lower 350case letter, it is converted to upper case. Then bit 6 of the character (hex 35140) is inverted. Thus \cA to \cZ become hex 01 to hex 1A (A is 41, Z is 5A), 352but \c{ becomes hex 3B ({ is 7B), and \c; becomes hex 7B (; is 3B). If the 353data item (byte or 16-bit value) following \c has a value greater than 127, a 354compile-time error occurs. This locks out non-ASCII characters in all modes. 355</P> 356<P> 357When PCRE is compiled in EBCDIC mode, \a, \e, \f, \n, \r, and \t 358generate the appropriate EBCDIC code values. The \c escape is processed 359as specified for Perl in the <b>perlebcdic</b> document. The only characters 360that are allowed after \c are A-Z, a-z, or one of @, [, \, ], ^, _, or ?. Any 361other character provokes a compile-time error. The sequence \@ encodes 362character code 0; the letters (in either case) encode characters 1-26 (hex 01 363to hex 1A); [, \, ], ^, and _ encode characters 27-31 (hex 1B to hex 1F), and 364\? becomes either 255 (hex FF) or 95 (hex 5F). 365</P> 366<P> 367Thus, apart from \?, these escapes generate the same character code values as 368they do in an ASCII environment, though the meanings of the values mostly 369differ. For example, \G always generates code value 7, which is BEL in ASCII 370but DEL in EBCDIC. 371</P> 372<P> 373The sequence \? generates DEL (127, hex 7F) in an ASCII environment, but 374because 127 is not a control character in EBCDIC, Perl makes it generate the 375APC character. Unfortunately, there are several variants of EBCDIC. In most of 376them the APC character has the value 255 (hex FF), but in the one Perl calls 377POSIX-BC its value is 95 (hex 5F). If certain other characters have POSIX-BC 378values, PCRE makes \? generate 95; otherwise it generates 255. 379</P> 380<P> 381After \0 up to two further octal digits are read. If there are fewer than two 382digits, just those that are present are used. Thus the sequence \0\x\015 383specifies two binary zeros followed by a CR character (code value 13). Make 384sure you supply two digits after the initial zero if the pattern character that 385follows is itself an octal digit. 386</P> 387<P> 388The escape \o must be followed by a sequence of octal digits, enclosed in 389braces. An error occurs if this is not the case. This escape is a recent 390addition to Perl; it provides way of specifying character code points as octal 391numbers greater than 0777, and it also allows octal numbers and back references 392to be unambiguously specified. 393</P> 394<P> 395For greater clarity and unambiguity, it is best to avoid following \ by a 396digit greater than zero. Instead, use \o{} or \x{} to specify character 397numbers, and \g{} to specify back references. The following paragraphs 398describe the old, ambiguous syntax. 399</P> 400<P> 401The handling of a backslash followed by a digit other than 0 is complicated, 402and Perl has changed in recent releases, causing PCRE also to change. Outside a 403character class, PCRE reads the digit and any following digits as a decimal 404number. If the number is less than 8, or if there have been at least that many 405previous capturing left parentheses in the expression, the entire sequence is 406taken as a <i>back reference</i>. A description of how this works is given 407<a href="#backreferences">later,</a> 408following the discussion of 409<a href="#subpattern">parenthesized subpatterns.</a> 410</P> 411<P> 412Inside a character class, or if the decimal number following \ is greater than 4137 and there have not been that many capturing subpatterns, PCRE handles \8 and 414\9 as the literal characters "8" and "9", and otherwise re-reads up to three 415octal digits following the backslash, using them to generate a data character. 416Any subsequent digits stand for themselves. For example: 417<pre> 418 \040 is another way of writing an ASCII space 419 \40 is the same, provided there are fewer than 40 previous capturing subpatterns 420 \7 is always a back reference 421 \11 might be a back reference, or another way of writing a tab 422 \011 is always a tab 423 \0113 is a tab followed by the character "3" 424 \113 might be a back reference, otherwise the character with octal code 113 425 \377 might be a back reference, otherwise the value 255 (decimal) 426 \81 is either a back reference, or the two characters "8" and "1" 427</pre> 428Note that octal values of 100 or greater that are specified using this syntax 429must not be introduced by a leading zero, because no more than three octal 430digits are ever read. 431</P> 432<P> 433By default, after \x that is not followed by {, from zero to two hexadecimal 434digits are read (letters can be in upper or lower case). Any number of 435hexadecimal digits may appear between \x{ and }. If a character other than 436a hexadecimal digit appears between \x{ and }, or if there is no terminating 437}, an error occurs. 438</P> 439<P> 440If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \x is 441as just described only when it is followed by two hexadecimal digits. 442Otherwise, it matches a literal "x" character. In JavaScript mode, support for 443code points greater than 256 is provided by \u, which must be followed by 444four hexadecimal digits; otherwise it matches a literal "u" character. 445</P> 446<P> 447Characters whose value is less than 256 can be defined by either of the two 448syntaxes for \x (or by \u in JavaScript mode). There is no difference in the 449way they are handled. For example, \xdc is exactly the same as \x{dc} (or 450\u00dc in JavaScript mode). 451</P> 452<br><b> 453Constraints on character values 454</b><br> 455<P> 456Characters that are specified using octal or hexadecimal numbers are 457limited to certain values, as follows: 458<pre> 459 8-bit non-UTF mode less than 0x100 460 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint 461 16-bit non-UTF mode less than 0x10000 462 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint 463 32-bit non-UTF mode less than 0x100000000 464 32-bit UTF-32 mode less than 0x10ffff and a valid codepoint 465</pre> 466Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called 467"surrogate" codepoints), and 0xffef. 468</P> 469<br><b> 470Escape sequences in character classes 471</b><br> 472<P> 473All the sequences that define a single character value can be used both inside 474and outside character classes. In addition, inside a character class, \b is 475interpreted as the backspace character (hex 08). 476</P> 477<P> 478\N is not allowed in a character class. \B, \R, and \X are not special 479inside a character class. Like other unrecognized escape sequences, they are 480treated as the literal characters "B", "R", and "X" by default, but cause an 481error if the PCRE_EXTRA option is set. Outside a character class, these 482sequences have different meanings. 483</P> 484<br><b> 485Unsupported escape sequences 486</b><br> 487<P> 488In Perl, the sequences \l, \L, \u, and \U are recognized by its string 489handler and used to modify the case of following characters. By default, PCRE 490does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT 491option is set, \U matches a "U" character, and \u can be used to define a 492character by code point, as described in the previous section. 493</P> 494<br><b> 495Absolute and relative back references 496</b><br> 497<P> 498The sequence \g followed by an unsigned or a negative number, optionally 499enclosed in braces, is an absolute or relative back reference. A named back 500reference can be coded as \g{name}. Back references are discussed 501<a href="#backreferences">later,</a> 502following the discussion of 503<a href="#subpattern">parenthesized subpatterns.</a> 504</P> 505<br><b> 506Absolute and relative subroutine calls 507</b><br> 508<P> 509For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or 510a number enclosed either in angle brackets or single quotes, is an alternative 511syntax for referencing a subpattern as a "subroutine". Details are discussed 512<a href="#onigurumasubroutines">later.</a> 513Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are <i>not</i> 514synonymous. The former is a back reference; the latter is a 515<a href="#subpatternsassubroutines">subroutine</a> 516call. 517<a name="genericchartypes"></a></P> 518<br><b> 519Generic character types 520</b><br> 521<P> 522Another use of backslash is for specifying generic character types: 523<pre> 524 \d any decimal digit 525 \D any character that is not a decimal digit 526 \h any horizontal white space character 527 \H any character that is not a horizontal white space character 528 \s any white space character 529 \S any character that is not a white space character 530 \v any vertical white space character 531 \V any character that is not a vertical white space character 532 \w any "word" character 533 \W any "non-word" character 534</pre> 535There is also the single sequence \N, which matches a non-newline character. 536This is the same as 537<a href="#fullstopdot">the "." metacharacter</a> 538when PCRE_DOTALL is not set. Perl also uses \N to match characters by name; 539PCRE does not support this. 540</P> 541<P> 542Each pair of lower and upper case escape sequences partitions the complete set 543of characters into two disjoint sets. Any given character matches one, and only 544one, of each pair. The sequences can appear both inside and outside character 545classes. They each match one character of the appropriate type. If the current 546matching point is at the end of the subject string, all of them fail, because 547there is no character to match. 548</P> 549<P> 550For compatibility with Perl, \s did not used to match the VT character (code 55111), which made it different from the the POSIX "space" class. However, Perl 552added VT at release 5.18, and PCRE followed suit at release 8.34. The default 553\s characters are now HT (9), LF (10), VT (11), FF (12), CR (13), and space 554(32), which are defined as white space in the "C" locale. This list may vary if 555locale-specific matching is taking place. For example, in some locales the 556"non-breaking space" character (\xA0) is recognized as white space, and in 557others the VT character is not. 558</P> 559<P> 560A "word" character is an underscore or any character that is a letter or digit. 561By default, the definition of letters and digits is controlled by PCRE's 562low-valued character tables, and may vary if locale-specific matching is taking 563place (see 564<a href="pcreapi.html#localesupport">"Locale support"</a> 565in the 566<a href="pcreapi.html"><b>pcreapi</b></a> 567page). For example, in a French locale such as "fr_FR" in Unix-like systems, 568or "french" in Windows, some character codes greater than 127 are used for 569accented letters, and these are then matched by \w. The use of locales with 570Unicode is discouraged. 571</P> 572<P> 573By default, characters whose code points are greater than 127 never match \d, 574\s, or \w, and always match \D, \S, and \W, although this may vary for 575characters in the range 128-255 when locale-specific matching is happening. 576These escape sequences retain their original meanings from before Unicode 577support was available, mainly for efficiency reasons. If PCRE is compiled with 578Unicode property support, and the PCRE_UCP option is set, the behaviour is 579changed so that Unicode properties are used to determine character types, as 580follows: 581<pre> 582 \d any character that matches \p{Nd} (decimal digit) 583 \s any character that matches \p{Z} or \h or \v 584 \w any character that matches \p{L} or \p{N}, plus underscore 585</pre> 586The upper case escapes match the inverse sets of characters. Note that \d 587matches only decimal digits, whereas \w matches any Unicode digit, as well as 588any Unicode letter, and underscore. Note also that PCRE_UCP affects \b, and 589\B because they are defined in terms of \w and \W. Matching these sequences 590is noticeably slower when PCRE_UCP is set. 591</P> 592<P> 593The sequences \h, \H, \v, and \V are features that were added to Perl at 594release 5.10. In contrast to the other sequences, which match only ASCII 595characters by default, these always match certain high-valued code points, 596whether or not PCRE_UCP is set. The horizontal space characters are: 597<pre> 598 U+0009 Horizontal tab (HT) 599 U+0020 Space 600 U+00A0 Non-break space 601 U+1680 Ogham space mark 602 U+180E Mongolian vowel separator 603 U+2000 En quad 604 U+2001 Em quad 605 U+2002 En space 606 U+2003 Em space 607 U+2004 Three-per-em space 608 U+2005 Four-per-em space 609 U+2006 Six-per-em space 610 U+2007 Figure space 611 U+2008 Punctuation space 612 U+2009 Thin space 613 U+200A Hair space 614 U+202F Narrow no-break space 615 U+205F Medium mathematical space 616 U+3000 Ideographic space 617</pre> 618The vertical space characters are: 619<pre> 620 U+000A Linefeed (LF) 621 U+000B Vertical tab (VT) 622 U+000C Form feed (FF) 623 U+000D Carriage return (CR) 624 U+0085 Next line (NEL) 625 U+2028 Line separator 626 U+2029 Paragraph separator 627</pre> 628In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are 629relevant. 630<a name="newlineseq"></a></P> 631<br><b> 632Newline sequences 633</b><br> 634<P> 635Outside a character class, by default, the escape sequence \R matches any 636Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent to the 637following: 638<pre> 639 (?>\r\n|\n|\x0b|\f|\r|\x85) 640</pre> 641This is an example of an "atomic group", details of which are given 642<a href="#atomicgroup">below.</a> 643This particular group matches either the two-character sequence CR followed by 644LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab, 645U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next 646line, U+0085). The two-character sequence is treated as a single unit that 647cannot be split. 648</P> 649<P> 650In other modes, two additional characters whose codepoints are greater than 255 651are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029). 652Unicode character property support is not needed for these characters to be 653recognized. 654</P> 655<P> 656It is possible to restrict \R to match only CR, LF, or CRLF (instead of the 657complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF 658either at compile time or when the pattern is matched. (BSR is an abbrevation 659for "backslash R".) This can be made the default when PCRE is built; if this is 660the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option. 661It is also possible to specify these settings by starting a pattern string with 662one of the following sequences: 663<pre> 664 (*BSR_ANYCRLF) CR, LF, or CRLF only 665 (*BSR_UNICODE) any Unicode newline sequence 666</pre> 667These override the default and the options given to the compiling function, but 668they can themselves be overridden by options given to a matching function. Note 669that these special settings, which are not Perl-compatible, are recognized only 670at the very start of a pattern, and that they must be in upper case. If more 671than one of them is present, the last one is used. They can be combined with a 672change of newline convention; for example, a pattern can start with: 673<pre> 674 (*ANY)(*BSR_ANYCRLF) 675</pre> 676They can also be combined with the (*UTF8), (*UTF16), (*UTF32), (*UTF) or 677(*UCP) special sequences. Inside a character class, \R is treated as an 678unrecognized escape sequence, and so matches the letter "R" by default, but 679causes an error if PCRE_EXTRA is set. 680<a name="uniextseq"></a></P> 681<br><b> 682Unicode character properties 683</b><br> 684<P> 685When PCRE is built with Unicode character property support, three additional 686escape sequences that match characters with specific properties are available. 687When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing 688characters whose codepoints are less than 256, but they do work in this mode. 689The extra escape sequences are: 690<pre> 691 \p{<i>xx</i>} a character with the <i>xx</i> property 692 \P{<i>xx</i>} a character without the <i>xx</i> property 693 \X a Unicode extended grapheme cluster 694</pre> 695The property names represented by <i>xx</i> above are limited to the Unicode 696script names, the general category properties, "Any", which matches any 697character (including newline), and some special PCRE properties (described 698in the 699<a href="#extraprops">next section).</a> 700Other Perl properties such as "InMusicalSymbols" are not currently supported by 701PCRE. Note that \P{Any} does not match any characters, so always causes a 702match failure. 