Annotation of embedaddon/pcre/doc/html/pcrepattern.html, revision 1.1.1.2

1.1       misho       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>
                      8: Return to the <a href="index.html">PCRE index page</a>.
                      9: </p>
                     10: <p>
                     11: This page is part of the PCRE HTML documentation. It was generated automatically
                     12: from the original man page. If there is any nonsense in it, please consult the
                     13: man page, in case the conversion went wrong.
                     14: <br>
                     15: <ul>
                     16: <li><a name="TOC1" href="#SEC1">PCRE REGULAR EXPRESSION DETAILS</a>
                     17: <li><a name="TOC2" href="#SEC2">NEWLINE CONVENTIONS</a>
                     18: <li><a name="TOC3" href="#SEC3">CHARACTERS AND METACHARACTERS</a>
                     19: <li><a name="TOC4" href="#SEC4">BACKSLASH</a>
                     20: <li><a name="TOC5" href="#SEC5">CIRCUMFLEX AND DOLLAR</a>
                     21: <li><a name="TOC6" href="#SEC6">FULL STOP (PERIOD, DOT) AND \N</a>
1.1.1.2 ! misho      22: <li><a name="TOC7" href="#SEC7">MATCHING A SINGLE DATA UNIT</a>
1.1       misho      23: <li><a name="TOC8" href="#SEC8">SQUARE BRACKETS AND CHARACTER CLASSES</a>
                     24: <li><a name="TOC9" href="#SEC9">POSIX CHARACTER CLASSES</a>
                     25: <li><a name="TOC10" href="#SEC10">VERTICAL BAR</a>
                     26: <li><a name="TOC11" href="#SEC11">INTERNAL OPTION SETTING</a>
                     27: <li><a name="TOC12" href="#SEC12">SUBPATTERNS</a>
                     28: <li><a name="TOC13" href="#SEC13">DUPLICATE SUBPATTERN NUMBERS</a>
                     29: <li><a name="TOC14" href="#SEC14">NAMED SUBPATTERNS</a>
                     30: <li><a name="TOC15" href="#SEC15">REPETITION</a>
                     31: <li><a name="TOC16" href="#SEC16">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a>
                     32: <li><a name="TOC17" href="#SEC17">BACK REFERENCES</a>
                     33: <li><a name="TOC18" href="#SEC18">ASSERTIONS</a>
                     34: <li><a name="TOC19" href="#SEC19">CONDITIONAL SUBPATTERNS</a>
                     35: <li><a name="TOC20" href="#SEC20">COMMENTS</a>
                     36: <li><a name="TOC21" href="#SEC21">RECURSIVE PATTERNS</a>
                     37: <li><a name="TOC22" href="#SEC22">SUBPATTERNS AS SUBROUTINES</a>
                     38: <li><a name="TOC23" href="#SEC23">ONIGURUMA SUBROUTINE SYNTAX</a>
                     39: <li><a name="TOC24" href="#SEC24">CALLOUTS</a>
                     40: <li><a name="TOC25" href="#SEC25">BACKTRACKING CONTROL</a>
                     41: <li><a name="TOC26" href="#SEC26">SEE ALSO</a>
                     42: <li><a name="TOC27" href="#SEC27">AUTHOR</a>
                     43: <li><a name="TOC28" href="#SEC28">REVISION</a>
                     44: </ul>
                     45: <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
                     46: <P>
                     47: The syntax and semantics of the regular expressions that are supported by PCRE
                     48: are described in detail below. There is a quick-reference syntax summary in the
                     49: <a href="pcresyntax.html"><b>pcresyntax</b></a>
                     50: page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
                     51: also supports some alternative regular expression syntax (which does not
                     52: conflict with the Perl syntax) in order to provide some compatibility with
                     53: regular expressions in Python, .NET, and Oniguruma.
                     54: </P>
                     55: <P>
                     56: Perl's regular expressions are described in its own documentation, and
                     57: regular expressions in general are covered in a number of books, some of which
                     58: have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
                     59: published by O'Reilly, covers regular expressions in great detail. This
                     60: description of PCRE's regular expressions is intended as reference material.
                     61: </P>
                     62: <P>
                     63: The original operation of PCRE was on strings of one-byte characters. However,
1.1.1.2 ! misho      64: there is now also support for UTF-8 strings in the original library, and a
        !            65: second library that supports 16-bit and UTF-16 character strings. To use these
        !            66: features, PCRE must be built to include appropriate support. When using UTF
        !            67: strings you must either call the compiling function with the PCRE_UTF8 or
        !            68: PCRE_UTF16 option, or the pattern must start with one of these special
        !            69: sequences:
1.1       misho      70: <pre>
                     71:   (*UTF8)
1.1.1.2 ! misho      72:   (*UTF16)
1.1       misho      73: </pre>
1.1.1.2 ! misho      74: Starting a pattern with such a sequence is equivalent to setting the relevant
        !            75: option. This feature is not Perl-compatible. How setting a UTF mode affects
1.1       misho      76: pattern matching is mentioned in several places below. There is also a summary
1.1.1.2 ! misho      77: of features in the
1.1       misho      78: <a href="pcreunicode.html"><b>pcreunicode</b></a>
                     79: page.
                     80: </P>
                     81: <P>
                     82: Another special sequence that may appear at the start of a pattern or in
1.1.1.2 ! misho      83: combination with (*UTF8) or (*UTF16) is:
1.1       misho      84: <pre>
                     85:   (*UCP)
                     86: </pre>
                     87: This has the same effect as setting the PCRE_UCP option: it causes sequences
                     88: such as \d and \w to use Unicode properties to determine character types,
                     89: instead of recognizing only characters with codes less than 128 via a lookup
                     90: table.
                     91: </P>
                     92: <P>
                     93: If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
                     94: PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
                     95: also some more of these special sequences that are concerned with the handling
                     96: of newlines; they are described below.
                     97: </P>
                     98: <P>
                     99: The remainder of this document discusses the patterns that are supported by
1.1.1.2 ! misho     100: PCRE when one its main matching functions, <b>pcre_exec()</b> (8-bit) or
        !           101: <b>pcre16_exec()</b> (16-bit), is used. PCRE also has alternative matching
        !           102: functions, <b>pcre_dfa_exec()</b> and <b>pcre16_dfa_exec()</b>, which match using
        !           103: a different algorithm that is not Perl-compatible. Some of the features
        !           104: discussed below are not available when DFA matching is used. The advantages and
        !           105: disadvantages of the alternative functions, and how they differ from the normal
        !           106: functions, are discussed in the
1.1       misho     107: <a href="pcrematching.html"><b>pcrematching</b></a>
                    108: page.
                    109: <a name="newlines"></a></P>
                    110: <br><a name="SEC2" href="#TOC1">NEWLINE CONVENTIONS</a><br>
                    111: <P>
                    112: PCRE supports five different conventions for indicating line breaks in
                    113: strings: a single CR (carriage return) character, a single LF (linefeed)
                    114: character, the two-character sequence CRLF, any of the three preceding, or any
                    115: Unicode newline sequence. The
                    116: <a href="pcreapi.html"><b>pcreapi</b></a>
                    117: page has
                    118: <a href="pcreapi.html#newlines">further discussion</a>
                    119: about newlines, and shows how to set the newline convention in the
                    120: <i>options</i> arguments for the compiling and matching functions.
                    121: </P>
                    122: <P>
                    123: It is also possible to specify a newline convention by starting a pattern
                    124: string with one of the following five sequences:
                    125: <pre>
                    126:   (*CR)        carriage return
                    127:   (*LF)        linefeed
                    128:   (*CRLF)      carriage return, followed by linefeed
                    129:   (*ANYCRLF)   any of the three above
                    130:   (*ANY)       all Unicode newline sequences
                    131: </pre>
1.1.1.2 ! misho     132: These override the default and the options given to the compiling function. For
        !           133: example, on a Unix system where LF is the default newline sequence, the pattern
1.1       misho     134: <pre>
                    135:   (*CR)a.b
                    136: </pre>
                    137: changes the convention to CR. That pattern matches "a\nb" because LF is no
                    138: longer a newline. Note that these special settings, which are not
                    139: Perl-compatible, are recognized only at the very start of a pattern, and that
                    140: they must be in upper case. If more than one of them is present, the last one
                    141: is used.
                    142: </P>
                    143: <P>
                    144: The newline convention affects the interpretation of the dot metacharacter when
                    145: PCRE_DOTALL is not set, and also the behaviour of \N. However, it does not
                    146: affect what the \R escape sequence matches. By default, this is any Unicode
                    147: newline sequence, for Perl compatibility. However, this can be changed; see the
                    148: description of \R in the section entitled
                    149: <a href="#newlineseq">"Newline sequences"</a>
                    150: below. A change of \R setting can be combined with a change of newline
                    151: convention.
                    152: </P>
                    153: <br><a name="SEC3" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
                    154: <P>
                    155: A regular expression is a pattern that is matched against a subject string from
                    156: left to right. Most characters stand for themselves in a pattern, and match the
                    157: corresponding characters in the subject. As a trivial example, the pattern
                    158: <pre>
                    159:   The quick brown fox
                    160: </pre>
                    161: matches a portion of a subject string that is identical to itself. When
                    162: caseless matching is specified (the PCRE_CASELESS option), letters are matched
1.1.1.2 ! misho     163: independently of case. In a UTF mode, PCRE always understands the concept of
1.1       misho     164: case for characters whose values are less than 128, so caseless matching is
                    165: always possible. For characters with higher values, the concept of case is
                    166: supported if PCRE is compiled with Unicode property support, but not otherwise.
                    167: If you want to use caseless matching for characters 128 and above, you must
                    168: ensure that PCRE is compiled with Unicode property support as well as with
1.1.1.2 ! misho     169: UTF support.
1.1       misho     170: </P>
                    171: <P>
                    172: The power of regular expressions comes from the ability to include alternatives
                    173: and repetitions in the pattern. These are encoded in the pattern by the use of
                    174: <i>metacharacters</i>, which do not stand for themselves but instead are
                    175: interpreted in some special way.
                    176: </P>
                    177: <P>
                    178: There are two different sets of metacharacters: those that are recognized
                    179: anywhere in the pattern except within square brackets, and those that are
                    180: recognized within square brackets. Outside square brackets, the metacharacters
                    181: are as follows:
                    182: <pre>
                    183:   \      general escape character with several uses
                    184:   ^      assert start of string (or line, in multiline mode)
                    185:   $      assert end of string (or line, in multiline mode)
                    186:   .      match any character except newline (by default)
                    187:   [      start character class definition
                    188:   |      start of alternative branch
                    189:   (      start subpattern
                    190:   )      end subpattern
                    191:   ?      extends the meaning of (
                    192:          also 0 or 1 quantifier
                    193:          also quantifier minimizer
                    194:   *      0 or more quantifier
                    195:   +      1 or more quantifier
                    196:          also "possessive quantifier"
                    197:   {      start min/max quantifier
                    198: </pre>
                    199: Part of a pattern that is in square brackets is called a "character class". In
                    200: a character class the only metacharacters are:
                    201: <pre>
                    202:   \      general escape character
                    203:   ^      negate the class, but only if the first character
                    204:   -      indicates character range
                    205:   [      POSIX character class (only if followed by POSIX syntax)
                    206:   ]      terminates the character class
                    207: </pre>
                    208: The following sections describe the use of each of the metacharacters.
                    209: </P>
                    210: <br><a name="SEC4" href="#TOC1">BACKSLASH</a><br>
                    211: <P>
                    212: The backslash character has several uses. Firstly, if it is followed by a
                    213: character that is not a number or a letter, it takes away any special meaning
                    214: that character may have. This use of backslash as an escape character applies
                    215: both inside and outside character classes.
                    216: </P>
                    217: <P>
                    218: For example, if you want to match a * character, you write \* in the pattern.
                    219: This escaping action applies whether or not the following character would
                    220: otherwise be interpreted as a metacharacter, so it is always safe to precede a
                    221: non-alphanumeric with backslash to specify that it stands for itself. In
                    222: particular, if you want to match a backslash, you write \\.
                    223: </P>
                    224: <P>
1.1.1.2 ! misho     225: In a UTF mode, only ASCII numbers and letters have any special meaning after a
1.1       misho     226: backslash. All other characters (in particular, those whose codepoints are
                    227: greater than 127) are treated as literals.
                    228: </P>
                    229: <P>
                    230: If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
                    231: pattern (other than in a character class) and characters between a # outside
                    232: a character class and the next newline are ignored. An escaping backslash can
                    233: be used to include a whitespace or # character as part of the pattern.
                    234: </P>
                    235: <P>
                    236: If you want to remove the special meaning from a sequence of characters, you
                    237: can do so by putting them between \Q and \E. This is different from Perl in
                    238: that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
                    239: Perl, $ and @ cause variable interpolation. Note the following examples:
                    240: <pre>
                    241:   Pattern            PCRE matches   Perl matches
                    242: 
                    243:   \Qabc$xyz\E        abc$xyz        abc followed by the contents of $xyz
                    244:   \Qabc\$xyz\E       abc\$xyz       abc\$xyz
                    245:   \Qabc\E\$\Qxyz\E   abc$xyz        abc$xyz
                    246: </pre>
                    247: The \Q...\E sequence is recognized both inside and outside character classes.
                    248: An isolated \E that is not preceded by \Q is ignored. If \Q is not followed
                    249: by \E later in the pattern, the literal interpretation continues to the end of
                    250: the pattern (that is, \E is assumed at the end). If the isolated \Q is inside
                    251: a character class, this causes an error, because the character class is not
                    252: terminated.
                    253: <a name="digitsafterbackslash"></a></P>
                    254: <br><b>
                    255: Non-printing characters
                    256: </b><br>
                    257: <P>
                    258: A second use of backslash provides a way of encoding non-printing characters
                    259: in patterns in a visible manner. There is no restriction on the appearance of
                    260: non-printing characters, apart from the binary zero that terminates a pattern,
                    261: but when a pattern is being prepared by text editing, it is often easier to use
                    262: one of the following escape sequences than the binary character it represents:
                    263: <pre>
                    264:   \a        alarm, that is, the BEL character (hex 07)
                    265:   \cx       "control-x", where x is any ASCII character
                    266:   \e        escape (hex 1B)
                    267:   \f        formfeed (hex 0C)
                    268:   \n        linefeed (hex 0A)
                    269:   \r        carriage return (hex 0D)
                    270:   \t        tab (hex 09)
                    271:   \ddd      character with octal code ddd, or back reference
                    272:   \xhh      character with hex code hh
                    273:   \x{hhh..} character with hex code hhh.. (non-JavaScript mode)
                    274:   \uhhhh    character with hex code hhhh (JavaScript mode only)
                    275: </pre>
                    276: The precise effect of \cx is as follows: if x is a lower case letter, it
                    277: is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
                    278: Thus \cz becomes hex 1A (z is 7A), but \c{ becomes hex 3B ({ is 7B), while
                    279: \c; becomes hex 7B (; is 3B). If the byte following \c has a value greater
                    280: than 127, a compile-time error occurs. This locks out non-ASCII characters in
1.1.1.2 ! misho     281: all modes. (When PCRE is compiled in EBCDIC mode, all byte values are valid. A
        !           282: lower case letter is converted to upper case, and then the 0xc0 bits are
        !           283: flipped.)
1.1       misho     284: </P>
                    285: <P>
                    286: By default, after \x, from zero to two hexadecimal digits are read (letters
                    287: can be in upper or lower case). Any number of hexadecimal digits may appear
1.1.1.2 ! misho     288: between \x{ and }, but the character code is constrained as follows:
        !           289: <pre>
        !           290:   8-bit non-UTF mode    less than 0x100
        !           291:   8-bit UTF-8 mode      less than 0x10ffff and a valid codepoint
        !           292:   16-bit non-UTF mode   less than 0x10000
        !           293:   16-bit UTF-16 mode    less than 0x10ffff and a valid codepoint
        !           294: </pre>
        !           295: Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
        !           296: "surrogate" codepoints).
1.1       misho     297: </P>
                    298: <P>
                    299: If characters other than hexadecimal digits appear between \x{ and }, or if
                    300: there is no terminating }, this form of escape is not recognized. Instead, the
                    301: initial \x will be interpreted as a basic hexadecimal escape, with no
                    302: following digits, giving a character whose value is zero.
                    303: </P>
                    304: <P>
                    305: If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \x is
                    306: as just described only when it is followed by two hexadecimal digits.
                    307: Otherwise, it matches a literal "x" character. In JavaScript mode, support for
                    308: code points greater than 256 is provided by \u, which must be followed by
                    309: four hexadecimal digits; otherwise it matches a literal "u" character.
                    310: </P>
                    311: <P>
                    312: Characters whose value is less than 256 can be defined by either of the two
                    313: syntaxes for \x (or by \u in JavaScript mode). There is no difference in the
                    314: way they are handled. For example, \xdc is exactly the same as \x{dc} (or
                    315: \u00dc in JavaScript mode).
                    316: </P>
                    317: <P>
                    318: After \0 up to two further octal digits are read. If there are fewer than two
                    319: digits, just those that are present are used. Thus the sequence \0\x\07
                    320: specifies two binary zeros followed by a BEL character (code value 7). Make
                    321: sure you supply two digits after the initial zero if the pattern character that
                    322: follows is itself an octal digit.
