libs/regex/doc/html/boost_regex/syntax/perl_syntax.html

   1 <html>
   2 <head>
   3 <meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
   4 <title>Perl Regular Expression Syntax</title>
   5 <link rel="stylesheet" href="../../../../../../doc/src/boostbook.css" type="text/css">
   6 <meta name="generator" content="DocBook XSL Stylesheets V1.79.1">
   7 <link rel="home" href="../../index.html" title="Boost.Regex 5.1.4">
   8 <link rel="up" href="../syntax.html" title="Regular Expression Syntax">
   9 <link rel="prev" href="../syntax.html" title="Regular Expression Syntax">
  10 <link rel="next" href="basic_extended.html" title="POSIX Extended Regular Expression Syntax">
  11 </head>
  12 <body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
  13 <table cellpadding="2" width="100%"><tr>
  14 <td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../../boost.png"></td>
  15 <td align="center"><a href="../../../../../../index.html">Home</a></td>
  16 <td align="center"><a href="../../../../../../libs/libraries.htm">Libraries</a></td>
  17 <td align="center"><a href="http://www.boost.org/users/people.html">People</a></td>
  18 <td align="center"><a href="http://www.boost.org/users/faq.html">FAQ</a></td>
  19 <td align="center"><a href="../../../../../../more/index.htm">More</a></td>
  20 </tr></table>
  21 <hr>
  22 <div class="spirit-nav">
  23 <a accesskey="p" href="../syntax.html"><img src="../../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../syntax.html"><img src="../../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../../index.html"><img src="../../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="basic_extended.html"><img src="../../../../../../doc/src/images/next.png" alt="Next"></a>
  24 </div>
  25 <div class="section">
  26 <div class="titlepage"><div><div><h3 class="title">
  27 <a name="boost_regex.syntax.perl_syntax"></a><a class="link" href="perl_syntax.html" title="Perl Regular Expression Syntax">Perl Regular Expression
  28       Syntax</a>
  29 </h3></div></div></div>
  30 <h4>
  31 <a name="boost_regex.syntax.perl_syntax.h0"></a>
  32         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.synopsis"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.synopsis">Synopsis</a>
  33       </h4>
  34 <p>
  35         The Perl regular expression syntax is based on that used by the programming
  36         language Perl . Perl regular expressions are the default behavior in Boost.Regex
  37         or you can pass the flag <code class="literal">perl</code> to the <a class="link" href="../ref/basic_regex.html" title="basic_regex"><code class="computeroutput"><span class="identifier">basic_regex</span></code></a> constructor, for example:
  38       </p>
  39 <pre class="programlisting"><span class="comment">// e1 is a case sensitive Perl regular expression: </span>
  40 <span class="comment">// since Perl is the default option there's no need to explicitly specify the syntax used here:</span>
  41 <span class="identifier">boost</span><span class="special">::</span><span class="identifier">regex</span> <span class="identifier">e1</span><span class="special">(</span><span class="identifier">my_expression</span><span class="special">);</span>
  42 <span class="comment">// e2 a case insensitive Perl regular expression:</span>
  43 <span class="identifier">boost</span><span class="special">::</span><span class="identifier">regex</span> <span class="identifier">e2</span><span class="special">(</span><span class="identifier">my_expression</span><span class="special">,</span> <span class="identifier">boost</span><span class="special">::</span><span class="identifier">regex</span><span class="special">::</span><span class="identifier">perl</span><span class="special">|</span><span class="identifier">boost</span><span class="special">::</span><span class="identifier">regex</span><span class="special">::</span><span class="identifier">icase</span><span class="special">);</span>
  44 </pre>
  45 <h4>
  46 <a name="boost_regex.syntax.perl_syntax.h1"></a>
  47         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.perl_regular_expression_syntax"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.perl_regular_expression_syntax">Perl
  48         Regular Expression Syntax</a>
  49       </h4>
  50 <p>
  51         In Perl regular expressions, all characters match themselves except for the
  52         following special characters:
  53       </p>
  54 <pre class="programlisting">.[{}()\*+?|^$</pre>
  55 <p>
  56         Other characters are special only in certain situations - for example <code class="computeroutput"><span class="special">]</span></code> is special only after an opening <code class="computeroutput"><span class="special">[</span></code>.
  57       </p>
  58 <h5>
  59 <a name="boost_regex.syntax.perl_syntax.h2"></a>
  60         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.wildcard"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.wildcard">Wildcard</a>
  61       </h5>
  62 <p>
  63         The single character '.' when used outside of a character set will match
  64         any single character except:
  65       </p>
  66 <div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; ">
  67 <li class="listitem">
  68             The NULL character when the <a class="link" href="../ref/match_flag_type.html" title="match_flag_type">flag
  69             <code class="literal">match_not_dot_null</code></a> is passed to the matching
  70             algorithms.
  71           </li>
  72 <li class="listitem">
  73             The newline character when the <a class="link" href="../ref/match_flag_type.html" title="match_flag_type">flag
  74             <code class="literal">match_not_dot_newline</code></a> is passed to the matching
  75             algorithms.
  76           </li>
  77 </ul></div>
  78 <h5>
  79 <a name="boost_regex.syntax.perl_syntax.h3"></a>
  80         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.anchors"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.anchors">Anchors</a>
  81       </h5>
  82 <p>
  83         A '^' character shall match the start of a line.
  84       </p>
  85 <p>
  86         A '$' character shall match the end of a line.
  87       </p>
  88 <h5>
  89 <a name="boost_regex.syntax.perl_syntax.h4"></a>
  90         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.marked_sub_expressions"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.marked_sub_expressions">Marked sub-expressions</a>
  91       </h5>
  92 <p>
  93         A section beginning <code class="literal">(</code> and ending <code class="literal">)</code>
  94         acts as a marked sub-expression. Whatever matched the sub-expression is split
  95         out in a separate field by the matching algorithms. Marked sub-expressions
  96         can also repeated, or referred to by a back-reference.
  97       </p>
  98 <h5>
  99 <a name="boost_regex.syntax.perl_syntax.h5"></a>
 100         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.non_marking_grouping"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.non_marking_grouping">Non-marking
 101         grouping</a>
 102       </h5>
 103 <p>
 104         A marked sub-expression is useful to lexically group part of a regular expression,
 105         but has the side-effect of spitting out an extra field in the result. As
 106         an alternative you can lexically group part of a regular expression, without
 107         generating a marked sub-expression by using <code class="literal">(?:</code> and <code class="literal">)</code>
 108         , for example <code class="literal">(?:ab)+</code> will repeat <code class="literal">ab</code>
 109         without splitting out any separate sub-expressions.
 110       </p>
 111 <h5>
 112 <a name="boost_regex.syntax.perl_syntax.h6"></a>
 113         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.repeats"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.repeats">Repeats</a>
 114       </h5>
 115 <p>
 116         Any atom (a single character, a marked sub-expression, or a character class)
 117         can be repeated with the <code class="literal">*</code>, <code class="literal">+</code>, <code class="literal">?</code>,
 118         and <code class="literal">{}</code> operators.
 119       </p>
 120 <p>
 121         The <code class="literal">*</code> operator will match the preceding atom zero or more
 122         times, for example the expression <code class="literal">a*b</code> will match any of
 123         the following:
 124       </p>
 125 <pre class="programlisting"><span class="identifier">b</span>
 126 <span class="identifier">ab</span>
 127 <span class="identifier">aaaaaaaab</span>
 128 </pre>
 129 <p>
 130         The <code class="literal">+</code> operator will match the preceding atom one or more
 131         times, for example the expression <code class="literal">a+b</code> will match any of
 132         the following:
 133       </p>
 134 <pre class="programlisting"><span class="identifier">ab</span>
 135 <span class="identifier">aaaaaaaab</span>
 136 </pre>
 137 <p>
 138         But will not match:
 139       </p>
 140 <pre class="programlisting"><span class="identifier">b</span>
 141 </pre>
 142 <p>
 143         The <code class="literal">?</code> operator will match the preceding atom zero or one
 144         times, for example the expression ca?b will match any of the following:
 145       </p>
 146 <pre class="programlisting"><span class="identifier">cb</span>
 147 <span class="identifier">cab</span>
 148 </pre>
 149 <p>
 150         But will not match:
 151       </p>
 152 <pre class="programlisting"><span class="identifier">caab</span>
 153 </pre>
 154 <p>
 155         An atom can also be repeated with a bounded repeat:
 156       </p>
 157 <p>
 158         <code class="literal">a{n}</code> Matches 'a' repeated exactly n times.
