1<html> 2<head> 3<title>pcrematching specification</title> 4</head> 5<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB"> 6<h1>pcrematching man page</h1> 7<p> 8Return to the <a href="index.html">PCRE index page</a>. 9</p> 10<p> 11This page is part of the PCRE HTML documentation. It was generated automatically 12from the original man page. If there is any nonsense in it, please consult the 13man page, in case the conversion went wrong. 14<br> 15<ul> 16<li><a name="TOC1" href="#SEC1">PCRE MATCHING ALGORITHMS</a> 17<li><a name="TOC2" href="#SEC2">REGULAR EXPRESSIONS AS TREES</a> 18<li><a name="TOC3" href="#SEC3">THE STANDARD MATCHING ALGORITHM</a> 19<li><a name="TOC4" href="#SEC4">THE ALTERNATIVE MATCHING ALGORITHM</a> 20<li><a name="TOC5" href="#SEC5">ADVANTAGES OF THE ALTERNATIVE ALGORITHM</a> 21<li><a name="TOC6" href="#SEC6">DISADVANTAGES OF THE ALTERNATIVE ALGORITHM</a> 22<li><a name="TOC7" href="#SEC7">AUTHOR</a> 23<li><a name="TOC8" href="#SEC8">REVISION</a> 24</ul> 25<br><a name="SEC1" href="#TOC1">PCRE MATCHING ALGORITHMS</a><br> 26<P> 27This document describes the two different algorithms that are available in PCRE 28for matching a compiled regular expression against a given subject string. The 29"standard" algorithm is the one provided by the <b>pcre_exec()</b> function. 30This works in the same was as Perl's matching function, and provides a 31Perl-compatible matching operation. 32</P> 33<P> 34An alternative algorithm is provided by the <b>pcre_dfa_exec()</b> function; 35this operates in a different way, and is not Perl-compatible. It has advantages 36and disadvantages compared with the standard algorithm, and these are described 37below. 38</P> 39<P> 40When there is only one possible way in which a given subject string can match a 41pattern, the two algorithms give the same answer. A difference arises, however, 42when there are multiple possibilities. For example, if the pattern 43<pre> 44 ^<.*> 45</pre> 46is matched against the string 47<pre> 48 <something> <something else> <something further> 49</pre> 50there are three possible answers. The standard algorithm finds only one of 51them, whereas the alternative algorithm finds all three. 52</P> 53<br><a name="SEC2" href="#TOC1">REGULAR EXPRESSIONS AS TREES</a><br> 54<P> 55The set of strings that are matched by a regular expression can be represented 56as a tree structure. An unlimited repetition in the pattern makes the tree of 57infinite size, but it is still a tree. Matching the pattern to a given subject 58string (from a given starting point) can be thought of as a search of the tree. 59There are two ways to search a tree: depth-first and breadth-first, and these 60correspond to the two matching algorithms provided by PCRE. 61</P> 62<br><a name="SEC3" href="#TOC1">THE STANDARD MATCHING ALGORITHM</a><br> 63<P> 64In the terminology of Jeffrey Friedl's book "Mastering Regular 65Expressions", the standard algorithm is an "NFA algorithm". It conducts a 66depth-first search of the pattern tree. That is, it proceeds along a single 67path through the tree, checking that the subject matches what is required. When 68there is a mismatch, the algorithm tries any alternatives at the current point, 69and if they all fail, it backs up to the previous branch point in the tree, and 70tries the next alternative branch at that level. This often involves backing up 71(moving to the left) in the subject string as well. The order in which 72repetition branches are tried is controlled by the greedy or ungreedy nature of 73the quantifier. 74</P> 75<P> 76If a leaf node is reached, a matching string has been found, and at that point 77the algorithm stops. Thus, if there is more than one possible match, this 78algorithm returns the first one that it finds. Whether this is the shortest, 79the longest, or some intermediate length depends on the way the greedy and 80ungreedy repetition quantifiers are specified in the pattern. 81</P> 82<P> 83Because it ends up with a single path through the tree, it is relatively 84straightforward for this algorithm to keep track of the substrings that are 85matched by portions of the pattern in parentheses. This provides support for 86capturing parentheses and back references. 87</P> 88<br><a name="SEC4" href="#TOC1">THE ALTERNATIVE MATCHING ALGORITHM</a><br> 89<P> 90This algorithm conducts a breadth-first search of the tree. Starting from the 91first matching point in the subject, it scans the subject string from left to 92right, once, character by character, and as it does this, it remembers all the 93paths through the tree that represent valid matches. In Friedl's terminology, 94this is a kind of "DFA algorithm", though it is not implemented as a 95traditional finite state machine (it keeps multiple states active 96simultaneously). 97</P> 98<P> 99Although the general principle of this matching algorithm is that it scans the 100subject string only once, without backtracking, there is one exception: when a 101lookaround assertion is encountered, the characters following or preceding the 102current point have to be independently inspected. 