703</P> 704<P> 705Sets of Unicode characters are defined as belonging to certain scripts. A 706character from one of these sets can be matched using a script name. For 707example: 708<pre> 709 \p{Greek} 710 \P{Han} 711</pre> 712Those that are not part of an identified script are lumped together as 713"Common". The current list of scripts is: 714</P> 715<P> 716Arabic, 717Armenian, 718Avestan, 719Balinese, 720Bamum, 721Bassa_Vah, 722Batak, 723Bengali, 724Bopomofo, 725Brahmi, 726Braille, 727Buginese, 728Buhid, 729Canadian_Aboriginal, 730Carian, 731Caucasian_Albanian, 732Chakma, 733Cham, 734Cherokee, 735Common, 736Coptic, 737Cuneiform, 738Cypriot, 739Cyrillic, 740Deseret, 741Devanagari, 742Duployan, 743Egyptian_Hieroglyphs, 744Elbasan, 745Ethiopic, 746Georgian, 747Glagolitic, 748Gothic, 749Grantha, 750Greek, 751Gujarati, 752Gurmukhi, 753Han, 754Hangul, 755Hanunoo, 756Hebrew, 757Hiragana, 758Imperial_Aramaic, 759Inherited, 760Inscriptional_Pahlavi, 761Inscriptional_Parthian, 762Javanese, 763Kaithi, 764Kannada, 765Katakana, 766Kayah_Li, 767Kharoshthi, 768Khmer, 769Khojki, 770Khudawadi, 771Lao, 772Latin, 773Lepcha, 774Limbu, 775Linear_A, 776Linear_B, 777Lisu, 778Lycian, 779Lydian, 780Mahajani, 781Malayalam, 782Mandaic, 783Manichaean, 784Meetei_Mayek, 785Mende_Kikakui, 786Meroitic_Cursive, 787Meroitic_Hieroglyphs, 788Miao, 789Modi, 790Mongolian, 791Mro, 792Myanmar, 793Nabataean, 794New_Tai_Lue, 795Nko, 796Ogham, 797Ol_Chiki, 798Old_Italic, 799Old_North_Arabian, 800Old_Permic, 801Old_Persian, 802Old_South_Arabian, 803Old_Turkic, 804Oriya, 805Osmanya, 806Pahawh_Hmong, 807Palmyrene, 808Pau_Cin_Hau, 809Phags_Pa, 810Phoenician, 811Psalter_Pahlavi, 812Rejang, 813Runic, 814Samaritan, 815Saurashtra, 816Sharada, 817Shavian, 818Siddham, 819Sinhala, 820Sora_Sompeng, 821Sundanese, 822Syloti_Nagri, 823Syriac, 824Tagalog, 825Tagbanwa, 826Tai_Le, 827Tai_Tham, 828Tai_Viet, 829Takri, 830Tamil, 831Telugu, 832Thaana, 833Thai, 834Tibetan, 835Tifinagh, 836Tirhuta, 837Ugaritic, 838Vai, 839Warang_Citi, 840Yi. 841</P> 842<P> 843Each character has exactly one Unicode general category property, specified by 844a two-letter abbreviation. For compatibility with Perl, negation can be 845specified by including a circumflex between the opening brace and the property 846name. For example, \p{^Lu} is the same as \P{Lu}. 847</P> 848<P> 849If only one letter is specified with \p or \P, it includes all the general 850category properties that start with that letter. In this case, in the absence 851of negation, the curly brackets in the escape sequence are optional; these two 852examples have the same effect: 853<pre> 854 \p{L} 855 \pL 856</pre> 857The following general category property codes are supported: 858<pre> 859 C Other 860 Cc Control 861 Cf Format 862 Cn Unassigned 863 Co Private use 864 Cs Surrogate 865 866 L Letter 867 Ll Lower case letter 868 Lm Modifier letter 869 Lo Other letter 870 Lt Title case letter 871 Lu Upper case letter 872 873 M Mark 874 Mc Spacing mark 875 Me Enclosing mark 876 Mn Non-spacing mark 877 878 N Number 879 Nd Decimal number 880 Nl Letter number 881 No Other number 882 883 P Punctuation 884 Pc Connector punctuation 885 Pd Dash punctuation 886 Pe Close punctuation 887 Pf Final punctuation 888 Pi Initial punctuation 889 Po Other punctuation 890 Ps Open punctuation 891 892 S Symbol 893 Sc Currency symbol 894 Sk Modifier symbol 895 Sm Mathematical symbol 896 So Other symbol 897 898 Z Separator 899 Zl Line separator 900 Zp Paragraph separator 901 Zs Space separator 902</pre> 903The special property L& is also supported: it matches a character that has 904the Lu, Ll, or Lt property, in other words, a letter that is not classified as 905a modifier or "other". 906</P> 907<P> 908The Cs (Surrogate) property applies only to characters in the range U+D800 to 909U+DFFF. Such characters are not valid in Unicode strings and so 910cannot be tested by PCRE, unless UTF validity checking has been turned off 911(see the discussion of PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK and 912PCRE_NO_UTF32_CHECK in the 913<a href="pcreapi.html"><b>pcreapi</b></a> 914page). Perl does not support the Cs property. 915</P> 916<P> 917The long synonyms for property names that Perl supports (such as \p{Letter}) 918are not supported by PCRE, nor is it permitted to prefix any of these 919properties with "Is". 920</P> 921<P> 922No character that is in the Unicode table has the Cn (unassigned) property. 923Instead, this property is assumed for any code point that is not in the 924Unicode table. 925</P> 926<P> 927Specifying caseless matching does not affect these escape sequences. For 928example, \p{Lu} always matches only upper case letters. This is different from 929the behaviour of current versions of Perl. 930</P> 931<P> 932Matching characters by Unicode property is not fast, because PCRE has to do a 933multistage table lookup in order to find a character's property. That is why 934the traditional escape sequences such as \d and \w do not use Unicode 935properties in PCRE by default, though you can make them do so by setting the 936PCRE_UCP option or by starting the pattern with (*UCP). 937</P> 938<br><b> 939Extended grapheme clusters 940</b><br> 941<P> 942The \X escape matches any number of Unicode characters that form an "extended 943grapheme cluster", and treats the sequence as an atomic group 944<a href="#atomicgroup">(see below).</a> 945Up to and including release 8.31, PCRE matched an earlier, simpler definition 946that was equivalent to 947<pre> 948 (?>\PM\pM*) 949</pre> 950That is, it matched a character without the "mark" property, followed by zero 951or more characters with the "mark" property. Characters with the "mark" 952property are typically non-spacing accents that affect the preceding character. 953</P> 954<P> 955This simple definition was extended in Unicode to include more complicated 956kinds of composite character by giving each character a grapheme breaking 957property, and creating rules that use these properties to define the boundaries 958of extended grapheme clusters. In releases of PCRE later than 8.31, \X matches 959one of these clusters. 960</P> 961<P> 962\X always matches at least one character. Then it decides whether to add 963additional characters according to the following rules for ending a cluster: 964</P> 965<P> 9661. End at the end of the subject string. 967</P> 968<P> 9692. Do not end between CR and LF; otherwise end after any control character. 970</P> 971<P> 9723. Do not break Hangul (a Korean script) syllable sequences. Hangul characters 973are of five types: L, V, T, LV, and LVT. An L character may be followed by an 974L, V, LV, or LVT character; an LV or V character may be followed by a V or T 975character; an LVT or T character may be follwed only by a T character. 976</P> 977<P> 9784. Do not end before extending characters or spacing marks. Characters with 979the "mark" property always have the "extend" grapheme breaking property. 980</P> 981<P> 9825. Do not end after prepend characters. 983</P> 984<P> 9856. Otherwise, end the cluster. 986<a name="extraprops"></a></P> 987<br><b> 988PCRE's additional properties 989</b><br> 990<P> 991As well as the standard Unicode properties described above, PCRE supports four 992more that make it possible to convert traditional escape sequences such as \w 993and \s to use Unicode properties. PCRE uses these non-standard, non-Perl 994properties internally when PCRE_UCP is set. However, they may also be used 995explicitly. These properties are: 996<pre> 997 Xan Any alphanumeric character 998 Xps Any POSIX space character 999 Xsp Any Perl space character 1000 Xwd Any Perl "word" character 1001</pre> 1002Xan matches characters that have either the L (letter) or the N (number) 1003property. Xps matches the characters tab, linefeed, vertical tab, form feed, or 1004carriage return, and any other character that has the Z (separator) property. 1005Xsp is the same as Xps; it used to exclude vertical tab, for Perl 1006compatibility, but Perl changed, and so PCRE followed at release 8.34. Xwd 1007matches the same characters as Xan, plus underscore. 1008</P> 1009<P> 1010There is another non-standard property, Xuc, which matches any character that 1011can be represented by a Universal Character Name in C++ and other programming 1012languages. These are the characters $, @, ` (grave accent), and all characters 1013with Unicode code points greater than or equal to U+00A0, except for the 1014surrogates U+D800 to U+DFFF. Note that most base (ASCII) characters are 1015excluded. (Universal Character Names are of the form \uHHHH or \UHHHHHHHH 1016where H is a hexadecimal digit. Note that the Xuc property does not match these 1017sequences but the characters that they represent.) 1018<a name="resetmatchstart"></a></P> 1019<br><b> 1020Resetting the match start 1021</b><br> 1022<P> 1023The escape sequence \K causes any previously matched characters not to be 1024included in the final matched sequence. For example, the pattern: 1025<pre> 1026 foo\Kbar 1027</pre> 1028matches "foobar", but reports that it has matched "bar". This feature is 1029similar to a lookbehind assertion 1030<a href="#lookbehind">(described below).</a> 1031However, in this case, the part of the subject before the real match does not 1032have to be of fixed length, as lookbehind assertions do. The use of \K does 1033not interfere with the setting of 1034<a href="#subpattern">captured substrings.</a> 1035For example, when the pattern 1036<pre> 1037 (foo)\Kbar 1038</pre> 1039matches "foobar", the first substring is still set to "foo". 1040</P> 1041<P> 1042Perl documents that the use of \K within assertions is "not well defined". In 1043PCRE, \K is acted upon when it occurs inside positive assertions, but is 1044ignored in negative assertions. Note that when a pattern such as (?=ab\K) 1045matches, the reported start of the match can be greater than the end of the 1046match. 1047<a name="smallassertions"></a></P> 1048<br><b> 1049Simple assertions 1050</b><br> 1051<P> 1052The final use of backslash is for certain simple assertions. An assertion 1053specifies a condition that has to be met at a particular point in a match, 1054without consuming any characters from the subject string. The use of 1055subpatterns for more complicated assertions is described 1056<a href="#bigassertions">below.</a> 1057The backslashed assertions are: 1058<pre> 1059 \b matches at a word boundary 1060 \B matches when not at a word boundary 1061 \A matches at the start of the subject 1062 \Z matches at the end of the subject 1063 also matches before a newline at the end of the subject 1064 \z matches only at the end of the subject 1065 \G matches at the first matching position in the subject 1066</pre> 1067Inside a character class, \b has a different meaning; it matches the backspace 1068character. If any other of these assertions appears in a character class, by 1069default it matches the corresponding literal character (for example, \B 1070matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid 1071escape sequence" error is generated instead. 1072</P> 1073<P> 1074A word boundary is a position in the subject string where the current character 1075and the previous character do not both match \w or \W (i.e. one matches 1076\w and the other matches \W), or the start or end of the string if the 1077first or last character matches \w, respectively. In a UTF mode, the meanings 1078of \w and \W can be changed by setting the PCRE_UCP option. When this is 1079done, it also affects \b and \B. Neither PCRE nor Perl has a separate "start 1080of word" or "end of word" metasequence. However, whatever follows \b normally 1081determines which it is. For example, the fragment \ba matches "a" at the start 1082of a word. 1083</P> 1084<P> 1085The \A, \Z, and \z assertions differ from the traditional circumflex and 1086dollar (described in the next section) in that they only ever match at the very 1087start and end of the subject string, whatever options are set. Thus, they are 1088independent of multiline mode. These three assertions are not affected by the 1089PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the 1090circumflex and dollar metacharacters. However, if the <i>startoffset</i> 1091argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start 1092at a point other than the beginning of the subject, \A can never match. The 1093difference between \Z and \z is that \Z matches before a newline at the end 1094of the string as well as at the very end, whereas \z matches only at the end. 1095</P> 1096<P> 1097The \G assertion is true only when the current matching position is at the 1098start point of the match, as specified by the <i>startoffset</i> argument of 1099<b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is 1100non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate 1101arguments, you can mimic Perl's /g option, and it is in this kind of 1102implementation where \G can be useful. 1103</P> 1104<P> 1105Note, however, that PCRE's interpretation of \G, as the start of the current 1106match, is subtly different from Perl's, which defines it as the end of the 1107previous match. In Perl, these can be different when the previously matched 1108string was empty. Because PCRE does just one match at a time, it cannot 1109reproduce this behaviour. 1110</P> 1111<P> 1112If all the alternatives of a pattern begin with \G, the expression is anchored 1113to the starting match position, and the "anchored" flag is set in the compiled 1114regular expression. 1115</P> 1116<br><a name="SEC6" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br> 1117<P> 1118The circumflex and dollar metacharacters are zero-width assertions. That is, 1119they test for a particular condition being true without consuming any 1120characters from the subject string. 1121</P> 1122<P> 1123Outside a character class, in the default matching mode, the circumflex 1124character is an assertion that is true only if the current matching point is at 1125the start of the subject string. If the <i>startoffset</i> argument of 1126<b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE 1127option is unset. Inside a character class, circumflex has an entirely different 1128meaning 1129<a href="#characterclass">(see below).</a> 1130</P> 1131<P> 1132Circumflex need not be the first character of the pattern if a number of 1133alternatives are involved, but it should be the first thing in each alternative 1134in which it appears if the pattern is ever to match that branch. If all 1135possible alternatives start with a circumflex, that is, if the pattern is 1136constrained to match only at the start of the subject, it is said to be an 1137"anchored" pattern. (There are also other constructs that can cause a pattern 1138to be anchored.) 1139</P> 1140<P> 1141The dollar character is an assertion that is true only if the current matching 1142point is at the end of the subject string, or immediately before a newline at 1143the end of the string (by default). Note, however, that it does not actually 1144match the newline. Dollar need not be the last character of the pattern if a 1145number of alternatives are involved, but it should be the last item in any 1146branch in which it appears. Dollar has no special meaning in a character class. 1147</P> 1148<P> 1149The meaning of dollar can be changed so that it matches only at the very end of 1150the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This 1151does not affect the \Z assertion. 1152</P> 1153<P> 1154The meanings of the circumflex and dollar characters are changed if the 1155PCRE_MULTILINE option is set. When this is the case, a circumflex matches 1156immediately after internal newlines as well as at the start of the subject 1157string. It does not match after a newline that ends the string. A dollar 1158matches before any newlines in the string, as well as at the very end, when 1159PCRE_MULTILINE is set. When newline is specified as the two-character 1160sequence CRLF, isolated CR and LF characters do not indicate newlines. 