 159       </p>
 160 <p>
 161         <code class="literal">a{n,}</code> Matches 'a' repeated n or more times.
 162       </p>
 163 <p>
 164         <code class="literal">a{n, m}</code> Matches 'a' repeated between n and m times inclusive.
 165       </p>
 166 <p>
 167         For example:
 168       </p>
 169 <pre class="programlisting">^a{2,3}$</pre>
 170 <p>
 171         Will match either of:
 172       </p>
 173 <pre class="programlisting"><span class="identifier">aa</span>
 174 <span class="identifier">aaa</span>
 175 </pre>
 176 <p>
 177         But neither of:
 178       </p>
 179 <pre class="programlisting"><span class="identifier">a</span>
 180 <span class="identifier">aaaa</span>
 181 </pre>
 182 <p>
 183         Note that the "{" and "}" characters will treated as
 184         ordinary literals when used in a context that is not a repeat: this matches
 185         Perl 5.x behavior. For example in the expressions "ab{1", "ab1}"
 186         and "a{b}c" the curly brackets are all treated as literals and
 187         <span class="emphasis"><em>no error will be raised</em></span>.
 188       </p>
 189 <p>
 190         It is an error to use a repeat operator, if the preceding construct can not
 191         be repeated, for example:
 192       </p>
 193 <pre class="programlisting"><span class="identifier">a</span><span class="special">(*)</span>
 194 </pre>
 195 <p>
 196         Will raise an error, as there is nothing for the <code class="literal">*</code> operator
 197         to be applied to.
 198       </p>
 199 <h5>
 200 <a name="boost_regex.syntax.perl_syntax.h7"></a>
 201         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.non_greedy_repeats"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.non_greedy_repeats">Non
 202         greedy repeats</a>
 203       </h5>
 204 <p>
 205         The normal repeat operators are "greedy", that is to say they will
 206         consume as much input as possible. There are non-greedy versions available
 207         that will consume as little input as possible while still producing a match.
 208       </p>
 209 <p>
 210         <code class="literal">*?</code> Matches the previous atom zero or more times, while
 211         consuming as little input as possible.
 212       </p>
 213 <p>
 214         <code class="literal">+?</code> Matches the previous atom one or more times, while
 215         consuming as little input as possible.
 216       </p>
 217 <p>
 218         <code class="literal">??</code> Matches the previous atom zero or one times, while
 219         consuming as little input as possible.
 220       </p>
 221 <p>
 222         <code class="literal">{n,}?</code> Matches the previous atom n or more times, while
 223         consuming as little input as possible.
 224       </p>
 225 <p>
 226         <code class="literal">{n,m}?</code> Matches the previous atom between n and m times,
 227         while consuming as little input as possible.
 228       </p>
 229 <h5>
 230 <a name="boost_regex.syntax.perl_syntax.h8"></a>
 231         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.possessive_repeats"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.possessive_repeats">Possessive
 232         repeats</a>
 233       </h5>
 234 <p>
 235         By default when a repeated pattern does not match then the engine will backtrack
 236         until a match is found. However, this behaviour can sometime be undesireble
 237         so there are also "possessive" repeats: these match as much as
 238         possible and do not then allow backtracking if the rest of the expression
 239         fails to match.
 240       </p>
 241 <p>
 242         <code class="literal">*+</code> Matches the previous atom zero or more times, while
 243         giving nothing back.
 244       </p>
 245 <p>
 246         <code class="literal">++</code> Matches the previous atom one or more times, while
 247         giving nothing back.
 248       </p>
 249 <p>
 250         <code class="literal">?+</code> Matches the previous atom zero or one times, while
 251         giving nothing back.
 252       </p>
 253 <p>
 254         <code class="literal">{n,}+</code> Matches the previous atom n or more times, while
 255         giving nothing back.
 256       </p>
 257 <p>
 258         <code class="literal">{n,m}+</code> Matches the previous atom between n and m times,
 259         while giving nothing back.
 260       </p>
 261 <h5>
 262 <a name="boost_regex.syntax.perl_syntax.h9"></a>
 263         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.back_references"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.back_references">Back
 264         references</a>
 265       </h5>
 266 <p>
 267         An escape character followed by a digit <span class="emphasis"><em>n</em></span>, where <span class="emphasis"><em>n</em></span>
 268         is in the range 1-9, matches the same string that was matched by sub-expression
 269         <span class="emphasis"><em>n</em></span>. For example the expression:
 270       </p>
 271 <pre class="programlisting">^(a*)[^a]*\1$</pre>
 272 <p>
 273         Will match the string:
 274       </p>
 275 <pre class="programlisting"><span class="identifier">aaabbaaa</span>
 276 </pre>
 277 <p>
 278         But not the string:
 279       </p>
 280 <pre class="programlisting"><span class="identifier">aaabba</span>
 281 </pre>
 282 <p>
 283         You can also use the \g escape for the same function, for example:
 284       </p>
 285 <div class="informaltable"><table class="table">
 286 <colgroup>
 287 <col>
 288 <col>
 289 </colgroup>
 290 <thead><tr>
 291 <th>
 292                 <p>
 293                   Escape
 294                 </p>
 295               </th>
 296 <th>
 297                 <p>
 298                   Meaning
 299                 </p>
 300               </th>
 301 </tr></thead>
 302 <tbody>
 303 <tr>
 304 <td>
 305                 <p>
 306                   <code class="literal">\g1</code>
 307                 </p>
 308               </td>
 309 <td>
 310                 <p>
 311                   Match whatever matched sub-expression 1
 312                 </p>
 313               </td>
 314 </tr>
 315 <tr>
 316 <td>
 317                 <p>
 318                   <code class="literal">\g{1}</code>
 319                 </p>
 320               </td>
 321 <td>
 322                 <p>
 323                   Match whatever matched sub-expression 1: this form allows for safer
 324                   parsing of the expression in cases like <code class="literal">\g{1}2</code>
 325                   or for indexes higher than 9 as in <code class="literal">\g{1234}</code>
 326                 </p>
 327               </td>
 328 </tr>
 329 <tr>
 330 <td>
 331                 <p>
 332                   <code class="literal">\g-1</code>
 333                 </p>
 334               </td>
 335 <td>
 336                 <p>
 337                   Match whatever matched the last opened sub-expression
 338                 </p>
 339               </td>
 340 </tr>
 341 <tr>
 342 <td>
 343                 <p>
 344                   <code class="literal">\g{-2}</code>
 345                 </p>
 346               </td>
 347 <td>
 348                 <p>
 349                   Match whatever matched the last but one opened sub-expression
 350                 </p>
 351               </td>
 352 </tr>
 353 <tr>
 354 <td>
 355                 <p>
 356                   <code class="literal">\g{one}</code>
 357                 </p>
 358               </td>
 359 <td>
 360                 <p>
 361                   Match whatever matched the sub-expression named "one"
 362                 </p>
 363               </td>
 364 </tr>
 365 </tbody>
 366 </table></div>
 367 <p>
 368         Finally the \k escape can be used to refer to named subexpressions, for example
 369         <code class="literal">\k&lt;two&gt;</code> will match whatever matched the subexpression
 370         named "two".
 371       </p>
 372 <h5>
 373 <a name="boost_regex.syntax.perl_syntax.h10"></a>
 374         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.alternation"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.alternation">Alternation</a>
 375       </h5>
 376 <p>
 377         The <code class="literal">|</code> operator will match either of its arguments, so
 378         for example: <code class="literal">abc|def</code> will match either "abc"
 379         or "def".
 380       </p>
 381 <p>
 382         Parenthesis can be used to group alternations, for example: <code class="literal">ab(d|ef)</code>
 383         will match either of "abd" or "abef".
 384       </p>
 385 <p>
 386         Empty alternatives are not allowed (these are almost always a mistake), but
 387         if you really want an empty alternative use <code class="literal">(?:)</code> as a
 388         placeholder, for example:
 389       </p>
 390 <p>
 391         <code class="literal">|abc</code> is not a valid expression, but
 392       </p>
 393 <p>
 394         <code class="literal">(?:)|abc</code> is and is equivalent, also the expression:
 395       </p>
 396 <p>
 397         <code class="literal">(?:abc)??</code> has exactly the same effect.