                    323: </P>
                    324: <P>
                    325: The handling of a backslash followed by a digit other than 0 is complicated.
                    326: Outside a character class, PCRE reads it and any following digits as a decimal
                    327: number. If the number is less than 10, or if there have been at least that many
                    328: previous capturing left parentheses in the expression, the entire sequence is
                    329: taken as a <i>back reference</i>. A description of how this works is given
                    330: <a href="#backreferences">later,</a>
                    331: following the discussion of
                    332: <a href="#subpattern">parenthesized subpatterns.</a>
                    333: </P>
                    334: <P>
                    335: Inside a character class, or if the decimal number is greater than 9 and there
                    336: have not been that many capturing subpatterns, PCRE re-reads up to three octal
                    337: digits following the backslash, and uses them to generate a data character. Any
1.1.1.2 ! misho     338: subsequent digits stand for themselves. The value of the character is
        !           339: constrained in the same way as characters specified in hexadecimal.
        !           340: For example:
1.1       misho     341: <pre>
                    342:   \040   is another way of writing a space
                    343:   \40    is the same, provided there are fewer than 40 previous capturing subpatterns
                    344:   \7     is always a back reference
                    345:   \11    might be a back reference, or another way of writing a tab
                    346:   \011   is always a tab
                    347:   \0113  is a tab followed by the character "3"
                    348:   \113   might be a back reference, otherwise the character with octal code 113
1.1.1.2 ! misho     349:   \377   might be a back reference, otherwise the value 255 (decimal)
1.1       misho     350:   \81    is either a back reference, or a binary zero followed by the two characters "8" and "1"
                    351: </pre>
                    352: Note that octal values of 100 or greater must not be introduced by a leading
                    353: zero, because no more than three octal digits are ever read.
                    354: </P>
                    355: <P>
                    356: All the sequences that define a single character value can be used both inside
                    357: and outside character classes. In addition, inside a character class, \b is
                    358: interpreted as the backspace character (hex 08).
                    359: </P>
                    360: <P>
                    361: \N is not allowed in a character class. \B, \R, and \X are not special
                    362: inside a character class. Like other unrecognized escape sequences, they are
                    363: treated as the literal characters "B", "R", and "X" by default, but cause an
                    364: error if the PCRE_EXTRA option is set. Outside a character class, these
                    365: sequences have different meanings.
                    366: </P>
                    367: <br><b>
                    368: Unsupported escape sequences
                    369: </b><br>
                    370: <P>
                    371: In Perl, the sequences \l, \L, \u, and \U are recognized by its string
                    372: handler and used to modify the case of following characters. By default, PCRE
                    373: does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
                    374: option is set, \U matches a "U" character, and \u can be used to define a
                    375: character by code point, as described in the previous section.
                    376: </P>
                    377: <br><b>
                    378: Absolute and relative back references
                    379: </b><br>
                    380: <P>
                    381: The sequence \g followed by an unsigned or a negative number, optionally
                    382: enclosed in braces, is an absolute or relative back reference. A named back
                    383: reference can be coded as \g{name}. Back references are discussed
                    384: <a href="#backreferences">later,</a>
                    385: following the discussion of
                    386: <a href="#subpattern">parenthesized subpatterns.</a>
                    387: </P>
                    388: <br><b>
                    389: Absolute and relative subroutine calls
                    390: </b><br>
                    391: <P>
                    392: For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
                    393: a number enclosed either in angle brackets or single quotes, is an alternative
                    394: syntax for referencing a subpattern as a "subroutine". Details are discussed
                    395: <a href="#onigurumasubroutines">later.</a>
                    396: Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
                    397: synonymous. The former is a back reference; the latter is a
                    398: <a href="#subpatternsassubroutines">subroutine</a>
                    399: call.
                    400: <a name="genericchartypes"></a></P>
                    401: <br><b>
                    402: Generic character types
                    403: </b><br>
                    404: <P>
                    405: Another use of backslash is for specifying generic character types:
                    406: <pre>
                    407:   \d     any decimal digit
                    408:   \D     any character that is not a decimal digit
                    409:   \h     any horizontal whitespace character
                    410:   \H     any character that is not a horizontal whitespace character
                    411:   \s     any whitespace character
                    412:   \S     any character that is not a whitespace character
                    413:   \v     any vertical whitespace character
                    414:   \V     any character that is not a vertical whitespace character
                    415:   \w     any "word" character
                    416:   \W     any "non-word" character
                    417: </pre>
                    418: There is also the single sequence \N, which matches a non-newline character.
                    419: This is the same as
                    420: <a href="#fullstopdot">the "." metacharacter</a>
                    421: when PCRE_DOTALL is not set. Perl also uses \N to match characters by name;
                    422: PCRE does not support this.
                    423: </P>
                    424: <P>
                    425: Each pair of lower and upper case escape sequences partitions the complete set
                    426: of characters into two disjoint sets. Any given character matches one, and only
                    427: one, of each pair. The sequences can appear both inside and outside character
                    428: classes. They each match one character of the appropriate type. If the current
                    429: matching point is at the end of the subject string, all of them fail, because
                    430: there is no character to match.
                    431: </P>
                    432: <P>
                    433: For compatibility with Perl, \s does not match the VT character (code 11).
                    434: This makes it different from the the POSIX "space" class. The \s characters
                    435: are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
                    436: included in a Perl script, \s may match the VT character. In PCRE, it never
                    437: does.
                    438: </P>
                    439: <P>
                    440: A "word" character is an underscore or any character that is a letter or digit.
                    441: By default, the definition of letters and digits is controlled by PCRE's
                    442: low-valued character tables, and may vary if locale-specific matching is taking
                    443: place (see
                    444: <a href="pcreapi.html#localesupport">"Locale support"</a>
                    445: in the
                    446: <a href="pcreapi.html"><b>pcreapi</b></a>
                    447: page). For example, in a French locale such as "fr_FR" in Unix-like systems,
                    448: or "french" in Windows, some character codes greater than 128 are used for
                    449: accented letters, and these are then matched by \w. The use of locales with
                    450: Unicode is discouraged.
                    451: </P>
                    452: <P>
1.1.1.2 ! misho     453: By default, in a UTF mode, characters with values greater than 128 never match
1.1       misho     454: \d, \s, or \w, and always match \D, \S, and \W. These sequences retain
1.1.1.2 ! misho     455: their original meanings from before UTF support was available, mainly for
1.1       misho     456: efficiency reasons. However, if PCRE is compiled with Unicode property support,
                    457: and the PCRE_UCP option is set, the behaviour is changed so that Unicode
                    458: properties are used to determine character types, as follows:
                    459: <pre>
                    460:   \d  any character that \p{Nd} matches (decimal digit)
                    461:   \s  any character that \p{Z} matches, plus HT, LF, FF, CR
                    462:   \w  any character that \p{L} or \p{N} matches, plus underscore
                    463: </pre>
                    464: The upper case escapes match the inverse sets of characters. Note that \d
                    465: matches only decimal digits, whereas \w matches any Unicode digit, as well as
                    466: any Unicode letter, and underscore. Note also that PCRE_UCP affects \b, and
                    467: \B because they are defined in terms of \w and \W. Matching these sequences
                    468: is noticeably slower when PCRE_UCP is set.
                    469: </P>
                    470: <P>
                    471: The sequences \h, \H, \v, and \V are features that were added to Perl at
                    472: release 5.10. In contrast to the other sequences, which match only ASCII
1.1.1.2 ! misho     473: characters by default, these always match certain high-valued codepoints,
        !           474: whether or not PCRE_UCP is set. The horizontal space characters are:
1.1       misho     475: <pre>
                    476:   U+0009     Horizontal tab
                    477:   U+0020     Space
                    478:   U+00A0     Non-break space
                    479:   U+1680     Ogham space mark
                    480:   U+180E     Mongolian vowel separator
                    481:   U+2000     En quad
                    482:   U+2001     Em quad
                    483:   U+2002     En space
                    484:   U+2003     Em space
                    485:   U+2004     Three-per-em space
                    486:   U+2005     Four-per-em space
                    487:   U+2006     Six-per-em space
                    488:   U+2007     Figure space
                    489:   U+2008     Punctuation space
                    490:   U+2009     Thin space
                    491:   U+200A     Hair space
                    492:   U+202F     Narrow no-break space
                    493:   U+205F     Medium mathematical space
                    494:   U+3000     Ideographic space
                    495: </pre>
                    496: The vertical space characters are:
                    497: <pre>
                    498:   U+000A     Linefeed
                    499:   U+000B     Vertical tab
                    500:   U+000C     Formfeed
                    501:   U+000D     Carriage return
                    502:   U+0085     Next line
                    503:   U+2028     Line separator
                    504:   U+2029     Paragraph separator
1.1.1.2 ! misho     505: </pre>
        !           506: In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
        !           507: relevant.
        !           508: <a name="newlineseq"></a></P>
1.1       misho     509: <br><b>
                    510: Newline sequences
                    511: </b><br>
                    512: <P>
                    513: Outside a character class, by default, the escape sequence \R matches any
1.1.1.2 ! misho     514: Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent to the
        !           515: following:
1.1       misho     516: <pre>
                    517:   (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
                    518: </pre>
                    519: This is an example of an "atomic group", details of which are given
                    520: <a href="#atomicgroup">below.</a>
                    521: This particular group matches either the two-character sequence CR followed by
                    522: LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
                    523: U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
                    524: line, U+0085). The two-character sequence is treated as a single unit that
                    525: cannot be split.
                    526: </P>
                    527: <P>
1.1.1.2 ! misho     528: In other modes, two additional characters whose codepoints are greater than 255
1.1       misho     529: are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
                    530: Unicode character property support is not needed for these characters to be
                    531: recognized.
                    532: </P>
                    533: <P>
                    534: It is possible to restrict \R to match only CR, LF, or CRLF (instead of the
                    535: complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
                    536: either at compile time or when the pattern is matched. (BSR is an abbrevation
                    537: for "backslash R".) This can be made the default when PCRE is built; if this is
                    538: the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
                    539: It is also possible to specify these settings by starting a pattern string with
                    540: one of the following sequences:
                    541: <pre>
                    542:   (*BSR_ANYCRLF)   CR, LF, or CRLF only
                    543:   (*BSR_UNICODE)   any Unicode newline sequence
                    544: </pre>
1.1.1.2 ! misho     545: These override the default and the options given to the compiling function, but
        !           546: they can themselves be overridden by options given to a matching function. Note
        !           547: that these special settings, which are not Perl-compatible, are recognized only
        !           548: at the very start of a pattern, and that they must be in upper case. If more
        !           549: than one of them is present, the last one is used. They can be combined with a
        !           550: change of newline convention; for example, a pattern can start with:
1.1       misho     551: <pre>
                    552:   (*ANY)(*BSR_ANYCRLF)
                    553: </pre>
1.1.1.2 ! misho     554: They can also be combined with the (*UTF8), (*UTF16), or (*UCP) special
        !           555: sequences. Inside a character class, \R is treated as an unrecognized escape
        !           556: sequence, and so matches the letter "R" by default, but causes an error if
        !           557: PCRE_EXTRA is set.
1.1       misho     558: <a name="uniextseq"></a></P>
                    559: <br><b>
                    560: Unicode character properties
                    561: </b><br>
                    562: <P>
                    563: When PCRE is built with Unicode character property support, three additional
                    564: escape sequences that match characters with specific properties are available.
1.1.1.2 ! misho     565: When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
1.1       misho     566: characters whose codepoints are less than 256, but they do work in this mode.
                    567: The extra escape sequences are:
                    568: <pre>
                    569:   \p{<i>xx</i>}   a character with the <i>xx</i> property
                    570:   \P{<i>xx</i>}   a character without the <i>xx</i> property
                    571:   \X       an extended Unicode sequence
                    572: </pre>
                    573: The property names represented by <i>xx</i> above are limited to the Unicode
                    574: script names, the general category properties, "Any", which matches any
                    575: character (including newline), and some special PCRE properties (described
                    576: in the
                    577: <a href="#extraprops">next section).</a>
                    578: Other Perl properties such as "InMusicalSymbols" are not currently supported by
                    579: PCRE. Note that \P{Any} does not match any characters, so always causes a
                    580: match failure.
                    581: </P>
                    582: <P>
                    583: Sets of Unicode characters are defined as belonging to certain scripts. A
                    584: character from one of these sets can be matched using a script name. For
                    585: example:
                    586: <pre>
                    587:   \p{Greek}
                    588:   \P{Han}
                    589: </pre>
                    590: Those that are not part of an identified script are lumped together as
                    591: "Common". The current list of scripts is:
                    592: </P>
                    593: <P>
                    594: Arabic,
                    595: Armenian,
                    596: Avestan,
                    597: Balinese,
                    598: Bamum,
                    599: Bengali,
                    600: Bopomofo,
                    601: Braille,
                    602: Buginese,
                    603: Buhid,
                    604: Canadian_Aboriginal,
                    605: Carian,
                    606: Cham,
                    607: Cherokee,
                    608: Common,
                    609: Coptic,
                    610: Cuneiform,
                    611: Cypriot,
                    612: Cyrillic,
                    613: Deseret,
                    614: Devanagari,
                    615: Egyptian_Hieroglyphs,
                    616: Ethiopic,
                    617: Georgian,
                    618: Glagolitic,
                    619: Gothic,
                    620: Greek,
                    621: Gujarati,
                    622: Gurmukhi,
                    623: Han,
                    624: Hangul,
                    625: Hanunoo,
                    626: Hebrew,
                    627: Hiragana,
                    628: Imperial_Aramaic,
                    629: Inherited,
                    630: Inscriptional_Pahlavi,
                    631: Inscriptional_Parthian,
                    632: Javanese,
                    633: Kaithi,
                    634: Kannada,
                    635: Katakana,
                    636: Kayah_Li,
                    637: Kharoshthi,
                    638: Khmer,
                    639: Lao,
                    640: Latin,
                    641: Lepcha,
                    642: Limbu,
                    643: Linear_B,
                    644: Lisu,
                    645: Lycian,
                    646: Lydian,
                    647: Malayalam,
                    648: Meetei_Mayek,
                    649: Mongolian,
                    650: Myanmar,
                    651: New_Tai_Lue,
                    652: Nko,
                    653: Ogham,
                    654: Old_Italic,
                    655: Old_Persian,
                    656: Old_South_Arabian,
                    657: Old_Turkic,
                    658: Ol_Chiki,
                    659: Oriya,
                    660: Osmanya,
                    661: Phags_Pa,
                    662: Phoenician,
                    663: Rejang,
                    664: Runic,
                    665: Samaritan,
                    666: Saurashtra,
                    667: Shavian,
                    668: Sinhala,
                    669: Sundanese,
                    670: Syloti_Nagri,
                    671: Syriac,
                    672: Tagalog,
                    673: Tagbanwa,
                    674: Tai_Le,
                    675: Tai_Tham,
                    676: Tai_Viet,
                    677: Tamil,
                    678: Telugu,
                    679: Thaana,
                    680: Thai,
                    681: Tibetan,
                    682: Tifinagh,
                    683: Ugaritic,
                    684: Vai,
                    685: Yi.
                    686: </P>
                    687: <P>
                    688: Each character has exactly one Unicode general category property, specified by
                    689: a two-letter abbreviation. For compatibility with Perl, negation can be
                    690: specified by including a circumflex between the opening brace and the property
                    691: name. For example, \p{^Lu} is the same as \P{Lu}.
                    692: </P>
                    693: <P>
                    694: If only one letter is specified with \p or \P, it includes all the general
                    695: category properties that start with that letter. In this case, in the absence
                    696: of negation, the curly brackets in the escape sequence are optional; these two
                    697: examples have the same effect:
                    698: <pre>
                    699:   \p{L}
                    700:   \pL
                    701: </pre>
                    702: The following general category property codes are supported:
                    703: <pre>
                    704:   C     Other
                    705:   Cc    Control
                    706:   Cf    Format
                    707:   Cn    Unassigned
                    708:   Co    Private use
                    709:   Cs    Surrogate
                    710: 
                    711:   L     Letter
                    712:   Ll    Lower case letter
                    713:   Lm    Modifier letter
                    714:   Lo    Other letter
                    715:   Lt    Title case letter
                    716:   Lu    Upper case letter
                    717: 
                    718:   M     Mark
                    719:   Mc    Spacing mark
                    720:   Me    Enclosing mark
                    721:   Mn    Non-spacing mark
                    722: 
                    723:   N     Number
                    724:   Nd    Decimal number
                    725:   Nl    Letter number
                    726:   No    Other number
                    727: 
                    728:   P     Punctuation
                    729:   Pc    Connector punctuation
                    730:   Pd    Dash punctuation
                    731:   Pe    Close punctuation
                    732:   Pf    Final punctuation
                    733:   Pi    Initial punctuation
                    734:   Po    Other punctuation
                    735:   Ps    Open punctuation
                    736: 
                    737:   S     Symbol
                    738:   Sc    Currency symbol
                    739:   Sk    Modifier symbol
                    740:   Sm    Mathematical symbol
                    741:   So    Other symbol
                    742: 
                    743:   Z     Separator
                    744:   Zl    Line separator
                    745:   Zp    Paragraph separator
                    746:   Zs    Space separator
                    747: </pre>
                    748: The special property L& is also supported: it matches a character that has
                    749: the Lu, Ll, or Lt property, in other words, a letter that is not classified as
                    750: a modifier or "other".