1161</P> 1162<P> 1163For example, the pattern /^abc$/ matches the subject string "def\nabc" (where 1164\n represents a newline) in multiline mode, but not otherwise. Consequently, 1165patterns that are anchored in single line mode because all branches start with 1166^ are not anchored in multiline mode, and a match for circumflex is possible 1167when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The 1168PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set. 1169</P> 1170<P> 1171Note that the sequences \A, \Z, and \z can be used to match the start and 1172end of the subject in both modes, and if all branches of a pattern start with 1173\A it is always anchored, whether or not PCRE_MULTILINE is set. 1174<a name="fullstopdot"></a></P> 1175<br><a name="SEC7" href="#TOC1">FULL STOP (PERIOD, DOT) AND \N</a><br> 1176<P> 1177Outside a character class, a dot in the pattern matches any one character in 1178the subject string except (by default) a character that signifies the end of a 1179line. 1180</P> 1181<P> 1182When a line ending is defined as a single character, dot never matches that 1183character; when the two-character sequence CRLF is used, dot does not match CR 1184if it is immediately followed by LF, but otherwise it matches all characters 1185(including isolated CRs and LFs). When any Unicode line endings are being 1186recognized, dot does not match CR or LF or any of the other line ending 1187characters. 1188</P> 1189<P> 1190The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL 1191option is set, a dot matches any one character, without exception. If the 1192two-character sequence CRLF is present in the subject string, it takes two dots 1193to match it. 1194</P> 1195<P> 1196The handling of dot is entirely independent of the handling of circumflex and 1197dollar, the only relationship being that they both involve newlines. Dot has no 1198special meaning in a character class. 1199</P> 1200<P> 1201The escape sequence \N behaves like a dot, except that it is not affected by 1202the PCRE_DOTALL option. In other words, it matches any character except one 1203that signifies the end of a line. Perl also uses \N to match characters by 1204name; PCRE does not support this. 1205</P> 1206<br><a name="SEC8" href="#TOC1">MATCHING A SINGLE DATA UNIT</a><br> 1207<P> 1208Outside a character class, the escape sequence \C matches any one data unit, 1209whether or not a UTF mode is set. In the 8-bit library, one data unit is one 1210byte; in the 16-bit library it is a 16-bit unit; in the 32-bit library it is 1211a 32-bit unit. Unlike a dot, \C always 1212matches line-ending characters. The feature is provided in Perl in order to 1213match individual bytes in UTF-8 mode, but it is unclear how it can usefully be 1214used. Because \C breaks up characters into individual data units, matching one 1215unit with \C in a UTF mode means that the rest of the string may start with a 1216malformed UTF character. This has undefined results, because PCRE assumes that 1217it is dealing with valid UTF strings (and by default it checks this at the 1218start of processing unless the PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK or 1219PCRE_NO_UTF32_CHECK option is used). 1220</P> 1221<P> 1222PCRE does not allow \C to appear in lookbehind assertions 1223<a href="#lookbehind">(described below)</a> 1224in a UTF mode, because this would make it impossible to calculate the length of 1225the lookbehind. 1226</P> 1227<P> 1228In general, the \C escape sequence is best avoided. However, one 1229way of using it that avoids the problem of malformed UTF characters is to use a 1230lookahead to check the length of the next character, as in this pattern, which 1231could be used with a UTF-8 string (ignore white space and line breaks): 1232<pre> 1233 (?| (?=[\x00-\x7f])(\C) | 1234 (?=[\x80-\x{7ff}])(\C)(\C) | 1235 (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) | 1236 (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C)) 1237</pre> 1238A group that starts with (?| resets the capturing parentheses numbers in each 1239alternative (see 1240<a href="#dupsubpatternnumber">"Duplicate Subpattern Numbers"</a> 1241below). The assertions at the start of each branch check the next UTF-8 1242character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The 1243character's individual bytes are then captured by the appropriate number of 1244groups. 1245<a name="characterclass"></a></P> 1246<br><a name="SEC9" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br> 1247<P> 1248An opening square bracket introduces a character class, terminated by a closing 1249square bracket. A closing square bracket on its own is not special by default. 1250However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square 1251bracket causes a compile-time error. If a closing square bracket is required as 1252a member of the class, it should be the first data character in the class 1253(after an initial circumflex, if present) or escaped with a backslash. 1254</P> 1255<P> 1256A character class matches a single character in the subject. In a UTF mode, the 1257character may be more than one data unit long. A matched character must be in 1258the set of characters defined by the class, unless the first character in the 1259class definition is a circumflex, in which case the subject character must not 1260be in the set defined by the class. If a circumflex is actually required as a 1261member of the class, ensure it is not the first character, or escape it with a 1262backslash. 1263</P> 1264<P> 1265For example, the character class [aeiou] matches any lower case vowel, while 1266[^aeiou] matches any character that is not a lower case vowel. Note that a 1267circumflex is just a convenient notation for specifying the characters that 1268are in the class by enumerating those that are not. A class that starts with a 1269circumflex is not an assertion; it still consumes a character from the subject 1270string, and therefore it fails if the current pointer is at the end of the 1271string. 1272</P> 1273<P> 1274In UTF-8 (UTF-16, UTF-32) mode, characters with values greater than 255 (0xffff) 1275can be included in a class as a literal string of data units, or by using the 1276\x{ escaping mechanism. 1277</P> 1278<P> 1279When caseless matching is set, any letters in a class represent both their 1280upper case and lower case versions, so for example, a caseless [aeiou] matches 1281"A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a 1282caseful version would. In a UTF mode, PCRE always understands the concept of 1283case for characters whose values are less than 128, so caseless matching is 1284always possible. For characters with higher values, the concept of case is 1285supported if PCRE is compiled with Unicode property support, but not otherwise. 1286If you want to use caseless matching in a UTF mode for characters 128 and 1287above, you must ensure that PCRE is compiled with Unicode property support as 1288well as with UTF support. 1289</P> 1290<P> 1291Characters that might indicate line breaks are never treated in any special way 1292when matching character classes, whatever line-ending sequence is in use, and 1293whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class 1294such as [^a] always matches one of these characters. 1295</P> 1296<P> 1297The minus (hyphen) character can be used to specify a range of characters in a 1298character class. For example, [d-m] matches any letter between d and m, 1299inclusive. If a minus character is required in a class, it must be escaped with 1300a backslash or appear in a position where it cannot be interpreted as 1301indicating a range, typically as the first or last character in the class, or 1302immediately after a range. For example, [b-d-z] matches letters in the range b 1303to d, a hyphen character, or z. 1304</P> 1305<P> 1306It is not possible to have the literal character "]" as the end character of a 1307range. A pattern such as [W-]46] is interpreted as a class of two characters 1308("W" and "-") followed by a literal string "46]", so it would match "W46]" or 1309"-46]". However, if the "]" is escaped with a backslash it is interpreted as 1310the end of range, so [W-\]46] is interpreted as a class containing a range 1311followed by two other characters. The octal or hexadecimal representation of 1312"]" can also be used to end a range. 1313</P> 1314<P> 1315An error is generated if a POSIX character class (see below) or an escape 1316sequence other than one that defines a single character appears at a point 1317where a range ending character is expected. For example, [z-\xff] is valid, 1318but [A-\d] and [A-[:digit:]] are not. 1319</P> 1320<P> 1321Ranges operate in the collating sequence of character values. They can also be 1322used for characters specified numerically, for example [\000-\037]. Ranges 1323can include any characters that are valid for the current mode. 1324</P> 1325<P> 1326If a range that includes letters is used when caseless matching is set, it 1327matches the letters in either case. For example, [W-c] is equivalent to 1328[][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character 1329tables for a French locale are in use, [\xc8-\xcb] matches accented E 1330characters in both cases. In UTF modes, PCRE supports the concept of case for 1331characters with values greater than 128 only when it is compiled with Unicode 1332property support. 1333</P> 1334<P> 1335The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v, 1336\V, \w, and \W may appear in a character class, and add the characters that 1337they match to the class. For example, [\dABCDEF] matches any hexadecimal 1338digit. In UTF modes, the PCRE_UCP option affects the meanings of \d, \s, \w 1339and their upper case partners, just as it does when they appear outside a 1340character class, as described in the section entitled 1341<a href="#genericchartypes">"Generic character types"</a> 1342above. The escape sequence \b has a different meaning inside a character 1343class; it matches the backspace character. The sequences \B, \N, \R, and \X 1344are not special inside a character class. Like any other unrecognized escape 1345sequences, they are treated as the literal characters "B", "N", "R", and "X" by 1346default, but cause an error if the PCRE_EXTRA option is set. 1347</P> 1348<P> 1349A circumflex can conveniently be used with the upper case character types to 1350specify a more restricted set of characters than the matching lower case type. 1351For example, the class [^\W_] matches any letter or digit, but not underscore, 1352whereas [\w] includes underscore. A positive character class should be read as 1353"something OR something OR ..." and a negative class as "NOT something AND NOT 1354something AND NOT ...". 1355</P> 1356<P> 1357The only metacharacters that are recognized in character classes are backslash, 1358hyphen (only where it can be interpreted as specifying a range), circumflex 1359(only at the start), opening square bracket (only when it can be interpreted as 1360introducing a POSIX class name, or for a special compatibility feature - see 1361the next two sections), and the terminating closing square bracket. However, 1362escaping other non-alphanumeric characters does no harm. 1363</P> 1364<br><a name="SEC10" href="#TOC1">POSIX CHARACTER CLASSES</a><br> 1365<P> 1366Perl supports the POSIX notation for character classes. This uses names 1367enclosed by [: and :] within the enclosing square brackets. PCRE also supports 1368this notation. For example, 1369<pre> 1370 [01[:alpha:]%] 1371</pre> 1372matches "0", "1", any alphabetic character, or "%". The supported class names 1373are: 1374<pre> 1375 alnum letters and digits 1376 alpha letters 1377 ascii character codes 0 - 127 1378 blank space or tab only 1379 cntrl control characters 1380 digit decimal digits (same as \d) 1381 graph printing characters, excluding space 1382 lower lower case letters 1383 print printing characters, including space 1384 punct printing characters, excluding letters and digits and space 1385 space white space (the same as \s from PCRE 8.34) 1386 upper upper case letters 1387 word "word" characters (same as \w) 1388 xdigit hexadecimal digits 1389</pre> 1390The default "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), 1391and space (32). If locale-specific matching is taking place, the list of space 1392characters may be different; there may be fewer or more of them. "Space" used 1393to be different to \s, which did not include VT, for Perl compatibility. 1394However, Perl changed at release 5.18, and PCRE followed at release 8.34. 1395"Space" and \s now match the same set of characters. 1396</P> 1397<P> 1398The name "word" is a Perl extension, and "blank" is a GNU extension from Perl 13995.8. Another Perl extension is negation, which is indicated by a ^ character 1400after the colon. For example, 1401<pre> 1402 [12[:^digit:]] 1403</pre> 1404matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX 1405syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not 1406supported, and an error is given if they are encountered. 1407</P> 1408<P> 1409By default, characters with values greater than 128 do not match any of the 1410POSIX character classes. However, if the PCRE_UCP option is passed to 1411<b>pcre_compile()</b>, some of the classes are changed so that Unicode character 1412properties are used. This is achieved by replacing certain POSIX classes by 1413other sequences, as follows: 1414<pre> 1415 [:alnum:] becomes \p{Xan} 1416 [:alpha:] becomes \p{L} 1417 [:blank:] becomes \h 1418 [:digit:] becomes \p{Nd} 1419 [:lower:] becomes \p{Ll} 1420 [:space:] becomes \p{Xps} 1421 [:upper:] becomes \p{Lu} 1422 [:word:] becomes \p{Xwd} 1423</pre> 1424Negated versions, such as [:^alpha:] use \P instead of \p. Three other POSIX 1425classes are handled specially in UCP mode: 1426</P> 1427<P> 1428[:graph:] 1429This matches characters that have glyphs that mark the page when printed. In 1430Unicode property terms, it matches all characters with the L, M, N, P, S, or Cf 1431properties, except for: 1432<pre> 1433 U+061C Arabic Letter Mark 1434 U+180E Mongolian Vowel Separator 1435 U+2066 - U+2069 Various "isolate"s 1436 1437</PRE> 1438</P> 1439<P> 1440[:print:] 1441This matches the same characters as [:graph:] plus space characters that are 1442not controls, that is, characters with the Zs property. 1443</P> 1444<P> 1445[:punct:] 1446This matches all characters that have the Unicode P (punctuation) property, 1447plus those characters whose code points are less than 128 that have the S 1448(Symbol) property. 1449</P> 1450<P> 1451The other POSIX classes are unchanged, and match only characters with code 1452points less than 128. 1453</P> 1454<br><a name="SEC11" href="#TOC1">COMPATIBILITY FEATURE FOR WORD BOUNDARIES</a><br> 1455<P> 1456In the POSIX.2 compliant library that was included in 4.4BSD Unix, the ugly 1457syntax [[:<:]] and [[:>:]] is used for matching "start of word" and "end of 1458word". PCRE treats these items as follows: 1459<pre> 1460 [[:<:]] is converted to \b(?=\w) 1461 [[:>:]] is converted to \b(?<=\w) 1462</pre> 1463Only these exact character sequences are recognized. A sequence such as 1464[a[:<:]b] provokes error for an unrecognized POSIX class name. This support is 1465not compatible with Perl. It is provided to help migrations from other 1466environments, and is best not used in any new patterns. Note that \b matches 1467at the start and the end of a word (see 1468<a href="#smallassertions">"Simple assertions"</a> 1469above), and in a Perl-style pattern the preceding or following character 1470normally shows which is wanted, without the need for the assertions that are 1471used above in order to give exactly the POSIX behaviour. 1472</P> 1473<br><a name="SEC12" href="#TOC1">VERTICAL BAR</a><br> 1474<P> 1475Vertical bar characters are used to separate alternative patterns. For example, 1476the pattern 1477<pre> 1478 gilbert|sullivan 1479</pre> 1480matches either "gilbert" or "sullivan". Any number of alternatives may appear, 1481and an empty alternative is permitted (matching the empty string). The matching 1482process tries each alternative in turn, from left to right, and the first one 1483that succeeds is used. If the alternatives are within a subpattern 1484<a href="#subpattern">(defined below),</a> 1485"succeeds" means matching the rest of the main pattern as well as the 1486alternative in the subpattern. 1487</P> 1488<br><a name="SEC13" href="#TOC1">INTERNAL OPTION SETTING</a><br> 1489<P> 1490The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and 1491PCRE_EXTENDED options (which are Perl-compatible) can be changed from within 1492the pattern by a sequence of Perl option letters enclosed between "(?" and ")". 1493The option letters are 1494<pre> 1495 i for PCRE_CASELESS 1496 m for PCRE_MULTILINE 1497 s for PCRE_DOTALL 1498 x for PCRE_EXTENDED 1499</pre> 1500For example, (?im) sets caseless, multiline matching. It is also possible to 1501unset these options by preceding the letter with a hyphen, and a combined 1502setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and 1503PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also 1504permitted. If a letter appears both before and after the hyphen, the option is 1505unset. 