 398       </p>
 399 <h5>
 400 <a name="boost_regex.syntax.perl_syntax.h11"></a>
 401         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.character_sets"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.character_sets">Character
 402         sets</a>
 403       </h5>
 404 <p>
 405         A character set is a bracket-expression starting with <code class="literal">[] and ending
 406         with <code class="literal"></code></code>, it defines a set of characters, and matches
 407         any single character that is a member of that set.
 408       </p>
 409 <p>
 410         A bracket expression may contain any combination of the following:
 411       </p>
 412 <h6>
 413 <a name="boost_regex.syntax.perl_syntax.h12"></a>
 414         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.single_characters"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.single_characters">Single
 415         characters</a>
 416       </h6>
 417 <p>
 418         For example <code class="literal">[abc]</code>, will match any of the characters 'a',
 419         'b', or 'c'.
 420       </p>
 421 <h6>
 422 <a name="boost_regex.syntax.perl_syntax.h13"></a>
 423         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.character_ranges"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.character_ranges">Character
 424         ranges</a>
 425       </h6>
 426 <p>
 427         For example <code class="literal">[a-c]</code> will match any single character in the
 428         range 'a' to 'c'. By default, for Perl regular expressions, a character x
 429         is within the range y to z, if the code point of the character lies within
 430         the codepoints of the endpoints of the range. Alternatively, if you set the
 431         <a class="link" href="../ref/syntax_option_type/syntax_option_type_perl.html" title="Options for Perl Regular Expressions"><code class="literal">collate</code>
 432         flag</a> when constructing the regular expression, then ranges are locale
 433         sensitive.
 434       </p>
 435 <h6>
 436 <a name="boost_regex.syntax.perl_syntax.h14"></a>
 437         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.negation"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.negation">Negation</a>
 438       </h6>
 439 <p>
 440         If the bracket-expression begins with the ^ character, then it matches the
 441         complement of the characters it contains, for example <code class="literal">[^a-c]</code>
 442         matches any character that is not in the range <code class="literal">a-c</code>.
 443       </p>
 444 <h6>
 445 <a name="boost_regex.syntax.perl_syntax.h15"></a>
 446         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.character_classes"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.character_classes">Character
 447         classes</a>
 448       </h6>
 449 <p>
 450         An expression of the form <code class="literal">[[:name:]]</code> matches the named
 451         character class "name", for example <code class="literal">[[:lower:]]</code>
 452         matches any lower case character. See <a class="link" href="character_classes.html" title="Character Class Names">character
 453         class names</a>.
 454       </p>
 455 <h6>
 456 <a name="boost_regex.syntax.perl_syntax.h16"></a>
 457         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.collating_elements"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.collating_elements">Collating
 458         Elements</a>
 459       </h6>
 460 <p>
 461         An expression of the form <code class="literal">[[.col.]]</code> matches the collating
 462         element <span class="emphasis"><em>col</em></span>. A collating element is any single character,
 463         or any sequence of characters that collates as a single unit. Collating elements
 464         may also be used as the end point of a range, for example: <code class="literal">[[.ae.]-c]</code>
 465         matches the character sequence "ae", plus any single character
 466         in the range "ae"-c, assuming that "ae" is treated as
 467         a single collating element in the current locale.
 468       </p>
 469 <p>
 470         As an extension, a collating element may also be specified via it's <a class="link" href="collating_names.html" title="Collating Names">symbolic name</a>, for example:
 471       </p>
 472 <pre class="programlisting"><span class="special">[[.</span><span class="identifier">NUL</span><span class="special">.]]</span>
 473 </pre>
 474 <p>
 475         matches a <code class="literal">\0</code> character.
 476       </p>
 477 <h6>
 478 <a name="boost_regex.syntax.perl_syntax.h17"></a>
 479         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.equivalence_classes"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.equivalence_classes">Equivalence
 480         classes</a>
 481       </h6>
 482 <p>
 483         An expression of the form <code class="literal">[[=col=]]</code>, matches any character
 484         or collating element whose primary sort key is the same as that for collating
 485         element <span class="emphasis"><em>col</em></span>, as with collating elements the name <span class="emphasis"><em>col</em></span>
 486         may be a <a class="link" href="collating_names.html" title="Collating Names">symbolic name</a>.
 487         A primary sort key is one that ignores case, accentation, or locale-specific
 488         tailorings; so for example <code class="computeroutput"><span class="special">[[=</span><span class="identifier">a</span><span class="special">=]]</span></code> matches
 489         any of the characters: a, &#192;, &#193;, &#194;, &#195;, &#196;, &#197;, A, &#224;, &#225;, &#226;, &#227;, &#228; and &#229;. Unfortunately implementation
 490         of this is reliant on the platform's collation and localisation support;
 491         this feature can not be relied upon to work portably across all platforms,
 492         or even all locales on one platform.
 493       </p>
 494 <h6>
 495 <a name="boost_regex.syntax.perl_syntax.h18"></a>
 496         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.escaped_characters"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.escaped_characters">Escaped
 497         Characters</a>
 498       </h6>
 499 <p>
 500         All the escape sequences that match a single character, or a single character
 501         class are permitted within a character class definition. For example <code class="computeroutput"><span class="special">[\[\]]</span></code> would match either of <code class="computeroutput"><span class="special">[</span></code> or <code class="computeroutput"><span class="special">]</span></code>
 502         while <code class="computeroutput"><span class="special">[\</span><span class="identifier">W</span><span class="special">\</span><span class="identifier">d</span><span class="special">]</span></code>
 503         would match any character that is either a "digit", <span class="emphasis"><em>or</em></span>
 504         is <span class="emphasis"><em>not</em></span> a "word" character.
 505       </p>
 506 <h6>
 507 <a name="boost_regex.syntax.perl_syntax.h19"></a>
 508         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.combinations"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.combinations">Combinations</a>
 509       </h6>
 510 <p>
 511         All of the above can be combined in one character set declaration, for example:
 512         <code class="literal">[[:digit:]a-c[.NUL.]]</code>.
 513       </p>
 514 <h5>
 515 <a name="boost_regex.syntax.perl_syntax.h20"></a>
 516         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.escapes"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.escapes">Escapes</a>
 517       </h5>
 518 <p>
 519         Any special character preceded by an escape shall match itself.
 520       </p>
 521 <p>
 522         The following escape sequences are all synonyms for single characters:
 523       </p>
 524 <div class="informaltable"><table class="table">
 525 <colgroup>
 526 <col>
 527 <col>
 528 </colgroup>
 529 <thead><tr>
 530 <th>
 531                 <p>
 532                   Escape
 533                 </p>
 534               </th>
 535 <th>
 536                 <p>
 537                   Character
 538                 </p>
 539               </th>
 540 </tr></thead>
 541 <tbody>
 542 <tr>
 543 <td>
 544                 <p>
 545                   <code class="literal">\a</code>
 546                 </p>
 547               </td>
 548 <td>
 549                 <p>
 550                   <code class="literal">\a</code>
 551                 </p>
 552               </td>
 553 </tr>
 554 <tr>
 555 <td>
 556                 <p>
 557                   <code class="literal">\e</code>
 558                 </p>
 559               </td>
 560 <td>
 561                 <p>
 562                   <code class="literal">0x1B</code>
 563                 </p>
 564               </td>
 565 </tr>
 566 <tr>
 567 <td>
 568                 <p>
 569                   <code class="literal">\f</code>
 570                 </p>
 571               </td>
 572 <td>
 573                 <p>
 574                   <code class="literal">\f</code>
 575                 </p>
 576               </td>
 577 </tr>
 578 <tr>
 579 <td>
 580                 <p>
 581                   <code class="literal">\n</code>
 582                 </p>
 583               </td>
 584 <td>
 585                 <p>
 586                   <code class="literal">\n</code>
 587                 </p>
 588               </td>
 589 </tr>
 590 <tr>
 591 <td>
 592                 <p>
 593                   <code class="literal">\r</code>
 594                 </p>
 595               </td>
 596 <td>
 597                 <p>
 598                   <code class="literal">\r</code>
 599                 </p>
 600               </td>
 601 </tr>
 602 <tr>
 603 <td>
 604                 <p>
 605                   <code class="literal">\t</code>
 606                 </p>
 607               </td>
 608 <td>
 609                 <p>
 610                   <code class="literal">\t</code>
 611                 </p>
 612               </td>
 613 </tr>
 614 <tr>
 615 <td>
 616                 <p>
 617                   <code class="literal">\v</code>
 618                 </p>
 619               </td>
 620 <td>
 621                 <p>
 622                   <code class="literal">\v</code>
 623                 </p>
 624               </td>
 625 </tr>
 626 <tr>
 627 <td>
 628                 <p>
 629                   <code class="literal">\b</code>
 630                 </p>
 631               </td>
 632 <td>
 633                 <p>
 634                   <code class="literal">\b</code> (but only inside a character class declaration).
