<?xml version="1.0" encoding="iso-8859-1"?>
<!-- $Revision: 1.13 $ -->
<!-- splitted from ./en/functions/pcre.xml, last change in rev 1.2 -->
  <refentry id="reference.pcre.pattern.syntax">
   <refnamediv>
    <refname>Pattern Syntax</refname>
    <refpurpose>Describes PCRE regex syntax</refpurpose>
   </refnamediv>

   <refsect1>
    <title>Description</title>
    <simpara>
     The PCRE library is a set of functions that implement regular
     expression pattern matching using the same syntax and semantics
     as Perl 5, with just a few differences (see below).  The current
     implementation corresponds to Perl 5.005.
    </simpara>
   </refsect1>

   <refsect1>
    <title>Differences From Perl</title>
    <para>
     The differences described here are with respect to Perl 5.005.
    <orderedlist>
     <listitem>
      <simpara>
       By default, a whitespace character is any character  that
       the C library function isspace() recognizes, though it is
       possible to compile PCRE  with alternative character type
       tables. Normally isspace() matches space, formfeed, newline,
       carriage return, horizontal tab, and vertical tab. Perl 5 no
       longer  includes vertical tab in its set of whitespace characters.
       The \v escape that was in the Perl documentation for
       a long time was never in fact recognized. However, the character
       itself was treated as whitespace at least up to 5.002.
       In 5.004 and 5.005 it does not match \s.
      </simpara>
     </listitem>
     <listitem>
      <simpara>
       PCRE does not allow repeat quantifiers on lookahead
       assertions. Perl permits them, but they do not mean what you
       might think. For example, (?!a){3} does not assert that  the
       next three characters are not "a". It just asserts that the
       next character is not "a" three times.
      </simpara>
     </listitem>
     <listitem>
      <simpara>
       Capturing subpatterns that occur inside negative
       lookahead assertions are counted, but their entries in the
       offsets vector are never set. Perl sets its numerical
       variables from any such patterns that are matched before the
       assertion fails to match something (thereby succeeding), but
       only  if  the negative lookahead assertion contains just one
       branch.
      </simpara>
     </listitem>
     <listitem>
      <simpara>
       Though binary zero characters are supported in  the  subject  string,
       they are not allowed in a pattern string because it is passed as a
       normal C string, terminated  by zero. The escape sequence "\x00" can
       be used in the pattern to represent a binary zero.
      </simpara>
      </listitem>
      <listitem>
      <simpara>
       The following Perl escape sequences  are  not  supported:
       \l,  \u,  \L,  \U. In fact these are implemented by
       Perl's general string-handling and are not part of its
       pattern matching engine.
      </simpara>
      </listitem>
      <listitem>
      <simpara>
       The Perl \G assertion is  not  supported  as  it  is  not
       relevant to single pattern matches.
      </simpara>
      </listitem>
      <listitem>
      <simpara>
       Fairly obviously, PCRE does not support the (?{code})
       construction.
      </simpara>
      </listitem>
      <listitem>
      <simpara>
       There are at the time of writing some  oddities  in  Perl
       5.005_02  concerned  with  the  settings of captured strings
       when part of a pattern is repeated.  For  example,  matching
       "aba"  against the pattern /^(a(b)?)+$/ sets $2 to the value
       "b", but matching "aabbaa" against /^(aa(bb)?)+$/ leaves  $2
       unset.    However,    if   the   pattern   is   changed   to
       /^(aa(b(b))?)+$/ then $2 (and $3) get set.
       In Perl 5.004 $2 is set in both cases, and that is also &true;
       of PCRE. If in the future Perl changes to a consistent state
       that is different, PCRE may change to follow.
      </simpara>
      </listitem>
      <listitem>
      <simpara>
       Another as yet unresolved discrepancy  is  that  in  Perl
       5.005_02  the  pattern /^(a)?(?(1)a|b)+$/ matches the string
       "a", whereas in PCRE it does not.  However, in both Perl and
       PCRE /^(a)?a/ matched against "a" leaves $1 unset.
      </simpara>
      </listitem>
      <listitem>
      <para>
       PCRE  provides  some  extensions  to  the  Perl  regular
       expression facilities:
        <orderedlist>
         <listitem>
          <simpara>
           Although lookbehind assertions must match  fixed  length
           strings,  each  alternative branch of a lookbehind assertion
           can match a different length of string. Perl 5.005  requires
           them all to have the same length.
         </simpara>
        </listitem>
        <listitem>
         <simpara>
          If <link linkend="reference.pcre.pattern.modifiers">PCRE_DOLLAR_ENDONLY</link>
          is set and <link linkend="reference.pcre.pattern.modifiers">PCRE_MULTILINE</link> is
          not set, the $ meta-character matches only at the very end of the
          string.
         </simpara>
        </listitem>
        <listitem>
         <simpara>
          If <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTRA</link> is
          set, a backslash followed by a letter with no special meaning is
          faulted.
         </simpara>
        </listitem>
        <listitem>
         <simpara>
          If <link linkend="reference.pcre.pattern.modifiers">PCRE_UNGREEDY</link> is
          set, the greediness of the repetition  quantifiers  is inverted,
          that is, by default they are not greedy, but if followed by a
          question mark they are.
         </simpara>
        </listitem>
       </orderedlist>
      </para>
     </listitem>
    </orderedlist>
   </para>
   </refsect1>

   <refsect1 id="regexp.reference">
    <title>Regular Expression Details</title>
    <refsect2 id="regexp.introduction">
     <title>Introduction</title>
     <para>
      The syntax and semantics of  the  regular  expressions
      supported  by PCRE are described below. Regular expressions are
      also described in the Perl documentation and in a number  of
      other  books,  some  of which have copious examples. Jeffrey
      Friedl's  "Mastering  Regular  Expressions",  published   by
      O'Reilly  (ISBN 1-56592-257-3), covers them in great detail.
      The description here is intended as reference documentation.
     </para>
     <para>
      A regular expression is a pattern that is matched against  a
      subject string from left to right. Most characters stand for
      themselves in a pattern, and match the corresponding
      characters in the subject. As a trivial example, the pattern
      <literal>The quick brown fox</literal>
      matches a portion of a subject string that is  identical  to
      itself.  
     </para>
    </refsect2>
    <refsect2 id="regexp.reference.meta">
     <title>Meta-characters</title>
     <para>     
      The  power  of  regular  expressions comes from the
      ability to include alternatives and repetitions in the
      pattern.  These  are encoded in the pattern by the use of 
      <emphasis>meta-characters</emphasis>, which do not stand for  themselves  but  instead
      are interpreted in some special way.
     </para>
     <para>
      There are two different sets of meta-characters: those  that
      are  recognized anywhere in the pattern except within square
      brackets, and those that are recognized in square brackets.
      Outside square brackets, the meta-characters are as follows:
      <variablelist>
       <varlistentry>
        <term><emphasis>\</emphasis></term>
        <listitem><simpara>general escape character with several uses</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>^</emphasis></term>
        <listitem><simpara>assert start of subject (or line, in multiline mode)</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>$</emphasis></term>
        <listitem><simpara>assert end of subject (or line, in multiline mode)</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>.</emphasis></term>
        <listitem><simpara>match any character except newline (by default)</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>[</emphasis></term>
        <listitem><simpara>start character class definition</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>]</emphasis></term>
        <listitem><simpara>end character class definition</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>|</emphasis></term>
        <listitem><simpara>start of alternative branch</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>(</emphasis></term>
        <listitem><simpara>start subpattern</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>)</emphasis></term>
        <listitem><simpara>end subpattern</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>?</emphasis></term>
        <listitem><simpara>extends the meaning of (, also 0 or 1 quantifier, also quantifier minimizer</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>*</emphasis></term>
        <listitem><simpara>0 or more quantifier</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>+</emphasis></term>
        <listitem><simpara>1 or more quantifier</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>{</emphasis></term>
        <listitem><simpara>start min/max quantifier</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>}</emphasis></term>
        <listitem><simpara>end min/max quantifier</simpara></listitem>
       </varlistentry>
      </variablelist>

      Part of a pattern that is in square brackets is called a
      "character class". In a character class the only
      meta-characters are:

      <variablelist>
       <varlistentry>
        <term><emphasis>\</emphasis></term>
        <listitem><simpara>general escape character</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>^</emphasis></term>
        <listitem><simpara>negate the class, but only if the first character</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>-</emphasis></term>
        <listitem><simpara>indicates character range</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>]</emphasis></term>
        <listitem><simpara>terminates the character class</simpara></listitem>
       </varlistentry>
      </variablelist>

      The following sections describe the use of each of the
      meta-characters.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.backslash">
     <title>Backslash</title>
     <para>
      The backslash character has several uses. Firstly, if it  is
      followed by a non-alphanumeric character, it takes away any
      special  meaning that character may have. This use of
      backslash as an escape character applies both inside and
      outside character classes.
     </para>
     <para>
      For example, if you want to match a "*" character, you write
      "\*" in the pattern. This applies whether or not the
      following character would otherwise be interpreted as a
      meta-character, so it is always safe to precede a non-alphanumeric
      with "\" to specify that it stands for itself.  In
      particular, if you want to match a backslash, you write "\\".
     </para>
     <note>
      <para>
       Single and double quoted PHP <link
       linkend="language.types.string.syntax">strings</link> have special
       meaning of backslash. Thus if \ has to be matched with a regular
       expression \\, then "\\\\" or '\\\\' must be used in PHP code.
      </para>
     </note>
     <para>
      If a pattern is compiled with the
      <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTENDED</link> option,
      whitespace in the pattern (other than in a character class) and
      characters between a "#" outside a character class and the next newline
      character are ignored. An escaping backslash can be used to include a
      whitespace or "#" character as part of the pattern.
     </para>
     <para>
      A second use of backslash provides a way of encoding
      non-printing characters in patterns in a visible manner. There
      is no restriction on the appearance of non-printing  characters,
      apart from the binary zero that terminates a pattern,
      but when a pattern is being prepared by text editing, it is
      usually  easier to use one of the following escape sequences
      than the binary character it represents:
     </para>
     <para>
      <variablelist>
       <varlistentry>
        <term><emphasis>\a</emphasis></term>
        <listitem><simpara>alarm, that is, the BEL character (hex 07)</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>\cx</emphasis></term>
        <listitem><simpara>"control-x", where x is any character</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>\e</emphasis></term>
        <listitem><simpara>escape (hex 1B)</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>\f</emphasis></term>
        <listitem><simpara>formfeed (hex 0C)</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>\n</emphasis></term>
        <listitem><simpara>newline (hex 0A)</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>\r</emphasis></term>
        <listitem><simpara>carriage return (hex 0D)</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>\t</emphasis></term>
        <listitem><simpara>tab (hex 09)</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>\xhh</emphasis></term>
        <listitem><simpara>character with hex code hh</simpara></listitem>
       </varlistentry>
       <varlistentry>
        <term><emphasis>\ddd</emphasis></term>
        <listitem><simpara>character with octal code ddd, or backreference</simpara></listitem>
       </varlistentry>
      </variablelist>
     </para>
    <para>
     The precise effect of "<literal>\cx</literal>" is as follows: 
     if "<literal>x</literal>" is a lower case  letter, it is converted
     to upper case. Then bit 6 of the character (hex 40) is inverted. 
     Thus "<literal>\cz</literal>" becomes  hex 1A, but
     "<literal>\c{</literal>" becomes hex 3B, while "<literal>\c;</literal>"
     becomes hex 7B.
    </para>
    <para>
     After "<literal>\x</literal>", up to two hexadecimal digits are
     read (letters can be in upper or lower case).
     In <emphasis>UTF-8 mode</emphasis>, "<literal>\x{...}</literal>" is
     allowed, where the contents of the braces is a string of hexadecimal
     digits. It is interpreted as a UTF-8 character whose code number is the
     given hexadecimal number. The original hexadecimal escape sequence,
     <literal>\xhh</literal>, matches a two-byte UTF-8 character if the value
     is greater than 127.
    </para>
    <para>
     After "<literal>\0</literal>" up to two further octal digits are read.
     In  both cases,  if  there are fewer than two digits, just those that
     are present are used. Thus the sequence "<literal>\0\x\07</literal>" 
     specifies two binary zeros followed by a BEL character. Make sure you
     supply two digits after the initial zero if the character
     that follows is itself an octal digit.
    </para>
    <para>
     The handling of a backslash followed by a digit other than 0
     is complicated. Outside a character class, PCRE reads it
     and any following digits as a decimal number. If the  number
     is  less  than  10, or if there have been at least that many
     previous capturing left parentheses in the  expression,  the
     entire  sequence is taken as a <emphasis>back</emphasis> 
     <emphasis>reference</emphasis>. A description
     of how this works is given later, following  the  discussion
     of parenthesized subpatterns.
    </para>
    <para>
     Inside a character  class,  or  if  the  decimal  number  is
     greater than 9 and there have not been that many capturing
     subpatterns, PCRE re-reads up to three octal digits following 
     the backslash, and generates a single byte from the
     least significant 8 bits of the value. Any subsequent digits
     stand for themselves.  For example:
    </para>
    <para>
     <variablelist>
      <varlistentry>
       <term><emphasis>\040</emphasis></term>
       <listitem><simpara>is another way of writing a space</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\40</emphasis></term>
       <listitem>
        <simpara>
         is the same, provided there are fewer than 40
         previous capturing subpatterns
        </simpara>
       </listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\7</emphasis></term>
       <listitem><simpara>is always a back reference</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\11</emphasis></term>
       <listitem>
        <simpara>
         might be a back reference, or another way of
         writing a tab
        </simpara>
       </listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\011</emphasis></term>
       <listitem><simpara>is always a tab</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\0113</emphasis></term>
       <listitem><simpara>is a tab followed by the character "3"</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\113</emphasis></term>
       <listitem>
        <simpara>
         is the character with octal code 113 (since there
         can be no more than 99 back references)
        </simpara>
       </listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\377</emphasis></term>
       <listitem><simpara>is a byte consisting entirely of 1 bits</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\81</emphasis></term>
       <listitem>
        <simpara>
         is either a back reference, or a binary zero
         followed by the two characters "8" and "1"
        </simpara>
       </listitem>
      </varlistentry>
     </variablelist>
    </para>
    <para>
     Note that octal values of 100 or greater must not be
     introduced by a leading zero, because no more than three octal
     digits are ever read.
    </para>
    <para>
     All the sequences that define a single byte value can  be
     used both inside and outside character classes. In addition,
     inside a character class, the sequence "<literal>\b</literal>"
     is interpreted as the backspace character (hex 08). Outside a character
     class it has a different meaning (see below).
    </para>
    <para>
     The third use of backslash is for specifying generic
     character types:
    </para>
    <para>
     <variablelist>
      <varlistentry>
       <term><emphasis>\d</emphasis></term>
       <listitem><simpara>any decimal digit</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\D</emphasis></term>
       <listitem><simpara>any character that is not a decimal digit</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\s</emphasis></term>
       <listitem><simpara>any whitespace character</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\S</emphasis></term>
       <listitem><simpara>any character that is not a whitespace character</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\w</emphasis></term>
       <listitem><simpara>any "word" character</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\W</emphasis></term>
       <listitem><simpara>any "non-word" character</simpara></listitem>
      </varlistentry>
     </variablelist>
    </para>
    <para>
     Each pair of escape sequences partitions the complete set of
     characters into two disjoint sets. Any given character
     matches one, and only one, of each pair.
    </para>
    <para>
     A "word" character is any letter or digit or the underscore
     character,  that  is,  any  character which can be part of a
     Perl "<literal>word</literal>". The definition of letters and digits is  
     controlled by PCRE's character tables, and may vary if locale-specific
     matching is taking place. For example, in the "fr" (French) locale, some
     character codes greater than 128 are used for accented letters,
     and these are matched by <literal>\w</literal>.
    </para>
    <para>
     These character type sequences can appear both inside and
     outside  character classes. They each match one character of
     the appropriate type. If the current matching  point is at
     the end of the subject string, all of them fail, since there
     is no character to match.
    </para>
    <para>
     The fourth use of backslash is  for  certain  simple
     assertions. An assertion specifies a condition that has to be met
     at a particular point in  a match, without consuming any
     characters from the subject string. The use of subpatterns
     for more complicated assertions is described below. The
     backslashed assertions are
    </para>
    <para>
     <variablelist>
      <varlistentry>
       <term><emphasis>\b</emphasis></term>
       <listitem><simpara>word boundary</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\B</emphasis></term>
       <listitem><simpara>not a word boundary</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\A</emphasis></term>
       <listitem><simpara>start of subject (independent of multiline mode)</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\Z</emphasis></term>
       <listitem>
        <simpara>
         end of subject or newline at end (independent of
         multiline mode)
        </simpara>
       </listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\z</emphasis></term>
       <listitem><simpara>end of subject (independent of multiline mode)</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\G</emphasis></term>
       <listitem><simpara>first matching position in subject</simpara></listitem>
      </varlistentry>
     </variablelist>
    </para>
    <para>
     These assertions may not appear in  character  classes  (but
     note that "<literal>\b</literal>" has a different meaning, namely the backspace
     character, inside a character class).
    </para>
    <para>
     A word boundary is a position in the subject string where
     the current character and the previous character do not both
     match <literal>\w</literal> or <literal>\W</literal> (i.e. one matches 
     <literal>\w</literal> and  the  other  matches
     <literal>\W</literal>), or the start or end of the string if the first
     or last character matches <literal>\w</literal>, respectively.
    </para>
    <para>
     The <literal>\A</literal>, <literal>\Z</literal>, and
     <literal>\z</literal> assertions differ  from  the  traditional
     circumflex  and  dollar  (described below) in that they only
     ever match at the very start and end of the subject  string,
     whatever  options  are  set.  They  are  not affected by the
     <link linkend="reference.pcre.pattern.modifiers">PCRE_MULTILINE</link> or
     <link linkend="reference.pcre.pattern.modifiers">PCRE_DOLLAR_ENDONLY</link>
     options. The  difference  between <literal>\Z</literal> and
     <literal>\z</literal>  is that <literal>\Z</literal> matches before a
     newline that is the last character of the string as well as at the end of
     the string, whereas <literal>\z</literal> matches only at the end.
     </para>
     <para>
      The <literal>\G</literal> assertion is true only when the current
      matching position is at the start point of the match, as specified by
      the <parameter>offset</parameter> argument of
      <function>preg_match</function>. It differs from <literal>\A</literal>
      when the value of <parameter>offset</parameter> is non-zero.
      It is available since PHP 4.3.3.
     </para>
     