1506</P> 1507<P> 1508The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be 1509changed in the same way as the Perl-compatible options by using the characters 1510J, U and X respectively. 1511</P> 1512<P> 1513When one of these option changes occurs at top level (that is, not inside 1514subpattern parentheses), the change applies to the remainder of the pattern 1515that follows. If the change is placed right at the start of a pattern, PCRE 1516extracts it into the global options (and it will therefore show up in data 1517extracted by the <b>pcre_fullinfo()</b> function). 1518</P> 1519<P> 1520An option change within a subpattern (see below for a description of 1521subpatterns) affects only that part of the subpattern that follows it, so 1522<pre> 1523 (a(?i)b)c 1524</pre> 1525matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used). 1526By this means, options can be made to have different settings in different 1527parts of the pattern. Any changes made in one alternative do carry on 1528into subsequent branches within the same subpattern. For example, 1529<pre> 1530 (a(?i)b|c) 1531</pre> 1532matches "ab", "aB", "c", and "C", even though when matching "C" the first 1533branch is abandoned before the option setting. This is because the effects of 1534option settings happen at compile time. There would be some very weird 1535behaviour otherwise. 1536</P> 1537<P> 1538<b>Note:</b> There are other PCRE-specific options that can be set by the 1539application when the compiling or matching functions are called. In some cases 1540the pattern can contain special leading sequences such as (*CRLF) to override 1541what the application has set or what has been defaulted. Details are given in 1542the section entitled 1543<a href="#newlineseq">"Newline sequences"</a> 1544above. There are also the (*UTF8), (*UTF16),(*UTF32), and (*UCP) leading 1545sequences that can be used to set UTF and Unicode property modes; they are 1546equivalent to setting the PCRE_UTF8, PCRE_UTF16, PCRE_UTF32 and the PCRE_UCP 1547options, respectively. The (*UTF) sequence is a generic version that can be 1548used with any of the libraries. However, the application can set the 1549PCRE_NEVER_UTF option, which locks out the use of the (*UTF) sequences. 1550<a name="subpattern"></a></P> 1551<br><a name="SEC14" href="#TOC1">SUBPATTERNS</a><br> 1552<P> 1553Subpatterns are delimited by parentheses (round brackets), which can be nested. 1554Turning part of a pattern into a subpattern does two things: 1555<br> 1556<br> 15571. It localizes a set of alternatives. For example, the pattern 1558<pre> 1559 cat(aract|erpillar|) 1560</pre> 1561matches "cataract", "caterpillar", or "cat". Without the parentheses, it would 1562match "cataract", "erpillar" or an empty string. 1563<br> 1564<br> 15652. It sets up the subpattern as a capturing subpattern. This means that, when 1566the whole pattern matches, that portion of the subject string that matched the 1567subpattern is passed back to the caller via the <i>ovector</i> argument of the 1568matching function. (This applies only to the traditional matching functions; 1569the DFA matching functions do not support capturing.) 1570</P> 1571<P> 1572Opening parentheses are counted from left to right (starting from 1) to obtain 1573numbers for the capturing subpatterns. For example, if the string "the red 1574king" is matched against the pattern 1575<pre> 1576 the ((red|white) (king|queen)) 1577</pre> 1578the captured substrings are "red king", "red", and "king", and are numbered 1, 15792, and 3, respectively. 1580</P> 1581<P> 1582The fact that plain parentheses fulfil two functions is not always helpful. 1583There are often times when a grouping subpattern is required without a 1584capturing requirement. If an opening parenthesis is followed by a question mark 1585and a colon, the subpattern does not do any capturing, and is not counted when 1586computing the number of any subsequent capturing subpatterns. For example, if 1587the string "the white queen" is matched against the pattern 1588<pre> 1589 the ((?:red|white) (king|queen)) 1590</pre> 1591the captured substrings are "white queen" and "queen", and are numbered 1 and 15922. The maximum number of capturing subpatterns is 65535. 1593</P> 1594<P> 1595As a convenient shorthand, if any option settings are required at the start of 1596a non-capturing subpattern, the option letters may appear between the "?" and 1597the ":". Thus the two patterns 1598<pre> 1599 (?i:saturday|sunday) 1600 (?:(?i)saturday|sunday) 1601</pre> 1602match exactly the same set of strings. Because alternative branches are tried 1603from left to right, and options are not reset until the end of the subpattern 1604is reached, an option setting in one branch does affect subsequent branches, so 1605the above patterns match "SUNDAY" as well as "Saturday". 1606<a name="dupsubpatternnumber"></a></P> 1607<br><a name="SEC15" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br> 1608<P> 1609Perl 5.10 introduced a feature whereby each alternative in a subpattern uses 1610the same numbers for its capturing parentheses. Such a subpattern starts with 1611(?| and is itself a non-capturing subpattern. For example, consider this 1612pattern: 1613<pre> 1614 (?|(Sat)ur|(Sun))day 1615</pre> 1616Because the two alternatives are inside a (?| group, both sets of capturing 1617parentheses are numbered one. Thus, when the pattern matches, you can look 1618at captured substring number one, whichever alternative matched. This construct 1619is useful when you want to capture part, but not all, of one of a number of 1620alternatives. Inside a (?| group, parentheses are numbered as usual, but the 1621number is reset at the start of each branch. The numbers of any capturing 1622parentheses that follow the subpattern start after the highest number used in 1623any branch. The following example is taken from the Perl documentation. The 1624numbers underneath show in which buffer the captured content will be stored. 1625<pre> 1626 # before ---------------branch-reset----------- after 1627 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x 1628 # 1 2 2 3 2 3 4 1629</pre> 1630A back reference to a numbered subpattern uses the most recent value that is 1631set for that number by any subpattern. The following pattern matches "abcabc" 1632or "defdef": 1633<pre> 1634 /(?|(abc)|(def))\1/ 1635</pre> 1636In contrast, a subroutine call to a numbered subpattern always refers to the 1637first one in the pattern with the given number. The following pattern matches 1638"abcabc" or "defabc": 1639<pre> 1640 /(?|(abc)|(def))(?1)/ 1641</pre> 1642If a 1643<a href="#conditions">condition test</a> 1644for a subpattern's having matched refers to a non-unique number, the test is 1645true if any of the subpatterns of that number have matched. 1646</P> 1647<P> 1648An alternative approach to using this "branch reset" feature is to use 1649duplicate named subpatterns, as described in the next section. 1650</P> 1651<br><a name="SEC16" href="#TOC1">NAMED SUBPATTERNS</a><br> 1652<P> 1653Identifying capturing parentheses by number is simple, but it can be very hard 1654to keep track of the numbers in complicated regular expressions. Furthermore, 1655if an expression is modified, the numbers may change. To help with this 1656difficulty, PCRE supports the naming of subpatterns. This feature was not 1657added to Perl until release 5.10. Python had the feature earlier, and PCRE 1658introduced it at release 4.0, using the Python syntax. PCRE now supports both 1659the Perl and the Python syntax. Perl allows identically numbered subpatterns to 1660have different names, but PCRE does not. 1661</P> 1662<P> 1663In PCRE, a subpattern can be named in one of three ways: (?<name>...) or 1664(?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing 1665parentheses from other parts of the pattern, such as 1666<a href="#backreferences">back references,</a> 1667<a href="#recursion">recursion,</a> 1668and 1669<a href="#conditions">conditions,</a> 1670can be made by name as well as by number. 1671</P> 1672<P> 1673Names consist of up to 32 alphanumeric characters and underscores, but must 1674start with a non-digit. Named capturing parentheses are still allocated numbers 1675as well as names, exactly as if the names were not present. The PCRE API 1676provides function calls for extracting the name-to-number translation table 1677from a compiled pattern. There is also a convenience function for extracting a 1678captured substring by name. 1679</P> 1680<P> 1681By default, a name must be unique within a pattern, but it is possible to relax 1682this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate 1683names are also always permitted for subpatterns with the same number, set up as 1684described in the previous section.) Duplicate names can be useful for patterns 1685where only one instance of the named parentheses can match. Suppose you want to 1686match the name of a weekday, either as a 3-letter abbreviation or as the full 1687name, and in both cases you want to extract the abbreviation. This pattern 1688(ignoring the line breaks) does the job: 1689<pre> 1690 (?<DN>Mon|Fri|Sun)(?:day)?| 1691 (?<DN>Tue)(?:sday)?| 1692 (?<DN>Wed)(?:nesday)?| 1693 (?<DN>Thu)(?:rsday)?| 1694 (?<DN>Sat)(?:urday)? 1695</pre> 1696There are five capturing substrings, but only one is ever set after a match. 1697(An alternative way of solving this problem is to use a "branch reset" 1698subpattern, as described in the previous section.) 1699</P> 1700<P> 1701The convenience function for extracting the data by name returns the substring 1702for the first (and in this example, the only) subpattern of that name that 1703matched. This saves searching to find which numbered subpattern it was. 1704</P> 1705<P> 1706If you make a back reference to a non-unique named subpattern from elsewhere in 1707the pattern, the subpatterns to which the name refers are checked in the order 1708in which they appear in the overall pattern. The first one that is set is used 1709for the reference. For example, this pattern matches both "foofoo" and 1710"barbar" but not "foobar" or "barfoo": 1711<pre> 1712 (?:(?<n>foo)|(?<n>bar))\k<n> 1713 1714</PRE> 1715</P> 1716<P> 1717If you make a subroutine call to a non-unique named subpattern, the one that 1718corresponds to the first occurrence of the name is used. In the absence of 1719duplicate numbers (see the previous section) this is the one with the lowest 1720number. 1721</P> 1722<P> 1723If you use a named reference in a condition 1724test (see the 1725<a href="#conditions">section about conditions</a> 1726below), either to check whether a subpattern has matched, or to check for 1727recursion, all subpatterns with the same name are tested. If the condition is 1728true for any one of them, the overall condition is true. This is the same 1729behaviour as testing by number. For further details of the interfaces for 1730handling named subpatterns, see the 1731<a href="pcreapi.html"><b>pcreapi</b></a> 1732documentation. 1733</P> 1734<P> 1735<b>Warning:</b> You cannot use different names to distinguish between two 1736subpatterns with the same number because PCRE uses only the numbers when 1737matching. For this reason, an error is given at compile time if different names 1738are given to subpatterns with the same number. However, you can always give the 1739same name to subpatterns with the same number, even when PCRE_DUPNAMES is not 1740set. 1741</P> 1742<br><a name="SEC17" href="#TOC1">REPETITION</a><br> 1743<P> 1744Repetition is specified by quantifiers, which can follow any of the following 1745items: 1746<pre> 1747 a literal data character 1748 the dot metacharacter 1749 the \C escape sequence 1750 the \X escape sequence 1751 the \R escape sequence 1752 an escape such as \d or \pL that matches a single character 1753 a character class 1754 a back reference (see next section) 1755 a parenthesized subpattern (including assertions) 1756 a subroutine call to a subpattern (recursive or otherwise) 1757</pre> 1758The general repetition quantifier specifies a minimum and maximum number of 1759permitted matches, by giving the two numbers in curly brackets (braces), 1760separated by a comma. The numbers must be less than 65536, and the first must 1761be less than or equal to the second. For example: 1762<pre> 1763 z{2,4} 1764</pre> 1765matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special 1766character. If the second number is omitted, but the comma is present, there is 1767no upper limit; if the second number and the comma are both omitted, the 1768quantifier specifies an exact number of required matches. Thus 1769<pre> 1770 [aeiou]{3,} 1771</pre> 1772matches at least 3 successive vowels, but may match many more, while 1773<pre> 1774 \d{8} 1775</pre> 1776matches exactly 8 digits. An opening curly bracket that appears in a position 1777where a quantifier is not allowed, or one that does not match the syntax of a 1778quantifier, is taken as a literal character. For example, {,6} is not a 1779quantifier, but a literal string of four characters. 1780</P> 1781<P> 1782In UTF modes, quantifiers apply to characters rather than to individual data 1783units. Thus, for example, \x{100}{2} matches two characters, each of 1784which is represented by a two-byte sequence in a UTF-8 string. Similarly, 1785\X{3} matches three Unicode extended grapheme clusters, each of which may be 1786several data units long (and they may be of different lengths). 1787</P> 1788<P> 1789The quantifier {0} is permitted, causing the expression to behave as if the 1790previous item and the quantifier were not present. This may be useful for 1791subpatterns that are referenced as 1792<a href="#subpatternsassubroutines">subroutines</a> 1793from elsewhere in the pattern (but see also the section entitled 1794<a href="#subdefine">"Defining subpatterns for use by reference only"</a> 1795below). Items other than subpatterns that have a {0} quantifier are omitted 1796from the compiled pattern. 1797</P> 1798<P> 1799For convenience, the three most common quantifiers have single-character 1800abbreviations: 1801<pre> 1802 * is equivalent to {0,} 1803 + is equivalent to {1,} 1804 ? is equivalent to {0,1} 1805</pre> 1806It is possible to construct infinite loops by following a subpattern that can 1807match no characters with a quantifier that has no upper limit, for example: 1808<pre> 1809 (a?)* 1810</pre> 1811Earlier versions of Perl and PCRE used to give an error at compile time for 1812such patterns. However, because there are cases where this can be useful, such 1813patterns are now accepted, but if any repetition of the subpattern does in fact 1814match no characters, the loop is forcibly broken. 1815</P> 1816<P> 1817By default, the quantifiers are "greedy", that is, they match as much as 1818possible (up to the maximum number of permitted times), without causing the 1819rest of the pattern to fail. The classic example of where this gives problems 1820is in trying to match comments in C programs. These appear between /* and */ 1821and within the comment, individual * and / characters may appear. An attempt to 1822match C comments by applying the pattern 1823<pre> 1824 /\*.*\*/ 1825</pre> 1826to the string 1827<pre> 1828 /* first comment */ not comment /* second comment */ 1829</pre> 1830fails, because it matches the entire string owing to the greediness of the .* 1831item. 1832</P> 1833<P> 1834However, if a quantifier is followed by a question mark, it ceases to be 1835greedy, and instead matches the minimum number of times possible, so the 1836pattern 1837<pre> 1838 /\*.*?\*/ 1839</pre> 1840does the right thing with the C comments. The meaning of the various 1841quantifiers is not otherwise changed, just the preferred number of matches. 1842Do not confuse this use of question mark with its use as a quantifier in its 1843own right. Because it has two uses, it can sometimes appear doubled, as in 1844<pre> 1845 \d??\d 1846</pre> 1847which matches one digit by preference, but can match two if that is the only 1848way the rest of the pattern matches. 1849</P> 1850<P> 1851If the PCRE_UNGREEDY option is set (an option that is not available in Perl), 1852the quantifiers are not greedy by default, but individual ones can be made 1853greedy by following them with a question mark. In other words, it inverts the 1854default behaviour. 1855</P> 1856<P> 1857When a parenthesized subpattern is quantified with a minimum repeat count that 1858is greater than 1 or with a limited maximum, more memory is required for the 1859compiled pattern, in proportion to the size of the minimum or maximum. 