                    751: </P>
                    752: <P>
                    753: The Cs (Surrogate) property applies only to characters in the range U+D800 to
1.1.1.2 ! misho     754: U+DFFF. Such characters are not valid in Unicode strings and so
        !           755: cannot be tested by PCRE, unless UTF validity checking has been turned off
        !           756: (see the discussion of PCRE_NO_UTF8_CHECK and PCRE_NO_UTF16_CHECK in the
1.1       misho     757: <a href="pcreapi.html"><b>pcreapi</b></a>
                    758: page). Perl does not support the Cs property.
                    759: </P>
                    760: <P>
                    761: The long synonyms for property names that Perl supports (such as \p{Letter})
                    762: are not supported by PCRE, nor is it permitted to prefix any of these
                    763: properties with "Is".
                    764: </P>
                    765: <P>
                    766: No character that is in the Unicode table has the Cn (unassigned) property.
                    767: Instead, this property is assumed for any code point that is not in the
                    768: Unicode table.
                    769: </P>
                    770: <P>
                    771: Specifying caseless matching does not affect these escape sequences. For
                    772: example, \p{Lu} always matches only upper case letters.
                    773: </P>
                    774: <P>
                    775: The \X escape matches any number of Unicode characters that form an extended
                    776: Unicode sequence. \X is equivalent to
                    777: <pre>
                    778:   (?&#62;\PM\pM*)
                    779: </pre>
                    780: That is, it matches a character without the "mark" property, followed by zero
                    781: or more characters with the "mark" property, and treats the sequence as an
                    782: atomic group
                    783: <a href="#atomicgroup">(see below).</a>
                    784: Characters with the "mark" property are typically accents that affect the
                    785: preceding character. None of them have codepoints less than 256, so in
1.1.1.2 ! misho     786: 8-bit non-UTF-8 mode \X matches any one character.
1.1       misho     787: </P>
                    788: <P>
                    789: Note that recent versions of Perl have changed \X to match what Unicode calls
                    790: an "extended grapheme cluster", which has a more complicated definition.
                    791: </P>
                    792: <P>
                    793: Matching characters by Unicode property is not fast, because PCRE has to search
                    794: a structure that contains data for over fifteen thousand characters. That is
                    795: why the traditional escape sequences such as \d and \w do not use Unicode
                    796: properties in PCRE by default, though you can make them do so by setting the
1.1.1.2 ! misho     797: PCRE_UCP option or by starting the pattern with (*UCP).
1.1       misho     798: <a name="extraprops"></a></P>
                    799: <br><b>
                    800: PCRE's additional properties
                    801: </b><br>
                    802: <P>
                    803: As well as the standard Unicode properties described in the previous
                    804: section, PCRE supports four more that make it possible to convert traditional
                    805: escape sequences such as \w and \s and POSIX character classes to use Unicode
                    806: properties. PCRE uses these non-standard, non-Perl properties internally when
                    807: PCRE_UCP is set. They are:
                    808: <pre>
                    809:   Xan   Any alphanumeric character
                    810:   Xps   Any POSIX space character
                    811:   Xsp   Any Perl space character
                    812:   Xwd   Any Perl "word" character
                    813: </pre>
                    814: Xan matches characters that have either the L (letter) or the N (number)
                    815: property. Xps matches the characters tab, linefeed, vertical tab, formfeed, or
                    816: carriage return, and any other character that has the Z (separator) property.
                    817: Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
                    818: same characters as Xan, plus underscore.
                    819: <a name="resetmatchstart"></a></P>
                    820: <br><b>
                    821: Resetting the match start
                    822: </b><br>
                    823: <P>
                    824: The escape sequence \K causes any previously matched characters not to be
                    825: included in the final matched sequence. For example, the pattern:
                    826: <pre>
                    827:   foo\Kbar
                    828: </pre>
                    829: matches "foobar", but reports that it has matched "bar". This feature is
                    830: similar to a lookbehind assertion
                    831: <a href="#lookbehind">(described below).</a>
                    832: However, in this case, the part of the subject before the real match does not
                    833: have to be of fixed length, as lookbehind assertions do. The use of \K does
                    834: not interfere with the setting of
                    835: <a href="#subpattern">captured substrings.</a>
                    836: For example, when the pattern
                    837: <pre>
                    838:   (foo)\Kbar
                    839: </pre>
                    840: matches "foobar", the first substring is still set to "foo".
                    841: </P>
                    842: <P>
                    843: Perl documents that the use of \K within assertions is "not well defined". In
                    844: PCRE, \K is acted upon when it occurs inside positive assertions, but is
                    845: ignored in negative assertions.
                    846: <a name="smallassertions"></a></P>
                    847: <br><b>
                    848: Simple assertions
                    849: </b><br>
                    850: <P>
                    851: The final use of backslash is for certain simple assertions. An assertion
                    852: specifies a condition that has to be met at a particular point in a match,
                    853: without consuming any characters from the subject string. The use of
                    854: subpatterns for more complicated assertions is described
                    855: <a href="#bigassertions">below.</a>
                    856: The backslashed assertions are:
                    857: <pre>
                    858:   \b     matches at a word boundary
                    859:   \B     matches when not at a word boundary
                    860:   \A     matches at the start of the subject
                    861:   \Z     matches at the end of the subject
                    862:           also matches before a newline at the end of the subject
                    863:   \z     matches only at the end of the subject
                    864:   \G     matches at the first matching position in the subject
                    865: </pre>
                    866: Inside a character class, \b has a different meaning; it matches the backspace
                    867: character. If any other of these assertions appears in a character class, by
                    868: default it matches the corresponding literal character (for example, \B
                    869: matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
                    870: escape sequence" error is generated instead.
                    871: </P>
                    872: <P>
                    873: A word boundary is a position in the subject string where the current character
                    874: and the previous character do not both match \w or \W (i.e. one matches
                    875: \w and the other matches \W), or the start or end of the string if the
1.1.1.2 ! misho     876: first or last character matches \w, respectively. In a UTF mode, the meanings
1.1       misho     877: of \w and \W can be changed by setting the PCRE_UCP option. When this is
                    878: done, it also affects \b and \B. Neither PCRE nor Perl has a separate "start
                    879: of word" or "end of word" metasequence. However, whatever follows \b normally
                    880: determines which it is. For example, the fragment \ba matches "a" at the start
                    881: of a word.
                    882: </P>
                    883: <P>
                    884: The \A, \Z, and \z assertions differ from the traditional circumflex and
                    885: dollar (described in the next section) in that they only ever match at the very
                    886: start and end of the subject string, whatever options are set. Thus, they are
                    887: independent of multiline mode. These three assertions are not affected by the
                    888: PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
                    889: circumflex and dollar metacharacters. However, if the <i>startoffset</i>
                    890: argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
                    891: at a point other than the beginning of the subject, \A can never match. The
                    892: difference between \Z and \z is that \Z matches before a newline at the end
                    893: of the string as well as at the very end, whereas \z matches only at the end.
                    894: </P>
                    895: <P>
                    896: The \G assertion is true only when the current matching position is at the
                    897: start point of the match, as specified by the <i>startoffset</i> argument of
                    898: <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
                    899: non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
                    900: arguments, you can mimic Perl's /g option, and it is in this kind of
                    901: implementation where \G can be useful.
                    902: </P>
                    903: <P>
                    904: Note, however, that PCRE's interpretation of \G, as the start of the current
                    905: match, is subtly different from Perl's, which defines it as the end of the
                    906: previous match. In Perl, these can be different when the previously matched
                    907: string was empty. Because PCRE does just one match at a time, it cannot
                    908: reproduce this behaviour.
                    909: </P>
                    910: <P>
                    911: If all the alternatives of a pattern begin with \G, the expression is anchored
                    912: to the starting match position, and the "anchored" flag is set in the compiled
                    913: regular expression.
                    914: </P>
                    915: <br><a name="SEC5" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
                    916: <P>
                    917: Outside a character class, in the default matching mode, the circumflex
                    918: character is an assertion that is true only if the current matching point is
                    919: at the start of the subject string. If the <i>startoffset</i> argument of
                    920: <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
                    921: option is unset. Inside a character class, circumflex has an entirely different
                    922: meaning
                    923: <a href="#characterclass">(see below).</a>
                    924: </P>
                    925: <P>
                    926: Circumflex need not be the first character of the pattern if a number of
                    927: alternatives are involved, but it should be the first thing in each alternative
                    928: in which it appears if the pattern is ever to match that branch. If all
                    929: possible alternatives start with a circumflex, that is, if the pattern is
                    930: constrained to match only at the start of the subject, it is said to be an
                    931: "anchored" pattern. (There are also other constructs that can cause a pattern
                    932: to be anchored.)
                    933: </P>
                    934: <P>
                    935: A dollar character is an assertion that is true only if the current matching
                    936: point is at the end of the subject string, or immediately before a newline
                    937: at the end of the string (by default). Dollar need not be the last character of
                    938: the pattern if a number of alternatives are involved, but it should be the last
                    939: item in any branch in which it appears. Dollar has no special meaning in a
                    940: character class.
                    941: </P>
                    942: <P>
                    943: The meaning of dollar can be changed so that it matches only at the very end of
                    944: the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
                    945: does not affect the \Z assertion.
                    946: </P>
                    947: <P>
                    948: The meanings of the circumflex and dollar characters are changed if the
                    949: PCRE_MULTILINE option is set. When this is the case, a circumflex matches
                    950: immediately after internal newlines as well as at the start of the subject
                    951: string. It does not match after a newline that ends the string. A dollar
                    952: matches before any newlines in the string, as well as at the very end, when
                    953: PCRE_MULTILINE is set. When newline is specified as the two-character
                    954: sequence CRLF, isolated CR and LF characters do not indicate newlines.
                    955: </P>
                    956: <P>
                    957: For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
                    958: \n represents a newline) in multiline mode, but not otherwise. Consequently,
                    959: patterns that are anchored in single line mode because all branches start with
                    960: ^ are not anchored in multiline mode, and a match for circumflex is possible
                    961: when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
                    962: PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
                    963: </P>
                    964: <P>
                    965: Note that the sequences \A, \Z, and \z can be used to match the start and
                    966: end of the subject in both modes, and if all branches of a pattern start with
                    967: \A it is always anchored, whether or not PCRE_MULTILINE is set.
                    968: <a name="fullstopdot"></a></P>
                    969: <br><a name="SEC6" href="#TOC1">FULL STOP (PERIOD, DOT) AND \N</a><br>
                    970: <P>
                    971: Outside a character class, a dot in the pattern matches any one character in
                    972: the subject string except (by default) a character that signifies the end of a
1.1.1.2 ! misho     973: line.
1.1       misho     974: </P>
                    975: <P>
                    976: When a line ending is defined as a single character, dot never matches that
                    977: character; when the two-character sequence CRLF is used, dot does not match CR
                    978: if it is immediately followed by LF, but otherwise it matches all characters
                    979: (including isolated CRs and LFs). When any Unicode line endings are being
                    980: recognized, dot does not match CR or LF or any of the other line ending
                    981: characters.
                    982: </P>
                    983: <P>
                    984: The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
                    985: option is set, a dot matches any one character, without exception. If the
                    986: two-character sequence CRLF is present in the subject string, it takes two dots
                    987: to match it.
                    988: </P>
                    989: <P>
                    990: The handling of dot is entirely independent of the handling of circumflex and
                    991: dollar, the only relationship being that they both involve newlines. Dot has no
                    992: special meaning in a character class.
                    993: </P>
                    994: <P>
                    995: The escape sequence \N behaves like a dot, except that it is not affected by
                    996: the PCRE_DOTALL option. In other words, it matches any character except one
                    997: that signifies the end of a line. Perl also uses \N to match characters by
                    998: name; PCRE does not support this.
                    999: </P>
1.1.1.2 ! misho    1000: <br><a name="SEC7" href="#TOC1">MATCHING A SINGLE DATA UNIT</a><br>
1.1       misho    1001: <P>
1.1.1.2 ! misho    1002: Outside a character class, the escape sequence \C matches any one data unit,
        !          1003: whether or not a UTF mode is set. In the 8-bit library, one data unit is one
        !          1004: byte; in the 16-bit library it is a 16-bit unit. Unlike a dot, \C always
        !          1005: matches line-ending characters. The feature is provided in Perl in order to
        !          1006: match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
        !          1007: used. Because \C breaks up characters into individual data units, matching one
        !          1008: unit with \C in a UTF mode means that the rest of the string may start with a
        !          1009: malformed UTF character. This has undefined results, because PCRE assumes that
        !          1010: it is dealing with valid UTF strings (and by default it checks this at the
        !          1011: start of processing unless the PCRE_NO_UTF8_CHECK option is used).
1.1       misho    1012: </P>
                   1013: <P>
                   1014: PCRE does not allow \C to appear in lookbehind assertions
                   1015: <a href="#lookbehind">(described below)</a>
1.1.1.2 ! misho    1016: in a UTF mode, because this would make it impossible to calculate the length of
1.1       misho    1017: the lookbehind.
                   1018: </P>
                   1019: <P>
1.1.1.2 ! misho    1020: In general, the \C escape sequence is best avoided. However, one
        !          1021: way of using it that avoids the problem of malformed UTF characters is to use a
        !          1022: lookahead to check the length of the next character, as in this pattern, which
        !          1023: could be used with a UTF-8 string (ignore white space and line breaks):
1.1       misho    1024: <pre>
                   1025:   (?| (?=[\x00-\x7f])(\C) |
                   1026:       (?=[\x80-\x{7ff}])(\C)(\C) |
                   1027:       (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) |
                   1028:       (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C))
                   1029: </pre>
                   1030: A group that starts with (?| resets the capturing parentheses numbers in each
                   1031: alternative (see
                   1032: <a href="#dupsubpatternnumber">"Duplicate Subpattern Numbers"</a>
                   1033: below). The assertions at the start of each branch check the next UTF-8
                   1034: character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
                   1035: character's individual bytes are then captured by the appropriate number of
                   1036: groups.
                   1037: <a name="characterclass"></a></P>
                   1038: <br><a name="SEC8" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
                   1039: <P>
                   1040: An opening square bracket introduces a character class, terminated by a closing
                   1041: square bracket. A closing square bracket on its own is not special by default.
                   1042: However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
                   1043: bracket causes a compile-time error. If a closing square bracket is required as
                   1044: a member of the class, it should be the first data character in the class
                   1045: (after an initial circumflex, if present) or escaped with a backslash.
                   1046: </P>
                   1047: <P>
1.1.1.2 ! misho    1048: A character class matches a single character in the subject. In a UTF mode, the
        !          1049: character may be more than one data unit long. A matched character must be in
        !          1050: the set of characters defined by the class, unless the first character in the
        !          1051: class definition is a circumflex, in which case the subject character must not
        !          1052: be in the set defined by the class. If a circumflex is actually required as a
        !          1053: member of the class, ensure it is not the first character, or escape it with a
1.1       misho    1054: backslash.
                   1055: </P>
                   1056: <P>
                   1057: For example, the character class [aeiou] matches any lower case vowel, while
                   1058: [^aeiou] matches any character that is not a lower case vowel. Note that a
                   1059: circumflex is just a convenient notation for specifying the characters that
                   1060: are in the class by enumerating those that are not. A class that starts with a
                   1061: circumflex is not an assertion; it still consumes a character from the subject
                   1062: string, and therefore it fails if the current pointer is at the end of the
                   1063: string.
                   1064: </P>
                   1065: <P>
1.1.1.2 ! misho    1066: In UTF-8 (UTF-16) mode, characters with values greater than 255 (0xffff) can be
        !          1067: included in a class as a literal string of data units, or by using the \x{
        !          1068: escaping mechanism.
1.1       misho    1069: </P>
                   1070: <P>
                   1071: When caseless matching is set, any letters in a class represent both their
                   1072: upper case and lower case versions, so for example, a caseless [aeiou] matches
                   1073: "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1.1.1.2 ! misho    1074: caseful version would. In a UTF mode, PCRE always understands the concept of
1.1       misho    1075: case for characters whose values are less than 128, so caseless matching is
                   1076: always possible. For characters with higher values, the concept of case is
                   1077: supported if PCRE is compiled with Unicode property support, but not otherwise.
1.1.1.2 ! misho    1078: If you want to use caseless matching in a UTF mode for characters 128 and
        !          1079: above, you must ensure that PCRE is compiled with Unicode property support as
        !          1080: well as with UTF support.
1.1       misho    1081: </P>
                   1082: <P>
                   1083: Characters that might indicate line breaks are never treated in any special way
                   1084: when matching character classes, whatever line-ending sequence is in use, and
                   1085: whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
                   1086: such as [^a] always matches one of these characters.
                   1087: </P>
                   1088: <P>
                   1089: The minus (hyphen) character can be used to specify a range of characters in a
                   1090: character class. For example, [d-m] matches any letter between d and m,
                   1091: inclusive. If a minus character is required in a class, it must be escaped with
                   1092: a backslash or appear in a position where it cannot be interpreted as
                   1093: indicating a range, typically as the first or last character in the class.