 635                 </p>
 636               </td>
 637 </tr>
 638 <tr>
 639 <td>
 640                 <p>
 641                   <code class="literal">\cX</code>
 642                 </p>
 643               </td>
 644 <td>
 645                 <p>
 646                   An ASCII escape sequence - the character whose code point is X
 647                   % 32
 648                 </p>
 649               </td>
 650 </tr>
 651 <tr>
 652 <td>
 653                 <p>
 654                   <code class="literal">\xdd</code>
 655                 </p>
 656               </td>
 657 <td>
 658                 <p>
 659                   A hexadecimal escape sequence - matches the single character whose
 660                   code point is 0xdd.
 661                 </p>
 662               </td>
 663 </tr>
 664 <tr>
 665 <td>
 666                 <p>
 667                   <code class="literal">\x{dddd}</code>
 668                 </p>
 669               </td>
 670 <td>
 671                 <p>
 672                   A hexadecimal escape sequence - matches the single character whose
 673                   code point is 0xdddd.
 674                 </p>
 675               </td>
 676 </tr>
 677 <tr>
 678 <td>
 679                 <p>
 680                   <code class="literal">\0ddd</code>
 681                 </p>
 682               </td>
 683 <td>
 684                 <p>
 685                   An octal escape sequence - matches the single character whose code
 686                   point is 0ddd.
 687                 </p>
 688               </td>
 689 </tr>
 690 <tr>
 691 <td>
 692                 <p>
 693                   <code class="literal">\N{name}</code>
 694                 </p>
 695               </td>
 696 <td>
 697                 <p>
 698                   Matches the single character which has the <a class="link" href="collating_names.html" title="Collating Names">symbolic
 699                   name</a> <span class="emphasis"><em>name</em></span>. For example <code class="literal">\N{newline}</code>
 700                   matches the single character \n.
 701                 </p>
 702               </td>
 703 </tr>
 704 </tbody>
 705 </table></div>
 706 <h6>
 707 <a name="boost_regex.syntax.perl_syntax.h21"></a>
 708         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.single_character_character_class"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.single_character_character_class">"Single
 709         character" character classes:</a>
 710       </h6>
 711 <p>
 712         Any escaped character <span class="emphasis"><em>x</em></span>, if <span class="emphasis"><em>x</em></span> is
 713         the name of a character class shall match any character that is a member
 714         of that class, and any escaped character <span class="emphasis"><em>X</em></span>, if <span class="emphasis"><em>x</em></span>
 715         is the name of a character class, shall match any character not in that class.
 716       </p>
 717 <p>
 718         The following are supported by default:
 719       </p>
 720 <div class="informaltable"><table class="table">
 721 <colgroup>
 722 <col>
 723 <col>
 724 </colgroup>
 725 <thead><tr>
 726 <th>
 727                 <p>
 728                   Escape sequence
 729                 </p>
 730               </th>
 731 <th>
 732                 <p>
 733                   Equivalent to
 734                 </p>
 735               </th>
 736 </tr></thead>
 737 <tbody>
 738 <tr>
 739 <td>
 740                 <p>
 741                   <code class="computeroutput"><span class="special">\</span><span class="identifier">d</span></code>
 742                 </p>
 743               </td>
 744 <td>
 745                 <p>
 746                   <code class="computeroutput"><span class="special">[[:</span><span class="identifier">digit</span><span class="special">:]]</span></code>
 747                 </p>
 748               </td>
 749 </tr>
 750 <tr>
 751 <td>
 752                 <p>
 753                   <code class="computeroutput"><span class="special">\</span><span class="identifier">l</span></code>
 754                 </p>
 755               </td>
 756 <td>
 757                 <p>
 758                   <code class="computeroutput"><span class="special">[[:</span><span class="identifier">lower</span><span class="special">:]]</span></code>
 759                 </p>
 760               </td>
 761 </tr>
 762 <tr>
 763 <td>
 764                 <p>
 765                   <code class="computeroutput"><span class="special">\</span><span class="identifier">s</span></code>
 766                 </p>
 767               </td>
 768 <td>
 769                 <p>
 770                   <code class="computeroutput"><span class="special">[[:</span><span class="identifier">space</span><span class="special">:]]</span></code>
 771                 </p>
 772               </td>
 773 </tr>
 774 <tr>
 775 <td>
 776                 <p>
 777                   <code class="computeroutput"><span class="special">\</span><span class="identifier">u</span></code>
 778                 </p>
 779               </td>
 780 <td>
 781                 <p>
 782                   <code class="computeroutput"><span class="special">[[:</span><span class="identifier">upper</span><span class="special">:]]</span></code>
 783                 </p>
 784               </td>
 785 </tr>
 786 <tr>
 787 <td>
 788                 <p>
 789                   <code class="computeroutput"><span class="special">\</span><span class="identifier">w</span></code>
 790                 </p>
 791               </td>
 792 <td>
 793                 <p>
 794                   <code class="computeroutput"><span class="special">[[:</span><span class="identifier">word</span><span class="special">:]]</span></code>
 795                 </p>
 796               </td>
 797 </tr>
 798 <tr>
 799 <td>
 800                 <p>
 801                   <code class="computeroutput"><span class="special">\</span><span class="identifier">h</span></code>
 802                 </p>
 803               </td>
 804 <td>
 805                 <p>
 806                   Horizontal whitespace
 807                 </p>
 808               </td>
 809 </tr>
 810 <tr>
 811 <td>
 812                 <p>
 813                   <code class="computeroutput"><span class="special">\</span><span class="identifier">v</span></code>
 814                 </p>
 815               </td>
 816 <td>
 817                 <p>
 818                   Vertical whitespace
 819                 </p>
 820               </td>
 821 </tr>
 822 <tr>
 823 <td>
 824                 <p>
 825                   <code class="computeroutput"><span class="special">\</span><span class="identifier">D</span></code>
 826                 </p>
 827               </td>
 828 <td>
 829                 <p>
 830                   <code class="computeroutput"><span class="special">[^[:</span><span class="identifier">digit</span><span class="special">:]]</span></code>
 831                 </p>
 832               </td>
 833 </tr>
 834 <tr>
 835 <td>
 836                 <p>
 837                   <code class="computeroutput"><span class="special">\</span><span class="identifier">L</span></code>
 838                 </p>
 839               </td>
 840 <td>
 841                 <p>
 842                   <code class="computeroutput"><span class="special">[^[:</span><span class="identifier">lower</span><span class="special">:]]</span></code>
 843                 </p>
 844               </td>
 845 </tr>
 846 <tr>
 847 <td>
 848                 <p>
 849                   <code class="computeroutput"><span class="special">\</span><span class="identifier">S</span></code>
 850                 </p>
 851               </td>
 852 <td>
 853                 <p>
 854                   <code class="computeroutput"><span class="special">[^[:</span><span class="identifier">space</span><span class="special">:]]</span></code>
 855                 </p>
 856               </td>
 857 </tr>
 858 <tr>
 859 <td>
 860                 <p>
 861                   <code class="computeroutput"><span class="special">\</span><span class="identifier">U</span></code>
 862                 </p>
 863               </td>
 864 <td>
 865                 <p>
 866                   <code class="computeroutput"><span class="special">[^[:</span><span class="identifier">upper</span><span class="special">:]]</span></code>
 867                 </p>
 868               </td>
 869 </tr>
 870 <tr>
 871 <td>
 872                 <p>
 873                   <code class="computeroutput"><span class="special">\</span><span class="identifier">W</span></code>
 874                 </p>
 875               </td>
 876 <td>
 877                 <p>
 878                   <code class="computeroutput"><span class="special">[^[:</span><span class="identifier">word</span><span class="special">:]]</span></code>
 879                 </p>
 880               </td>
 881 </tr>
 882 <tr>
 883 <td>
 884                 <p>
 885                   <code class="computeroutput"><span class="special">\</span><span class="identifier">H</span></code>
 886                 </p>
 887               </td>
 888 <td>
 889                 <p>
 890                   Not Horizontal whitespace
 891                 </p>
 892               </td>
 893 </tr>
 894 <tr>
 895 <td>
 896                 <p>
 897                   <code class="computeroutput"><span class="special">\</span><span class="identifier">V</span></code>
 898                 </p>
 899               </td>
 900 <td>
 901                 <p>
 902                   Not Vertical whitespace
 903                 </p>
 904               </td>
 905 </tr>
 906 </tbody>
 907 </table></div>
 908 <h6>
 909 <a name="boost_regex.syntax.perl_syntax.h22"></a>
 910         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.character_properties"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.character_properties">Character
 911         Properties</a>
 912       </h6>
 913 <p>
 914         The character property names in the following table are all equivalent to
 915         the <a class="link" href="character_classes.html" title="Character Class Names">names used in character
 916         classes</a>.