1860</P> 1861<P> 1862If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent 1863to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is 1864implicitly anchored, because whatever follows will be tried against every 1865character position in the subject string, so there is no point in retrying the 1866overall match at any position after the first. PCRE normally treats such a 1867pattern as though it were preceded by \A. 1868</P> 1869<P> 1870In cases where it is known that the subject string contains no newlines, it is 1871worth setting PCRE_DOTALL in order to obtain this optimization, or 1872alternatively using ^ to indicate anchoring explicitly. 1873</P> 1874<P> 1875However, there are some cases where the optimization cannot be used. When .* 1876is inside capturing parentheses that are the subject of a back reference 1877elsewhere in the pattern, a match at the start may fail where a later one 1878succeeds. Consider, for example: 1879<pre> 1880 (.*)abc\1 1881</pre> 1882If the subject is "xyz123abc123" the match point is the fourth character. For 1883this reason, such a pattern is not implicitly anchored. 1884</P> 1885<P> 1886Another case where implicit anchoring is not applied is when the leading .* is 1887inside an atomic group. Once again, a match at the start may fail where a later 1888one succeeds. Consider this pattern: 1889<pre> 1890 (?>.*?a)b 1891</pre> 1892It matches "ab" in the subject "aab". The use of the backtracking control verbs 1893(*PRUNE) and (*SKIP) also disable this optimization. 1894</P> 1895<P> 1896When a capturing subpattern is repeated, the value captured is the substring 1897that matched the final iteration. For example, after 1898<pre> 1899 (tweedle[dume]{3}\s*)+ 1900</pre> 1901has matched "tweedledum tweedledee" the value of the captured substring is 1902"tweedledee". However, if there are nested capturing subpatterns, the 1903corresponding captured values may have been set in previous iterations. For 1904example, after 1905<pre> 1906 /(a|(b))+/ 1907</pre> 1908matches "aba" the value of the second captured substring is "b". 1909<a name="atomicgroup"></a></P> 1910<br><a name="SEC18" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br> 1911<P> 1912With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy") 1913repetition, failure of what follows normally causes the repeated item to be 1914re-evaluated to see if a different number of repeats allows the rest of the 1915pattern to match. Sometimes it is useful to prevent this, either to change the 1916nature of the match, or to cause it fail earlier than it otherwise might, when 1917the author of the pattern knows there is no point in carrying on. 1918</P> 1919<P> 1920Consider, for example, the pattern \d+foo when applied to the subject line 1921<pre> 1922 123456bar 1923</pre> 1924After matching all 6 digits and then failing to match "foo", the normal 1925action of the matcher is to try again with only 5 digits matching the \d+ 1926item, and then with 4, and so on, before ultimately failing. "Atomic grouping" 1927(a term taken from Jeffrey Friedl's book) provides the means for specifying 1928that once a subpattern has matched, it is not to be re-evaluated in this way. 1929</P> 1930<P> 1931If we use atomic grouping for the previous example, the matcher gives up 1932immediately on failing to match "foo" the first time. The notation is a kind of 1933special parenthesis, starting with (?> as in this example: 1934<pre> 1935 (?>\d+)foo 1936</pre> 1937This kind of parenthesis "locks up" the part of the pattern it contains once 1938it has matched, and a failure further into the pattern is prevented from 1939backtracking into it. Backtracking past it to previous items, however, works as 1940normal. 1941</P> 1942<P> 1943An alternative description is that a subpattern of this type matches the string 1944of characters that an identical standalone pattern would match, if anchored at 1945the current point in the subject string. 1946</P> 1947<P> 1948Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as 1949the above example can be thought of as a maximizing repeat that must swallow 1950everything it can. So, while both \d+ and \d+? are prepared to adjust the 1951number of digits they match in order to make the rest of the pattern match, 1952(?>\d+) can only match an entire sequence of digits. 1953</P> 1954<P> 1955Atomic groups in general can of course contain arbitrarily complicated 1956subpatterns, and can be nested. However, when the subpattern for an atomic 1957group is just a single repeated item, as in the example above, a simpler 1958notation, called a "possessive quantifier" can be used. This consists of an 1959additional + character following a quantifier. Using this notation, the 1960previous example can be rewritten as 1961<pre> 1962 \d++foo 1963</pre> 1964Note that a possessive quantifier can be used with an entire group, for 1965example: 1966<pre> 1967 (abc|xyz){2,3}+ 1968</pre> 1969Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY 1970option is ignored. They are a convenient notation for the simpler forms of 1971atomic group. However, there is no difference in the meaning of a possessive 1972quantifier and the equivalent atomic group, though there may be a performance 1973difference; possessive quantifiers should be slightly faster. 1974</P> 1975<P> 1976The possessive quantifier syntax is an extension to the Perl 5.8 syntax. 1977Jeffrey Friedl originated the idea (and the name) in the first edition of his 1978book. Mike McCloskey liked it, so implemented it when he built Sun's Java 1979package, and PCRE copied it from there. It ultimately found its way into Perl 1980at release 5.10. 1981</P> 1982<P> 1983PCRE has an optimization that automatically "possessifies" certain simple 1984pattern constructs. For example, the sequence A+B is treated as A++B because 1985there is no point in backtracking into a sequence of A's when B must follow. 1986</P> 1987<P> 1988When a pattern contains an unlimited repeat inside a subpattern that can itself 1989be repeated an unlimited number of times, the use of an atomic group is the 1990only way to avoid some failing matches taking a very long time indeed. The 1991pattern 1992<pre> 1993 (\D+|<\d+>)*[!?] 1994</pre> 1995matches an unlimited number of substrings that either consist of non-digits, or 1996digits enclosed in <>, followed by either ! or ?. When it matches, it runs 1997quickly. However, if it is applied to 1998<pre> 1999 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa 2000</pre> 2001it takes a long time before reporting failure. This is because the string can 2002be divided between the internal \D+ repeat and the external * repeat in a 2003large number of ways, and all have to be tried. (The example uses [!?] rather 2004than a single character at the end, because both PCRE and Perl have an 2005optimization that allows for fast failure when a single character is used. They 2006remember the last single character that is required for a match, and fail early 2007if it is not present in the string.) If the pattern is changed so that it uses 2008an atomic group, like this: 2009<pre> 2010 ((?>\D+)|<\d+>)*[!?] 2011</pre> 2012sequences of non-digits cannot be broken, and failure happens quickly. 2013<a name="backreferences"></a></P> 2014<br><a name="SEC19" href="#TOC1">BACK REFERENCES</a><br> 2015<P> 2016Outside a character class, a backslash followed by a digit greater than 0 (and 2017possibly further digits) is a back reference to a capturing subpattern earlier 2018(that is, to its left) in the pattern, provided there have been that many 2019previous capturing left parentheses. 2020</P> 2021<P> 2022However, if the decimal number following the backslash is less than 10, it is 2023always taken as a back reference, and causes an error only if there are not 2024that many capturing left parentheses in the entire pattern. In other words, the 2025parentheses that are referenced need not be to the left of the reference for 2026numbers less than 10. A "forward back reference" of this type can make sense 2027when a repetition is involved and the subpattern to the right has participated 2028in an earlier iteration. 2029</P> 2030<P> 2031It is not possible to have a numerical "forward back reference" to a subpattern 2032whose number is 10 or more using this syntax because a sequence such as \50 is 2033interpreted as a character defined in octal. See the subsection entitled 2034"Non-printing characters" 2035<a href="#digitsafterbackslash">above</a> 2036for further details of the handling of digits following a backslash. There is 2037no such problem when named parentheses are used. A back reference to any 2038subpattern is possible using named parentheses (see below). 2039</P> 2040<P> 2041Another way of avoiding the ambiguity inherent in the use of digits following a 2042backslash is to use the \g escape sequence. This escape must be followed by an 2043unsigned number or a negative number, optionally enclosed in braces. These 2044examples are all identical: 2045<pre> 2046 (ring), \1 2047 (ring), \g1 2048 (ring), \g{1} 2049</pre> 2050An unsigned number specifies an absolute reference without the ambiguity that 2051is present in the older syntax. It is also useful when literal digits follow 2052the reference. A negative number is a relative reference. Consider this 2053example: 2054<pre> 2055 (abc(def)ghi)\g{-1} 2056</pre> 2057The sequence \g{-1} is a reference to the most recently started capturing 2058subpattern before \g, that is, is it equivalent to \2 in this example. 2059Similarly, \g{-2} would be equivalent to \1. The use of relative references 2060can be helpful in long patterns, and also in patterns that are created by 2061joining together fragments that contain references within themselves. 2062</P> 2063<P> 2064A back reference matches whatever actually matched the capturing subpattern in 2065the current subject string, rather than anything matching the subpattern 2066itself (see 2067<a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a> 2068below for a way of doing that). So the pattern 2069<pre> 2070 (sens|respons)e and \1ibility 2071</pre> 2072matches "sense and sensibility" and "response and responsibility", but not 2073"sense and responsibility". If caseful matching is in force at the time of the 2074back reference, the case of letters is relevant. For example, 2075<pre> 2076 ((?i)rah)\s+\1 2077</pre> 2078matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original 2079capturing subpattern is matched caselessly. 2080</P> 2081<P> 2082There are several different ways of writing back references to named 2083subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or 2084\k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified 2085back reference syntax, in which \g can be used for both numeric and named 2086references, is also supported. We could rewrite the above example in any of 2087the following ways: 2088<pre> 2089 (?<p1>(?i)rah)\s+\k<p1> 2090 (?'p1'(?i)rah)\s+\k{p1} 2091 (?P<p1>(?i)rah)\s+(?P=p1) 2092 (?<p1>(?i)rah)\s+\g{p1} 2093</pre> 2094A subpattern that is referenced by name may appear in the pattern before or 2095after the reference. 2096</P> 2097<P> 2098There may be more than one back reference to the same subpattern. If a 2099subpattern has not actually been used in a particular match, any back 2100references to it always fail by default. For example, the pattern 2101<pre> 2102 (a|(bc))\2 2103</pre> 2104always fails if it starts to match "a" rather than "bc". However, if the 2105PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an 2106unset value matches an empty string. 2107</P> 2108<P> 2109Because there may be many capturing parentheses in a pattern, all digits 2110following a backslash are taken as part of a potential back reference number. 2111If the pattern continues with a digit character, some delimiter must be used to 2112terminate the back reference. If the PCRE_EXTENDED option is set, this can be 2113white space. Otherwise, the \g{ syntax or an empty comment (see 2114<a href="#comments">"Comments"</a> 2115below) can be used. 2116</P> 2117<br><b> 2118Recursive back references 2119</b><br> 2120<P> 2121A back reference that occurs inside the parentheses to which it refers fails 2122when the subpattern is first used, so, for example, (a\1) never matches. 2123However, such references can be useful inside repeated subpatterns. For 2124example, the pattern 2125<pre> 2126 (a|b\1)+ 2127</pre> 2128matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of 2129the subpattern, the back reference matches the character string corresponding 2130to the previous iteration. In order for this to work, the pattern must be such 2131that the first iteration does not need to match the back reference. This can be 2132done using alternation, as in the example above, or by a quantifier with a 2133minimum of zero. 2134</P> 2135<P> 2136Back references of this type cause the group that they reference to be treated 2137as an 2138<a href="#atomicgroup">atomic group.</a> 2139Once the whole group has been matched, a subsequent matching failure cannot 2140cause backtracking into the middle of the group. 2141<a name="bigassertions"></a></P> 2142<br><a name="SEC20" href="#TOC1">ASSERTIONS</a><br> 2143<P> 2144An assertion is a test on the characters following or preceding the current 2145matching point that does not actually consume any characters. The simple 2146assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described 2147<a href="#smallassertions">above.</a> 2148</P> 2149<P> 2150More complicated assertions are coded as subpatterns. There are two kinds: 2151those that look ahead of the current position in the subject string, and those 2152that look behind it. An assertion subpattern is matched in the normal way, 2153except that it does not cause the current matching position to be changed. 2154</P> 2155<P> 2156Assertion subpatterns are not capturing subpatterns. If such an assertion 2157contains capturing subpatterns within it, these are counted for the purposes of 2158numbering the capturing subpatterns in the whole pattern. However, substring 2159capturing is carried out only for positive assertions. (Perl sometimes, but not 2160always, does do capturing in negative assertions.) 2161</P> 2162<P> 2163For compatibility with Perl, assertion subpatterns may be repeated; though 2164it makes no sense to assert the same thing several times, the side effect of 2165capturing parentheses may occasionally be useful. In practice, there only three 2166cases: 2167<br> 2168<br> 2169(1) If the quantifier is {0}, the assertion is never obeyed during matching. 2170However, it may contain internal capturing parenthesized groups that are called 2171from elsewhere via the 2172<a href="#subpatternsassubroutines">subroutine mechanism.</a> 2173<br> 2174<br> 2175(2) If quantifier is {0,n} where n is greater than zero, it is treated as if it 2176were {0,1}. At run time, the rest of the pattern match is tried with and 2177without the assertion, the order depending on the greediness of the quantifier. 2178<br> 2179<br> 2180(3) If the minimum repetition is greater than zero, the quantifier is ignored. 2181The assertion is obeyed just once when encountered during matching. 2182</P> 2183<br><b> 2184Lookahead assertions 2185</b><br> 2186<P> 2187Lookahead assertions start with (?= for positive assertions and (?! for 2188negative assertions. For example, 2189<pre> 2190 \w+(?=;) 2191</pre> 2192matches a word followed by a semicolon, but does not include the semicolon in 2193the match, and 2194<pre> 2195 foo(?!bar) 2196</pre> 2197matches any occurrence of "foo" that is not followed by "bar". Note that the 2198apparently similar pattern 2199<pre> 2200 (?!foo)bar 2201</pre> 2202does not find an occurrence of "bar" that is preceded by something other than 2203"foo"; it finds any occurrence of "bar" whatsoever, because the assertion 2204(?!foo) is always true when the next three characters are "bar". A 2205lookbehind assertion is needed to achieve the other effect. 2206</P> 2207<P> 2208If you want to force a matching failure at some point in a pattern, the most 2209convenient way to do it is with (?!) because an empty string always matches, so 2210an assertion that requires there not to be an empty string must always fail. 2211The backtracking control verb (*FAIL) or (*F) is a synonym for (?!). 2212<a name="lookbehind"></a></P> 2213<br><b> 2214Lookbehind assertions 2215</b><br> 2216<P> 2217Lookbehind assertions start with (?<= for positive assertions and (?<! for 2218negative assertions. For example, 2219<pre> 2220 (?