                   1094: </P>
                   1095: <P>
                   1096: It is not possible to have the literal character "]" as the end character of a
                   1097: range. A pattern such as [W-]46] is interpreted as a class of two characters
                   1098: ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
                   1099: "-46]". However, if the "]" is escaped with a backslash it is interpreted as
                   1100: the end of range, so [W-\]46] is interpreted as a class containing a range
                   1101: followed by two other characters. The octal or hexadecimal representation of
                   1102: "]" can also be used to end a range.
                   1103: </P>
                   1104: <P>
                   1105: Ranges operate in the collating sequence of character values. They can also be
1.1.1.2 ! misho    1106: used for characters specified numerically, for example [\000-\037]. Ranges
        !          1107: can include any characters that are valid for the current mode.
1.1       misho    1108: </P>
                   1109: <P>
                   1110: If a range that includes letters is used when caseless matching is set, it
                   1111: matches the letters in either case. For example, [W-c] is equivalent to
1.1.1.2 ! misho    1112: [][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1.1       misho    1113: tables for a French locale are in use, [\xc8-\xcb] matches accented E
1.1.1.2 ! misho    1114: characters in both cases. In UTF modes, PCRE supports the concept of case for
1.1       misho    1115: characters with values greater than 128 only when it is compiled with Unicode
                   1116: property support.
                   1117: </P>
                   1118: <P>
                   1119: The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v,
                   1120: \V, \w, and \W may appear in a character class, and add the characters that
                   1121: they match to the class. For example, [\dABCDEF] matches any hexadecimal
1.1.1.2 ! misho    1122: digit. In UTF modes, the PCRE_UCP option affects the meanings of \d, \s, \w
1.1       misho    1123: and their upper case partners, just as it does when they appear outside a
                   1124: character class, as described in the section entitled
                   1125: <a href="#genericchartypes">"Generic character types"</a>
                   1126: above. The escape sequence \b has a different meaning inside a character
                   1127: class; it matches the backspace character. The sequences \B, \N, \R, and \X
                   1128: are not special inside a character class. Like any other unrecognized escape
                   1129: sequences, they are treated as the literal characters "B", "N", "R", and "X" by
                   1130: default, but cause an error if the PCRE_EXTRA option is set.
                   1131: </P>
                   1132: <P>
                   1133: A circumflex can conveniently be used with the upper case character types to
                   1134: specify a more restricted set of characters than the matching lower case type.
                   1135: For example, the class [^\W_] matches any letter or digit, but not underscore,
                   1136: whereas [\w] includes underscore. A positive character class should be read as
                   1137: "something OR something OR ..." and a negative class as "NOT something AND NOT
                   1138: something AND NOT ...".
                   1139: </P>
                   1140: <P>
                   1141: The only metacharacters that are recognized in character classes are backslash,
                   1142: hyphen (only where it can be interpreted as specifying a range), circumflex
                   1143: (only at the start), opening square bracket (only when it can be interpreted as
                   1144: introducing a POSIX class name - see the next section), and the terminating
                   1145: closing square bracket. However, escaping other non-alphanumeric characters
                   1146: does no harm.
                   1147: </P>
                   1148: <br><a name="SEC9" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
                   1149: <P>
                   1150: Perl supports the POSIX notation for character classes. This uses names
                   1151: enclosed by [: and :] within the enclosing square brackets. PCRE also supports
                   1152: this notation. For example,
                   1153: <pre>
                   1154:   [01[:alpha:]%]
                   1155: </pre>
                   1156: matches "0", "1", any alphabetic character, or "%". The supported class names
                   1157: are:
                   1158: <pre>
                   1159:   alnum    letters and digits
                   1160:   alpha    letters
                   1161:   ascii    character codes 0 - 127
                   1162:   blank    space or tab only
                   1163:   cntrl    control characters
                   1164:   digit    decimal digits (same as \d)
                   1165:   graph    printing characters, excluding space
                   1166:   lower    lower case letters
                   1167:   print    printing characters, including space
                   1168:   punct    printing characters, excluding letters and digits and space
                   1169:   space    white space (not quite the same as \s)
                   1170:   upper    upper case letters
                   1171:   word     "word" characters (same as \w)
                   1172:   xdigit   hexadecimal digits
                   1173: </pre>
                   1174: The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
                   1175: space (32). Notice that this list includes the VT character (code 11). This
                   1176: makes "space" different to \s, which does not include VT (for Perl
                   1177: compatibility).
                   1178: </P>
                   1179: <P>
                   1180: The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
                   1181: 5.8. Another Perl extension is negation, which is indicated by a ^ character
                   1182: after the colon. For example,
                   1183: <pre>
                   1184:   [12[:^digit:]]
                   1185: </pre>
                   1186: matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
                   1187: syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
                   1188: supported, and an error is given if they are encountered.
                   1189: </P>
                   1190: <P>
1.1.1.2 ! misho    1191: By default, in UTF modes, characters with values greater than 128 do not match
1.1       misho    1192: any of the POSIX character classes. However, if the PCRE_UCP option is passed
                   1193: to <b>pcre_compile()</b>, some of the classes are changed so that Unicode
                   1194: character properties are used. This is achieved by replacing the POSIX classes
                   1195: by other sequences, as follows:
                   1196: <pre>
                   1197:   [:alnum:]  becomes  \p{Xan}
                   1198:   [:alpha:]  becomes  \p{L}
                   1199:   [:blank:]  becomes  \h
                   1200:   [:digit:]  becomes  \p{Nd}
                   1201:   [:lower:]  becomes  \p{Ll}
                   1202:   [:space:]  becomes  \p{Xps}
                   1203:   [:upper:]  becomes  \p{Lu}
                   1204:   [:word:]   becomes  \p{Xwd}
                   1205: </pre>
                   1206: Negated versions, such as [:^alpha:] use \P instead of \p. The other POSIX
                   1207: classes are unchanged, and match only characters with code points less than
                   1208: 128.
                   1209: </P>
                   1210: <br><a name="SEC10" href="#TOC1">VERTICAL BAR</a><br>
                   1211: <P>
                   1212: Vertical bar characters are used to separate alternative patterns. For example,
                   1213: the pattern
                   1214: <pre>
                   1215:   gilbert|sullivan
                   1216: </pre>
                   1217: matches either "gilbert" or "sullivan". Any number of alternatives may appear,
                   1218: and an empty alternative is permitted (matching the empty string). The matching
                   1219: process tries each alternative in turn, from left to right, and the first one
                   1220: that succeeds is used. If the alternatives are within a subpattern
                   1221: <a href="#subpattern">(defined below),</a>
                   1222: "succeeds" means matching the rest of the main pattern as well as the
                   1223: alternative in the subpattern.
                   1224: </P>
                   1225: <br><a name="SEC11" href="#TOC1">INTERNAL OPTION SETTING</a><br>
                   1226: <P>
                   1227: The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
                   1228: PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
                   1229: the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
                   1230: The option letters are
                   1231: <pre>
                   1232:   i  for PCRE_CASELESS
                   1233:   m  for PCRE_MULTILINE
                   1234:   s  for PCRE_DOTALL
                   1235:   x  for PCRE_EXTENDED
                   1236: </pre>
                   1237: For example, (?im) sets caseless, multiline matching. It is also possible to
                   1238: unset these options by preceding the letter with a hyphen, and a combined
                   1239: setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
                   1240: PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
                   1241: permitted. If a letter appears both before and after the hyphen, the option is
                   1242: unset.
                   1243: </P>
                   1244: <P>
                   1245: The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
                   1246: changed in the same way as the Perl-compatible options by using the characters
                   1247: J, U and X respectively.
                   1248: </P>
                   1249: <P>
                   1250: When one of these option changes occurs at top level (that is, not inside
                   1251: subpattern parentheses), the change applies to the remainder of the pattern
                   1252: that follows. If the change is placed right at the start of a pattern, PCRE
                   1253: extracts it into the global options (and it will therefore show up in data
                   1254: extracted by the <b>pcre_fullinfo()</b> function).
                   1255: </P>
                   1256: <P>
                   1257: An option change within a subpattern (see below for a description of
                   1258: subpatterns) affects only that part of the subpattern that follows it, so
                   1259: <pre>
                   1260:   (a(?i)b)c
                   1261: </pre>
                   1262: matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
                   1263: By this means, options can be made to have different settings in different
                   1264: parts of the pattern. Any changes made in one alternative do carry on
                   1265: into subsequent branches within the same subpattern. For example,
                   1266: <pre>
                   1267:   (a(?i)b|c)
                   1268: </pre>
                   1269: matches "ab", "aB", "c", and "C", even though when matching "C" the first
                   1270: branch is abandoned before the option setting. This is because the effects of
                   1271: option settings happen at compile time. There would be some very weird
                   1272: behaviour otherwise.
                   1273: </P>
                   1274: <P>
                   1275: <b>Note:</b> There are other PCRE-specific options that can be set by the
1.1.1.2 ! misho    1276: application when the compiling or matching functions are called. In some cases
        !          1277: the pattern can contain special leading sequences such as (*CRLF) to override
        !          1278: what the application has set or what has been defaulted. Details are given in
        !          1279: the section entitled
1.1       misho    1280: <a href="#newlineseq">"Newline sequences"</a>
1.1.1.2 ! misho    1281: above. There are also the (*UTF8), (*UTF16), and (*UCP) leading sequences that
        !          1282: can be used to set UTF and Unicode property modes; they are equivalent to
        !          1283: setting the PCRE_UTF8, PCRE_UTF16, and the PCRE_UCP options, respectively.
1.1       misho    1284: <a name="subpattern"></a></P>
                   1285: <br><a name="SEC12" href="#TOC1">SUBPATTERNS</a><br>
                   1286: <P>
                   1287: Subpatterns are delimited by parentheses (round brackets), which can be nested.
                   1288: Turning part of a pattern into a subpattern does two things:
                   1289: <br>
                   1290: <br>
                   1291: 1. It localizes a set of alternatives. For example, the pattern
                   1292: <pre>
                   1293:   cat(aract|erpillar|)
                   1294: </pre>
                   1295: matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
                   1296: match "cataract", "erpillar" or an empty string.
                   1297: <br>
                   1298: <br>
                   1299: 2. It sets up the subpattern as a capturing subpattern. This means that, when
                   1300: the whole pattern matches, that portion of the subject string that matched the
1.1.1.2 ! misho    1301: subpattern is passed back to the caller via the <i>ovector</i> argument of the
        !          1302: matching function. (This applies only to the traditional matching functions;
        !          1303: the DFA matching functions do not support capturing.)
        !          1304: </P>
        !          1305: <P>
        !          1306: Opening parentheses are counted from left to right (starting from 1) to obtain
        !          1307: numbers for the capturing subpatterns. For example, if the string "the red
        !          1308: king" is matched against the pattern
1.1       misho    1309: <pre>
                   1310:   the ((red|white) (king|queen))
                   1311: </pre>
                   1312: the captured substrings are "red king", "red", and "king", and are numbered 1,
                   1313: 2, and 3, respectively.
                   1314: </P>
                   1315: <P>
                   1316: The fact that plain parentheses fulfil two functions is not always helpful.
                   1317: There are often times when a grouping subpattern is required without a
                   1318: capturing requirement. If an opening parenthesis is followed by a question mark
                   1319: and a colon, the subpattern does not do any capturing, and is not counted when
                   1320: computing the number of any subsequent capturing subpatterns. For example, if
                   1321: the string "the white queen" is matched against the pattern
                   1322: <pre>
                   1323:   the ((?:red|white) (king|queen))
                   1324: </pre>
                   1325: the captured substrings are "white queen" and "queen", and are numbered 1 and
                   1326: 2. The maximum number of capturing subpatterns is 65535.
                   1327: </P>
                   1328: <P>
                   1329: As a convenient shorthand, if any option settings are required at the start of
                   1330: a non-capturing subpattern, the option letters may appear between the "?" and
                   1331: the ":". Thus the two patterns
                   1332: <pre>
                   1333:   (?i:saturday|sunday)
                   1334:   (?:(?i)saturday|sunday)
                   1335: </pre>
                   1336: match exactly the same set of strings. Because alternative branches are tried
                   1337: from left to right, and options are not reset until the end of the subpattern
                   1338: is reached, an option setting in one branch does affect subsequent branches, so
                   1339: the above patterns match "SUNDAY" as well as "Saturday".
                   1340: <a name="dupsubpatternnumber"></a></P>
                   1341: <br><a name="SEC13" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br>
                   1342: <P>
                   1343: Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
                   1344: the same numbers for its capturing parentheses. Such a subpattern starts with
                   1345: (?| and is itself a non-capturing subpattern. For example, consider this
                   1346: pattern:
                   1347: <pre>
                   1348:   (?|(Sat)ur|(Sun))day
                   1349: </pre>
                   1350: Because the two alternatives are inside a (?| group, both sets of capturing
                   1351: parentheses are numbered one. Thus, when the pattern matches, you can look
                   1352: at captured substring number one, whichever alternative matched. This construct
                   1353: is useful when you want to capture part, but not all, of one of a number of
                   1354: alternatives. Inside a (?| group, parentheses are numbered as usual, but the
                   1355: number is reset at the start of each branch. The numbers of any capturing
                   1356: parentheses that follow the subpattern start after the highest number used in
                   1357: any branch. The following example is taken from the Perl documentation. The
                   1358: numbers underneath show in which buffer the captured content will be stored.
                   1359: <pre>
                   1360:   # before  ---------------branch-reset----------- after
                   1361:   / ( a )  (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
                   1362:   # 1            2         2  3        2     3     4
                   1363: </pre>
                   1364: A back reference to a numbered subpattern uses the most recent value that is
                   1365: set for that number by any subpattern. The following pattern matches "abcabc"
                   1366: or "defdef":
                   1367: <pre>
                   1368:   /(?|(abc)|(def))\1/
                   1369: </pre>
                   1370: In contrast, a subroutine call to a numbered subpattern always refers to the
                   1371: first one in the pattern with the given number. The following pattern matches
                   1372: "abcabc" or "defabc":
                   1373: <pre>
                   1374:   /(?|(abc)|(def))(?1)/
                   1375: </pre>
                   1376: If a
                   1377: <a href="#conditions">condition test</a>
                   1378: for a subpattern's having matched refers to a non-unique number, the test is
                   1379: true if any of the subpatterns of that number have matched.
                   1380: </P>
                   1381: <P>
                   1382: An alternative approach to using this "branch reset" feature is to use
                   1383: duplicate named subpatterns, as described in the next section.
                   1384: </P>
                   1385: <br><a name="SEC14" href="#TOC1">NAMED SUBPATTERNS</a><br>
                   1386: <P>
                   1387: Identifying capturing parentheses by number is simple, but it can be very hard
                   1388: to keep track of the numbers in complicated regular expressions. Furthermore,
                   1389: if an expression is modified, the numbers may change. To help with this
                   1390: difficulty, PCRE supports the naming of subpatterns. This feature was not
                   1391: added to Perl until release 5.10. Python had the feature earlier, and PCRE
                   1392: introduced it at release 4.0, using the Python syntax. PCRE now supports both
                   1393: the Perl and the Python syntax. Perl allows identically numbered subpatterns to
                   1394: have different names, but PCRE does not.
                   1395: </P>
                   1396: <P>
                   1397: In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
                   1398: (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
                   1399: parentheses from other parts of the pattern, such as
                   1400: <a href="#backreferences">back references,</a>
                   1401: <a href="#recursion">recursion,</a>
                   1402: and
                   1403: <a href="#conditions">conditions,</a>
                   1404: can be made by name as well as by number.
                   1405: </P>
                   1406: <P>
                   1407: Names consist of up to 32 alphanumeric characters and underscores. Named
                   1408: capturing parentheses are still allocated numbers as well as names, exactly as
                   1409: if the names were not present. The PCRE API provides function calls for
                   1410: extracting the name-to-number translation table from a compiled pattern. There
                   1411: is also a convenience function for extracting a captured substring by name.
                   1412: </P>
                   1413: <P>
                   1414: By default, a name must be unique within a pattern, but it is possible to relax
                   1415: this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
                   1416: names are also always permitted for subpatterns with the same number, set up as
                   1417: described in the previous section.) Duplicate names can be useful for patterns
                   1418: where only one instance of the named parentheses can match. Suppose you want to
                   1419: match the name of a weekday, either as a 3-letter abbreviation or as the full
                   1420: name, and in both cases you want to extract the abbreviation. This pattern
                   1421: (ignoring the line breaks) does the job:
                   1422: <pre>
                   1423:   (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
                   1424:   (?&#60;DN&#62;Tue)(?:sday)?|
                   1425:   (?&#60;DN&#62;Wed)(?:nesday)?|
                   1426:   (?&#60;DN&#62;Thu)(?:rsday)?|
                   1427:   (?&#60;DN&#62;Sat)(?:urday)?
                   1428: </pre>
                   1429: There are five capturing substrings, but only one is ever set after a match.
                   1430: (An alternative way of solving this problem is to use a "branch reset"
                   1431: subpattern, as described in the previous section.)
                   1432: </P>
                   1433: <P>
                   1434: The convenience function for extracting the data by name returns the substring
                   1435: for the first (and in this example, the only) subpattern of that name that
                   1436: matched. This saves searching to find which numbered subpattern it was.