     <para>
      <literal>\Q</literal> and <literal>\E</literal> can be used to ignore
      regexp metacharacters in the pattern since PHP 4.3.3. For example:
      <literal>\w+\Q.$.\E$</literal> will match one or more word characters,
      followed by literals <literal>.$.</literal> and anchored at the end of
      the string.
     </para>
     
    </refsect2>

    <refsect2 id="regexp.reference.unicode">
     <title>Unicode character properties</title>
     <para>
      Since PHP 4.4.0 and 5.1.0, three
      additional escape sequences to match generic character types are available
      when <emphasis>UTF-8 mode</emphasis> is selected. They are:
     </para>
     <variablelist>
      <varlistentry>
       <term><emphasis>\p{xx}</emphasis></term>
       <listitem><simpara>a character with the xx property</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\P{xx}</emphasis></term>
       <listitem><simpara>a character without the xx property</simpara></listitem>
      </varlistentry>
      <varlistentry>
       <term><emphasis>\X</emphasis></term>
       <listitem><simpara>an extended Unicode sequence</simpara></listitem>
      </varlistentry>
     </variablelist>
     <para>
      The property names represented by <literal>xx</literal> above are limited to the Unicode
      general category properties. Each character has exactly one such
      property, specified by a two-letter abbreviation. For compatibility with
      Perl, negation can be specified by including a circumflex between the
      opening brace and the property name. For example, <literal>\p{^Lu}</literal> is the same
      as <literal>\P{Lu}</literal>.
     </para>
     <para>
      If only one letter is specified with <literal>\p</literal> or <literal>\P</literal>, it includes all the
      properties that start with that letter. In this case, in the absence of
      negation, the curly brackets in the escape sequence are optional; these
      two examples have the same effect:
     </para>
     <literallayout>
      \p{L}
      \pL
     </literallayout>
     <table>
      <title>Supported property codes</title>
      <tgroup cols="2">
       <tbody>
        <row><entry><literal>C</literal></entry><entry>Other</entry></row>
        <row><entry><literal>Cc</literal></entry><entry>Control</entry></row>
        <row><entry><literal>Cf</literal></entry><entry>Format</entry></row>
        <row><entry><literal>Cn</literal></entry><entry>Unassigned</entry></row>
        <row><entry><literal>Co</literal></entry><entry>Private use</entry></row>
        <row rowsep="1"><entry><literal>Cs</literal></entry><entry>Surrogate</entry></row>
        <row><entry><literal>L</literal></entry><entry>Letter</entry></row>
        <row><entry><literal>Ll</literal></entry><entry>Lower case letter</entry></row>
        <row><entry><literal>Lm</literal></entry><entry>Modifier letter</entry></row>
        <row><entry><literal>Lo</literal></entry><entry>Other letter</entry></row>
        <row><entry><literal>Lt</literal></entry><entry>Title case letter</entry></row>
        <row rowsep="1"><entry><literal>Lu</literal></entry><entry>Upper case letter</entry></row>
        <row><entry><literal>M</literal></entry><entry>Mark</entry></row>
        <row><entry><literal>Mc</literal></entry><entry>Spacing mark</entry></row>
        <row><entry><literal>Me</literal></entry><entry>Enclosing mark</entry></row>
        <row rowsep="1"><entry><literal>Mn</literal></entry><entry>Non-spacing mark</entry></row>
        <row><entry><literal>N</literal></entry><entry>Number</entry></row>
        <row><entry><literal>Nd</literal></entry><entry>Decimal number</entry></row>
        <row><entry><literal>Nl</literal></entry><entry>Letter number</entry></row>
        <row rowsep="1"><entry><literal>No</literal></entry><entry>Other number</entry></row>
        <row><entry><literal>P</literal></entry><entry>Punctuation</entry></row>
        <row><entry><literal>Pc</literal></entry><entry>Connector punctuation</entry></row>
        <row><entry><literal>Pd</literal></entry><entry>Dash punctuation</entry></row>
        <row><entry><literal>Pe</literal></entry><entry>Close punctuation</entry></row>
        <row><entry><literal>Pf</literal></entry><entry>Final punctuation</entry></row>
        <row><entry><literal>Pi</literal></entry><entry>Initial punctuation</entry></row>
        <row><entry><literal>Po</literal></entry><entry>Other punctuation</entry></row>
        <row rowsep="1"><entry><literal>Ps</literal></entry><entry>Open punctuation</entry></row>
        <row><entry><literal>S</literal></entry><entry>Symbol</entry></row>
        <row><entry><literal>Sc</literal></entry><entry>Currency symbol</entry></row>
        <row><entry><literal>Sk</literal></entry><entry>Modifier symbol</entry></row>
        <row><entry><literal>Sm</literal></entry><entry>Mathematical symbol</entry></row>
        <row rowsep="1"><entry><literal>So</literal></entry><entry>Other symbol</entry></row>
        <row><entry><literal>Z</literal></entry><entry>Separator</entry></row>
        <row><entry><literal>Zl</literal></entry><entry>Line separator</entry></row>
        <row><entry><literal>Zp</literal></entry><entry>Paragraph separator</entry></row>
        <row><entry><literal>Zs</literal></entry><entry>Space separator</entry></row>
       </tbody>
      </tgroup>
     </table>
     <para>
      Extended properties such as "Greek" or "InMusicalSymbols" are not
      supported by PCRE.
     </para>
     <para>
      Specifying caseless matching does not affect these escape sequences.
      For example, <literal>\p{Lu}</literal> always matches only upper case letters.
     </para>
     <para>
      The <literal>\X</literal> escape matches any number of Unicode characters that form an
      extended Unicode sequence. <literal>\X</literal> is equivalent to
      <literal>(?>\PM\pM*)</literal>.
     </para>
     <para>
      That is, it matches a character without the "mark" property, followed
      by zero or more characters with the "mark" property, and treats the
      sequence as an atomic group (see below). Characters with the "mark"
      property are typically accents that affect the preceding character.
     </para>
     <para>
      Matching characters by Unicode property is not fast, because PCRE has
      to search a structure that contains data for over fifteen thousand
      characters. That is why the traditional escape sequences such as <literal>\d</literal> and
      <literal>\w</literal> do not use Unicode properties in PCRE.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.circudollar">
     <title>Circumflex and dollar</title>
     <para>
      Outside a character class, in the default matching mode, the
      circumflex  character  is an assertion which is true only if
      the current matching point is at the start  of  the  subject
      string. Inside a character class, circumflex has an entirely
      different meaning (see below).
     </para>
     <para>
      Circumflex need not be the first character of the pattern if
      a number of alternatives are involved, but it should be the
      first thing in each alternative in which it appears  if  the
      pattern is ever to match that branch. If all possible
      alternatives start with a circumflex, that is, if the pattern is
      constrained to match only at the start of the subject, it is
      said to be an "anchored" pattern. (There are also other
      constructs that can cause a pattern to be anchored.)
     </para>
     <para>
      A dollar character is an assertion which is &true; only if the
      current  matching point is at the end of the subject string,
      or immediately before a newline character that is  the  last
      character in the string (by default). Dollar need not be the
      last character of the pattern if a  number  of  alternatives
      are  involved,  but it should be the last item in any branch
      in which it appears.  Dollar has no  special  meaning  in  a
      character class.
     </para>
     <para>
      The meaning of dollar can be changed so that it matches only
      at the very end of the string, by setting the