<!foo)bar 2221</pre> 2222does find an occurrence of "bar" that is not preceded by "foo". The contents of 2223a lookbehind assertion are restricted such that all the strings it matches must 2224have a fixed length. However, if there are several top-level alternatives, they 2225do not all have to have the same fixed length. Thus 2226<pre> 2227 (?<=bullock|donkey) 2228</pre> 2229is permitted, but 2230<pre> 2231 (?<!dogs?|cats?) 2232</pre> 2233causes an error at compile time. Branches that match different length strings 2234are permitted only at the top level of a lookbehind assertion. This is an 2235extension compared with Perl, which requires all branches to match the same 2236length of string. An assertion such as 2237<pre> 2238 (?<=ab(c|de)) 2239</pre> 2240is not permitted, because its single top-level branch can match two different 2241lengths, but it is acceptable to PCRE if rewritten to use two top-level 2242branches: 2243<pre> 2244 (?<=abc|abde) 2245</pre> 2246In some cases, the escape sequence \K 2247<a href="#resetmatchstart">(see above)</a> 2248can be used instead of a lookbehind assertion to get round the fixed-length 2249restriction. 2250</P> 2251<P> 2252The implementation of lookbehind assertions is, for each alternative, to 2253temporarily move the current position back by the fixed length and then try to 2254match. If there are insufficient characters before the current position, the 2255assertion fails. 2256</P> 2257<P> 2258In a UTF mode, PCRE does not allow the \C escape (which matches a single data 2259unit even in a UTF mode) to appear in lookbehind assertions, because it makes 2260it impossible to calculate the length of the lookbehind. The \X and \R 2261escapes, which can match different numbers of data units, are also not 2262permitted. 2263</P> 2264<P> 2265<a href="#subpatternsassubroutines">"Subroutine"</a> 2266calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long 2267as the subpattern matches a fixed-length string. 2268<a href="#recursion">Recursion,</a> 2269however, is not supported. 2270</P> 2271<P> 2272Possessive quantifiers can be used in conjunction with lookbehind assertions to 2273specify efficient matching of fixed-length strings at the end of subject 2274strings. Consider a simple pattern such as 2275<pre> 2276 abcd$ 2277</pre> 2278when applied to a long string that does not match. Because matching proceeds 2279from left to right, PCRE will look for each "a" in the subject and then see if 2280what follows matches the rest of the pattern. If the pattern is specified as 2281<pre> 2282 ^.*abcd$ 2283</pre> 2284the initial .* matches the entire string at first, but when this fails (because 2285there is no following "a"), it backtracks to match all but the last character, 2286then all but the last two characters, and so on. Once again the search for "a" 2287covers the entire string, from right to left, so we are no better off. However, 2288if the pattern is written as 2289<pre> 2290 ^.*+(?<=abcd) 2291</pre> 2292there can be no backtracking for the .*+ item; it can match only the entire 2293string. The subsequent lookbehind assertion does a single test on the last four 2294characters. If it fails, the match fails immediately. For long strings, this 2295approach makes a significant difference to the processing time. 2296</P> 2297<br><b> 2298Using multiple assertions 2299</b><br> 2300<P> 2301Several assertions (of any sort) may occur in succession. For example, 2302<pre> 2303 (?<=\d{3})(?<!999)foo 2304</pre> 2305matches "foo" preceded by three digits that are not "999". Notice that each of 2306the assertions is applied independently at the same point in the subject 2307string. First there is a check that the previous three characters are all 2308digits, and then there is a check that the same three characters are not "999". 2309This pattern does <i>not</i> match "foo" preceded by six characters, the first 2310of which are digits and the last three of which are not "999". For example, it 2311doesn't match "123abcfoo". A pattern to do that is 2312<pre> 2313 (?<=\d{3}...)(?<!999)foo 2314</pre> 2315This time the first assertion looks at the preceding six characters, checking 2316that the first three are digits, and then the second assertion checks that the 2317preceding three characters are not "999". 2318</P> 2319<P> 2320Assertions can be nested in any combination. For example, 2321<pre> 2322 (?<=(?<!foo)bar)baz 2323</pre> 2324matches an occurrence of "baz" that is preceded by "bar" which in turn is not 2325preceded by "foo", while 2326<pre> 2327 (?<=\d{3}(?!999)...)foo 2328</pre> 2329is another pattern that matches "foo" preceded by three digits and any three 2330characters that are not "999". 2331<a name="conditions"></a></P> 2332<br><a name="SEC21" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br> 2333<P> 2334It is possible to cause the matching process to obey a subpattern 2335conditionally or to choose between two alternative subpatterns, depending on 2336the result of an assertion, or whether a specific capturing subpattern has 2337already been matched. The two possible forms of conditional subpattern are: 2338<pre> 2339 (?(condition)yes-pattern) 2340 (?(condition)yes-pattern|no-pattern) 2341</pre> 2342If the condition is satisfied, the yes-pattern is used; otherwise the 2343no-pattern (if present) is used. If there are more than two alternatives in the 2344subpattern, a compile-time error occurs. Each of the two alternatives may 2345itself contain nested subpatterns of any form, including conditional 2346subpatterns; the restriction to two alternatives applies only at the level of 2347the condition. This pattern fragment is an example where the alternatives are 2348complex: 2349<pre> 2350 (?(1) (A|B|C) | (D | (?(2)E|F) | E) ) 2351 2352</PRE> 2353</P> 2354<P> 2355There are four kinds of condition: references to subpatterns, references to 2356recursion, a pseudo-condition called DEFINE, and assertions. 2357</P> 2358<br><b> 2359Checking for a used subpattern by number 2360</b><br> 2361<P> 2362If the text between the parentheses consists of a sequence of digits, the 2363condition is true if a capturing subpattern of that number has previously 2364matched. If there is more than one capturing subpattern with the same number 2365(see the earlier 2366<a href="#recursion">section about duplicate subpattern numbers),</a> 2367the condition is true if any of them have matched. An alternative notation is 2368to precede the digits with a plus or minus sign. In this case, the subpattern 2369number is relative rather than absolute. The most recently opened parentheses 2370can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside 2371loops it can also make sense to refer to subsequent groups. The next 2372parentheses to be opened can be referenced as (?(+1), and so on. (The value 2373zero in any of these forms is not used; it provokes a compile-time error.) 2374</P> 2375<P> 2376Consider the following pattern, which contains non-significant white space to 2377make it more readable (assume the PCRE_EXTENDED option) and to divide it into 2378three parts for ease of discussion: 2379<pre> 2380 ( \( )? [^()]+ (?(1) \) ) 2381</pre> 2382The first part matches an optional opening parenthesis, and if that 2383character is present, sets it as the first captured substring. The second part 2384matches one or more characters that are not parentheses. The third part is a 2385conditional subpattern that tests whether or not the first set of parentheses 2386matched. If they did, that is, if subject started with an opening parenthesis, 2387the condition is true, and so the yes-pattern is executed and a closing 2388parenthesis is required. Otherwise, since no-pattern is not present, the 2389subpattern matches nothing. In other words, this pattern matches a sequence of 2390non-parentheses, optionally enclosed in parentheses. 2391</P> 2392<P> 2393If you were embedding this pattern in a larger one, you could use a relative 2394reference: 2395<pre> 2396 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ... 2397</pre> 2398This makes the fragment independent of the parentheses in the larger pattern. 2399</P> 2400<br><b> 2401Checking for a used subpattern by name 2402</b><br> 2403<P> 2404Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used 2405subpattern by name. For compatibility with earlier versions of PCRE, which had 2406this facility before Perl, the syntax (?(name)...) is also recognized. 2407</P> 2408<P> 2409Rewriting the above example to use a named subpattern gives this: 2410<pre> 2411 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) ) 2412</pre> 2413If the name used in a condition of this kind is a duplicate, the test is 2414applied to all subpatterns of the same name, and is true if any one of them has 2415matched. 2416</P> 2417<br><b> 2418Checking for pattern recursion 2419</b><br> 2420<P> 2421If the condition is the string (R), and there is no subpattern with the name R, 2422the condition is true if a recursive call to the whole pattern or any 2423subpattern has been made. If digits or a name preceded by ampersand follow the 2424letter R, for example: 2425<pre> 2426 (?(R3)...) or (?(R&name)...) 2427</pre> 2428the condition is true if the most recent recursion is into a subpattern whose 2429number or name is given. This condition does not check the entire recursion 2430stack. If the name used in a condition of this kind is a duplicate, the test is 2431applied to all subpatterns of the same name, and is true if any one of them is 2432the most recent recursion. 2433</P> 2434<P> 2435At "top level", all these recursion test conditions are false. 2436<a href="#recursion">The syntax for recursive patterns</a> 2437is described below. 2438<a name="subdefine"></a></P> 2439<br><b> 2440Defining subpatterns for use by reference only 2441</b><br> 2442<P> 2443If the condition is the string (DEFINE), and there is no subpattern with the 2444name DEFINE, the condition is always false. In this case, there may be only one 2445alternative in the subpattern. It is always skipped if control reaches this 2446point in the pattern; the idea of DEFINE is that it can be used to define 2447subroutines that can be referenced from elsewhere. (The use of 2448<a href="#subpatternsassubroutines">subroutines</a> 2449is described below.) For example, a pattern to match an IPv4 address such as 2450"192.168.23.245" could be written like this (ignore white space and line 2451breaks): 2452<pre> 2453 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) ) 2454 \b (?&byte) (\.(?&byte)){3} \b 2455</pre> 2456The first part of the pattern is a DEFINE group inside which a another group 2457named "byte" is defined. This matches an individual component of an IPv4 2458address (a number less than 256). When matching takes place, this part of the 2459pattern is skipped because DEFINE acts like a false condition. The rest of the 2460pattern uses references to the named group to match the four dot-separated 2461components of an IPv4 address, insisting on a word boundary at each end. 2462</P> 2463<br><b> 2464Assertion conditions 2465</b><br> 2466<P> 2467If the condition is not in any of the above formats, it must be an assertion. 2468This may be a positive or negative lookahead or lookbehind assertion. Consider 2469this pattern, again containing non-significant white space, and with the two 2470alternatives on the second line: 2471<pre> 2472 (?(?=[^a-z]*[a-z]) 2473 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} ) 2474</pre> 2475The condition is a positive lookahead assertion that matches an optional 2476sequence of non-letters followed by a letter. In other words, it tests for the 2477presence of at least one letter in the subject. If a letter is found, the 2478subject is matched against the first alternative; otherwise it is matched 2479against the second. This pattern matches strings in one of the two forms 2480dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits. 2481<a name="comments"></a></P> 2482<br><a name="SEC22" href="#TOC1">COMMENTS</a><br> 2483<P> 2484There are two ways of including comments in patterns that are processed by 2485PCRE. In both cases, the start of the comment must not be in a character class, 2486nor in the middle of any other sequence of related characters such as (?: or a 2487subpattern name or number. The characters that make up a comment play no part 2488in the pattern matching. 2489</P> 2490<P> 2491The sequence (?# marks the start of a comment that continues up to the next 2492closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED 2493option is set, an unescaped # character also introduces a comment, which in 2494this case continues to immediately after the next newline character or 2495character sequence in the pattern. Which characters are interpreted as newlines 2496is controlled by the options passed to a compiling function or by a special 2497sequence at the start of the pattern, as described in the section entitled 2498<a href="#newlines">"Newline conventions"</a> 2499above. Note that the end of this type of comment is a literal newline sequence 2500in the pattern; escape sequences that happen to represent a newline do not 2501count. For example, consider this pattern when PCRE_EXTENDED is set, and the 2502default newline convention is in force: 2503<pre> 2504 abc #comment \n still comment 2505</pre> 2506On encountering the # character, <b>pcre_compile()</b> skips along, looking for 2507a newline in the pattern. The sequence \n is still literal at this stage, so 2508it does not terminate the comment. Only an actual character with the code value 25090x0a (the default newline) does so. 2510<a name="recursion"></a></P> 2511<br><a name="SEC23" href="#TOC1">RECURSIVE PATTERNS</a><br> 2512<P> 2513Consider the problem of matching a string in parentheses, allowing for 2514unlimited nested parentheses. Without the use of recursion, the best that can 2515be done is to use a pattern that matches up to some fixed depth of nesting. It 2516is not possible to handle an arbitrary nesting depth. 2517</P> 2518<P> 2519For some time, Perl has provided a facility that allows regular expressions to 2520recurse (amongst other things). It does this by interpolating Perl code in the 2521expression at run time, and the code can refer to the expression itself. A Perl 2522pattern using code interpolation to solve the parentheses problem can be 2523created like this: 2524<pre> 2525 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x; 2526</pre> 2527The (?p{...}) item interpolates Perl code at run time, and in this case refers 2528recursively to the pattern in which it appears. 2529</P> 2530<P> 2531Obviously, PCRE cannot support the interpolation of Perl code. Instead, it 2532supports special syntax for recursion of the entire pattern, and also for 2533individual subpattern recursion. After its introduction in PCRE and Python, 2534this kind of recursion was subsequently introduced into Perl at release 5.10. 2535</P> 2536<P> 2537A special item that consists of (? followed by a number greater than zero and a 2538closing parenthesis is a recursive subroutine call of the subpattern of the 2539given number, provided that it occurs inside that subpattern. (If not, it is a 2540<a href="#subpatternsassubroutines">non-recursive subroutine</a> 2541call, which is described in the next section.) The special item (?R) or (?0) is 2542a recursive call of the entire regular expression. 2543</P> 2544<P> 2545This PCRE pattern solves the nested parentheses problem (assume the 2546PCRE_EXTENDED option is set so that white space is ignored): 2547<pre> 2548 \( ( [^()]++ | (?R) )* \) 2549</pre> 2550First it matches an opening parenthesis. Then it matches any number of 2551substrings which can either be a sequence of non-parentheses, or a recursive 2552match of the pattern itself (that is, a correctly parenthesized substring). 2553Finally there is a closing parenthesis. Note the use of a possessive quantifier 2554to avoid backtracking into sequences of non-parentheses. 2555</P> 2556<P> 2557If this were part of a larger pattern, you would not want to recurse the entire 2558pattern, so instead you could use this: 2559<pre> 2560 ( \( ( [^()]++ | (?1) )* \) ) 2561</pre> 2562We have put the pattern into parentheses, and caused the recursion to refer to 2563them instead of the whole pattern. 2564</P> 2565<P> 2566In a larger pattern, keeping track of parenthesis numbers can be tricky. This 2567is made easier by the use of relative references. Instead of (?1) in the 2568pattern above you can write (?-2) to refer to the second most recently opened 2569parentheses preceding the recursion. In other words, a negative number counts 2570capturing parentheses leftwards from the point at which it is encountered. 