                   1437: </P>
                   1438: <P>
                   1439: If you make a back reference to a non-unique named subpattern from elsewhere in
                   1440: the pattern, the one that corresponds to the first occurrence of the name is
                   1441: used. In the absence of duplicate numbers (see the previous section) this is
                   1442: the one with the lowest number. If you use a named reference in a condition
                   1443: test (see the
                   1444: <a href="#conditions">section about conditions</a>
                   1445: below), either to check whether a subpattern has matched, or to check for
                   1446: recursion, all subpatterns with the same name are tested. If the condition is
                   1447: true for any one of them, the overall condition is true. This is the same
                   1448: behaviour as testing by number. For further details of the interfaces for
                   1449: handling named subpatterns, see the
                   1450: <a href="pcreapi.html"><b>pcreapi</b></a>
                   1451: documentation.
                   1452: </P>
                   1453: <P>
                   1454: <b>Warning:</b> You cannot use different names to distinguish between two
                   1455: subpatterns with the same number because PCRE uses only the numbers when
                   1456: matching. For this reason, an error is given at compile time if different names
                   1457: are given to subpatterns with the same number. However, you can give the same
                   1458: name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
                   1459: </P>
                   1460: <br><a name="SEC15" href="#TOC1">REPETITION</a><br>
                   1461: <P>
                   1462: Repetition is specified by quantifiers, which can follow any of the following
                   1463: items:
                   1464: <pre>
                   1465:   a literal data character
                   1466:   the dot metacharacter
                   1467:   the \C escape sequence
1.1.1.2 ! misho    1468:   the \X escape sequence
1.1       misho    1469:   the \R escape sequence
                   1470:   an escape such as \d or \pL that matches a single character
                   1471:   a character class
                   1472:   a back reference (see next section)
                   1473:   a parenthesized subpattern (including assertions)
                   1474:   a subroutine call to a subpattern (recursive or otherwise)
                   1475: </pre>
                   1476: The general repetition quantifier specifies a minimum and maximum number of
                   1477: permitted matches, by giving the two numbers in curly brackets (braces),
                   1478: separated by a comma. The numbers must be less than 65536, and the first must
                   1479: be less than or equal to the second. For example:
                   1480: <pre>
                   1481:   z{2,4}
                   1482: </pre>
                   1483: matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
                   1484: character. If the second number is omitted, but the comma is present, there is
                   1485: no upper limit; if the second number and the comma are both omitted, the
                   1486: quantifier specifies an exact number of required matches. Thus
                   1487: <pre>
                   1488:   [aeiou]{3,}
                   1489: </pre>
                   1490: matches at least 3 successive vowels, but may match many more, while
                   1491: <pre>
                   1492:   \d{8}
                   1493: </pre>
                   1494: matches exactly 8 digits. An opening curly bracket that appears in a position
                   1495: where a quantifier is not allowed, or one that does not match the syntax of a
                   1496: quantifier, is taken as a literal character. For example, {,6} is not a
                   1497: quantifier, but a literal string of four characters.
                   1498: </P>
                   1499: <P>
1.1.1.2 ! misho    1500: In UTF modes, quantifiers apply to characters rather than to individual data
        !          1501: units. Thus, for example, \x{100}{2} matches two characters, each of
        !          1502: which is represented by a two-byte sequence in a UTF-8 string. Similarly,
        !          1503: \X{3} matches three Unicode extended sequences, each of which may be several
        !          1504: data units long (and they may be of different lengths).
1.1       misho    1505: </P>
                   1506: <P>
                   1507: The quantifier {0} is permitted, causing the expression to behave as if the
                   1508: previous item and the quantifier were not present. This may be useful for
                   1509: subpatterns that are referenced as
                   1510: <a href="#subpatternsassubroutines">subroutines</a>
                   1511: from elsewhere in the pattern (but see also the section entitled
                   1512: <a href="#subdefine">"Defining subpatterns for use by reference only"</a>
                   1513: below). Items other than subpatterns that have a {0} quantifier are omitted
                   1514: from the compiled pattern.
                   1515: </P>
                   1516: <P>
                   1517: For convenience, the three most common quantifiers have single-character
                   1518: abbreviations:
                   1519: <pre>
                   1520:   *    is equivalent to {0,}
                   1521:   +    is equivalent to {1,}
                   1522:   ?    is equivalent to {0,1}
                   1523: </pre>
                   1524: It is possible to construct infinite loops by following a subpattern that can
                   1525: match no characters with a quantifier that has no upper limit, for example:
                   1526: <pre>
                   1527:   (a?)*
                   1528: </pre>
                   1529: Earlier versions of Perl and PCRE used to give an error at compile time for
                   1530: such patterns. However, because there are cases where this can be useful, such
                   1531: patterns are now accepted, but if any repetition of the subpattern does in fact
                   1532: match no characters, the loop is forcibly broken.
                   1533: </P>
                   1534: <P>
                   1535: By default, the quantifiers are "greedy", that is, they match as much as
                   1536: possible (up to the maximum number of permitted times), without causing the
                   1537: rest of the pattern to fail. The classic example of where this gives problems
                   1538: is in trying to match comments in C programs. These appear between /* and */
                   1539: and within the comment, individual * and / characters may appear. An attempt to
                   1540: match C comments by applying the pattern
                   1541: <pre>
                   1542:   /\*.*\*/
                   1543: </pre>
                   1544: to the string
                   1545: <pre>
                   1546:   /* first comment */  not comment  /* second comment */
                   1547: </pre>
                   1548: fails, because it matches the entire string owing to the greediness of the .*
                   1549: item.
                   1550: </P>
                   1551: <P>
                   1552: However, if a quantifier is followed by a question mark, it ceases to be
                   1553: greedy, and instead matches the minimum number of times possible, so the
                   1554: pattern
                   1555: <pre>
                   1556:   /\*.*?\*/
                   1557: </pre>
                   1558: does the right thing with the C comments. The meaning of the various
                   1559: quantifiers is not otherwise changed, just the preferred number of matches.
                   1560: Do not confuse this use of question mark with its use as a quantifier in its
                   1561: own right. Because it has two uses, it can sometimes appear doubled, as in
                   1562: <pre>
                   1563:   \d??\d
                   1564: </pre>
                   1565: which matches one digit by preference, but can match two if that is the only
                   1566: way the rest of the pattern matches.
                   1567: </P>
                   1568: <P>
                   1569: If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
                   1570: the quantifiers are not greedy by default, but individual ones can be made
                   1571: greedy by following them with a question mark. In other words, it inverts the
                   1572: default behaviour.
                   1573: </P>
                   1574: <P>
                   1575: When a parenthesized subpattern is quantified with a minimum repeat count that
                   1576: is greater than 1 or with a limited maximum, more memory is required for the
                   1577: compiled pattern, in proportion to the size of the minimum or maximum.
                   1578: </P>
                   1579: <P>
                   1580: If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
                   1581: to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
                   1582: implicitly anchored, because whatever follows will be tried against every
                   1583: character position in the subject string, so there is no point in retrying the
                   1584: overall match at any position after the first. PCRE normally treats such a
                   1585: pattern as though it were preceded by \A.
                   1586: </P>
                   1587: <P>
                   1588: In cases where it is known that the subject string contains no newlines, it is
                   1589: worth setting PCRE_DOTALL in order to obtain this optimization, or
                   1590: alternatively using ^ to indicate anchoring explicitly.
                   1591: </P>
                   1592: <P>
                   1593: However, there is one situation where the optimization cannot be used. When .*
                   1594: is inside capturing parentheses that are the subject of a back reference
                   1595: elsewhere in the pattern, a match at the start may fail where a later one
                   1596: succeeds. Consider, for example:
                   1597: <pre>
                   1598:   (.*)abc\1
                   1599: </pre>
                   1600: If the subject is "xyz123abc123" the match point is the fourth character. For
                   1601: this reason, such a pattern is not implicitly anchored.
                   1602: </P>
                   1603: <P>
                   1604: When a capturing subpattern is repeated, the value captured is the substring
                   1605: that matched the final iteration. For example, after
                   1606: <pre>
                   1607:   (tweedle[dume]{3}\s*)+
                   1608: </pre>
                   1609: has matched "tweedledum tweedledee" the value of the captured substring is
                   1610: "tweedledee". However, if there are nested capturing subpatterns, the
                   1611: corresponding captured values may have been set in previous iterations. For
                   1612: example, after
                   1613: <pre>
                   1614:   /(a|(b))+/
                   1615: </pre>
                   1616: matches "aba" the value of the second captured substring is "b".
                   1617: <a name="atomicgroup"></a></P>
                   1618: <br><a name="SEC16" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
                   1619: <P>
                   1620: With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
                   1621: repetition, failure of what follows normally causes the repeated item to be
                   1622: re-evaluated to see if a different number of repeats allows the rest of the
                   1623: pattern to match. Sometimes it is useful to prevent this, either to change the
                   1624: nature of the match, or to cause it fail earlier than it otherwise might, when
                   1625: the author of the pattern knows there is no point in carrying on.
                   1626: </P>
                   1627: <P>
                   1628: Consider, for example, the pattern \d+foo when applied to the subject line
                   1629: <pre>
                   1630:   123456bar
                   1631: </pre>
                   1632: After matching all 6 digits and then failing to match "foo", the normal
                   1633: action of the matcher is to try again with only 5 digits matching the \d+
                   1634: item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
                   1635: (a term taken from Jeffrey Friedl's book) provides the means for specifying
                   1636: that once a subpattern has matched, it is not to be re-evaluated in this way.
                   1637: </P>
                   1638: <P>
                   1639: If we use atomic grouping for the previous example, the matcher gives up
                   1640: immediately on failing to match "foo" the first time. The notation is a kind of
                   1641: special parenthesis, starting with (?&#62; as in this example:
                   1642: <pre>
                   1643:   (?&#62;\d+)foo
                   1644: </pre>
                   1645: This kind of parenthesis "locks up" the  part of the pattern it contains once
                   1646: it has matched, and a failure further into the pattern is prevented from
                   1647: backtracking into it. Backtracking past it to previous items, however, works as
                   1648: normal.
                   1649: </P>
                   1650: <P>
                   1651: An alternative description is that a subpattern of this type matches the string
                   1652: of characters that an identical standalone pattern would match, if anchored at
                   1653: the current point in the subject string.
                   1654: </P>
                   1655: <P>
                   1656: Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
                   1657: the above example can be thought of as a maximizing repeat that must swallow
                   1658: everything it can. So, while both \d+ and \d+? are prepared to adjust the
                   1659: number of digits they match in order to make the rest of the pattern match,
                   1660: (?&#62;\d+) can only match an entire sequence of digits.
                   1661: </P>
                   1662: <P>
                   1663: Atomic groups in general can of course contain arbitrarily complicated
                   1664: subpatterns, and can be nested. However, when the subpattern for an atomic
                   1665: group is just a single repeated item, as in the example above, a simpler
                   1666: notation, called a "possessive quantifier" can be used. This consists of an
                   1667: additional + character following a quantifier. Using this notation, the
                   1668: previous example can be rewritten as
                   1669: <pre>
                   1670:   \d++foo
                   1671: </pre>
                   1672: Note that a possessive quantifier can be used with an entire group, for
                   1673: example:
                   1674: <pre>
                   1675:   (abc|xyz){2,3}+
                   1676: </pre>
                   1677: Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
                   1678: option is ignored. They are a convenient notation for the simpler forms of
                   1679: atomic group. However, there is no difference in the meaning of a possessive
                   1680: quantifier and the equivalent atomic group, though there may be a performance
                   1681: difference; possessive quantifiers should be slightly faster.
                   1682: </P>
                   1683: <P>
                   1684: The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
                   1685: Jeffrey Friedl originated the idea (and the name) in the first edition of his
                   1686: book. Mike McCloskey liked it, so implemented it when he built Sun's Java
                   1687: package, and PCRE copied it from there. It ultimately found its way into Perl
                   1688: at release 5.10.
                   1689: </P>
                   1690: <P>
                   1691: PCRE has an optimization that automatically "possessifies" certain simple
                   1692: pattern constructs. For example, the sequence A+B is treated as A++B because
                   1693: there is no point in backtracking into a sequence of A's when B must follow.
                   1694: </P>
                   1695: <P>
                   1696: When a pattern contains an unlimited repeat inside a subpattern that can itself
                   1697: be repeated an unlimited number of times, the use of an atomic group is the
                   1698: only way to avoid some failing matches taking a very long time indeed. The
                   1699: pattern
                   1700: <pre>
                   1701:   (\D+|&#60;\d+&#62;)*[!?]
                   1702: </pre>
                   1703: matches an unlimited number of substrings that either consist of non-digits, or
                   1704: digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
                   1705: quickly. However, if it is applied to
                   1706: <pre>
                   1707:   aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
                   1708: </pre>
                   1709: it takes a long time before reporting failure. This is because the string can
                   1710: be divided between the internal \D+ repeat and the external * repeat in a
                   1711: large number of ways, and all have to be tried. (The example uses [!?] rather
                   1712: than a single character at the end, because both PCRE and Perl have an
                   1713: optimization that allows for fast failure when a single character is used. They
                   1714: remember the last single character that is required for a match, and fail early
                   1715: if it is not present in the string.) If the pattern is changed so that it uses
                   1716: an atomic group, like this:
                   1717: <pre>
                   1718:   ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
                   1719: </pre>
                   1720: sequences of non-digits cannot be broken, and failure happens quickly.
                   1721: <a name="backreferences"></a></P>
                   1722: <br><a name="SEC17" href="#TOC1">BACK REFERENCES</a><br>
                   1723: <P>
                   1724: Outside a character class, a backslash followed by a digit greater than 0 (and
                   1725: possibly further digits) is a back reference to a capturing subpattern earlier
                   1726: (that is, to its left) in the pattern, provided there have been that many
                   1727: previous capturing left parentheses.
                   1728: </P>
                   1729: <P>
                   1730: However, if the decimal number following the backslash is less than 10, it is
                   1731: always taken as a back reference, and causes an error only if there are not
                   1732: that many capturing left parentheses in the entire pattern. In other words, the
                   1733: parentheses that are referenced need not be to the left of the reference for
                   1734: numbers less than 10. A "forward back reference" of this type can make sense
                   1735: when a repetition is involved and the subpattern to the right has participated
                   1736: in an earlier iteration.
                   1737: </P>
                   1738: <P>
                   1739: It is not possible to have a numerical "forward back reference" to a subpattern
                   1740: whose number is 10 or more using this syntax because a sequence such as \50 is
                   1741: interpreted as a character defined in octal. See the subsection entitled
                   1742: "Non-printing characters"
                   1743: <a href="#digitsafterbackslash">above</a>
                   1744: for further details of the handling of digits following a backslash. There is
                   1745: no such problem when named parentheses are used. A back reference to any
                   1746: subpattern is possible using named parentheses (see below).
                   1747: </P>
                   1748: <P>
                   1749: Another way of avoiding the ambiguity inherent in the use of digits following a
                   1750: backslash is to use the \g escape sequence. This escape must be followed by an
                   1751: unsigned number or a negative number, optionally enclosed in braces. These
                   1752: examples are all identical:
                   1753: <pre>
                   1754:   (ring), \1
                   1755:   (ring), \g1
                   1756:   (ring), \g{1}
                   1757: </pre>
                   1758: An unsigned number specifies an absolute reference without the ambiguity that
                   1759: is present in the older syntax. It is also useful when literal digits follow
                   1760: the reference. A negative number is a relative reference. Consider this
                   1761: example:
                   1762: <pre>
                   1763:   (abc(def)ghi)\g{-1}
                   1764: </pre>
                   1765: The sequence \g{-1} is a reference to the most recently started capturing
                   1766: subpattern before \g, that is, is it equivalent to \2 in this example.
                   1767: Similarly, \g{-2} would be equivalent to \1. The use of relative references
                   1768: can be helpful in long patterns, and also in patterns that are created by
                   1769: joining together fragments that contain references within themselves.
                   1770: </P>
                   1771: <P>
                   1772: A back reference matches whatever actually matched the capturing subpattern in
                   1773: the current subject string, rather than anything matching the subpattern
                   1774: itself (see
                   1775: <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
                   1776: below for a way of doing that). So the pattern
                   1777: <pre>
                   1778:   (sens|respons)e and \1ibility
                   1779: </pre>
                   1780: matches "sense and sensibility" and "response and responsibility", but not
                   1781: "sense and responsibility". If caseful matching is in force at the time of the
                   1782: back reference, the case of letters is relevant. For example,
                   1783: <pre>
                   1784:   ((?i)rah)\s+\1
                   1785: </pre>
                   1786: matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
                   1787: capturing subpattern is matched caselessly.
                   1788: </P>
                   1789: <P>
                   1790: There are several different ways of writing back references to named
                   1791: subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
                   1792: \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
                   1793: back reference syntax, in which \g can be used for both numeric and named
                   1794: references, is also supported. We could rewrite the above example in any of
                   1795: the following ways:
                   1796: <pre>
                   1797:   (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
                   1798:   (?'p1'(?i)rah)\s+\k{p1}
                   1799:   (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
                   1800:   (?&#60;p1&#62;(?i)rah)\s+\g{p1}
                   1801: </pre>
                   1802: A subpattern that is referenced by name may appear in the pattern before or
                   1803: after the reference.