 917       </p>
 918 <div class="informaltable"><table class="table">
 919 <colgroup>
 920 <col>
 921 <col>
 922 <col>
 923 </colgroup>
 924 <thead><tr>
 925 <th>
 926                 <p>
 927                   Form
 928                 </p>
 929               </th>
 930 <th>
 931                 <p>
 932                   Description
 933                 </p>
 934               </th>
 935 <th>
 936                 <p>
 937                   Equivalent character set form
 938                 </p>
 939               </th>
 940 </tr></thead>
 941 <tbody>
 942 <tr>
 943 <td>
 944                 <p>
 945                   <code class="computeroutput"><span class="special">\</span><span class="identifier">pX</span></code>
 946                 </p>
 947               </td>
 948 <td>
 949                 <p>
 950                   Matches any character that has the property X.
 951                 </p>
 952               </td>
 953 <td>
 954                 <p>
 955                   <code class="computeroutput"><span class="special">[[:</span><span class="identifier">X</span><span class="special">:]]</span></code>
 956                 </p>
 957               </td>
 958 </tr>
 959 <tr>
 960 <td>
 961                 <p>
 962                   <code class="computeroutput"><span class="special">\</span><span class="identifier">p</span><span class="special">{</span><span class="identifier">Name</span><span class="special">}</span></code>
 963                 </p>
 964               </td>
 965 <td>
 966                 <p>
 967                   Matches any character that has the property Name.
 968                 </p>
 969               </td>
 970 <td>
 971                 <p>
 972                   <code class="computeroutput"><span class="special">[[:</span><span class="identifier">Name</span><span class="special">:]]</span></code>
 973                 </p>
 974               </td>
 975 </tr>
 976 <tr>
 977 <td>
 978                 <p>
 979                   <code class="computeroutput"><span class="special">\</span><span class="identifier">PX</span></code>
 980                 </p>
 981               </td>
 982 <td>
 983                 <p>
 984                   Matches any character that does not have the property X.
 985                 </p>
 986               </td>
 987 <td>
 988                 <p>
 989                   <code class="computeroutput"><span class="special">[^[:</span><span class="identifier">X</span><span class="special">:]]</span></code>
 990                 </p>
 991               </td>
 992 </tr>
 993 <tr>
 994 <td>
 995                 <p>
 996                   <code class="computeroutput"><span class="special">\</span><span class="identifier">P</span><span class="special">{</span><span class="identifier">Name</span><span class="special">}</span></code>
 997                 </p>
 998               </td>
 999 <td>
1000                 <p>
1001                   Matches any character that does not have the property Name.
1002                 </p>
1003               </td>
1004 <td>
1005                 <p>
1006                   <code class="computeroutput"><span class="special">[^[:</span><span class="identifier">Name</span><span class="special">:]]</span></code>
1007                 </p>
1008               </td>
1009 </tr>
1010 </tbody>
1011 </table></div>
1012 <p>
1013         For example <code class="literal">\pd</code> matches any "digit" character,
1014         as does <code class="literal">\p{digit}</code>.
1015       </p>
1016 <h6>
1017 <a name="boost_regex.syntax.perl_syntax.h23"></a>
1018         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.word_boundaries"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.word_boundaries">Word
1019         Boundaries</a>
1020       </h6>
1021 <p>
1022         The following escape sequences match the boundaries of words:
1023       </p>
1024 <p>
1025         <code class="literal">&lt;</code> Matches the start of a word.
1026       </p>
1027 <p>
1028         <code class="literal">&gt;</code> Matches the end of a word.
1029       </p>
1030 <p>
1031         <code class="literal">\b</code> Matches a word boundary (the start or end of a word).
1032       </p>
1033 <p>
1034         <code class="literal">\B</code> Matches only when not at a word boundary.
1035       </p>
1036 <h6>
1037 <a name="boost_regex.syntax.perl_syntax.h24"></a>
1038         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.buffer_boundaries"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.buffer_boundaries">Buffer
1039         boundaries</a>
1040       </h6>
1041 <p>
1042         The following match only at buffer boundaries: a "buffer" in this
1043         context is the whole of the input text that is being matched against (note
1044         that ^ and $ may match embedded newlines within the text).
1045       </p>
1046 <p>
1047         \` Matches at the start of a buffer only.
1048       </p>
1049 <p>
1050         \' Matches at the end of a buffer only.
1051       </p>
1052 <p>
1053         \A Matches at the start of a buffer only (the same as <code class="literal">\`</code>).
1054       </p>
1055 <p>
1056         \z Matches at the end of a buffer only (the same as <code class="literal">\'</code>).
1057       </p>
1058 <p>
1059         \Z Matches a zero-width assertion consisting of an optional sequence of newlines
1060         at the end of a buffer: equivalent to the regular expression <code class="literal">(?=\v*\z)</code>.
1061         Note that this is subtly different from Perl which behaves as if matching
1062         <code class="literal">(?=\n?\z)</code>.
1063       </p>
1064 <h6>
1065 <a name="boost_regex.syntax.perl_syntax.h25"></a>
1066         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.continuation_escape"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.continuation_escape">Continuation
1067         Escape</a>
1068       </h6>
1069 <p>
1070         The sequence <code class="literal">\G</code> matches only at the end of the last match
1071         found, or at the start of the text being matched if no previous match was
1072         found. This escape useful if you're iterating over the matches contained
1073         within a text, and you want each subsequence match to start where the last
1074         one ended.
1075       </p>
1076 <h6>
1077 <a name="boost_regex.syntax.perl_syntax.h26"></a>
1078         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.quoting_escape"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.quoting_escape">Quoting
1079         escape</a>
1080       </h6>
1081 <p>
1082         The escape sequence <code class="literal">\Q</code> begins a "quoted sequence":
1083         all the subsequent characters are treated as literals, until either the end
1084         of the regular expression or \E is found. For example the expression: <code class="literal">\Q*+\Ea+</code>
1085         would match either of:
1086       </p>
1087 <pre class="programlisting"><span class="special">\*+</span><span class="identifier">a</span>
1088 <span class="special">\*+</span><span class="identifier">aaa</span>
1089 </pre>
1090 <h6>
1091 <a name="boost_regex.syntax.perl_syntax.h27"></a>
1092         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.unicode_escapes"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.unicode_escapes">Unicode
1093         escapes</a>
1094       </h6>
1095 <p>
1096         <code class="literal">\C</code> Matches a single code point: in Boost regex this has
1097         exactly the same effect as a "." operator. <code class="literal">\X</code>
1098         Matches a combining character sequence: that is any non-combining character
1099         followed by a sequence of zero or more combining characters.
1100       </p>
1101 <h6>
1102 <a name="boost_regex.syntax.perl_syntax.h28"></a>
1103         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.matching_line_endings"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.matching_line_endings">Matching Line
1104         Endings</a>
1105       </h6>
1106 <p>
1107         The escape sequence <code class="literal">\R</code> matches any line ending character
1108         sequence, specifically it is identical to the expression <code class="literal">(?&gt;\x0D\x0A?|[\x0A-\x0C\x85\x{2028}\x{2029}])</code>.
1109       </p>
1110 <h6>
1111 <a name="boost_regex.syntax.perl_syntax.h29"></a>
1112         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.keeping_back_some_text"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.keeping_back_some_text">Keeping back
1113         some text</a>
1114       </h6>
1115 <p>
1116         <code class="literal">\K</code> Resets the start location of $0 to the current text
1117         position: in other words everything to the left of \K is "kept back"
1118         and does not form part of the regular expression match. $` is updated accordingly.