      <link linkend="reference.pcre.pattern.modifiers">PCRE_DOLLAR_ENDONLY</link>
      option at compile or matching time. This does not affect the \Z assertion.
     </para>
     <para>
      The meanings of the circumflex and dollar characters are
      changed if the
      <link linkend="reference.pcre.pattern.modifiers">PCRE_MULTILINE</link> option
      is set. When this is the case, they match immediately after and
      immediately before an internal "\n" character, respectively, in addition
      to matching at the start and end of the subject string. For example, the
      pattern /^abc$/ matches the subject string "def\nabc" in multiline mode,
      but not otherwise. Consequently, patterns that are anchored in single
      line mode because all branches start with "^" are not anchored in
      multiline mode. The
      <link linkend="reference.pcre.pattern.modifiers">PCRE_DOLLAR_ENDONLY</link>
      option is ignored if
      <link linkend="reference.pcre.pattern.modifiers">PCRE_MULTILINE</link> is
      set.
     </para>
     <para>
      Note that the sequences \A, \Z, and \z can be used to  match
      the  start  and end of the subject in both modes, and if all
      branches of a pattern start with \A is it  always  anchored,
      whether <link linkend="reference.pcre.pattern.modifiers">PCRE_MULTILINE</link>  is set or not.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.dot">
     <title>Full stop</title>
     <para>
     Outside a character class, a dot in the pattern matches  any
     one  character  in  the  subject,  including  a non-printing
     character, but not (by default) newline.  If the
     <link linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link> 
     option  is  set,  then dots match newlines as well. The
     handling of dot is entirely independent of the handling of
     circumflex  and  dollar,  the only relationship being that they
     both involve newline characters.  Dot has no special meaning
     in a character class.
     </para>
     <para>
      <emphasis>\C</emphasis> can be used to match single byte. It makes sense
      in <emphasis>UTF-8 mode</emphasis> where full stop matches the whole
      character which can consist of multiple bytes.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.squarebrackets">
     <title>Square brackets</title>
     <para>
      An opening square bracket introduces a character class,
      terminated  by  a  closing  square  bracket.  A  closing square
      bracket on its own is  not  special.  If  a  closing  square
      bracket  is  required as a member of the class, it should be
      the first data character in the class (after an initial
      circumflex, if present) or escaped with a backslash.
     </para>
     <para>
      A character class matches a single character in the subject;
      the  character  must  be in the set of characters defined by
      the class, unless the first character in the class is a
      circumflex,  in which case the subject character must not be in
      the set defined by the class. If a  circumflex  is  actually
      required  as  a  member  of  the class, ensure it is not the
      first character, or escape it with a backslash.
     </para>
     <para>
      For example, the character class [aeiou] matches  any  lower
      case vowel, while [^aeiou] matches any character that is not
      a lower case vowel. Note that a circumflex is  just  a
      convenient  notation for specifying the characters which are in
      the class by enumerating those that are not. It  is  not  an
      assertion:  it  still  consumes a character from the subject
      string, and fails if the current pointer is at  the  end  of
      the string.
     </para>
     <para>
      When caseless matching  is  set,  any  letters  in  a  class
      represent  both their upper case and lower case versions, so
      for example, a caseless [aeiou] matches "A" as well as  "a",
      and  a caseless [^aeiou] does not match "A", whereas a
      caseful version would.
     </para>
     <para>
      The newline character is never treated in any special way in
      character  classes,  whatever the setting of the <link
       linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link> 
      or <link linkend="reference.pcre.pattern.modifiers">PCRE_MULTILINE</link>
      options is. A class such as [^a] will always match a newline.
     </para>
     <para>
      The minus (hyphen) character can be used to specify a  range
      of  characters  in  a  character  class.  For example, [d-m]
      matches any letter between d and m, inclusive.  If  a  minus
      character  is required in a class, it must be escaped with a
      backslash or appear in a position where it cannot be
      interpreted as indicating a range, typically as the first or last
      character in the class.
     </para>
     <para>
      It is not possible to have the literal character "]" as  the
      end  character  of  a  range.  A  pattern such as [W-]46] is
      interpreted as a class of two characters ("W" and "-")
      followed by a literal string "46]", so it would match "W46]" or
      "-46]". However, if the "]" is escaped with a  backslash  it
      is  interpreted  as  the end of range, so [W-\]46] is
      interpreted as a single class containing a range followed by  two
      separate characters. The octal or hexadecimal representation
      of "]" can also be used to end a range.
     </para>
     <para>
      Ranges operate in ASCII collating sequence. They can also be
      used  for  characters  specified  numerically,  for  example
      [\000-\037]. If a range that includes letters is  used  when
      caseless  matching  is set, it matches the letters in either
      case. For example, [W-c] is equivalent  to  [][\^_`wxyzabc],
      matched  caselessly,  and  if  character tables for the "fr"
      locale are in use, [\xc8-\xcb] matches accented E characters
      in both cases.
     </para>
     <para>
      The character types \d, \D, \s, \S,  \w,  and  \W  may  also
      appear  in  a  character  class, and add the characters that
      they match to the class. For example, [\dABCDEF] matches any
      hexadecimal  digit.  A  circumflex  can conveniently be used
      with the upper case character types to specify a  more
      restricted set of characters than the matching lower case type.
      For example, the class [^\W_] matches any letter  or  digit,
      but not underscore.
     </para>
     <para>
      All non-alphanumeric characters other than \,  -,  ^  (at  the
      start)  and  the  terminating ] are non-special in character
      classes, but it does no harm if they are escaped.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.verticalbar">
     <title>Vertical bar</title>
     <para>
     Vertical bar characters are  used  to  separate  alternative
     patterns. For example, the pattern
      <literal>gilbert|sullivan</literal>
     matches either "gilbert" or "sullivan". Any number of alternatives
     may  appear,  and an empty alternative is permitted
     (matching the empty string).   The  matching  process  tries
     each  alternative in turn, from left to right, and the first
     one that succeeds is used. If the alternatives are within  a
     subpattern  (defined  below),  "succeeds" means matching the
     rest of the main pattern as well as the alternative  in  the
     subpattern.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.internal-options">
     <title>Internal option setting</title>
     <para>
      The settings of <link linkend="reference.pcre.pattern.modifiers">PCRE_CASELESS</link>, 
      <link linkend="reference.pcre.pattern.modifiers">PCRE_MULTILINE</link>,  
      <link linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link>,
      <link linkend="reference.pcre.pattern.modifiers">PCRE_UNGREEDY</link>,
      <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTRA</link>,
      and  <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTENDED</link>
      can be changed from within the pattern by
      a sequence of Perl option letters enclosed between "(?"  and
      ")". The option letters are:

      <table>
       <title>Internal option letters</title>
       <tgroup cols="2">
        <tbody>
         <row>
          <entry><literal>i</literal></entry>
          <entry>for <link linkend="reference.pcre.pattern.modifiers">PCRE_CASELESS</link></entry>
         </row>
         <row>
          <entry><literal>m</literal></entry>
          <entry>for <link linkend="reference.pcre.pattern.modifiers">PCRE_MULTILINE</link></entry>
         </row>
         <row>
          <entry><literal>s</literal></entry>
          <entry>for <link linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link></entry>
         </row>
         <row>
          <entry><literal>x</literal></entry>
          <entry>for <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTENDED</link></entry>
         </row>
         <row>
          <entry><literal>U</literal></entry>
          <entry>for <link linkend="reference.pcre.pattern.modifiers">PCRE_UNGREEDY</link></entry>
         </row>
         <row>
          <entry><literal>X</literal></entry>
          <entry>for <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTRA</link></entry>
         </row>
        </tbody>
       </tgroup>
      </table>
     </para>
     <para>
      For example, (?im) sets caseless, multiline matching. It  is
      also possible to unset these options by preceding the letter
      with a hyphen, and a combined setting and unsetting such  as
      (?im-sx),  which sets <link
       linkend="reference.pcre.pattern.modifiers">PCRE_CASELESS</link> and
      <link linkend="reference.pcre.pattern.modifiers">PCRE_MULTILINE</link>
      while unsetting <link
       linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link> and
      <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTENDED</link>,
      is also  permitted. If  a  letter  appears both before and after the
      hyphen, the option is unset.
     </para>
     <para>
      When an option change occurs at top level (that is, not inside
      subpattern parentheses), the change applies to the remainder of the
      pattern that follows. So <literal>/ab(?i)c/</literal> matches only "abc"
      and "abC". This behaviour has been changed in PCRE 4.0, which is bundled
      since PHP 4.3.3. Before those versions, <literal>/ab(?i)c/</literal> would
      perform as <literal>/abc/i</literal> (e.g. matching "ABC" and "aBc").
     </para>
     <para>
      If an option change occurs inside a subpattern,  the  effect
      is  different.  This is a change of behaviour in Perl 5.005.
      An option change inside a subpattern affects only that  part
      of the subpattern that follows it, so

        <literal>(a(?i)b)c</literal>

      matches  abc  and  aBc  and  no  other   strings   (assuming <link
       linkend="reference.pcre.pattern.modifiers">PCRE_CASELESS</link> is not
      used). By this means, options can be made to have different settings in
      different parts of the pattern. Any changes made in one alternative do
      carry on into subsequent branches within the same subpattern. For
      example,

        <literal>(a(?i)b|c)</literal>

      matches "ab", "aB", "c", and "C", even though when  matching
      "C" the first branch is abandoned before the option setting.
      This is because the effects of  option  settings  happen  at
      compile  time. There would be some very weird behaviour otherwise.
     </para>
     <para>
      The PCRE-specific options <link linkend="reference.pcre.pattern.modifiers">PCRE_UNGREEDY</link>  and  
      <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTRA</link>   can
      be changed in the same way as the Perl-compatible options by
      using the characters U and X  respectively.  The  (?X)  flag
      setting  is  special in that it must always occur earlier in
      the pattern than any of the additional features it turns on,
      even when it is at top level. It is best put at the start.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.subpatterns">
     <title>Subpatterns</title>
     <para>
     Subpatterns are delimited by parentheses  (round  brackets),
     which can be nested.  Marking part of a pattern as a subpattern
     does two things:
    </para>
    <para>
     1. It localizes a set of alternatives. For example, the
     pattern

       <literal>cat(aract|erpillar|)</literal>

     matches one of the words "cat",  "cataract",  or  "caterpillar".
     Without  the  parentheses, it would match "cataract",
     "erpillar" or the empty string.
    </para>
    <para>
     2. It sets up the subpattern as a capturing  subpattern  (as
     defined  above).   When the whole pattern matches, that portion
     of the subject string that matched  the  subpattern  is
     passed  back  to  the  caller  via  the  <emphasis>ovector</emphasis>
     argument of
     <function>pcre_exec</function>. Opening parentheses are counted
     from  left  to right (starting from 1) to obtain the numbers of the
     capturing subpatterns.
    </para>
    <para>
     For example, if the string "the red king" is matched against
     the pattern

       <literal>the ((red|white) (king|queen))</literal>

     the captured substrings are "red king", "red",  and  "king",
     and are numbered 1, 2, and 3.
    </para>
    <para>
     The fact that plain parentheses fulfil two functions is  not
     always  helpful.  There are often times when a grouping subpattern
     is required without a capturing requirement.  If  an
     opening parenthesis is followed by "?:", the subpattern does
     not do any capturing, and is not counted when computing  the
     number of any subsequent capturing subpatterns. For example,
     if the string "the  white  queen"  is  matched  against  the
     pattern

       <literal>the ((?:red|white) (king|queen))</literal>

     the captured substrings are "white queen" and  "queen",  and
     are  numbered  1  and 2. The maximum number of captured substrings
     is 99, and the maximum number  of  all  subpatterns,
     both capturing and non-capturing, is 200.
    </para>
    <para>
     As a  convenient  shorthand,  if  any  option  settings  are
     required  at  the  start  of a non-capturing subpattern, the
     option letters may appear between the "?" and the ":".  Thus
     the two patterns
    </para>

    <literallayout>
       (?i:saturday|sunday)
       (?:(?i)saturday|sunday)
    </literallayout>

    <para>
     match exactly the same set of strings.  Because  alternative
     branches  are  tried from left to right, and options are not
     reset until the end of the subpattern is reached, an  option
     setting  in  one  branch does affect subsequent branches, so
     the above patterns match "SUNDAY" as well as "Saturday".
     </para>
     
     <para>
      It is possible to name the subpattern with
      <literal>(?P&lt;name&gt;pattern)</literal> since PHP 4.3.3. Array with matches will
      contain the match indexed by the string alongside the match indexed by
      a number, then.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.repetition">
     <title>Repetition</title>
     <para>
     Repetition is specified by quantifiers, which can follow any
     of the following items:

      <itemizedlist>
       <listitem><simpara>a single character, possibly escaped</simpara></listitem>
       <listitem><simpara>the . metacharacter</simpara></listitem>
       <listitem><simpara>a character class</simpara></listitem>
       <listitem><simpara>a back reference (see next section)</simpara></listitem>
       <listitem><simpara>a parenthesized subpattern (unless it is  an  assertion  -
     see below)</simpara></listitem>
      </itemizedlist>
    </para>
    <para>
     The general repetition quantifier specifies  a  minimum  and
     maximum  number  of  permitted  matches,  by  giving the two
     numbers in curly brackets (braces), separated  by  a  comma.
     The  numbers  must be less than 65536, and the first must be
     less than or equal to the second. For example:

       <literal>z{2,4}</literal>

     matches "zz", "zzz", or "zzzz". A closing brace on  its  own
     is not a special character. If the second number is omitted,
     but the comma is present, there is no upper  limit;  if  the
     second number and the comma are both omitted, the quantifier
     specifies an exact number of required matches. Thus

       <literal>[aeiou]{3,}</literal>

     matches at least 3 successive vowels,  but  may  match  many
     more, while

       <literal>\d{8}</literal>

     matches exactly 8 digits.  An  opening  curly  bracket  that
     appears  in a position where a quantifier is not allowed, or
     one that does not match the syntax of a quantifier, is taken
     as  a literal character. For example, {,6} is not a quantifier,
     but a literal string of four characters.
    </para>
    <para>
     The quantifier {0} is permitted, causing the  expression  to
     behave  as  if the previous item and the quantifier were not
     present.
    </para>
    <para>
     For convenience (and  historical  compatibility)  the  three
     most common quantifiers have single-character abbreviations:

     <table>
      <title>Single-character quantifiers</title>
      <tgroup cols="2">
       <tbody>
        <row>
         <entry><literal>*</literal></entry>
         <entry>equivalent to <literal>{0,}</literal></entry>
        </row>
        <row>
         <entry><literal>+</literal></entry>
         <entry>equivalent to <literal>{1,}</literal></entry>
        </row>
        <row>
         <entry><literal>?</literal></entry>
         <entry>equivalent to <literal>{0,1}</literal></entry>
        </row>
       </tbody>
      </tgroup>
     </table>
    </para>
    <para>
     It is possible to construct infinite loops  by  following  a
     subpattern  that  can  match no characters with a quantifier
     that has no upper limit, for example:

       <literal>(a?)*</literal>
    </para>
    <para>
     Earlier versions of Perl and PCRE used to give an  error  at
     compile  time  for such patterns. However, because there are
     cases where this  can  be  useful,  such  patterns  are  now
     accepted,  but  if  any repetition of the subpattern does in
     fact match no characters, the loop is forcibly broken.
    </para>
    <para>
     By default, the quantifiers  are  "greedy",  that  is,  they
     match  as much as possible (up to the maximum number of permitted
     times), without causing the rest of  the  pattern  to
     fail. The classic example of where this gives problems is in
     trying to match comments in C programs. These appear between
     the  sequences /* and */ and within the sequence, individual
     * and / characters may appear. An attempt to  match  C  comments
     by applying the pattern

       <literal>/\*.*\*/</literal>

     to the string

       <literal>/* first comment */  not comment  /* second comment */</literal>

     fails, because it matches  the  entire  string  due  to  the
     greediness of the .*  item.
    </para>
    <para>
     However, if a quantifier is followed  by  a  question  mark,
     then it ceases to be greedy, and instead matches the minimum
     number of times possible, so the pattern

       <literal>/\*.*?\*/</literal>

     does the right thing with the C comments. The meaning of the
     various  quantifiers is not otherwise changed, just the preferred
     number of matches.  Do not confuse this use of
     question  mark  with  its  use as a quantifier in its own right.
     Because it has two uses, it can sometimes appear doubled, as
     in

       <literal>\d??\d</literal>

     which matches one digit by preference, but can match two  if
     that is the only way the rest of the pattern matches.
    </para>
    <para>
     If the <link linkend="reference.pcre.pattern.modifiers">PCRE_UNGREEDY</link>  option is set (an option which  is  not
     available  in  Perl)  then the quantifiers are not greedy by
     default, but individual ones can be made greedy by following
     them  with  a  question mark. In other words, it inverts the
     default behaviour.
    </para>
    <para>
     Quantifiers followed by <literal>+</literal> are "possessive". They eat
     as many characters as possible and don't return to match the rest of the
     pattern. Thus <literal>.*abc</literal> matches "aabc" but
     <literal>.*+abc</literal> doesn't because <literal>.*+</literal> eats the
     whole string. Possessive quantifiers can be used to speed up processing since PHP 4.3.3.
    </para>
    <para>
     When a parenthesized subpattern is quantified with a minimum
     repeat  count  that is greater than 1 or with a limited maximum,
     more store is required for the  compiled  pattern,  in
     proportion to the size of the minimum or maximum.
    </para>
    <para>
     If a pattern starts with .* or  .{0,}  and  the  <link linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link> 
     option (equivalent to Perl's /s) is set, thus allowing the .
     to match newlines, then the pattern is implicitly  anchored,
     because whatever follows will be tried against every character
     position in the subject string, so there is no point  in
     retrying  the overall match at any position after the first.
     PCRE treats such a pattern as though it were preceded by \A.
     In  cases where it is known that the subject string contains
     no newlines, it is worth setting <link linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link>  when  the  pattern begins with .* in order to
     obtain this optimization, or
     alternatively using ^ to indicate anchoring explicitly.
    </para>
    <para>
     When a capturing subpattern is repeated, the value  captured
     is the substring that matched the final iteration. For example, after

       <literal>(tweedle[dume]{3}\s*)+</literal>

     has matched "tweedledum tweedledee" the value  of  the  captured
     substring  is  "tweedledee".  However,  if  there are
     nested capturing  subpatterns,  the  corresponding  captured
     values  may  have been set in previous iterations. For example,
     after
     
       <literal>/(a|(b))+/</literal>

     matches "aba" the value of the second captured substring  is
     "b".
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.back-references">
     <title>Back references</title>
     <para>
     Outside a character class, a backslash followed by  a  digit
     greater  than  0  (and  possibly  further  digits) is a back
     reference to a capturing subpattern  earlier  (i.e.  to  its
     left)  in  the  pattern,  provided there have been that many
     previous capturing left parentheses.
    </para>
    <para>
     However, if the decimal number following  the  backslash  is
     less  than  10,  it is always taken as a back reference, and
     causes an error only if there are not  that  many  capturing
     left  parentheses in the entire pattern. In other words, the
     parentheses that are referenced need not be to the  left  of
     the  reference  for  numbers  less  than 10. See the section
     entitled "Backslash" above for further details of  the  handling
     of digits following a backslash.
    </para>
    <para>
     A back reference matches whatever actually matched the  capturing
     subpattern in the current subject string, rather than
     anything matching the subpattern itself. So the pattern

       <literal>(sens|respons)e and \1ibility</literal>

     matches "sense and sensibility" and "response and  responsibility",
     but  not  "sense  and  responsibility". If caseful
     matching is in force at the time of the back reference, then
     the case of letters is relevant. For example,

       <literal>((?i)rah)\s+\1</literal>

     matches "rah rah" and "RAH RAH", but  not  "RAH  rah",  even
     though  the  original  capturing subpattern is matched caselessly.
    </para>
    <para>
     There may be more than one back reference to the  same  subpattern.
     If  a  subpattern  has not actually been used in a
     particular match, then any  back  references  to  it  always
     fail. For example, the pattern

       <literal>(a|(bc))\2</literal>

     always fails if it starts to match  "a"  rather  than  "bc".
     Because  there  may  be up to 99 back references, all digits
     following the backslash are taken as  part  of  a  potential
     back reference number. If the pattern continues with a digit
     character, then some delimiter must be used to terminate the
     back reference. If the <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTENDED</link>  option is set, this can
     be whitespace.  Otherwise an empty comment can be used.
    </para>
    <para>
     A back reference that occurs inside the parentheses to which
     it  refers  fails when the subpattern is first used, so, for
     example, (a\1) never matches.  However, such references  can
     be useful inside repeated subpatterns. For example, the pattern

       <literal>(a|b\1)+</literal>

     matches any number of "a"s and also "aba", "ababaa" etc.  At
     each iteration of the subpattern, the back reference matches
     the character string corresponding to  the  previous  iteration.
     In order for this to work, the pattern must be such
     that the first iteration does not need  to  match  the  back
     reference.  This  can  be  done using alternation, as in the
     example above, or by a quantifier with a minimum of zero.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.assertions">
     <title>Assertions</title>
     <para>
     An assertion is  a  test  on  the  characters  following  or
     preceding  the current matching point that does not actually
     consume any characters. The simple assertions coded  as  \b,
     \B,  \A,  \Z,  \z, ^ and $ are described above. More complicated
     assertions are coded as  subpatterns.  There  are  two
     kinds:  those that look ahead of the current position in the
     subject string, and those that look behind it.
    </para>
    <para>
     An assertion subpattern is matched in the normal way, except
     that  it  does not cause the current matching position to be
     changed. Lookahead assertions start with  (?=  for  positive
     assertions and (?! for negative assertions. For example,

       <literal>\w+(?=;)</literal>

     matches a word followed by a semicolon, but does not include
     the semicolon in the match, and

       <literal>foo(?!bar)</literal>

     matches any occurrence of "foo"  that  is  not  followed  by
     "bar". Note that the apparently similar pattern

       <literal>(?!foo)bar</literal>

     does not find an occurrence of "bar"  that  is  preceded  by
     something other than "foo"; it finds any occurrence of "bar"
     whatsoever, because the assertion  (?!foo)  is  always  &true;
     when  the  next  three  characters  are  "bar". A lookbehind
     assertion is needed to achieve this effect.
    </para>
    <para>
     Lookbehind assertions start with (?&lt;=  for  positive  assertions
     and (?&lt;! for negative assertions. For example,

       <literal>(?&lt;!foo)bar</literal>

     does find an occurrence of "bar" that  is  not  preceded  by
     "foo". The contents of a lookbehind assertion are restricted
     such that all the strings  it  matches  must  have  a  fixed
     length.  However, if there are several alternatives, they do
     not all have to have the same fixed length. Thus

       <literal>(?&lt;=bullock|donkey)</literal>

     is permitted, but

       <literal>(?&lt;!dogs?|cats?)</literal>

     causes an error at compile time. Branches  that  match  different
     length strings are permitted only at the top level of
     a lookbehind assertion. This is an extension  compared  with
     Perl  5.005,  which  requires all branches to match the same
     length of string. An assertion such as

       <literal>(?&lt;=ab(c|de))</literal>

     is not permitted, because its single  top-level  branch  can
     match two different lengths, but it is acceptable if rewritten
     to use two top-level branches:

       <literal>(?&lt;=abc|abde)</literal>

     The implementation of lookbehind  assertions  is,  for  each
     alternative,  to  temporarily move the current position back
     by the fixed width and then  try  to  match.  If  there  are
     insufficient  characters  before  the  current position, the
     match is deemed to fail.  Lookbehinds  in  conjunction  with
     once-only  subpatterns can be particularly useful for matching
     at the ends of strings; an example is given at  the  end
     of the section on once-only subpatterns.
    </para>
    <para>
     Several assertions (of any sort) may  occur  in  succession.
     For example,

       <literal>(?&lt;=\d{3})(?&lt;!999)foo</literal>

     matches "foo" preceded by three digits that are  not  "999".
     Notice  that each of the assertions is applied independently
     at the same point in the subject string. First  there  is  a
     check  that  the  previous  three characters are all digits,
     then there is a check that the same three characters are not
     "999".   This  pattern  does not match "foo" preceded by six
     characters, the first of which are digits and the last three
     of  which  are  not  "999".  For  example,  it doesn't match
     "123abcfoo". A pattern to do that is

       <literal>(?&lt;=\d{3}...)(?&lt;!999)foo</literal>
    </para>
    <para>
     This time the first assertion looks  at  the  preceding  six
     characters,  checking  that  the first three are digits, and
     then the second assertion checks that  the  preceding  three
     characters are not "999".
    </para>
    <para>
     Assertions can be nested in any combination. For example,

       <literal>(?&lt;=(?&lt;!foo)bar)baz</literal>

     matches an occurrence of "baz" that  is  preceded  by  "bar"
     which in turn is not preceded by "foo", while

       <literal>(?&lt;=\d{3}...(?&lt;!999))foo</literal>

     is another pattern which matches  "foo"  preceded  by  three
     digits and any three characters that are not "999".
    </para>
    <para>
     Assertion subpatterns are not capturing subpatterns, and may
     not  be  repeated,  because  it makes no sense to assert the
     same thing several times. If any kind of assertion  contains
     capturing  subpatterns  within it, these are counted for the
     purposes of numbering the capturing subpatterns in the whole
     pattern.   However,  substring capturing is carried out only
     for positive assertions, because it does not make sense  for
     negative assertions.
    </para>
    <para>
     Assertions count towards the maximum  of  200  parenthesized
     subpatterns.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.onlyonce">
     <title>Once-only subpatterns</title>
     <para>
     With both maximizing and minimizing repetition,  failure  of
     what  follows  normally  causes  the repeated item to be
     re-evaluated to see if a different number of repeats allows the
     rest  of  the  pattern  to  match. Sometimes it is useful to
     prevent this, either to change the nature of the  match,  or
     to  cause  it fail earlier than it otherwise might, when the
     author of the pattern knows there is no  point  in  carrying
     on.
    </para>
    <para>
     Consider, for example, the pattern \d+foo  when  applied  to
     the subject line

       <literal>123456bar</literal>
    </para>
    <para>
     After matching all 6 digits and then failing to match "foo",
     the normal action of the matcher is to try again with only 5
     digits matching the \d+ item, and then with 4,  and  so  on,
     before ultimately failing. Once-only subpatterns provide the
     means for specifying that once a portion of the pattern  has
     matched,  it  is  not to be re-evaluated in this way, so the
     matcher would give up immediately on failing to match  "foo"
     the  first  time.  The  notation  is another kind of special
     parenthesis, starting with (?&gt; as in this example:

       <literal>(?&gt;\d+)bar</literal>
    </para>
    <para>
     This kind of parenthesis "locks up" the  part of the pattern
     it  contains once it has matched, and a failure further into
     the pattern is prevented from backtracking  into  it.
     Backtracking  past  it to previous items, however, works as normal.
    </para>
    <para>
     An alternative description is that a subpattern of this type
     matches  the  string  of  characters that an identical standalone
     pattern would match, if anchored at the current point
     in the subject string.
    </para>
    <para>
     Once-only subpatterns are not capturing subpatterns.  Simple
     cases  such as the above example can be thought of as a maximizing
     repeat that must  swallow  everything  it  can.  So,
     while both \d+ and \d+? are prepared to adjust the number of
     digits they match in order to make the rest of  the  pattern
     match, (?&gt;\d+) can only match an entire sequence of digits.
    </para>
    <para>
     This construction can of course contain arbitrarily  complicated
     subpatterns, and it can be nested.
    </para>
    <para>
     Once-only subpatterns can be used in conjunction with
     look-behind  assertions  to specify efficient matching at the end
     of the subject string. Consider a simple pattern such as

       <literal>abcd$</literal>

     when applied to a long string which does not match.  Because
     matching  proceeds  from  left  to right, PCRE will look for
     each "a" in the subject and then see if what follows matches
     the rest of the pattern. If the pattern is specified as

       <literal>^.*abcd$</literal>

     then the initial .* matches the entire string at first,  but
     when  this  fails  (because  there  is no following "a"), it
     backtracks to match all but the last character, then all but
     the  last  two  characters, and so on. Once again the search
     for "a" covers the entire string, from right to left, so  we
     are no better off. However, if the pattern is written as

       <literal>^(?>.*)(?&lt;=abcd)</literal>

     then there can be no backtracking for the .*  item;  it  can
     match  only  the  entire  string.  The subsequent lookbehind
     assertion does a single test on the last four characters. If
     it  fails,  the  match  fails immediately. For long strings,
     this approach makes a significant difference to the processing time.
    </para>
    <para>
     When a pattern contains an unlimited repeat inside a subpattern
     that can itself be repeated an unlimited number of
     times, the use of a once-only subpattern is the only way  to
     avoid  some  failing matches taking a very long time indeed.
     The pattern

       <literal>(\D+|&lt;\d+>)*[!?]</literal>

     matches an unlimited number of substrings that  either  consist
     of  non-digits,  or digits enclosed in &lt;>, followed by
     either ! or ?. When it matches, it runs quickly. However, if
     it is applied to

       <literal>aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa</literal>

     it takes a long  time  before  reporting  failure.  This  is
     because the string can be divided between the two repeats in
     a large number of ways, and all have to be tried. (The example
     used  [!?]  rather  than a single character at the end,
     because both PCRE and Perl have an optimization that  allows
     for  fast  failure  when  a  single  character is used. They
     remember the last single character that is  required  for  a
     match,  and  fail early if it is not present in the string.)
     If the pattern is changed to

       <literal>((?>\D+)|&lt;\d+>)*[!?]</literal>

     sequences of non-digits cannot be broken, and  failure  happens quickly.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.conditional">
     <title>Conditional subpatterns</title>
     <para>
     It is possible to cause the matching process to obey a  subpattern 
     conditionally  or to choose between two alternative
     subpatterns, depending on the result  of  an  assertion,  or
     whether  a previous capturing subpattern matched or not. The
     two possible forms of conditional subpattern are
    </para>

    <literallayout>
       (?(condition)yes-pattern)
       (?(condition)yes-pattern|no-pattern)
    </literallayout>
    <para>
     If the condition is satisfied, the yes-pattern is used; otherwise
     the  no-pattern  (if  present) is used. If there are
     more than two alternatives in the subpattern, a compile-time
     error occurs.
    </para>
    <para>
     There are two kinds of condition. If the  text  between  the
     parentheses  consists  of  a  sequence  of  digits, then the
     condition is satisfied if the capturing subpattern  of  that
     number  has  previously matched. Consider the following pattern,
     which contains non-significant white space to make  it
     more  readable  (assume  the  <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTENDED</link>   option)  and to
     divide it into three parts for ease of discussion:

       <literal>( \( )?    [^()]+    (?(1) \) )</literal>
    </para>
    <para>
     The first part matches an optional opening parenthesis,  and
     if  that character is present, sets it as the first captured
     substring. The second part matches one  or  more  characters
     that  are  not  parentheses. The third part is a conditional
     subpattern that tests whether the first set  of  parentheses
     matched  or  not.  If  they did, that is, if subject started
     with an opening parenthesis, the condition is &true;,  and  so
     the  yes-pattern  is  executed  and a closing parenthesis is
     required. Otherwise, since no-pattern is  not  present,  the
     subpattern  matches  nothing.  In  other words, this pattern
     matches a sequence of non-parentheses,  optionally  enclosed
     in parentheses.
    </para>
    <para>
     If the condition is the string <literal>(R)</literal>, it is satisfied if
     a recursive call to the pattern or subpattern has been made. At "top
     level", the condition is false.
    </para>
    <para>
     If the condition is not a sequence of digits or (R), it must be  an
     assertion.  This  may be a positive or negative lookahead or
     lookbehind assertion. Consider this pattern, again  containing
     non-significant  white space, and with the two alternatives on
     the second line:
    </para>