                   1804: </P>
                   1805: <P>
                   1806: There may be more than one back reference to the same subpattern. If a
                   1807: subpattern has not actually been used in a particular match, any back
                   1808: references to it always fail by default. For example, the pattern
                   1809: <pre>
                   1810:   (a|(bc))\2
                   1811: </pre>
                   1812: always fails if it starts to match "a" rather than "bc". However, if the
                   1813: PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
                   1814: unset value matches an empty string.
                   1815: </P>
                   1816: <P>
                   1817: Because there may be many capturing parentheses in a pattern, all digits
                   1818: following a backslash are taken as part of a potential back reference number.
                   1819: If the pattern continues with a digit character, some delimiter must be used to
                   1820: terminate the back reference. If the PCRE_EXTENDED option is set, this can be
                   1821: whitespace. Otherwise, the \g{ syntax or an empty comment (see
                   1822: <a href="#comments">"Comments"</a>
                   1823: below) can be used.
                   1824: </P>
                   1825: <br><b>
                   1826: Recursive back references
                   1827: </b><br>
                   1828: <P>
                   1829: A back reference that occurs inside the parentheses to which it refers fails
                   1830: when the subpattern is first used, so, for example, (a\1) never matches.
                   1831: However, such references can be useful inside repeated subpatterns. For
                   1832: example, the pattern
                   1833: <pre>
                   1834:   (a|b\1)+
                   1835: </pre>
                   1836: matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
                   1837: the subpattern, the back reference matches the character string corresponding
                   1838: to the previous iteration. In order for this to work, the pattern must be such
                   1839: that the first iteration does not need to match the back reference. This can be
                   1840: done using alternation, as in the example above, or by a quantifier with a
                   1841: minimum of zero.
                   1842: </P>
                   1843: <P>
                   1844: Back references of this type cause the group that they reference to be treated
                   1845: as an
                   1846: <a href="#atomicgroup">atomic group.</a>
                   1847: Once the whole group has been matched, a subsequent matching failure cannot
                   1848: cause backtracking into the middle of the group.
                   1849: <a name="bigassertions"></a></P>
                   1850: <br><a name="SEC18" href="#TOC1">ASSERTIONS</a><br>
                   1851: <P>
                   1852: An assertion is a test on the characters following or preceding the current
                   1853: matching point that does not actually consume any characters. The simple
                   1854: assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
                   1855: <a href="#smallassertions">above.</a>
                   1856: </P>
                   1857: <P>
                   1858: More complicated assertions are coded as subpatterns. There are two kinds:
                   1859: those that look ahead of the current position in the subject string, and those
                   1860: that look behind it. An assertion subpattern is matched in the normal way,
                   1861: except that it does not cause the current matching position to be changed.
                   1862: </P>
                   1863: <P>
                   1864: Assertion subpatterns are not capturing subpatterns. If such an assertion
                   1865: contains capturing subpatterns within it, these are counted for the purposes of
                   1866: numbering the capturing subpatterns in the whole pattern. However, substring
                   1867: capturing is carried out only for positive assertions, because it does not make
                   1868: sense for negative assertions.
                   1869: </P>
                   1870: <P>
                   1871: For compatibility with Perl, assertion subpatterns may be repeated; though
                   1872: it makes no sense to assert the same thing several times, the side effect of
                   1873: capturing parentheses may occasionally be useful. In practice, there only three
                   1874: cases:
                   1875: <br>
                   1876: <br>
                   1877: (1) If the quantifier is {0}, the assertion is never obeyed during matching.
                   1878: However, it may contain internal capturing parenthesized groups that are called
                   1879: from elsewhere via the
                   1880: <a href="#subpatternsassubroutines">subroutine mechanism.</a>
                   1881: <br>
                   1882: <br>
                   1883: (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
                   1884: were {0,1}. At run time, the rest of the pattern match is tried with and
                   1885: without the assertion, the order depending on the greediness of the quantifier.
                   1886: <br>
                   1887: <br>
                   1888: (3) If the minimum repetition is greater than zero, the quantifier is ignored.
                   1889: The assertion is obeyed just once when encountered during matching.
                   1890: </P>
                   1891: <br><b>
                   1892: Lookahead assertions
                   1893: </b><br>
                   1894: <P>
                   1895: Lookahead assertions start with (?= for positive assertions and (?! for
                   1896: negative assertions. For example,
                   1897: <pre>
                   1898:   \w+(?=;)
                   1899: </pre>
                   1900: matches a word followed by a semicolon, but does not include the semicolon in
                   1901: the match, and
                   1902: <pre>
                   1903:   foo(?!bar)
                   1904: </pre>
                   1905: matches any occurrence of "foo" that is not followed by "bar". Note that the
                   1906: apparently similar pattern
                   1907: <pre>
                   1908:   (?!foo)bar
                   1909: </pre>
                   1910: does not find an occurrence of "bar" that is preceded by something other than
                   1911: "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
                   1912: (?!foo) is always true when the next three characters are "bar". A
                   1913: lookbehind assertion is needed to achieve the other effect.
                   1914: </P>
                   1915: <P>
                   1916: If you want to force a matching failure at some point in a pattern, the most
                   1917: convenient way to do it is with (?!) because an empty string always matches, so
                   1918: an assertion that requires there not to be an empty string must always fail.
                   1919: The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
                   1920: <a name="lookbehind"></a></P>
                   1921: <br><b>
                   1922: Lookbehind assertions
                   1923: </b><br>
                   1924: <P>
                   1925: Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
                   1926: negative assertions. For example,
                   1927: <pre>
                   1928:   (?&#60;!foo)bar
                   1929: </pre>
                   1930: does find an occurrence of "bar" that is not preceded by "foo". The contents of
                   1931: a lookbehind assertion are restricted such that all the strings it matches must
                   1932: have a fixed length. However, if there are several top-level alternatives, they
                   1933: do not all have to have the same fixed length. Thus
                   1934: <pre>
                   1935:   (?&#60;=bullock|donkey)
                   1936: </pre>
                   1937: is permitted, but
                   1938: <pre>
                   1939:   (?&#60;!dogs?|cats?)
                   1940: </pre>
                   1941: causes an error at compile time. Branches that match different length strings
                   1942: are permitted only at the top level of a lookbehind assertion. This is an
                   1943: extension compared with Perl, which requires all branches to match the same
                   1944: length of string. An assertion such as
                   1945: <pre>
                   1946:   (?&#60;=ab(c|de))
                   1947: </pre>
                   1948: is not permitted, because its single top-level branch can match two different
                   1949: lengths, but it is acceptable to PCRE if rewritten to use two top-level
                   1950: branches:
                   1951: <pre>
                   1952:   (?&#60;=abc|abde)
                   1953: </pre>
                   1954: In some cases, the escape sequence \K
                   1955: <a href="#resetmatchstart">(see above)</a>
                   1956: can be used instead of a lookbehind assertion to get round the fixed-length
                   1957: restriction.
                   1958: </P>
                   1959: <P>
                   1960: The implementation of lookbehind assertions is, for each alternative, to
                   1961: temporarily move the current position back by the fixed length and then try to
                   1962: match. If there are insufficient characters before the current position, the
                   1963: assertion fails.
                   1964: </P>
                   1965: <P>
1.1.1.2 ! misho    1966: In a UTF mode, PCRE does not allow the \C escape (which matches a single data
        !          1967: unit even in a UTF mode) to appear in lookbehind assertions, because it makes
        !          1968: it impossible to calculate the length of the lookbehind. The \X and \R
        !          1969: escapes, which can match different numbers of data units, are also not
        !          1970: permitted.
1.1       misho    1971: </P>
                   1972: <P>
                   1973: <a href="#subpatternsassubroutines">"Subroutine"</a>
                   1974: calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
                   1975: as the subpattern matches a fixed-length string.
                   1976: <a href="#recursion">Recursion,</a>
                   1977: however, is not supported.
                   1978: </P>
                   1979: <P>
                   1980: Possessive quantifiers can be used in conjunction with lookbehind assertions to
                   1981: specify efficient matching of fixed-length strings at the end of subject
                   1982: strings. Consider a simple pattern such as
                   1983: <pre>
                   1984:   abcd$
                   1985: </pre>
                   1986: when applied to a long string that does not match. Because matching proceeds
                   1987: from left to right, PCRE will look for each "a" in the subject and then see if
                   1988: what follows matches the rest of the pattern. If the pattern is specified as
                   1989: <pre>
                   1990:   ^.*abcd$
                   1991: </pre>
                   1992: the initial .* matches the entire string at first, but when this fails (because
                   1993: there is no following "a"), it backtracks to match all but the last character,
                   1994: then all but the last two characters, and so on. Once again the search for "a"
                   1995: covers the entire string, from right to left, so we are no better off. However,
                   1996: if the pattern is written as
                   1997: <pre>
                   1998:   ^.*+(?&#60;=abcd)
                   1999: </pre>
                   2000: there can be no backtracking for the .*+ item; it can match only the entire
                   2001: string. The subsequent lookbehind assertion does a single test on the last four
                   2002: characters. If it fails, the match fails immediately. For long strings, this
                   2003: approach makes a significant difference to the processing time.
                   2004: </P>
                   2005: <br><b>
                   2006: Using multiple assertions
                   2007: </b><br>
                   2008: <P>
                   2009: Several assertions (of any sort) may occur in succession. For example,
                   2010: <pre>
                   2011:   (?&#60;=\d{3})(?&#60;!999)foo
                   2012: </pre>
                   2013: matches "foo" preceded by three digits that are not "999". Notice that each of
                   2014: the assertions is applied independently at the same point in the subject
                   2015: string. First there is a check that the previous three characters are all
                   2016: digits, and then there is a check that the same three characters are not "999".
                   2017: This pattern does <i>not</i> match "foo" preceded by six characters, the first
                   2018: of which are digits and the last three of which are not "999". For example, it
                   2019: doesn't match "123abcfoo". A pattern to do that is
                   2020: <pre>
                   2021:   (?&#60;=\d{3}...)(?&#60;!999)foo
                   2022: </pre>
                   2023: This time the first assertion looks at the preceding six characters, checking
                   2024: that the first three are digits, and then the second assertion checks that the
                   2025: preceding three characters are not "999".
                   2026: </P>
                   2027: <P>
                   2028: Assertions can be nested in any combination. For example,
                   2029: <pre>
                   2030:   (?&#60;=(?&#60;!foo)bar)baz
                   2031: </pre>
                   2032: matches an occurrence of "baz" that is preceded by "bar" which in turn is not
                   2033: preceded by "foo", while
                   2034: <pre>
                   2035:   (?&#60;=\d{3}(?!999)...)foo
                   2036: </pre>
                   2037: is another pattern that matches "foo" preceded by three digits and any three
                   2038: characters that are not "999".
                   2039: <a name="conditions"></a></P>
                   2040: <br><a name="SEC19" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
                   2041: <P>
                   2042: It is possible to cause the matching process to obey a subpattern
                   2043: conditionally or to choose between two alternative subpatterns, depending on
                   2044: the result of an assertion, or whether a specific capturing subpattern has
                   2045: already been matched. The two possible forms of conditional subpattern are:
                   2046: <pre>
                   2047:   (?(condition)yes-pattern)
                   2048:   (?(condition)yes-pattern|no-pattern)
                   2049: </pre>
                   2050: If the condition is satisfied, the yes-pattern is used; otherwise the
                   2051: no-pattern (if present) is used. If there are more than two alternatives in the
                   2052: subpattern, a compile-time error occurs. Each of the two alternatives may
                   2053: itself contain nested subpatterns of any form, including conditional
                   2054: subpatterns; the restriction to two alternatives applies only at the level of
                   2055: the condition. This pattern fragment is an example where the alternatives are
                   2056: complex:
                   2057: <pre>
                   2058:   (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
                   2059: 
                   2060: </PRE>
                   2061: </P>
                   2062: <P>
                   2063: There are four kinds of condition: references to subpatterns, references to
                   2064: recursion, a pseudo-condition called DEFINE, and assertions.
                   2065: </P>
                   2066: <br><b>
                   2067: Checking for a used subpattern by number
                   2068: </b><br>
                   2069: <P>
                   2070: If the text between the parentheses consists of a sequence of digits, the
                   2071: condition is true if a capturing subpattern of that number has previously
                   2072: matched. If there is more than one capturing subpattern with the same number
                   2073: (see the earlier
                   2074: <a href="#recursion">section about duplicate subpattern numbers),</a>
                   2075: the condition is true if any of them have matched. An alternative notation is
                   2076: to precede the digits with a plus or minus sign. In this case, the subpattern
                   2077: number is relative rather than absolute. The most recently opened parentheses
                   2078: can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
                   2079: loops it can also make sense to refer to subsequent groups. The next
                   2080: parentheses to be opened can be referenced as (?(+1), and so on. (The value
                   2081: zero in any of these forms is not used; it provokes a compile-time error.)
                   2082: </P>
                   2083: <P>
                   2084: Consider the following pattern, which contains non-significant white space to
                   2085: make it more readable (assume the PCRE_EXTENDED option) and to divide it into
                   2086: three parts for ease of discussion:
                   2087: <pre>
                   2088:   ( \( )?    [^()]+    (?(1) \) )
                   2089: </pre>
                   2090: The first part matches an optional opening parenthesis, and if that
                   2091: character is present, sets it as the first captured substring. The second part
                   2092: matches one or more characters that are not parentheses. The third part is a
                   2093: conditional subpattern that tests whether or not the first set of parentheses
                   2094: matched. If they did, that is, if subject started with an opening parenthesis,
                   2095: the condition is true, and so the yes-pattern is executed and a closing
                   2096: parenthesis is required. Otherwise, since no-pattern is not present, the
                   2097: subpattern matches nothing. In other words, this pattern matches a sequence of
                   2098: non-parentheses, optionally enclosed in parentheses.
                   2099: </P>
                   2100: <P>
                   2101: If you were embedding this pattern in a larger one, you could use a relative
                   2102: reference:
                   2103: <pre>
                   2104:   ...other stuff... ( \( )?    [^()]+    (?(-1) \) ) ...
                   2105: </pre>
                   2106: This makes the fragment independent of the parentheses in the larger pattern.
                   2107: </P>
                   2108: <br><b>
                   2109: Checking for a used subpattern by name
                   2110: </b><br>
                   2111: <P>
                   2112: Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
                   2113: subpattern by name. For compatibility with earlier versions of PCRE, which had
                   2114: this facility before Perl, the syntax (?(name)...) is also recognized. However,
                   2115: there is a possible ambiguity with this syntax, because subpattern names may
                   2116: consist entirely of digits. PCRE looks first for a named subpattern; if it
                   2117: cannot find one and the name consists entirely of digits, PCRE looks for a
                   2118: subpattern of that number, which must be greater than zero. Using subpattern
                   2119: names that consist entirely of digits is not recommended.
                   2120: </P>
                   2121: <P>
                   2122: Rewriting the above example to use a named subpattern gives this:
                   2123: <pre>
                   2124:   (?&#60;OPEN&#62; \( )?    [^()]+    (?(&#60;OPEN&#62;) \) )
                   2125: </pre>
                   2126: If the name used in a condition of this kind is a duplicate, the test is
                   2127: applied to all subpatterns of the same name, and is true if any one of them has
                   2128: matched.
                   2129: </P>
                   2130: <br><b>
                   2131: Checking for pattern recursion
                   2132: </b><br>
                   2133: <P>
                   2134: If the condition is the string (R), and there is no subpattern with the name R,
                   2135: the condition is true if a recursive call to the whole pattern or any
                   2136: subpattern has been made. If digits or a name preceded by ampersand follow the
                   2137: letter R, for example:
                   2138: <pre>
                   2139:   (?(R3)...) or (?(R&name)...)
                   2140: </pre>
                   2141: the condition is true if the most recent recursion is into a subpattern whose
                   2142: number or name is given. This condition does not check the entire recursion
                   2143: stack. If the name used in a condition of this kind is a duplicate, the test is
                   2144: applied to all subpatterns of the same name, and is true if any one of them is
                   2145: the most recent recursion.
                   2146: </P>
                   2147: <P>
                   2148: At "top level", all these recursion test conditions are false.
                   2149: <a href="#recursion">The syntax for recursive patterns</a>
                   2150: is described below.
                   2151: <a name="subdefine"></a></P>
                   2152: <br><b>
                   2153: Defining subpatterns for use by reference only
                   2154: </b><br>
                   2155: <P>
                   2156: If the condition is the string (DEFINE), and there is no subpattern with the
                   2157: name DEFINE, the condition is always false. In this case, there may be only one
                   2158: alternative in the subpattern. It is always skipped if control reaches this
                   2159: point in the pattern; the idea of DEFINE is that it can be used to define
                   2160: subroutines that can be referenced from elsewhere. (The use of
                   2161: <a href="#subpatternsassubroutines">subroutines</a>
                   2162: is described below.) For example, a pattern to match an IPv4 address such as
                   2163: "192.168.23.245" could be written like this (ignore whitespace and line
                   2164: breaks):
                   2165: <pre>
                   2166:   (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
                   2167:   \b (?&byte) (\.(?&byte)){3} \b
                   2168: </pre>
                   2169: The first part of the pattern is a DEFINE group inside which a another group
                   2170: named "byte" is defined. This matches an individual component of an IPv4
                   2171: address (a number less than 256). When matching takes place, this part of the
                   2172: pattern is skipped because DEFINE acts like a false condition. The rest of the
                   2173: pattern uses references to the named group to match the four dot-separated
                   2174: components of an IPv4 address, insisting on a word boundary at each end.