1119       </p>
1120 <p>
1121         For example <code class="literal">foo\Kbar</code> matched against the text "foobar"
1122         would return the match "bar" for $0 and "foo" for $`.
1123         This can be used to simulate variable width lookbehind assertions.
1124       </p>
1125 <h6>
1126 <a name="boost_regex.syntax.perl_syntax.h30"></a>
1127         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.any_other_escape"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.any_other_escape">Any
1128         other escape</a>
1129       </h6>
1130 <p>
1131         Any other escape sequence matches the character that is escaped, for example
1132         \@ matches a literal '@'.
1133       </p>
1134 <h5>
1135 <a name="boost_regex.syntax.perl_syntax.h31"></a>
1136         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.perl_extended_patterns"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.perl_extended_patterns">Perl Extended
1137         Patterns</a>
1138       </h5>
1139 <p>
1140         Perl-specific extensions to the regular expression syntax all start with
1141         <code class="literal">(?</code>.
1142       </p>
1143 <h6>
1144 <a name="boost_regex.syntax.perl_syntax.h32"></a>
1145         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.named_subexpressions"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.named_subexpressions">Named
1146         Subexpressions</a>
1147       </h6>
1148 <p>
1149         You can create a named subexpression using:
1150       </p>
1151 <pre class="programlisting"><span class="special">(?&lt;</span><span class="identifier">NAME</span><span class="special">&gt;</span><span class="identifier">expression</span><span class="special">)</span>
1152 </pre>
1153 <p>
1154         Which can be then be referred to by the name <span class="emphasis"><em>NAME</em></span>. Alternatively
1155         you can delimit the name using 'NAME' as in:
1156       </p>
1157 <pre class="programlisting"><span class="special">(?</span><span class="char">'NAME'</span><span class="identifier">expression</span><span class="special">)</span>
1158 </pre>
1159 <p>
1160         These named subexpressions can be referred to in a backreference using either
1161         <code class="literal">\g{NAME}</code> or <code class="literal">\k&lt;NAME&gt;</code> and can
1162         also be referred to by name in a <a class="link" href="../format/perl_format.html" title="Perl Format String Syntax">Perl</a>
1163         format string for search and replace operations, or in the <a class="link" href="../ref/match_results.html" title="match_results"><code class="computeroutput"><span class="identifier">match_results</span></code></a> member functions.
1164       </p>
1165 <h6>
1166 <a name="boost_regex.syntax.perl_syntax.h33"></a>
1167         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.comments"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.comments">Comments</a>
1168       </h6>
1169 <p>
1170         <code class="literal">(?# ... )</code> is treated as a comment, it's contents are ignored.
1171       </p>
1172 <h6>
1173 <a name="boost_regex.syntax.perl_syntax.h34"></a>
1174         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.modifiers"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.modifiers">Modifiers</a>
1175       </h6>
1176 <p>
1177         <code class="literal">(?imsx-imsx ... )</code> alters which of the perl modifiers are
1178         in effect within the pattern, changes take effect from the point that the
1179         block is first seen and extend to any enclosing <code class="literal">)</code>. Letters
1180         before a '-' turn that perl modifier on, letters afterward, turn it off.
1181       </p>
1182 <p>
1183         <code class="literal">(?imsx-imsx:pattern)</code> applies the specified modifiers to
1184         pattern only.
1185       </p>
1186 <h6>
1187 <a name="boost_regex.syntax.perl_syntax.h35"></a>
1188         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.non_marking_groups"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.non_marking_groups">Non-marking
1189         groups</a>
1190       </h6>
1191 <p>
1192         <code class="literal">(?:pattern)</code> lexically groups pattern, without generating
1193         an additional sub-expression.
1194       </p>
1195 <h6>
1196 <a name="boost_regex.syntax.perl_syntax.h36"></a>
1197         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.branch_reset"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.branch_reset">Branch
1198         reset</a>
1199       </h6>
1200 <p>
1201         <code class="literal">(?|pattern)</code> resets the subexpression count at the start
1202         of each "|" alternative within <span class="emphasis"><em>pattern</em></span>.
1203       </p>
1204 <p>
1205         The sub-expression count following this construct is that of whichever branch
1206         had the largest number of sub-expressions. This construct is useful when
1207         you want to capture one of a number of alternative matches in a single sub-expression
1208         index.
1209       </p>
1210 <p>
1211         In the following example the index of each sub-expression is shown below
1212         the expression:
1213       </p>
1214 <pre class="programlisting"># before  ---------------branch-reset----------- after
1215 / ( a )  (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1216 # 1            2         2  3        2     3     4
1217 </pre>
1218 <h6>
1219 <a name="boost_regex.syntax.perl_syntax.h37"></a>
1220         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.lookahead"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.lookahead">Lookahead</a>
1221       </h6>
1222 <p>
1223         <code class="literal">(?=pattern)</code> consumes zero characters, only if pattern
1224         matches.
1225       </p>
1226 <p>
1227         <code class="literal">(?!pattern)</code> consumes zero characters, only if pattern
1228         does not match.
1229       </p>
1230 <p>
1231         Lookahead is typically used to create the logical AND of two regular expressions,
1232         for example if a password must contain a lower case letter, an upper case
1233         letter, a punctuation symbol, and be at least 6 characters long, then the
1234         expression:
1235       </p>
1236 <pre class="programlisting"><span class="special">(?=.*[[:</span><span class="identifier">lower</span><span class="special">:]])(?=.*[[:</span><span class="identifier">upper</span><span class="special">:]])(?=.*[[:</span><span class="identifier">punct</span><span class="special">:]]).{</span><span class="number">6</span><span class="special">,}</span>
1237 </pre>
1238 <p>
1239         could be used to validate the password.
1240       </p>
1241 <h6>
1242 <a name="boost_regex.syntax.perl_syntax.h38"></a>
1243         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.lookbehind"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.lookbehind">Lookbehind</a>
1244       </h6>
1245 <p>
1246         <code class="literal">(?&lt;=pattern)</code> consumes zero characters, only if pattern
1247         could be matched against the characters preceding the current position (pattern
1248         must be of fixed length).
1249       </p>
1250 <p>
1251         <code class="literal">(?&lt;!pattern)</code> consumes zero characters, only if pattern
1252         could not be matched against the characters preceding the current position
1253         (pattern must be of fixed length).
1254       </p>
1255 <h6>
1256 <a name="boost_regex.syntax.perl_syntax.h39"></a>
1257         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.independent_sub_expressions"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.independent_sub_expressions">Independent
1258         sub-expressions</a>
1259       </h6>
1260 <p>
1261         <code class="literal">(?&gt;pattern)</code> <span class="emphasis"><em>pattern</em></span> is matched
1262         independently of the surrounding patterns, the expression will never backtrack
1263         into <span class="emphasis"><em>pattern</em></span>. Independent sub-expressions are typically
1264         used to improve performance; only the best possible match for pattern will
1265         be considered, if this doesn't allow the expression as a whole to match then
1266         no match is found at all.
1267       </p>
1268 <h6>
1269 <a name="boost_regex.syntax.perl_syntax.h40"></a>
1270         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.recursive_expressions"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.recursive_expressions">Recursive
1271         Expressions</a>
1272       </h6>
1273 <p>
1274         <code class="literal">(?<span class="emphasis"><em>N</em></span>) (?-<span class="emphasis"><em>N</em></span>) (?+<span class="emphasis"><em>N</em></span>)
1275         (?R) (?0) (?&amp;NAME)</code>
1276       </p>
1277 <p>
1278         <code class="literal">(?R)</code> and <code class="literal">(?0)</code> recurse to the start
1279         of the entire pattern.
1280       </p>
1281 <p>
1282         <code class="literal">(?<span class="emphasis"><em>N</em></span>)</code> executes sub-expression <span class="emphasis"><em>N</em></span>
1283         recursively, for example <code class="literal">(?2)</code> will recurse to sub-expression
1284         2.
1285       </p>
1286 <p>
1287         <code class="literal">(?-<span class="emphasis"><em>N</em></span>)</code> and <code class="literal">(?+<span class="emphasis"><em>N</em></span>)</code>
1288         are relative recursions, so for example <code class="literal">(?-1)</code> recurses
1289         to the last sub-expression to be declared, and <code class="literal">(?+1)</code> recurses
1290         to the next sub-expression to be declared.