    <literallayout>
       (?(?=[^a-z]*[a-z])
       \d{2}-[a-z]{3}-\d{2}  |  \d{2}-\d{2}-\d{2} )
    </literallayout>
    <para>
     The condition is a positive lookahead assertion that matches
     an optional sequence of non-letters followed by a letter. In
     other words, it tests for  the  presence  of  at  least  one
     letter  in the subject. If a letter is found, the subject is
     matched against  the  first  alternative;  otherwise  it  is
     matched  against the second. This pattern matches strings in
     one of the two forms dd-aaa-dd or dd-dd-dd,  where  aaa  are
     letters and dd are digits.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.comments">
     <title>Comments</title>
     <para>
     The  sequence  (?#  marks  the  start  of  a  comment  which
     continues   up  to  the  next  closing  parenthesis.  Nested
     parentheses are not permitted. The characters that make up a
     comment play no part in the pattern matching at all.
    </para>
    <para>
     If the <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTENDED</link>  option is set, an unescaped # character
     outside  a character class introduces a comment that
     continues up to the next newline character in the pattern.
     </para>
    </refsect2>

    <refsect2 id="regexp.reference.recursive">
     <title>Recursive patterns</title>
     <para>
     Consider the problem of matching a  string  in  parentheses,
     allowing  for  unlimited nested parentheses. Without the use
     of recursion, the best that can be done is to use a  pattern
     that  matches  up  to some fixed depth of nesting. It is not
     possible to handle an arbitrary nesting depth. Perl 5.6  has
     provided   an  experimental  facility  that  allows  regular
     expressions to recurse (among other things).  The  special 
     item (?R) is  provided for  the specific  case of recursion. 
     This PCRE  pattern  solves the  parentheses  problem (assume 
     the <link linkend="reference.pcre.pattern.modifiers">PCRE_EXTENDED</link>
     option is set so that white space is 
     ignored):

       <literal>\( ( (?>[^()]+) | (?R) )* \)</literal>
    </para>
    <para>
     First it matches an opening parenthesis. Then it matches any
     number  of substrings which can either be a sequence of
     non-parentheses, or a recursive  match  of  the  pattern  itself
     (i.e. a correctly parenthesized substring). Finally there is
     a closing parenthesis.
    </para>
    <para>
     This particular example pattern  contains  nested  unlimited
     repeats, and so the use of a once-only subpattern for matching
     strings of non-parentheses is  important  when  applying
     the  pattern to strings that do not match. For example, when
     it is applied to

       <literal>(aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()</literal>

     it yields "no match" quickly. However, if a  once-only  subpattern
     is  not  used,  the match runs for a very long time
     indeed because there are so many different ways the + and  *
     repeats  can carve up the subject, and all have to be tested
     before failure can be reported.
    </para>
    <para>
     The values set for any capturing subpatterns are those  from
     the outermost level of the recursion at which the subpattern
     value is set. If the pattern above is matched against

       <literal>(ab(cd)ef)</literal>

     the value for the capturing parentheses is  "ef",  which  is
     the  last  value  taken  on  at the top level. If additional
     parentheses are added, giving

       <literal>\( ( ( (?>[^()]+) | (?R) )* ) \)</literal>
     then the string they capture
     is "ab(cd)ef", the contents of the top level parentheses. If
     there are more than 15 capturing parentheses in  a  pattern,
     PCRE  has  to  obtain  extra  memory  to store data during a
     recursion, which it does by using  pcre_malloc,  freeing  it
     via  pcre_free  afterwards. If no memory can be obtained, it
     saves data for the first 15 capturing parentheses  only,  as
     there is no way to give an out-of-memory error from within a
     recursion.
     </para>
     
     <para>
      Since PHP 4.3.3, <literal>(?1)</literal>, <literal>(?2)</literal> and so on can be used
      for recursive subpatterns too. It is also possible to use named
      subpatterns: <literal>(?P>foo)</literal>.
     </para>
     <para>
      If the syntax for a recursive subpattern reference (either by number or
      by name) is used outside the parentheses to which it refers, it operates
      like a subroutine in a programming language. An earlier example
      pointed out that the pattern
      <literal>(sens|respons)e and \1ibility</literal>
      matches "sense and sensibility" and "response and responsibility", but
      not "sense and responsibility". If instead the pattern
      <literal>(sens|respons)e and (?1)ibility</literal>
      is used, it does match "sense and responsibility" as well as the other
      two strings. Such references must, however, follow the subpattern to
      which they refer.
     </para>
     
    </refsect2>

    <refsect2 id="regexp.reference.performances">
     <title>Performances</title>
     <para>
     Certain items that may appear in patterns are more efficient
     than  others.  It is more efficient to use a character class
     like [aeiou] than a set of alternatives such as (a|e|i|o|u).
     In  general,  the  simplest  construction  that provides the
     required behaviour is usually the  most  efficient.  Jeffrey
     Friedl's  book contains a lot of discussion about optimizing
     regular expressions for efficient performance.
    </para>
    <para>
     When a pattern begins with .* and the <link linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link>  option  is
     set,  the  pattern  is implicitly anchored by PCRE, since it
     can match only at the start of a subject string. However, if
     <link linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link>   is not set, PCRE cannot make this optimization,
     because the . metacharacter does not then match  a  newline,
     and if the subject string contains newlines, the pattern may
     match from the character immediately following one  of  them
     instead of from the very start. For example, the pattern

       <literal>(.*) second</literal>

     matches the subject "first\nand second" (where \n stands for
     a newline character) with the first captured substring being
     "and". In order to do this, PCRE  has  to  retry  the  match
     starting after every newline in the subject.
    </para>
    <para>
     If you are using such a pattern with subject strings that do
     not  contain  newlines,  the best performance is obtained by
     setting <link linkend="reference.pcre.pattern.modifiers">PCRE_DOTALL</link>, or starting the  pattern  with  ^.*  to
     indicate  explicit anchoring. That saves PCRE from having to
     scan along the subject looking for a newline to restart at.
    </para>
    <para>
     Beware of patterns that contain nested  indefinite  repeats.
     These  can  take a long time to run when applied to a string
     that does not match. Consider the pattern fragment

       <literal>(a+)*</literal>
    </para>
    <para>
     This can match "aaaa" in 33 different ways, and this  number
     increases  very  rapidly  as  the string gets longer. (The *
     repeat can match 0, 1, 2, 3, or 4 times,  and  for  each  of
     those  cases other than 0, the + repeats can match different
     numbers of times.) When the remainder of the pattern is such
     that  the entire match is going to fail, PCRE has in principle
     to try every possible variation, and this  can  take  an
     extremely long time.
    </para>
    <para>
     An optimization catches some of the more simple  cases  such
     as

       <literal>(a+)*b</literal>

     where a literal character follows. Before embarking  on  the
     standard matching procedure, PCRE checks that there is a "b"
     later in the subject string, and if there is not,  it  fails
     the  match  immediately. However, when there is no following
     literal this optimization cannot be used. You  can  see  the
     difference by comparing the behaviour of

       <literal>(a+)*\d</literal>

     with the pattern above. The former gives  a  failure  almost
     instantly  when  applied  to a whole line of "a" characters,
     whereas the latter takes an appreciable  time  with  strings
     longer than about 20 characters.
     </para>
    </refsect2>
   </refsect1>
  </refentry>

<!-- Keep this comment at the end of the file
Local variables:
mode: sgml
sgml-omittag:t
sgml-shorttag:t
sgml-minimize-attributes:nil
sgml-always-quote-attributes:t
sgml-indent-step:1
sgml-indent-data:t
indent-tabs-mode:nil
sgml-parent-document:nil
sgml-default-dtd-file:"../../../../manual.ced"
sgml-exposed-tags:nil
sgml-local-catalogs:nil
sgml-local-ecat-files:nil
End:
vim600: syn=xml fen fdm=syntax fdl=2 si
vim: et tw=78 syn=sgml
vi: ts=1 sw=1
-->