                   2175: </P>
                   2176: <br><b>
                   2177: Assertion conditions
                   2178: </b><br>
                   2179: <P>
                   2180: If the condition is not in any of the above formats, it must be an assertion.
                   2181: This may be a positive or negative lookahead or lookbehind assertion. Consider
                   2182: this pattern, again containing non-significant white space, and with the two
                   2183: alternatives on the second line:
                   2184: <pre>
                   2185:   (?(?=[^a-z]*[a-z])
                   2186:   \d{2}-[a-z]{3}-\d{2}  |  \d{2}-\d{2}-\d{2} )
                   2187: </pre>
                   2188: The condition is a positive lookahead assertion that matches an optional
                   2189: sequence of non-letters followed by a letter. In other words, it tests for the
                   2190: presence of at least one letter in the subject. If a letter is found, the
                   2191: subject is matched against the first alternative; otherwise it is matched
                   2192: against the second. This pattern matches strings in one of the two forms
                   2193: dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
                   2194: <a name="comments"></a></P>
                   2195: <br><a name="SEC20" href="#TOC1">COMMENTS</a><br>
                   2196: <P>
                   2197: There are two ways of including comments in patterns that are processed by
                   2198: PCRE. In both cases, the start of the comment must not be in a character class,
                   2199: nor in the middle of any other sequence of related characters such as (?: or a
                   2200: subpattern name or number. The characters that make up a comment play no part
                   2201: in the pattern matching.
                   2202: </P>
                   2203: <P>
                   2204: The sequence (?# marks the start of a comment that continues up to the next
                   2205: closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
                   2206: option is set, an unescaped # character also introduces a comment, which in
                   2207: this case continues to immediately after the next newline character or
                   2208: character sequence in the pattern. Which characters are interpreted as newlines
1.1.1.2 ! misho    2209: is controlled by the options passed to a compiling function or by a special
1.1       misho    2210: sequence at the start of the pattern, as described in the section entitled
                   2211: <a href="#newlines">"Newline conventions"</a>
                   2212: above. Note that the end of this type of comment is a literal newline sequence
                   2213: in the pattern; escape sequences that happen to represent a newline do not
                   2214: count. For example, consider this pattern when PCRE_EXTENDED is set, and the
                   2215: default newline convention is in force:
                   2216: <pre>
                   2217:   abc #comment \n still comment
                   2218: </pre>
                   2219: On encountering the # character, <b>pcre_compile()</b> skips along, looking for
                   2220: a newline in the pattern. The sequence \n is still literal at this stage, so
                   2221: it does not terminate the comment. Only an actual character with the code value
                   2222: 0x0a (the default newline) does so.
                   2223: <a name="recursion"></a></P>
                   2224: <br><a name="SEC21" href="#TOC1">RECURSIVE PATTERNS</a><br>
                   2225: <P>
                   2226: Consider the problem of matching a string in parentheses, allowing for
                   2227: unlimited nested parentheses. Without the use of recursion, the best that can
                   2228: be done is to use a pattern that matches up to some fixed depth of nesting. It
                   2229: is not possible to handle an arbitrary nesting depth.
                   2230: </P>
                   2231: <P>
                   2232: For some time, Perl has provided a facility that allows regular expressions to
                   2233: recurse (amongst other things). It does this by interpolating Perl code in the
                   2234: expression at run time, and the code can refer to the expression itself. A Perl
                   2235: pattern using code interpolation to solve the parentheses problem can be
                   2236: created like this:
                   2237: <pre>
                   2238:   $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
                   2239: </pre>
                   2240: The (?p{...}) item interpolates Perl code at run time, and in this case refers
                   2241: recursively to the pattern in which it appears.
                   2242: </P>
                   2243: <P>
                   2244: Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
                   2245: supports special syntax for recursion of the entire pattern, and also for
                   2246: individual subpattern recursion. After its introduction in PCRE and Python,
                   2247: this kind of recursion was subsequently introduced into Perl at release 5.10.
                   2248: </P>
                   2249: <P>
                   2250: A special item that consists of (? followed by a number greater than zero and a
                   2251: closing parenthesis is a recursive subroutine call of the subpattern of the
                   2252: given number, provided that it occurs inside that subpattern. (If not, it is a
                   2253: <a href="#subpatternsassubroutines">non-recursive subroutine</a>
                   2254: call, which is described in the next section.) The special item (?R) or (?0) is
                   2255: a recursive call of the entire regular expression.
                   2256: </P>
                   2257: <P>
                   2258: This PCRE pattern solves the nested parentheses problem (assume the
                   2259: PCRE_EXTENDED option is set so that white space is ignored):
                   2260: <pre>
                   2261:   \( ( [^()]++ | (?R) )* \)
                   2262: </pre>
                   2263: First it matches an opening parenthesis. Then it matches any number of
                   2264: substrings which can either be a sequence of non-parentheses, or a recursive
                   2265: match of the pattern itself (that is, a correctly parenthesized substring).
                   2266: Finally there is a closing parenthesis. Note the use of a possessive quantifier
                   2267: to avoid backtracking into sequences of non-parentheses.
                   2268: </P>
                   2269: <P>
                   2270: If this were part of a larger pattern, you would not want to recurse the entire
                   2271: pattern, so instead you could use this:
                   2272: <pre>
                   2273:   ( \( ( [^()]++ | (?1) )* \) )
                   2274: </pre>
                   2275: We have put the pattern into parentheses, and caused the recursion to refer to
                   2276: them instead of the whole pattern.
                   2277: </P>
                   2278: <P>
                   2279: In a larger pattern, keeping track of parenthesis numbers can be tricky. This
                   2280: is made easier by the use of relative references. Instead of (?1) in the
                   2281: pattern above you can write (?-2) to refer to the second most recently opened
                   2282: parentheses preceding the recursion. In other words, a negative number counts
                   2283: capturing parentheses leftwards from the point at which it is encountered.
                   2284: </P>
                   2285: <P>
                   2286: It is also possible to refer to subsequently opened parentheses, by writing
                   2287: references such as (?+2). However, these cannot be recursive because the
                   2288: reference is not inside the parentheses that are referenced. They are always
                   2289: <a href="#subpatternsassubroutines">non-recursive subroutine</a>
                   2290: calls, as described in the next section.
                   2291: </P>
                   2292: <P>
                   2293: An alternative approach is to use named parentheses instead. The Perl syntax
                   2294: for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
                   2295: could rewrite the above example as follows:
                   2296: <pre>
                   2297:   (?&#60;pn&#62; \( ( [^()]++ | (?&pn) )* \) )
                   2298: </pre>
                   2299: If there is more than one subpattern with the same name, the earliest one is
                   2300: used.
                   2301: </P>
                   2302: <P>
                   2303: This particular example pattern that we have been looking at contains nested
                   2304: unlimited repeats, and so the use of a possessive quantifier for matching
                   2305: strings of non-parentheses is important when applying the pattern to strings
                   2306: that do not match. For example, when this pattern is applied to
                   2307: <pre>
                   2308:   (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
                   2309: </pre>
                   2310: it yields "no match" quickly. However, if a possessive quantifier is not used,
                   2311: the match runs for a very long time indeed because there are so many different
                   2312: ways the + and * repeats can carve up the subject, and all have to be tested
                   2313: before failure can be reported.
                   2314: </P>
                   2315: <P>
                   2316: At the end of a match, the values of capturing parentheses are those from
                   2317: the outermost level. If you want to obtain intermediate values, a callout
                   2318: function can be used (see below and the
                   2319: <a href="pcrecallout.html"><b>pcrecallout</b></a>
                   2320: documentation). If the pattern above is matched against
                   2321: <pre>
                   2322:   (ab(cd)ef)
                   2323: </pre>
                   2324: the value for the inner capturing parentheses (numbered 2) is "ef", which is
                   2325: the last value taken on at the top level. If a capturing subpattern is not
                   2326: matched at the top level, its final captured value is unset, even if it was
                   2327: (temporarily) set at a deeper level during the matching process.
                   2328: </P>
                   2329: <P>
                   2330: If there are more than 15 capturing parentheses in a pattern, PCRE has to
                   2331: obtain extra memory to store data during a recursion, which it does by using
                   2332: <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no memory can
                   2333: be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
                   2334: </P>
                   2335: <P>
                   2336: Do not confuse the (?R) item with the condition (R), which tests for recursion.
                   2337: Consider this pattern, which matches text in angle brackets, allowing for
                   2338: arbitrary nesting. Only digits are allowed in nested brackets (that is, when
                   2339: recursing), whereas any characters are permitted at the outer level.
                   2340: <pre>
                   2341:   &#60; (?: (?(R) \d++  | [^&#60;&#62;]*+) | (?R)) * &#62;
                   2342: </pre>
                   2343: In this pattern, (?(R) is the start of a conditional subpattern, with two
                   2344: different alternatives for the recursive and non-recursive cases. The (?R) item
                   2345: is the actual recursive call.
                   2346: <a name="recursiondifference"></a></P>
                   2347: <br><b>
                   2348: Differences in recursion processing between PCRE and Perl
                   2349: </b><br>
                   2350: <P>
                   2351: Recursion processing in PCRE differs from Perl in two important ways. In PCRE
                   2352: (like Python, but unlike Perl), a recursive subpattern call is always treated
                   2353: as an atomic group. That is, once it has matched some of the subject string, it
                   2354: is never re-entered, even if it contains untried alternatives and there is a
                   2355: subsequent matching failure. This can be illustrated by the following pattern,
                   2356: which purports to match a palindromic string that contains an odd number of
                   2357: characters (for example, "a", "aba", "abcba", "abcdcba"):
                   2358: <pre>
                   2359:   ^(.|(.)(?1)\2)$
                   2360: </pre>
                   2361: The idea is that it either matches a single character, or two identical
                   2362: characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
                   2363: it does not if the pattern is longer than three characters. Consider the
                   2364: subject string "abcba":
                   2365: </P>
                   2366: <P>
                   2367: At the top level, the first character is matched, but as it is not at the end
                   2368: of the string, the first alternative fails; the second alternative is taken
                   2369: and the recursion kicks in. The recursive call to subpattern 1 successfully
                   2370: matches the next character ("b"). (Note that the beginning and end of line
                   2371: tests are not part of the recursion).
                   2372: </P>
                   2373: <P>
                   2374: Back at the top level, the next character ("c") is compared with what
                   2375: subpattern 2 matched, which was "a". This fails. Because the recursion is
                   2376: treated as an atomic group, there are now no backtracking points, and so the
                   2377: entire match fails. (Perl is able, at this point, to re-enter the recursion and
                   2378: try the second alternative.) However, if the pattern is written with the
                   2379: alternatives in the other order, things are different:
                   2380: <pre>
                   2381:   ^((.)(?1)\2|.)$
                   2382: </pre>
                   2383: This time, the recursing alternative is tried first, and continues to recurse
                   2384: until it runs out of characters, at which point the recursion fails. But this
                   2385: time we do have another alternative to try at the higher level. That is the big
                   2386: difference: in the previous case the remaining alternative is at a deeper
                   2387: recursion level, which PCRE cannot use.
                   2388: </P>
                   2389: <P>
                   2390: To change the pattern so that it matches all palindromic strings, not just
                   2391: those with an odd number of characters, it is tempting to change the pattern to
                   2392: this:
                   2393: <pre>
                   2394:   ^((.)(?1)\2|.?)$
                   2395: </pre>
                   2396: Again, this works in Perl, but not in PCRE, and for the same reason. When a
                   2397: deeper recursion has matched a single character, it cannot be entered again in
                   2398: order to match an empty string. The solution is to separate the two cases, and
                   2399: write out the odd and even cases as alternatives at the higher level:
                   2400: <pre>
                   2401:   ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
                   2402: </pre>
                   2403: If you want to match typical palindromic phrases, the pattern has to ignore all
                   2404: non-word characters, which can be done like this:
                   2405: <pre>
                   2406:   ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
                   2407: </pre>
                   2408: If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
                   2409: man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
                   2410: the use of the possessive quantifier *+ to avoid backtracking into sequences of
                   2411: non-word characters. Without this, PCRE takes a great deal longer (ten times or
                   2412: more) to match typical phrases, and Perl takes so long that you think it has
                   2413: gone into a loop.
                   2414: </P>
                   2415: <P>
                   2416: <b>WARNING</b>: The palindrome-matching patterns above work only if the subject
                   2417: string does not start with a palindrome that is shorter than the entire string.
                   2418: For example, although "abcba" is correctly matched, if the subject is "ababa",
                   2419: PCRE finds the palindrome "aba" at the start, then fails at top level because
                   2420: the end of the string does not follow. Once again, it cannot jump back into the
                   2421: recursion to try other alternatives, so the entire match fails.
                   2422: </P>
                   2423: <P>
                   2424: The second way in which PCRE and Perl differ in their recursion processing is
                   2425: in the handling of captured values. In Perl, when a subpattern is called
                   2426: recursively or as a subpattern (see the next section), it has no access to any
                   2427: values that were captured outside the recursion, whereas in PCRE these values
                   2428: can be referenced. Consider this pattern:
                   2429: <pre>
                   2430:   ^(.)(\1|a(?2))
                   2431: </pre>
                   2432: In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
                   2433: then in the second group, when the back reference \1 fails to match "b", the
                   2434: second alternative matches "a" and then recurses. In the recursion, \1 does
                   2435: now match "b" and so the whole match succeeds. In Perl, the pattern fails to
                   2436: match because inside the recursive call \1 cannot access the externally set
                   2437: value.
                   2438: <a name="subpatternsassubroutines"></a></P>
                   2439: <br><a name="SEC22" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
                   2440: <P>
                   2441: If the syntax for a recursive subpattern call (either by number or by
                   2442: name) is used outside the parentheses to which it refers, it operates like a
                   2443: subroutine in a programming language. The called subpattern may be defined
                   2444: before or after the reference. A numbered reference can be absolute or
                   2445: relative, as in these examples:
                   2446: <pre>
                   2447:   (...(absolute)...)...(?2)...
                   2448:   (...(relative)...)...(?-1)...
                   2449:   (...(?+1)...(relative)...
                   2450: </pre>
                   2451: An earlier example pointed out that the pattern
                   2452: <pre>
                   2453:   (sens|respons)e and \1ibility
                   2454: </pre>
                   2455: matches "sense and sensibility" and "response and responsibility", but not
                   2456: "sense and responsibility". If instead the pattern
                   2457: <pre>
                   2458:   (sens|respons)e and (?1)ibility
                   2459: </pre>
                   2460: is used, it does match "sense and responsibility" as well as the other two
                   2461: strings. Another example is given in the discussion of DEFINE above.
                   2462: </P>
                   2463: <P>
                   2464: All subroutine calls, whether recursive or not, are always treated as atomic
                   2465: groups. That is, once a subroutine has matched some of the subject string, it
                   2466: is never re-entered, even if it contains untried alternatives and there is a
                   2467: subsequent matching failure. Any capturing parentheses that are set during the
                   2468: subroutine call revert to their previous values afterwards.
                   2469: </P>
                   2470: <P>
                   2471: Processing options such as case-independence are fixed when a subpattern is
                   2472: defined, so if it is used as a subroutine, such options cannot be changed for
                   2473: different calls. For example, consider this pattern:
                   2474: <pre>
                   2475:   (abc)(?i:(?-1))
                   2476: </pre>
                   2477: It matches "abcabc". It does not match "abcABC" because the change of
                   2478: processing option does not affect the called subpattern.
                   2479: <a name="onigurumasubroutines"></a></P>
                   2480: <br><a name="SEC23" href="#TOC1">ONIGURUMA SUBROUTINE SYNTAX</a><br>
                   2481: <P>
                   2482: For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
                   2483: a number enclosed either in angle brackets or single quotes, is an alternative
                   2484: syntax for referencing a subpattern as a subroutine, possibly recursively. Here
                   2485: are two of the examples used above, rewritten using this syntax:
                   2486: <pre>
                   2487:   (?&#60;pn&#62; \( ( (?&#62;[^()]+) | \g&#60;pn&#62; )* \) )
                   2488:   (sens|respons)e and \g'1'ibility
                   2489: </pre>
                   2490: PCRE supports an extension to Oniguruma: if a number is preceded by a
                   2491: plus or a minus sign it is taken as a relative reference. For example:
                   2492: <pre>
                   2493:   (abc)(?i:\g&#60;-1&#62;)
                   2494: </pre>
                   2495: Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
                   2496: synonymous. The former is a back reference; the latter is a subroutine call.
                   2497: </P>
                   2498: <br><a name="SEC24" href="#TOC1">CALLOUTS</a><br>
                   2499: <P>
                   2500: Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
                   2501: code to be obeyed in the middle of matching a regular expression. This makes it
                   2502: possible, amongst other things, to extract different substrings that match the
                   2503: same pair of parentheses when there is a repetition.