1291       </p>
1292 <p>
1293         <code class="literal">(?&amp;NAME)</code> recurses to named sub-expression <span class="emphasis"><em>NAME</em></span>.
1294       </p>
1295 <h6>
1296 <a name="boost_regex.syntax.perl_syntax.h41"></a>
1297         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.conditional_expressions"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.conditional_expressions">Conditional
1298         Expressions</a>
1299       </h6>
1300 <p>
1301         <code class="literal">(?(condition)yes-pattern|no-pattern)</code> attempts to match
1302         <span class="emphasis"><em>yes-pattern</em></span> if the <span class="emphasis"><em>condition</em></span> is
1303         true, otherwise attempts to match <span class="emphasis"><em>no-pattern</em></span>.
1304       </p>
1305 <p>
1306         <code class="literal">(?(condition)yes-pattern)</code> attempts to match <span class="emphasis"><em>yes-pattern</em></span>
1307         if the <span class="emphasis"><em>condition</em></span> is true, otherwise matches the NULL
1308         string.
1309       </p>
1310 <p>
1311         <span class="emphasis"><em>condition</em></span> may be either: a forward lookahead assert,
1312         the index of a marked sub-expression (the condition becomes true if the sub-expression
1313         has been matched), or an index of a recursion (the condition become true
1314         if we are executing directly inside the specified recursion).
1315       </p>
1316 <p>
1317         Here is a summary of the possible predicates:
1318       </p>
1319 <div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; ">
1320 <li class="listitem">
1321             <code class="literal">(?(?=assert)yes-pattern|no-pattern)</code> Executes <span class="emphasis"><em>yes-pattern</em></span>
1322             if the forward look-ahead assert matches, otherwise executes <span class="emphasis"><em>no-pattern</em></span>.
1323           </li>
1324 <li class="listitem">
1325             <code class="literal">(?(?!assert)yes-pattern|no-pattern)</code> Executes <span class="emphasis"><em>yes-pattern</em></span>
1326             if the forward look-ahead assert does not match, otherwise executes
1327             <span class="emphasis"><em>no-pattern</em></span>.
1328           </li>
1329 <li class="listitem">
1330             <code class="literal">(?(<span class="emphasis"><em>N</em></span>)yes-pattern|no-pattern)</code>
1331             Executes <span class="emphasis"><em>yes-pattern</em></span> if subexpression <span class="emphasis"><em>N</em></span>
1332             has been matched, otherwise executes <span class="emphasis"><em>no-pattern</em></span>.
1333           </li>
1334 <li class="listitem">
1335             <code class="literal">(?(&lt;<span class="emphasis"><em>name</em></span>&gt;)yes-pattern|no-pattern)</code>
1336             Executes <span class="emphasis"><em>yes-pattern</em></span> if named subexpression <span class="emphasis"><em>name</em></span>
1337             has been matched, otherwise executes <span class="emphasis"><em>no-pattern</em></span>.
1338           </li>
1339 <li class="listitem">
1340             <code class="literal">(?('<span class="emphasis"><em>name</em></span>')yes-pattern|no-pattern)</code>
1341             Executes <span class="emphasis"><em>yes-pattern</em></span> if named subexpression <span class="emphasis"><em>name</em></span>
1342             has been matched, otherwise executes <span class="emphasis"><em>no-pattern</em></span>.
1343           </li>
1344 <li class="listitem">
1345             <code class="literal">(?(R)yes-pattern|no-pattern)</code> Executes <span class="emphasis"><em>yes-pattern</em></span>
1346             if we are executing inside a recursion, otherwise executes <span class="emphasis"><em>no-pattern</em></span>.
1347           </li>
1348 <li class="listitem">
1349             <code class="literal">(?(R<span class="emphasis"><em>N</em></span>)yes-pattern|no-pattern)</code>
1350             Executes <span class="emphasis"><em>yes-pattern</em></span> if we are executing inside
1351             a recursion to sub-expression <span class="emphasis"><em>N</em></span>, otherwise executes
1352             <span class="emphasis"><em>no-pattern</em></span>.
1353           </li>
1354 <li class="listitem">
1355             <code class="literal">(?(R&amp;<span class="emphasis"><em>name</em></span>)yes-pattern|no-pattern)</code>
1356             Executes <span class="emphasis"><em>yes-pattern</em></span> if we are executing inside
1357             a recursion to named sub-expression <span class="emphasis"><em>name</em></span>, otherwise
1358             executes <span class="emphasis"><em>no-pattern</em></span>.
1359           </li>
1360 <li class="listitem">
1361             <code class="literal">(?(DEFINE)never-exectuted-pattern)</code> Defines a block
1362             of code that is never executed and matches no characters: this is usually
1363             used to define one or more named sub-expressions which are referred to
1364             from elsewhere in the pattern.
1365           </li>
1366 </ul></div>
1367 <h6>
1368 <a name="boost_regex.syntax.perl_syntax.h42"></a>
1369         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.backtracking_control_verbs"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.backtracking_control_verbs">Backtracking
1370         Control Verbs</a>
1371       </h6>
1372 <p>
1373         This library has partial support for Perl's backtracking control verbs, in
1374         particular (*MARK) is not supported. There may also be detail differences
1375         in behaviour between this library and Perl, not least because Perl's behaviour
1376         is rather under-documented and often somewhat random in how it behaves in
1377         practice. The verbs supported are:
1378       </p>
1379 <div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; ">
1380 <li class="listitem">
1381             <code class="literal">(*PRUNE)</code> Has no effect unless backtracked onto, in
1382             which case all the backtracking information prior to this point is discarded.
1383           </li>
1384 <li class="listitem">
1385             <code class="literal">(*SKIP)</code> Behaves the same as <code class="literal">(*PRUNE)</code>
1386             except that it is assumed that no match can possibly occur prior to the
1387             current point in the string being searched. This can be used to optimize
1388             searches by skipping over chunks of text that have already been determined
1389             can not form a match.
1390           </li>
1391 <li class="listitem">
1392             <code class="literal">(*THEN)</code> Has no effect unless backtracked onto, in
1393             which case all subsequent alternatives in a group of alternations are
1394             discarded.
1395           </li>
1396 <li class="listitem">
1397             <code class="literal">(*COMMIT)</code> Has no effect unless backtracked onto, in
1398             which case all subsequent matching/searching attempts are abandoned.
1399           </li>
1400 <li class="listitem">
1401             <code class="literal">(*FAIL)</code> Causes the match to fail unconditionally at
1402             this point, can be used to force the engine to backtrack.
1403           </li>
1404 <li class="listitem">
1405             <code class="literal">(*ACCEPT)</code> Causes the pattern to be considered matched
1406             at the current point. Any half-open sub-expressions are closed at the
1407             current point.
1408           </li>
1409 </ul></div>
1410 <h5>
1411 <a name="boost_regex.syntax.perl_syntax.h43"></a>
1412         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.operator_precedence"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.operator_precedence">Operator
1413         precedence</a>
1414       </h5>
1415 <p>
1416         The order of precedence for of operators is as follows:
1417       </p>
1418 <div class="orderedlist"><ol class="orderedlist" type="1">
1419 <li class="listitem">
1420             Collation-related bracket symbols <code class="computeroutput"><span class="special">[==]</span>
1421             <span class="special">[::]</span> <span class="special">[..]</span></code>
1422           </li>
1423 <li class="listitem">
1424             Escaped characters <code class="literal">\</code>
1425           </li>
1426 <li class="listitem">
1427             Character set (bracket expression) <code class="computeroutput"><span class="special">[]</span></code>
1428           </li>
1429 <li class="listitem">
1430             Grouping <code class="literal">()</code>
1431           </li>
1432 <li class="listitem">
1433             Single-character-ERE duplication <code class="literal">* + ? {m,n}</code>
1434           </li>
1435 <li class="listitem">
1436             Concatenation
1437           </li>
1438 <li class="listitem">
1439             Anchoring ^$
1440           </li>
1441 <li class="listitem">
1442             Alternation |
1443           </li>
1444 </ol></div>
1445 <h4>
1446 <a name="boost_regex.syntax.perl_syntax.h44"></a>
1447         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.what_gets_matched"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.what_gets_matched">What
1448         gets matched</a>
1449       </h4>
1450 <p>
1451         If you view the regular expression as a directed (possibly cyclic) graph,
1452         then the best match found is the first match found by a depth-first-search
1453         performed on that graph, while matching the input text.