2571</P> 2572<P> 2573It is also possible to refer to subsequently opened parentheses, by writing 2574references such as (?+2). However, these cannot be recursive because the 2575reference is not inside the parentheses that are referenced. They are always 2576<a href="#subpatternsassubroutines">non-recursive subroutine</a> 2577calls, as described in the next section. 2578</P> 2579<P> 2580An alternative approach is to use named parentheses instead. The Perl syntax 2581for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We 2582could rewrite the above example as follows: 2583<pre> 2584 (?<pn> \( ( [^()]++ | (?&pn) )* \) ) 2585</pre> 2586If there is more than one subpattern with the same name, the earliest one is 2587used. 2588</P> 2589<P> 2590This particular example pattern that we have been looking at contains nested 2591unlimited repeats, and so the use of a possessive quantifier for matching 2592strings of non-parentheses is important when applying the pattern to strings 2593that do not match. For example, when this pattern is applied to 2594<pre> 2595 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa() 2596</pre> 2597it yields "no match" quickly. However, if a possessive quantifier is not used, 2598the match runs for a very long time indeed because there are so many different 2599ways the + and * repeats can carve up the subject, and all have to be tested 2600before failure can be reported. 2601</P> 2602<P> 2603At the end of a match, the values of capturing parentheses are those from 2604the outermost level. If you want to obtain intermediate values, a callout 2605function can be used (see below and the 2606<a href="pcrecallout.html"><b>pcrecallout</b></a> 2607documentation). If the pattern above is matched against 2608<pre> 2609 (ab(cd)ef) 2610</pre> 2611the value for the inner capturing parentheses (numbered 2) is "ef", which is 2612the last value taken on at the top level. If a capturing subpattern is not 2613matched at the top level, its final captured value is unset, even if it was 2614(temporarily) set at a deeper level during the matching process. 2615</P> 2616<P> 2617If there are more than 15 capturing parentheses in a pattern, PCRE has to 2618obtain extra memory to store data during a recursion, which it does by using 2619<b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no memory can 2620be obtained, the match fails with the PCRE_ERROR_NOMEMORY error. 2621</P> 2622<P> 2623Do not confuse the (?R) item with the condition (R), which tests for recursion. 2624Consider this pattern, which matches text in angle brackets, allowing for 2625arbitrary nesting. Only digits are allowed in nested brackets (that is, when 2626recursing), whereas any characters are permitted at the outer level. 2627<pre> 2628 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * > 2629</pre> 2630In this pattern, (?(R) is the start of a conditional subpattern, with two 2631different alternatives for the recursive and non-recursive cases. The (?R) item 2632is the actual recursive call. 2633<a name="recursiondifference"></a></P> 2634<br><b> 2635Differences in recursion processing between PCRE and Perl 2636</b><br> 2637<P> 2638Recursion processing in PCRE differs from Perl in two important ways. In PCRE 2639(like Python, but unlike Perl), a recursive subpattern call is always treated 2640as an atomic group. That is, once it has matched some of the subject string, it 2641is never re-entered, even if it contains untried alternatives and there is a 2642subsequent matching failure. This can be illustrated by the following pattern, 2643which purports to match a palindromic string that contains an odd number of 2644characters (for example, "a", "aba", "abcba", "abcdcba"): 2645<pre> 2646 ^(.|(.)(?1)\2)$ 2647</pre> 2648The idea is that it either matches a single character, or two identical 2649characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE 2650it does not if the pattern is longer than three characters. Consider the 2651subject string "abcba": 2652</P> 2653<P> 2654At the top level, the first character is matched, but as it is not at the end 2655of the string, the first alternative fails; the second alternative is taken 2656and the recursion kicks in. The recursive call to subpattern 1 successfully 2657matches the next character ("b"). (Note that the beginning and end of line 2658tests are not part of the recursion). 2659</P> 2660<P> 2661Back at the top level, the next character ("c") is compared with what 2662subpattern 2 matched, which was "a". This fails. Because the recursion is 2663treated as an atomic group, there are now no backtracking points, and so the 2664entire match fails. (Perl is able, at this point, to re-enter the recursion and 2665try the second alternative.) However, if the pattern is written with the 2666alternatives in the other order, things are different: 2667<pre> 2668 ^((.)(?1)\2|.)$ 2669</pre> 2670This time, the recursing alternative is tried first, and continues to recurse 2671until it runs out of characters, at which point the recursion fails. But this 2672time we do have another alternative to try at the higher level. That is the big 2673difference: in the previous case the remaining alternative is at a deeper 2674recursion level, which PCRE cannot use. 2675</P> 2676<P> 2677To change the pattern so that it matches all palindromic strings, not just 2678those with an odd number of characters, it is tempting to change the pattern to 2679this: 2680<pre> 2681 ^((.)(?1)\2|.?)$ 2682</pre> 2683Again, this works in Perl, but not in PCRE, and for the same reason. When a 2684deeper recursion has matched a single character, it cannot be entered again in 2685order to match an empty string. The solution is to separate the two cases, and 2686write out the odd and even cases as alternatives at the higher level: 2687<pre> 2688 ^(?:((.)(?1)\2|)|((.)(?3)\4|.)) 2689</pre> 2690If you want to match typical palindromic phrases, the pattern has to ignore all 2691non-word characters, which can be done like this: 2692<pre> 2693 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$ 2694</pre> 2695If run with the PCRE_CASELESS option, this pattern matches phrases such as "A 2696man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note 2697the use of the possessive quantifier *+ to avoid backtracking into sequences of 2698non-word characters. Without this, PCRE takes a great deal longer (ten times or 2699more) to match typical phrases, and Perl takes so long that you think it has 2700gone into a loop. 2701</P> 2702<P> 2703<b>WARNING</b>: The palindrome-matching patterns above work only if the subject 2704string does not start with a palindrome that is shorter than the entire string. 2705For example, although "abcba" is correctly matched, if the subject is "ababa", 2706PCRE finds the palindrome "aba" at the start, then fails at top level because 2707the end of the string does not follow. Once again, it cannot jump back into the 2708recursion to try other alternatives, so the entire match fails. 2709</P> 2710<P> 2711The second way in which PCRE and Perl differ in their recursion processing is 2712in the handling of captured values. In Perl, when a subpattern is called 2713recursively or as a subpattern (see the next section), it has no access to any 2714values that were captured outside the recursion, whereas in PCRE these values 2715can be referenced. Consider this pattern: 2716<pre> 2717 ^(.)(\1|a(?2)) 2718</pre> 2719In PCRE, this pattern matches "bab". The first capturing parentheses match "b", 2720then in the second group, when the back reference \1 fails to match "b", the 2721second alternative matches "a" and then recurses. In the recursion, \1 does 2722now match "b" and so the whole match succeeds. In Perl, the pattern fails to 2723match because inside the recursive call \1 cannot access the externally set 2724value. 2725<a name="subpatternsassubroutines"></a></P> 2726<br><a name="SEC24" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br> 2727<P> 2728If the syntax for a recursive subpattern call (either by number or by 2729name) is used outside the parentheses to which it refers, it operates like a 2730subroutine in a programming language. The called subpattern may be defined 2731before or after the reference. A numbered reference can be absolute or 2732relative, as in these examples: 2733<pre> 2734 (...(absolute)...)...(?2)... 2735 (...(relative)...)...(?-1)... 2736 (...(?+1)...(relative)... 2737</pre> 2738An earlier example pointed out that the pattern 2739<pre> 2740 (sens|respons)e and \1ibility 2741</pre> 2742matches "sense and sensibility" and "response and responsibility", but not 2743"sense and responsibility". If instead the pattern 2744<pre> 2745 (sens|respons)e and (?1)ibility 2746</pre> 2747is used, it does match "sense and responsibility" as well as the other two 2748strings. Another example is given in the discussion of DEFINE above. 2749</P> 2750<P> 2751All subroutine calls, whether recursive or not, are always treated as atomic 2752groups. That is, once a subroutine has matched some of the subject string, it 2753is never re-entered, even if it contains untried alternatives and there is a 2754subsequent matching failure. Any capturing parentheses that are set during the 2755subroutine call revert to their previous values afterwards. 2756</P> 2757<P> 2758Processing options such as case-independence are fixed when a subpattern is 2759defined, so if it is used as a subroutine, such options cannot be changed for 2760different calls. For example, consider this pattern: 2761<pre> 2762 (abc)(?i:(?-1)) 2763</pre> 2764It matches "abcabc". It does not match "abcABC" because the change of 2765processing option does not affect the called subpattern. 2766<a name="onigurumasubroutines"></a></P> 2767<br><a name="SEC25" href="#TOC1">ONIGURUMA SUBROUTINE SYNTAX</a><br> 2768<P> 2769For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or 2770a number enclosed either in angle brackets or single quotes, is an alternative 2771syntax for referencing a subpattern as a subroutine, possibly recursively. Here 2772are two of the examples used above, rewritten using this syntax: 2773<pre> 2774 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) ) 2775 (sens|respons)e and \g'1'ibility 2776</pre> 2777PCRE supports an extension to Oniguruma: if a number is preceded by a 2778plus or a minus sign it is taken as a relative reference. For example: 2779<pre> 2780 (abc)(?i:\g<-1>) 2781</pre> 2782Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are <i>not</i> 2783synonymous. The former is a back reference; the latter is a subroutine call. 2784</P> 2785<br><a name="SEC26" href="#TOC1">CALLOUTS</a><br> 2786<P> 2787Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl 2788code to be obeyed in the middle of matching a regular expression. This makes it 2789possible, amongst other things, to extract different substrings that match the 2790same pair of parentheses when there is a repetition. 2791</P> 2792<P> 2793PCRE provides a similar feature, but of course it cannot obey arbitrary Perl 2794code. The feature is called "callout". The caller of PCRE provides an external 2795function by putting its entry point in the global variable <i>pcre_callout</i> 2796(8-bit library) or <i>pcre[16|32]_callout</i> (16-bit or 32-bit library). 2797By default, this variable contains NULL, which disables all calling out. 2798</P> 2799<P> 2800Within a regular expression, (?C) indicates the points at which the external 2801function is to be called. If you want to identify different callout points, you 2802can put a number less than 256 after the letter C. The default value is zero. 2803For example, this pattern has two callout points: 2804<pre> 2805 (?C1)abc(?C2)def 2806</pre> 2807If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are 2808automatically installed before each item in the pattern. They are all numbered 2809255. If there is a conditional group in the pattern whose condition is an 2810assertion, an additional callout is inserted just before the condition. An 2811explicit callout may also be set at this position, as in this example: 2812<pre> 2813 (?(?C9)(?=a)abc|def) 2814</pre> 2815Note that this applies only to assertion conditions, not to other types of 2816condition. 2817</P> 2818<P> 2819During matching, when PCRE reaches a callout point, the external function is 2820called. It is provided with the number of the callout, the position in the 2821pattern, and, optionally, one item of data originally supplied by the caller of 2822the matching function. The callout function may cause matching to proceed, to 2823backtrack, or to fail altogether. 2824</P> 2825<P> 2826By default, PCRE implements a number of optimizations at compile time and 2827matching time, and one side-effect is that sometimes callouts are skipped. If 2828you need all possible callouts to happen, you need to set options that disable 2829the relevant optimizations. More details, and a complete description of the 2830interface to the callout function, are given in the 2831<a href="pcrecallout.html"><b>pcrecallout</b></a> 2832documentation. 2833<a name="backtrackcontrol"></a></P> 2834<br><a name="SEC27" href="#TOC1">BACKTRACKING CONTROL</a><br> 2835<P> 2836Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which 2837are still described in the Perl documentation as "experimental and subject to 2838change or removal in a future version of Perl". It goes on to say: "Their usage 2839in production code should be noted to avoid problems during upgrades." The same 2840remarks apply to the PCRE features described in this section. 2841</P> 2842<P> 2843The new verbs make use of what was previously invalid syntax: an opening 2844parenthesis followed by an asterisk. They are generally of the form 2845(*VERB) or (*VERB:NAME). Some may take either form, possibly behaving 2846differently depending on whether or not a name is present. A name is any 2847sequence of characters that does not include a closing parenthesis. The maximum 2848length of name is 255 in the 8-bit library and 65535 in the 16-bit and 32-bit 2849libraries. If the name is empty, that is, if the closing parenthesis 2850immediately follows the colon, the effect is as if the colon were not there. 2851Any number of these verbs may occur in a pattern. 2852</P> 2853<P> 2854Since these verbs are specifically related to backtracking, most of them can be 2855used only when the pattern is to be matched using one of the traditional 2856matching functions, because these use a backtracking algorithm. With the 2857exception of (*FAIL), which behaves like a failing negative assertion, the 2858backtracking control verbs cause an error if encountered by a DFA matching 2859function. 2860</P> 2861<P> 2862The behaviour of these verbs in 2863<a href="#btrepeat">repeated groups,</a> 2864<a href="#btassert">assertions,</a> 2865and in 2866<a href="#btsub">subpatterns called as subroutines</a> 2867(whether or not recursively) is documented below. 2868<a name="nooptimize"></a></P> 2869<br><b> 2870Optimizations that affect backtracking verbs 2871</b><br> 2872<P> 2873PCRE contains some optimizations that are used to speed up matching by running 2874some checks at the start of each match attempt. For example, it may know the 2875minimum length of matching subject, or that a particular character must be 2876present. When one of these optimizations bypasses the running of a match, any 2877included backtracking verbs will not, of course, be processed. You can suppress 2878the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option 2879when calling <b>pcre_compile()</b> or <b>pcre_exec()</b>, or by starting the 2880pattern with (*NO_START_OPT). There is more discussion of this option in the 2881section entitled 2882<a href="pcreapi.html#execoptions">"Option bits for <b>pcre_exec()</b>"</a> 2883in the 2884<a href="pcreapi.html"><b>pcreapi</b></a> 2885documentation. 2886</P> 2887<P> 2888Experiments with Perl suggest that it too has similar optimizations, sometimes 2889leading to anomalous results. 2890</P> 2891<br><b> 2892Verbs that act immediately 2893</b><br> 2894<P> 2895The following verbs act as soon as they are encountered. They may not be 2896followed by a name. 2897<pre> 2898 (*ACCEPT) 2899</pre> 2900This verb causes the match to end successfully, skipping the remainder of the 2901pattern. However, when it is inside a subpattern that is called as a 2902subroutine, only that subpattern is ended successfully. Matching then continues 2903at the outer level. If (*ACCEPT) in triggered in a positive assertion, the 2904assertion succeeds; in a negative assertion, the assertion fails. 