                   2504: </P>
                   2505: <P>
                   2506: PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
                   2507: code. The feature is called "callout". The caller of PCRE provides an external
1.1.1.2 ! misho    2508: function by putting its entry point in the global variable <i>pcre_callout</i>
        !          2509: (8-bit library) or <i>pcre16_callout</i> (16-bit library). By default, this
        !          2510: variable contains NULL, which disables all calling out.
1.1       misho    2511: </P>
                   2512: <P>
                   2513: Within a regular expression, (?C) indicates the points at which the external
                   2514: function is to be called. If you want to identify different callout points, you
                   2515: can put a number less than 256 after the letter C. The default value is zero.
                   2516: For example, this pattern has two callout points:
                   2517: <pre>
                   2518:   (?C1)abc(?C2)def
                   2519: </pre>
1.1.1.2 ! misho    2520: If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
1.1       misho    2521: automatically installed before each item in the pattern. They are all numbered
                   2522: 255.
                   2523: </P>
                   2524: <P>
1.1.1.2 ! misho    2525: During matching, when PCRE reaches a callout point, the external function is
        !          2526: called. It is provided with the number of the callout, the position in the
        !          2527: pattern, and, optionally, one item of data originally supplied by the caller of
        !          2528: the matching function. The callout function may cause matching to proceed, to
        !          2529: backtrack, or to fail altogether. A complete description of the interface to
        !          2530: the callout function is given in the
1.1       misho    2531: <a href="pcrecallout.html"><b>pcrecallout</b></a>
                   2532: documentation.
                   2533: <a name="backtrackcontrol"></a></P>
                   2534: <br><a name="SEC25" href="#TOC1">BACKTRACKING CONTROL</a><br>
                   2535: <P>
                   2536: Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
                   2537: are described in the Perl documentation as "experimental and subject to change
                   2538: or removal in a future version of Perl". It goes on to say: "Their usage in
                   2539: production code should be noted to avoid problems during upgrades." The same
                   2540: remarks apply to the PCRE features described in this section.
                   2541: </P>
                   2542: <P>
                   2543: Since these verbs are specifically related to backtracking, most of them can be
1.1.1.2 ! misho    2544: used only when the pattern is to be matched using one of the traditional
        !          2545: matching functions, which use a backtracking algorithm. With the exception of
        !          2546: (*FAIL), which behaves like a failing negative assertion, they cause an error
        !          2547: if encountered by a DFA matching function.
1.1       misho    2548: </P>
                   2549: <P>
                   2550: If any of these verbs are used in an assertion or in a subpattern that is
                   2551: called as a subroutine (whether or not recursively), their effect is confined
                   2552: to that subpattern; it does not extend to the surrounding pattern, with one
                   2553: exception: the name from a *(MARK), (*PRUNE), or (*THEN) that is encountered in
                   2554: a successful positive assertion <i>is</i> passed back when a match succeeds
                   2555: (compare capturing parentheses in assertions). Note that such subpatterns are
                   2556: processed as anchored at the point where they are tested. Note also that Perl's
                   2557: treatment of subroutines is different in some cases.
                   2558: </P>
                   2559: <P>
                   2560: The new verbs make use of what was previously invalid syntax: an opening
                   2561: parenthesis followed by an asterisk. They are generally of the form
                   2562: (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
                   2563: depending on whether or not an argument is present. A name is any sequence of
                   2564: characters that does not include a closing parenthesis. If the name is empty,
                   2565: that is, if the closing parenthesis immediately follows the colon, the effect
                   2566: is as if the colon were not there. Any number of these verbs may occur in a
                   2567: pattern.
                   2568: </P>
                   2569: <P>
                   2570: PCRE contains some optimizations that are used to speed up matching by running
                   2571: some checks at the start of each match attempt. For example, it may know the
                   2572: minimum length of matching subject, or that a particular character must be
                   2573: present. When one of these optimizations suppresses the running of a match, any
                   2574: included backtracking verbs will not, of course, be processed. You can suppress
                   2575: the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
                   2576: when calling <b>pcre_compile()</b> or <b>pcre_exec()</b>, or by starting the
                   2577: pattern with (*NO_START_OPT).
                   2578: </P>
                   2579: <P>
                   2580: Experiments with Perl suggest that it too has similar optimizations, sometimes
                   2581: leading to anomalous results.
                   2582: </P>
                   2583: <br><b>
                   2584: Verbs that act immediately
                   2585: </b><br>
                   2586: <P>
                   2587: The following verbs act as soon as they are encountered. They may not be
                   2588: followed by a name.
                   2589: <pre>
                   2590:    (*ACCEPT)
                   2591: </pre>
                   2592: This verb causes the match to end successfully, skipping the remainder of the
                   2593: pattern. However, when it is inside a subpattern that is called as a
                   2594: subroutine, only that subpattern is ended successfully. Matching then continues
                   2595: at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
                   2596: far is captured. For example:
                   2597: <pre>
                   2598:   A((?:A|B(*ACCEPT)|C)D)
                   2599: </pre>
                   2600: This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
                   2601: the outer parentheses.
                   2602: <pre>
                   2603:   (*FAIL) or (*F)
                   2604: </pre>
                   2605: This verb causes a matching failure, forcing backtracking to occur. It is
                   2606: equivalent to (?!) but easier to read. The Perl documentation notes that it is
                   2607: probably useful only when combined with (?{}) or (??{}). Those are, of course,
                   2608: Perl features that are not present in PCRE. The nearest equivalent is the
                   2609: callout feature, as for example in this pattern:
                   2610: <pre>
                   2611:   a+(?C)(*FAIL)
                   2612: </pre>
                   2613: A match with the string "aaaa" always fails, but the callout is taken before
                   2614: each backtrack happens (in this example, 10 times).
                   2615: </P>
                   2616: <br><b>
                   2617: Recording which path was taken
                   2618: </b><br>
                   2619: <P>
                   2620: There is one verb whose main purpose is to track how a match was arrived at,
                   2621: though it also has a secondary use in conjunction with advancing the match
                   2622: starting point (see (*SKIP) below).
                   2623: <pre>
                   2624:   (*MARK:NAME) or (*:NAME)
                   2625: </pre>
                   2626: A name is always required with this verb. There may be as many instances of
                   2627: (*MARK) as you like in a pattern, and their names do not have to be unique.
                   2628: </P>
                   2629: <P>
                   2630: When a match succeeds, the name of the last-encountered (*MARK) on the matching
1.1.1.2 ! misho    2631: path is passed back to the caller as described in the section entitled
        !          2632: <a href="pcreapi.html#extradata">"Extra data for <b>pcre_exec()</b>"</a>
1.1       misho    2633: in the
                   2634: <a href="pcreapi.html"><b>pcreapi</b></a>
                   2635: documentation. Here is an example of <b>pcretest</b> output, where the /K
                   2636: modifier requests the retrieval and outputting of (*MARK) data:
                   2637: <pre>
                   2638:     re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
                   2639:   data&#62; XY
                   2640:    0: XY
                   2641:   MK: A
                   2642:   XZ
                   2643:    0: XZ
                   2644:   MK: B
                   2645: </pre>
                   2646: The (*MARK) name is tagged with "MK:" in this output, and in this example it
                   2647: indicates which of the two alternatives matched. This is a more efficient way
                   2648: of obtaining this information than putting each alternative in its own
                   2649: capturing parentheses.
                   2650: </P>
                   2651: <P>
                   2652: If (*MARK) is encountered in a positive assertion, its name is recorded and
                   2653: passed back if it is the last-encountered. This does not happen for negative
                   2654: assertions.
                   2655: </P>
                   2656: <P>
                   2657: After a partial match or a failed match, the name of the last encountered
                   2658: (*MARK) in the entire match process is returned. For example:
                   2659: <pre>
                   2660:     re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
                   2661:   data&#62; XP
                   2662:   No match, mark = B
                   2663: </pre>
                   2664: Note that in this unanchored example the mark is retained from the match
                   2665: attempt that started at the letter "X". Subsequent match attempts starting at
                   2666: "P" and then with an empty string do not get as far as the (*MARK) item, but
                   2667: nevertheless do not reset it.
                   2668: </P>
                   2669: <br><b>
                   2670: Verbs that act after backtracking
                   2671: </b><br>
                   2672: <P>
                   2673: The following verbs do nothing when they are encountered. Matching continues
                   2674: with what follows, but if there is no subsequent match, causing a backtrack to
                   2675: the verb, a failure is forced. That is, backtracking cannot pass to the left of
                   2676: the verb. However, when one of these verbs appears inside an atomic group, its
                   2677: effect is confined to that group, because once the group has been matched,
                   2678: there is never any backtracking into it. In this situation, backtracking can
                   2679: "jump back" to the left of the entire atomic group. (Remember also, as stated
                   2680: above, that this localization also applies in subroutine calls and assertions.)
                   2681: </P>
                   2682: <P>
                   2683: These verbs differ in exactly what kind of failure occurs when backtracking
                   2684: reaches them.
                   2685: <pre>
                   2686:   (*COMMIT)
                   2687: </pre>
                   2688: This verb, which may not be followed by a name, causes the whole match to fail
                   2689: outright if the rest of the pattern does not match. Even if the pattern is
                   2690: unanchored, no further attempts to find a match by advancing the starting point
                   2691: take place. Once (*COMMIT) has been passed, <b>pcre_exec()</b> is committed to
                   2692: finding a match at the current starting point, or not at all. For example:
                   2693: <pre>
                   2694:   a+(*COMMIT)b
                   2695: </pre>
                   2696: This matches "xxaab" but not "aacaab". It can be thought of as a kind of
                   2697: dynamic anchor, or "I've started, so I must finish." The name of the most
                   2698: recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
                   2699: match failure.
                   2700: </P>
                   2701: <P>
                   2702: Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
                   2703: unless PCRE's start-of-match optimizations are turned off, as shown in this
                   2704: <b>pcretest</b> example:
                   2705: <pre>
                   2706:     re&#62; /(*COMMIT)abc/
                   2707:   data&#62; xyzabc
                   2708:    0: abc
                   2709:   xyzabc\Y
                   2710:   No match
                   2711: </pre>
                   2712: PCRE knows that any match must start with "a", so the optimization skips along
                   2713: the subject to "a" before running the first match attempt, which succeeds. When
                   2714: the optimization is disabled by the \Y escape in the second subject, the match
                   2715: starts at "x" and so the (*COMMIT) causes it to fail without trying any other
                   2716: starting points.
                   2717: <pre>
                   2718:   (*PRUNE) or (*PRUNE:NAME)
                   2719: </pre>
                   2720: This verb causes the match to fail at the current starting position in the
                   2721: subject if the rest of the pattern does not match. If the pattern is
                   2722: unanchored, the normal "bumpalong" advance to the next starting character then
                   2723: happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
                   2724: reached, or when matching to the right of (*PRUNE), but if there is no match to
                   2725: the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
                   2726: (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
                   2727: but there are some uses of (*PRUNE) that cannot be expressed in any other way.
                   2728: The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE). In an
                   2729: anchored pattern (*PRUNE) has the same effect as (*COMMIT).
                   2730: <pre>
                   2731:   (*SKIP)
                   2732: </pre>
                   2733: This verb, when given without a name, is like (*PRUNE), except that if the
                   2734: pattern is unanchored, the "bumpalong" advance is not to the next character,
                   2735: but to the position in the subject where (*SKIP) was encountered. (*SKIP)
                   2736: signifies that whatever text was matched leading up to it cannot be part of a
                   2737: successful match. Consider:
                   2738: <pre>
                   2739:   a+(*SKIP)b
                   2740: </pre>
                   2741: If the subject is "aaaac...", after the first match attempt fails (starting at
                   2742: the first character in the string), the starting point skips on to start the
                   2743: next attempt at "c". Note that a possessive quantifer does not have the same
                   2744: effect as this example; although it would suppress backtracking during the
                   2745: first match attempt, the second attempt would start at the second character
                   2746: instead of skipping on to "c".
                   2747: <pre>
                   2748:   (*SKIP:NAME)
                   2749: </pre>
                   2750: When (*SKIP) has an associated name, its behaviour is modified. If the
                   2751: following pattern fails to match, the previous path through the pattern is
                   2752: searched for the most recent (*MARK) that has the same name. If one is found,
                   2753: the "bumpalong" advance is to the subject position that corresponds to that
                   2754: (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
                   2755: matching name is found, the (*SKIP) is ignored.
                   2756: <pre>
                   2757:   (*THEN) or (*THEN:NAME)
                   2758: </pre>
                   2759: This verb causes a skip to the next innermost alternative if the rest of the
                   2760: pattern does not match. That is, it cancels pending backtracking, but only
                   2761: within the current alternative. Its name comes from the observation that it can
                   2762: be used for a pattern-based if-then-else block:
                   2763: <pre>
                   2764:   ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
                   2765: </pre>
                   2766: If the COND1 pattern matches, FOO is tried (and possibly further items after
                   2767: the end of the group if FOO succeeds); on failure, the matcher skips to the
                   2768: second alternative and tries COND2, without backtracking into COND1. The
                   2769: behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN).
                   2770: If (*THEN) is not inside an alternation, it acts like (*PRUNE).
                   2771: </P>
                   2772: <P>
                   2773: Note that a subpattern that does not contain a | character is just a part of
                   2774: the enclosing alternative; it is not a nested alternation with only one
                   2775: alternative. The effect of (*THEN) extends beyond such a subpattern to the
                   2776: enclosing alternative. Consider this pattern, where A, B, etc. are complex
                   2777: pattern fragments that do not contain any | characters at this level:
                   2778: <pre>
                   2779:   A (B(*THEN)C) | D
                   2780: </pre>
                   2781: If A and B are matched, but there is a failure in C, matching does not
                   2782: backtrack into A; instead it moves to the next alternative, that is, D.
                   2783: However, if the subpattern containing (*THEN) is given an alternative, it
                   2784: behaves differently:
                   2785: <pre>
                   2786:   A (B(*THEN)C | (*FAIL)) | D
                   2787: </pre>
                   2788: The effect of (*THEN) is now confined to the inner subpattern. After a failure
                   2789: in C, matching moves to (*FAIL), which causes the whole subpattern to fail
                   2790: because there are no more alternatives to try. In this case, matching does now
                   2791: backtrack into A.
                   2792: </P>
                   2793: <P>
                   2794: Note also that a conditional subpattern is not considered as having two
                   2795: alternatives, because only one is ever used. In other words, the | character in
                   2796: a conditional subpattern has a different meaning. Ignoring white space,
                   2797: consider:
                   2798: <pre>
                   2799:   ^.*? (?(?=a) a | b(*THEN)c )
                   2800: </pre>
                   2801: If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
                   2802: it initially matches zero characters. The condition (?=a) then fails, the
                   2803: character "b" is matched, but "c" is not. At this point, matching does not
                   2804: backtrack to .*? as might perhaps be expected from the presence of the |
                   2805: character. The conditional subpattern is part of the single alternative that
                   2806: comprises the whole pattern, and so the match fails. (If there was a backtrack
                   2807: into .*?, allowing it to match "b", the match would succeed.)
                   2808: </P>
                   2809: <P>
                   2810: The verbs just described provide four different "strengths" of control when
                   2811: subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
                   2812: next alternative. (*PRUNE) comes next, failing the match at the current
                   2813: starting position, but allowing an advance to the next character (for an
                   2814: unanchored pattern). (*SKIP) is similar, except that the advance may be more
                   2815: than one character. (*COMMIT) is the strongest, causing the entire match to
                   2816: fail.
                   2817: </P>
                   2818: <P>
                   2819: If more than one such verb is present in a pattern, the "strongest" one wins.
                   2820: For example, consider this pattern, where A, B, etc. are complex pattern
                   2821: fragments:
                   2822: <pre>
                   2823:   (A(*COMMIT)B(*THEN)C|D)
                   2824: </pre>
                   2825: Once A has matched, PCRE is committed to this match, at the current starting
                   2826: position. If subsequently B matches, but C does not, the normal (*THEN) action
                   2827: of trying the next alternative (that is, D) does not happen because (*COMMIT)
                   2828: overrides.
                   2829: </P>
                   2830: <br><a name="SEC26" href="#TOC1">SEE ALSO</a><br>
                   2831: <P>
                   2832: <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3),
1.1.1.2 ! misho    2833: <b>pcresyntax</b>(3), <b>pcre</b>(3), <b>pcre16(3)</b>.
1.1       misho    2834: </P>
                   2835: <br><a name="SEC27" href="#TOC1">AUTHOR</a><br>
                   2836: <P>
                   2837: Philip Hazel
                   2838: <br>
                   2839: University Computing Service
                   2840: <br>
                   2841: Cambridge CB2 3QH, England.
                   2842: <br>
                   2843: </P>
                   2844: <br><a name="SEC28" href="#TOC1">REVISION</a><br>
                   2845: <P>
1.1.1.2 ! misho    2846: Last updated: 09 January 2012
1.1       misho    2847: <br>
1.1.1.2 ! misho    2848: Copyright &copy; 1997-2012 University of Cambridge.
1.1       misho    2849: <br>
                   2850: <p>
                   2851: Return to the <a href="index.html">PCRE index page</a>.
                   2852: </p>

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>