103</P> 104<P> 105The scan continues until either the end of the subject is reached, or there are 106no more unterminated paths. At this point, terminated paths represent the 107different matching possibilities (if there are none, the match has failed). 108Thus, if there is more than one possible match, this algorithm finds all of 109them, and in particular, it finds the longest. There is an option to stop the 110algorithm after the first match (which is necessarily the shortest) is found. 111</P> 112<P> 113Note that all the matches that are found start at the same point in the 114subject. If the pattern 115<pre> 116 cat(er(pillar)?) 117</pre> 118is matched against the string "the caterpillar catchment", the result will be 119the three strings "cat", "cater", and "caterpillar" that start at the fourth 120character of the subject. The algorithm does not automatically move on to find 121matches that start at later positions. 122</P> 123<P> 124There are a number of features of PCRE regular expressions that are not 125supported by the alternative matching algorithm. They are as follows: 126</P> 127<P> 1281. Because the algorithm finds all possible matches, the greedy or ungreedy 129nature of repetition quantifiers is not relevant. Greedy and ungreedy 130quantifiers are treated in exactly the same way. However, possessive 131quantifiers can make a difference when what follows could also match what is 132quantified, for example in a pattern like this: 133<pre> 134 ^a++\w! 135</pre> 136This pattern matches "aaab!" but not "aaa!", which would be matched by a 137non-possessive quantifier. Similarly, if an atomic group is present, it is 138matched as if it were a standalone pattern at the current point, and the 139longest match is then "locked in" for the rest of the overall pattern. 140</P> 141<P> 1422. When dealing with multiple paths through the tree simultaneously, it is not 143straightforward to keep track of captured substrings for the different matching 144possibilities, and PCRE's implementation of this algorithm does not attempt to 145do this. This means that no captured substrings are available. 146</P> 147<P> 1483. Because no substrings are captured, back references within the pattern are 149not supported, and cause errors if encountered. 150</P> 151<P> 1524. For the same reason, conditional expressions that use a backreference as the 153condition or test for a specific group recursion are not supported. 154</P> 155<P> 1565. Because many paths through the tree may be active, the \K escape sequence, 157which resets the start of the match when encountered (but may be on some paths 158and not on others), is not supported. It causes an error if encountered. 159</P> 160<P> 1616. Callouts are supported, but the value of the <i>capture_top</i> field is 162always 1, and the value of the <i>capture_last</i> field is always -1. 163</P> 164<P> 1657. The \C escape sequence, which (in the standard algorithm) matches a single 166byte, even in UTF-8 mode, is not supported because the alternative algorithm 167moves through the subject string one character at a time, for all active paths 168through the tree. 169</P> 170<P> 1718. Except for (*FAIL), the backtracking control verbs such as (*PRUNE) are not 172supported. (*FAIL) is supported, and behaves like a failing negative assertion. 173</P> 174<br><a name="SEC5" href="#TOC1">ADVANTAGES OF THE ALTERNATIVE ALGORITHM</a><br> 175<P> 176Using the alternative matching algorithm provides the following advantages: 177</P> 178<P> 1791. All possible matches (at a single point in the subject) are automatically 180found, and in particular, the longest match is found. To find more than one 181match using the standard algorithm, you have to do kludgy things with 182callouts. 183</P> 184<P> 1852. Because the alternative algorithm scans the subject string just once, and 186never needs to backtrack, it is possible to pass very long subject strings to 187the matching function in several pieces, checking for partial matching each 188time. The 189<a href="pcrepartial.html"><b>pcrepartial</b></a> 190documentation gives details of partial matching. 191</P> 192<br><a name="SEC6" href="#TOC1">DISADVANTAGES OF THE ALTERNATIVE ALGORITHM</a><br> 193<P> 194The alternative algorithm suffers from a number of disadvantages: 195</P> 196<P> 1971. It is substantially slower than the standard algorithm. This is partly 198because it has to search for all possible matches, but is also because it is 199less susceptible to optimization. 200</P> 201<P> 2022. Capturing parentheses and back references are not supported. 203</P> 204<P> 2053. Although atomic groups are supported, their use does not provide the 206performance advantage that it does for the standard algorithm. 207</P> 208<br><a name="SEC7" href="#TOC1">AUTHOR</a><br> 209<P> 210Philip Hazel 211<br> 212University Computing Service 213<br> 214Cambridge CB2 3QH, England. 215<br> 216</P> 217<br><a name="SEC8" href="#TOC1">REVISION</a><br> 218<P> 219Last updated: 29 September 2009 220<br> 221Copyright © 1997-2009 University of Cambridge. 222<br> 223<p> 224Return to the <a href="index.html">PCRE index page</a>. 225</p> 226