1454       </p>
1455 <p>
1456         Alternatively:
1457       </p>
1458 <p>
1459         The best match found is the <a class="link" href="leftmost_longest_rule.html" title="The Leftmost Longest Rule">leftmost
1460         match</a>, with individual elements matched as follows;
1461       </p>
1462 <div class="informaltable"><table class="table">
1463 <colgroup>
1464 <col>
1465 <col>
1466 </colgroup>
1467 <thead><tr>
1468 <th>
1469                 <p>
1470                   Construct
1471                 </p>
1472               </th>
1473 <th>
1474                 <p>
1475                   What gets matched
1476                 </p>
1477               </th>
1478 </tr></thead>
1479 <tbody>
1480 <tr>
1481 <td>
1482                 <p>
1483                   <code class="literal">AtomA AtomB</code>
1484                 </p>
1485               </td>
1486 <td>
1487                 <p>
1488                   Locates the best match for <span class="emphasis"><em>AtomA</em></span> that has
1489                   a following match for <span class="emphasis"><em>AtomB</em></span>.
1490                 </p>
1491               </td>
1492 </tr>
1493 <tr>
1494 <td>
1495                 <p>
1496                   <code class="literal">Expression1 | Expression2</code>
1497                 </p>
1498               </td>
1499 <td>
1500                 <p>
1501                   If <span class="emphasis"><em>Expresion1</em></span> can be matched then returns
1502                   that match, otherwise attempts to match <span class="emphasis"><em>Expression2</em></span>.
1503                 </p>
1504               </td>
1505 </tr>
1506 <tr>
1507 <td>
1508                 <p>
1509                   <code class="literal">S{N}</code>
1510                 </p>
1511               </td>
1512 <td>
1513                 <p>
1514                   Matches <span class="emphasis"><em>S</em></span> repeated exactly N times.
1515                 </p>
1516               </td>
1517 </tr>
1518 <tr>
1519 <td>
1520                 <p>
1521                   <code class="literal">S{N,M}</code>
1522                 </p>
1523               </td>
1524 <td>
1525                 <p>
1526                   Matches S repeated between N and M times, and as many times as
1527                   possible.
1528                 </p>
1529               </td>
1530 </tr>
1531 <tr>
1532 <td>
1533                 <p>
1534                   <code class="literal">S{N,M}?</code>
1535                 </p>
1536               </td>
1537 <td>
1538                 <p>
1539                   Matches S repeated between N and M times, and as few times as possible.
1540                 </p>
1541               </td>
1542 </tr>
1543 <tr>
1544 <td>
1545                 <p>
1546                   <code class="literal">S?, S*, S+</code>
1547                 </p>
1548               </td>
1549 <td>
1550                 <p>
1551                   The same as <code class="literal">S{0,1}</code>, <code class="literal">S{0,UINT_MAX}</code>,
1552                   <code class="literal">S{1,UINT_MAX}</code> respectively.
1553                 </p>
1554               </td>
1555 </tr>
1556 <tr>
1557 <td>
1558                 <p>
1559                   <code class="literal">S??, S*?, S+?</code>
1560                 </p>
1561               </td>
1562 <td>
1563                 <p>
1564                   The same as <code class="literal">S{0,1}?</code>, <code class="literal">S{0,UINT_MAX}?</code>,
1565                   <code class="literal">S{1,UINT_MAX}?</code> respectively.
1566                 </p>
1567               </td>
1568 </tr>
1569 <tr>
1570 <td>
1571                 <p>
1572                   <code class="literal">(?&gt;S)</code>
1573                 </p>
1574               </td>
1575 <td>
1576                 <p>
1577                   Matches the best match for <span class="emphasis"><em>S</em></span>, and only that.
1578                 </p>
1579               </td>
1580 </tr>
1581 <tr>
1582 <td>
1583                 <p>
1584                   <code class="literal">(?=S), (?&lt;=S)</code>
1585                 </p>
1586               </td>
1587 <td>
1588                 <p>
1589                   Matches only the best match for <span class="emphasis"><em>S</em></span> (this is
1590                   only visible if there are capturing parenthesis within <span class="emphasis"><em>S</em></span>).
1591                 </p>
1592               </td>
1593 </tr>
1594 <tr>
1595 <td>
1596                 <p>
1597                   <code class="literal">(?!S), (?&lt;!S)</code>
1598                 </p>
1599               </td>
1600 <td>
1601                 <p>
1602                   Considers only whether a match for S exists or not.
1603                 </p>
1604               </td>
1605 </tr>
1606 <tr>
1607 <td>
1608                 <p>
1609                   <code class="literal">(?(condition)yes-pattern | no-pattern)</code>
1610                 </p>
1611               </td>
1612 <td>
1613                 <p>
1614                   If condition is true, then only yes-pattern is considered, otherwise
1615                   only no-pattern is considered.
1616                 </p>
1617               </td>
1618 </tr>
1619 </tbody>
1620 </table></div>
1621 <h4>
1622 <a name="boost_regex.syntax.perl_syntax.h45"></a>
1623         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.variations"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.variations">Variations</a>
1624       </h4>
1625 <p>
1626         The <a class="link" href="../ref/syntax_option_type/syntax_option_type_perl.html" title="Options for Perl Regular Expressions">options
1627         <code class="literal">normal</code>, <code class="literal">ECMAScript</code>, <code class="literal">JavaScript</code>
1628         and <code class="literal">JScript</code></a> are all synonyms for <code class="literal">perl</code>.
1629       </p>
1630 <h4>
1631 <a name="boost_regex.syntax.perl_syntax.h46"></a>
1632         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.options"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.options">Options</a>
1633       </h4>
1634 <p>
1635         There are a <a class="link" href="../ref/syntax_option_type/syntax_option_type_perl.html" title="Options for Perl Regular Expressions">variety
1636         of flags</a> that may be combined with the <code class="literal">perl</code> option
1637         when constructing the regular expression, in particular note that the <code class="literal">newline_alt</code>
1638         option alters the syntax, while the <code class="literal">collate</code>, <code class="literal">nosubs</code>
1639         and <code class="literal">icase</code> options modify how the case and locale sensitivity
1640         are to be applied.
1641       </p>
1642 <h4>
1643 <a name="boost_regex.syntax.perl_syntax.h47"></a>
1644         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.pattern_modifiers"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.pattern_modifiers">Pattern
1645         Modifiers</a>
1646       </h4>
1647 <p>
1648         The perl <code class="literal">smix</code> modifiers can either be applied using a
1649         <code class="literal">(?smix-smix)</code> prefix to the regular expression, or with
1650         one of the <a class="link" href="../ref/syntax_option_type/syntax_option_type_perl.html" title="Options for Perl Regular Expressions">regex-compile
1651         time flags <code class="literal">no_mod_m</code>, <code class="literal">mod_x</code>, <code class="literal">mod_s</code>,
1652         and <code class="literal">no_mod_s</code></a>.
1653       </p>
1654 <h4>
1655 <a name="boost_regex.syntax.perl_syntax.h48"></a>
1656         <span class="phrase"><a name="boost_regex.syntax.perl_syntax.references"></a></span><a class="link" href="perl_syntax.html#boost_regex.syntax.perl_syntax.references">References</a>
1657       </h4>
1658 <p>
1659         <a href="http://perldoc.perl.org/perlre.html" target="_top">Perl 5.8</a>.
1660       </p>
1661 </div>
1662 <table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
1663 <td align="left"></td>
1664 <td align="right"><div class="copyright-footer">Copyright &#169; 1998-2013 John Maddock<p>
1665         Distributed under the Boost Software License, Version 1.0. (See accompanying
1666         file LICENSE_1_0.txt or copy at <a href="http://www.boost.org/LICENSE_1_0.txt" target="_top">http://www.boost.org/LICENSE_1_0.txt</a>)
1667       </p>
1668 </div></td>
1669 </tr></table>
1670 <hr>
1671 <div class="spirit-nav">
1672 <a accesskey="p" href="../syntax.html"><img src="../../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../syntax.html"><img src="../../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../../index.html"><img src="../../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="basic_extended.html"><img src="../../../../../../doc/src/images/next.png" alt="Next"></a>
1673 </div>
1674 </body>
1675 </html>