2905</P> 2906<P> 2907If (*ACCEPT) is inside capturing parentheses, the data so far is captured. For 2908example: 2909<pre> 2910 A((?:A|B(*ACCEPT)|C)D) 2911</pre> 2912This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by 2913the outer parentheses. 2914<pre> 2915 (*FAIL) or (*F) 2916</pre> 2917This verb causes a matching failure, forcing backtracking to occur. It is 2918equivalent to (?!) but easier to read. The Perl documentation notes that it is 2919probably useful only when combined with (?{}) or (??{}). Those are, of course, 2920Perl features that are not present in PCRE. The nearest equivalent is the 2921callout feature, as for example in this pattern: 2922<pre> 2923 a+(?C)(*FAIL) 2924</pre> 2925A match with the string "aaaa" always fails, but the callout is taken before 2926each backtrack happens (in this example, 10 times). 2927</P> 2928<br><b> 2929Recording which path was taken 2930</b><br> 2931<P> 2932There is one verb whose main purpose is to track how a match was arrived at, 2933though it also has a secondary use in conjunction with advancing the match 2934starting point (see (*SKIP) below). 2935<pre> 2936 (*MARK:NAME) or (*:NAME) 2937</pre> 2938A name is always required with this verb. There may be as many instances of 2939(*MARK) as you like in a pattern, and their names do not have to be unique. 2940</P> 2941<P> 2942When a match succeeds, the name of the last-encountered (*MARK:NAME), 2943(*PRUNE:NAME), or (*THEN:NAME) on the matching path is passed back to the 2944caller as described in the section entitled 2945<a href="pcreapi.html#extradata">"Extra data for <b>pcre_exec()</b>"</a> 2946in the 2947<a href="pcreapi.html"><b>pcreapi</b></a> 2948documentation. Here is an example of <b>pcretest</b> output, where the /K 2949modifier requests the retrieval and outputting of (*MARK) data: 2950<pre> 2951 re> /X(*MARK:A)Y|X(*MARK:B)Z/K 2952 data> XY 2953 0: XY 2954 MK: A 2955 XZ 2956 0: XZ 2957 MK: B 2958</pre> 2959The (*MARK) name is tagged with "MK:" in this output, and in this example it 2960indicates which of the two alternatives matched. This is a more efficient way 2961of obtaining this information than putting each alternative in its own 2962capturing parentheses. 2963</P> 2964<P> 2965If a verb with a name is encountered in a positive assertion that is true, the 2966name is recorded and passed back if it is the last-encountered. This does not 2967happen for negative assertions or failing positive assertions. 2968</P> 2969<P> 2970After a partial match or a failed match, the last encountered name in the 2971entire match process is returned. For example: 2972<pre> 2973 re> /X(*MARK:A)Y|X(*MARK:B)Z/K 2974 data> XP 2975 No match, mark = B 2976</pre> 2977Note that in this unanchored example the mark is retained from the match 2978attempt that started at the letter "X" in the subject. Subsequent match 2979attempts starting at "P" and then with an empty string do not get as far as the 2980(*MARK) item, but nevertheless do not reset it. 2981</P> 2982<P> 2983If you are interested in (*MARK) values after failed matches, you should 2984probably set the PCRE_NO_START_OPTIMIZE option 2985<a href="#nooptimize">(see above)</a> 2986to ensure that the match is always attempted. 2987</P> 2988<br><b> 2989Verbs that act after backtracking 2990</b><br> 2991<P> 2992The following verbs do nothing when they are encountered. Matching continues 2993with what follows, but if there is no subsequent match, causing a backtrack to 2994the verb, a failure is forced. That is, backtracking cannot pass to the left of 2995the verb. However, when one of these verbs appears inside an atomic group or an 2996assertion that is true, its effect is confined to that group, because once the 2997group has been matched, there is never any backtracking into it. In this 2998situation, backtracking can "jump back" to the left of the entire atomic group 2999or assertion. (Remember also, as stated above, that this localization also 3000applies in subroutine calls.) 3001</P> 3002<P> 3003These verbs differ in exactly what kind of failure occurs when backtracking 3004reaches them. The behaviour described below is what happens when the verb is 3005not in a subroutine or an assertion. Subsequent sections cover these special 3006cases. 3007<pre> 3008 (*COMMIT) 3009</pre> 3010This verb, which may not be followed by a name, causes the whole match to fail 3011outright if there is a later matching failure that causes backtracking to reach 3012it. Even if the pattern is unanchored, no further attempts to find a match by 3013advancing the starting point take place. If (*COMMIT) is the only backtracking 3014verb that is encountered, once it has been passed <b>pcre_exec()</b> is 3015committed to finding a match at the current starting point, or not at all. For 3016example: 3017<pre> 3018 a+(*COMMIT)b 3019</pre> 3020This matches "xxaab" but not "aacaab". It can be thought of as a kind of 3021dynamic anchor, or "I've started, so I must finish." The name of the most 3022recently passed (*MARK) in the path is passed back when (*COMMIT) forces a 3023match failure. 3024</P> 3025<P> 3026If there is more than one backtracking verb in a pattern, a different one that 3027follows (*COMMIT) may be triggered first, so merely passing (*COMMIT) during a 3028match does not always guarantee that a match must be at this starting point. 3029</P> 3030<P> 3031Note that (*COMMIT) at the start of a pattern is not the same as an anchor, 3032unless PCRE's start-of-match optimizations are turned off, as shown in this 3033output from <b>pcretest</b>: 3034<pre> 3035 re> /(*COMMIT)abc/ 3036 data> xyzabc 3037 0: abc 3038 data> xyzabc\Y 3039 No match 3040</pre> 3041For this pattern, PCRE knows that any match must start with "a", so the 3042optimization skips along the subject to "a" before applying the pattern to the 3043first set of data. The match attempt then succeeds. In the second set of data, 3044the escape sequence \Y is interpreted by the <b>pcretest</b> program. It causes 3045the PCRE_NO_START_OPTIMIZE option to be set when <b>pcre_exec()</b> is called. 3046This disables the optimization that skips along to the first character. The 3047pattern is now applied starting at "x", and so the (*COMMIT) causes the match 3048to fail without trying any other starting points. 3049<pre> 3050 (*PRUNE) or (*PRUNE:NAME) 3051</pre> 3052This verb causes the match to fail at the current starting position in the 3053subject if there is a later matching failure that causes backtracking to reach 3054it. If the pattern is unanchored, the normal "bumpalong" advance to the next 3055starting character then happens. Backtracking can occur as usual to the left of 3056(*PRUNE), before it is reached, or when matching to the right of (*PRUNE), but 3057if there is no match to the right, backtracking cannot cross (*PRUNE). In 3058simple cases, the use of (*PRUNE) is just an alternative to an atomic group or 3059possessive quantifier, but there are some uses of (*PRUNE) that cannot be 3060expressed in any other way. In an anchored pattern (*PRUNE) has the same effect 3061as (*COMMIT). 3062</P> 3063<P> 3064The behaviour of (*PRUNE:NAME) is the not the same as (*MARK:NAME)(*PRUNE). 3065It is like (*MARK:NAME) in that the name is remembered for passing back to the 3066caller. However, (*SKIP:NAME) searches only for names set with (*MARK). 3067<pre> 3068 (*SKIP) 3069</pre> 3070This verb, when given without a name, is like (*PRUNE), except that if the 3071pattern is unanchored, the "bumpalong" advance is not to the next character, 3072but to the position in the subject where (*SKIP) was encountered. (*SKIP) 3073signifies that whatever text was matched leading up to it cannot be part of a 3074successful match. Consider: 3075<pre> 3076 a+(*SKIP)b 3077</pre> 3078If the subject is "aaaac...", after the first match attempt fails (starting at 3079the first character in the string), the starting point skips on to start the 3080next attempt at "c". Note that a possessive quantifer does not have the same 3081effect as this example; although it would suppress backtracking during the 3082first match attempt, the second attempt would start at the second character 3083instead of skipping on to "c". 3084<pre> 3085 (*SKIP:NAME) 3086</pre> 3087When (*SKIP) has an associated name, its behaviour is modified. When it is 3088triggered, the previous path through the pattern is searched for the most 3089recent (*MARK) that has the same name. If one is found, the "bumpalong" advance 3090is to the subject position that corresponds to that (*MARK) instead of to where 3091(*SKIP) was encountered. If no (*MARK) with a matching name is found, the 3092(*SKIP) is ignored. 3093</P> 3094<P> 3095Note that (*SKIP:NAME) searches only for names set by (*MARK:NAME). It ignores 3096names that are set by (*PRUNE:NAME) or (*THEN:NAME). 3097<pre> 3098 (*THEN) or (*THEN:NAME) 3099</pre> 3100This verb causes a skip to the next innermost alternative when backtracking 3101reaches it. That is, it cancels any further backtracking within the current 3102alternative. Its name comes from the observation that it can be used for a 3103pattern-based if-then-else block: 3104<pre> 3105 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ... 3106</pre> 3107If the COND1 pattern matches, FOO is tried (and possibly further items after 3108the end of the group if FOO succeeds); on failure, the matcher skips to the 3109second alternative and tries COND2, without backtracking into COND1. If that 3110succeeds and BAR fails, COND3 is tried. If subsequently BAZ fails, there are no 3111more alternatives, so there is a backtrack to whatever came before the entire 3112group. If (*THEN) is not inside an alternation, it acts like (*PRUNE). 3113</P> 3114<P> 3115The behaviour of (*THEN:NAME) is the not the same as (*MARK:NAME)(*THEN). 3116It is like (*MARK:NAME) in that the name is remembered for passing back to the 3117caller. However, (*SKIP:NAME) searches only for names set with (*MARK). 3118</P> 3119<P> 3120A subpattern that does not contain a | character is just a part of the 3121enclosing alternative; it is not a nested alternation with only one 3122alternative. The effect of (*THEN) extends beyond such a subpattern to the 3123enclosing alternative. Consider this pattern, where A, B, etc. are complex 3124pattern fragments that do not contain any | characters at this level: 3125<pre> 3126 A (B(*THEN)C) | D 3127</pre> 3128If A and B are matched, but there is a failure in C, matching does not 3129backtrack into A; instead it moves to the next alternative, that is, D. 3130However, if the subpattern containing (*THEN) is given an alternative, it 3131behaves differently: 3132<pre> 3133 A (B(*THEN)C | (*FAIL)) | D 3134</pre> 3135The effect of (*THEN) is now confined to the inner subpattern. After a failure 3136in C, matching moves to (*FAIL), which causes the whole subpattern to fail 3137because there are no more alternatives to try. In this case, matching does now 3138backtrack into A. 3139</P> 3140<P> 3141Note that a conditional subpattern is not considered as having two 3142alternatives, because only one is ever used. In other words, the | character in 3143a conditional subpattern has a different meaning. Ignoring white space, 3144consider: 3145<pre> 3146 ^.*? (?(?=a) a | b(*THEN)c ) 3147</pre> 3148If the subject is "ba", this pattern does not match. Because .*? is ungreedy, 3149it initially matches zero characters. The condition (?=a) then fails, the 3150character "b" is matched, but "c" is not. At this point, matching does not 3151backtrack to .*? as might perhaps be expected from the presence of the | 3152character. The conditional subpattern is part of the single alternative that 3153comprises the whole pattern, and so the match fails. (If there was a backtrack 3154into .*?, allowing it to match "b", the match would succeed.) 3155</P> 3156<P> 3157The verbs just described provide four different "strengths" of control when 3158subsequent matching fails. (*THEN) is the weakest, carrying on the match at the 3159next alternative. (*PRUNE) comes next, failing the match at the current 3160starting position, but allowing an advance to the next character (for an 3161unanchored pattern). (*SKIP) is similar, except that the advance may be more 3162than one character. (*COMMIT) is the strongest, causing the entire match to 3163fail. 3164</P> 3165<br><b> 3166More than one backtracking verb 3167</b><br> 3168<P> 3169If more than one backtracking verb is present in a pattern, the one that is 3170backtracked onto first acts. For example, consider this pattern, where A, B, 3171etc. are complex pattern fragments: 3172<pre> 3173 (A(*COMMIT)B(*THEN)C|ABD) 3174</pre> 3175If A matches but B fails, the backtrack to (*COMMIT) causes the entire match to 3176fail. However, if A and B match, but C fails, the backtrack to (*THEN) causes 3177the next alternative (ABD) to be tried. This behaviour is consistent, but is 3178not always the same as Perl's. It means that if two or more backtracking verbs 3179appear in succession, all the the last of them has no effect. Consider this 3180example: 3181<pre> 3182 ...(*COMMIT)(*PRUNE)... 3183</pre> 3184If there is a matching failure to the right, backtracking onto (*PRUNE) causes 3185it to be triggered, and its action is taken. There can never be a backtrack 3186onto (*COMMIT). 3187<a name="btrepeat"></a></P> 3188<br><b> 3189Backtracking verbs in repeated groups 3190</b><br> 3191<P> 3192PCRE differs from Perl in its handling of backtracking verbs in repeated 3193groups. For example, consider: 3194<pre> 3195 /(a(*COMMIT)b)+ac/ 3196</pre> 3197If the subject is "abac", Perl matches, but PCRE fails because the (*COMMIT) in 3198the second repeat of the group acts. 3199<a name="btassert"></a></P> 3200<br><b> 3201Backtracking verbs in assertions 3202</b><br> 3203<P> 3204(*FAIL) in an assertion has its normal effect: it forces an immediate backtrack. 3205</P> 3206<P> 3207(*ACCEPT) in a positive assertion causes the assertion to succeed without any 3208further processing. In a negative assertion, (*ACCEPT) causes the assertion to 3209fail without any further processing. 3210</P> 3211<P> 3212The other backtracking verbs are not treated specially if they appear in a 3213positive assertion. In particular, (*THEN) skips to the next alternative in the 3214innermost enclosing group that has alternations, whether or not this is within 3215the assertion. 3216</P> 3217<P> 3218Negative assertions are, however, different, in order to ensure that changing a 3219positive assertion into a negative assertion changes its result. Backtracking 3220into (*COMMIT), (*SKIP), or (*PRUNE) causes a negative assertion to be true, 3221without considering any further alternative branches in the assertion. 3222Backtracking into (*THEN) causes it to skip to the next enclosing alternative 3223within the assertion (the normal behaviour), but if the assertion does not have 3224such an alternative, (*THEN) behaves like (*PRUNE). 3225<a name="btsub"></a></P> 3226<br><b> 3227Backtracking verbs in subroutines 3228</b><br> 3229<P> 3230These behaviours occur whether or not the subpattern is called recursively. 3231Perl's treatment of subroutines is different in some cases. 3232</P> 3233<P> 3234(*FAIL) in a subpattern called as a subroutine has its normal effect: it forces 3235an immediate backtrack. 3236</P> 3237<P> 3238(*ACCEPT) in a subpattern called as a subroutine causes the subroutine match to 3239succeed without any further processing. Matching then continues after the 3240subroutine call. 3241</P> 3242<P> 3243(*COMMIT), (*SKIP), and (*PRUNE) in a subpattern called as a subroutine cause 3244the subroutine match to fail. 3245</P> 3246<P> 3247(*THEN) skips to the next alternative in the innermost enclosing group within 3248the subpattern that has alternatives. If there is no such group within the 3249subpattern, (*THEN) causes the subroutine match to fail. 3250</P> 3251<br><a name="SEC28" href="#TOC1">SEE ALSO</a><br> 3252<P> 3253<b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3), 3254<b>pcresyntax</b>(3), <b>pcre</b>(3), <b>pcre16(3)</b>, <b>pcre32(3)</b>. 3255</P> 3256<br><a name="SEC29" href="#TOC1">AUTHOR</a><br> 3257<P> 3258Philip Hazel 3259<br> 3260University Computing Service 3261<br> 3262Cambridge CB2 3QH, England. 3263<br> 3264</P> 3265<br><a name="SEC30" href="#TOC1">REVISION</a><br> 3266<P> 3267Last updated: 14 June 2015 3268<br> 3269Copyright © 1997-2015 University of Cambridge. 3270<br> 3271<p> 3272Return to the <a href="index.html">PCRE index page</a>. 3273</p> 3274