# Contents of /code/trunk/doc/html/pcrematching.html

Tue Sep 1 16:10:16 2009 UTC (3 years, 8 months ago) by ph10
File MIME type: text/html
File size: 9390 byte(s)
```Add pcredemo man page, containing a listing of pcredemo.c.

```
 1 nigel 77 2 3 pcrematching specification 4 5 6

pcrematching man page

7

10 ph10 111

11 nigel 77 This page is part of the PCRE HTML documentation. It was generated automatically 12 from the original man page. If there is any nonsense in it, please consult the 13 man page, in case the conversion went wrong. 14 ph10 111
15 nigel 77

25
PCRE MATCHING ALGORITHMS
26

27 This document describes the two different algorithms that are available in PCRE 28 for matching a compiled regular expression against a given subject string. The 29 "standard" algorithm is the one provided by the pcre_exec() function. 30 This works in the same was as Perl's matching function, and provides a 31 Perl-compatible matching operation. 32

33

34 An alternative algorithm is provided by the pcre_dfa_exec() function; 35 this operates in a different way, and is not Perl-compatible. It has advantages 36 and disadvantages compared with the standard algorithm, and these are described 37 below. 38

39

40 When there is only one possible way in which a given subject string can match a 41 pattern, the two algorithms give the same answer. A difference arises, however, 42 when there are multiple possibilities. For example, if the pattern 43

44              ^<.*>
45
46 is matched against the string 47
48              <something> <something else> <something further>
49
50 there are three possible answers. The standard algorithm finds only one of 51 nigel 93 them, whereas the alternative algorithm finds all three. 52 nigel 77

53
REGULAR EXPRESSIONS AS TREES
54

55 The set of strings that are matched by a regular expression can be represented 56 as a tree structure. An unlimited repetition in the pattern makes the tree of 57 infinite size, but it is still a tree. Matching the pattern to a given subject 58 string (from a given starting point) can be thought of as a search of the tree. 59 nigel 91 There are two ways to search a tree: depth-first and breadth-first, and these 60 correspond to the two matching algorithms provided by PCRE. 61 nigel 77

62
THE STANDARD MATCHING ALGORITHM
63

64 ph10 148 In the terminology of Jeffrey Friedl's book "Mastering Regular 65 Expressions", the standard algorithm is an "NFA algorithm". It conducts a 66 nigel 77 depth-first search of the pattern tree. That is, it proceeds along a single 67 path through the tree, checking that the subject matches what is required. When 68 there is a mismatch, the algorithm tries any alternatives at the current point, 69 and if they all fail, it backs up to the previous branch point in the tree, and 70 tries 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 72 repetition branches are tried is controlled by the greedy or ungreedy nature of 73 the quantifier. 74

75

76 If a leaf node is reached, a matching string has been found, and at that point 77 the algorithm stops. Thus, if there is more than one possible match, this 78 algorithm returns the first one that it finds. Whether this is the shortest, 79 the longest, or some intermediate length depends on the way the greedy and 80 ungreedy repetition quantifiers are specified in the pattern. 81

82

83 Because it ends up with a single path through the tree, it is relatively 84 straightforward for this algorithm to keep track of the substrings that are 85 matched by portions of the pattern in parentheses. This provides support for 86 capturing parentheses and back references. 87

88 nigel 93
THE ALTERNATIVE MATCHING ALGORITHM
89 nigel 77

90 nigel 93 This algorithm conducts a breadth-first search of the tree. Starting from the 91 first matching point in the subject, it scans the subject string from left to 92 right, once, character by character, and as it does this, it remembers all the 93 paths through the tree that represent valid matches. In Friedl's terminology, 94 this is a kind of "DFA algorithm", though it is not implemented as a 95 traditional finite state machine (it keeps multiple states active 96 simultaneously). 97 nigel 77

98

99 The scan continues until either the end of the subject is reached, or there are 100 no more unterminated paths. At this point, terminated paths represent the 101 different matching possibilities (if there are none, the match has failed). 102 Thus, if there is more than one possible match, this algorithm finds all of 103 them, and in particular, it finds the longest. In PCRE, there is an option to 104 stop the algorithm after the first match (which is necessarily the shortest) 105 has been found. 106

107

108 Note that all the matches that are found start at the same point in the 109 subject. If the pattern 110

111              cat(er(pillar)?)
112
113 is matched against the string "the caterpillar catchment", the result will be 114 the three strings "cat", "cater", and "caterpillar" that start at the fourth 115 character of the subject. The algorithm does not automatically move on to find 116 matches that start at later positions. 117

118

119 There are a number of features of PCRE regular expressions that are not 120 nigel 93 supported by the alternative matching algorithm. They are as follows: 121 nigel 77

122

123 1. Because the algorithm finds all possible matches, the greedy or ungreedy 124 nature of repetition quantifiers is not relevant. Greedy and ungreedy 125 nigel 93 quantifiers are treated in exactly the same way. However, possessive 126 quantifiers can make a difference when what follows could also match what is 127 quantified, for example in a pattern like this: 128

129              ^a++\w!
130
131 This pattern matches "aaab!" but not "aaa!", which would be matched by a 132 non-possessive quantifier. Similarly, if an atomic group is present, it is 133 matched as if it were a standalone pattern at the current point, and the 134 longest match is then "locked in" for the rest of the overall pattern. 135 nigel 77

136

137 2. When dealing with multiple paths through the tree simultaneously, it is not 138 straightforward to keep track of captured substrings for the different matching 139 possibilities, and PCRE's implementation of this algorithm does not attempt to 140 do this. This means that no captured substrings are available. 141

142

143 3. Because no substrings are captured, back references within the pattern are 144 not supported, and cause errors if encountered. 145

146

147 4. For the same reason, conditional expressions that use a backreference as the 148 nigel 93 condition or test for a specific group recursion are not supported. 149 nigel 77

150

151 ph10 172 5. Because many paths through the tree may be active, the \K escape sequence, 152 which resets the start of the match when encountered (but may be on some paths 153 and not on others), is not supported. It causes an error if encountered. 154

155

156 6. Callouts are supported, but the value of the capture_top field is 157 nigel 77 always 1, and the value of the capture_last field is always -1. 158

159

160 ph10 211 7. The \C escape sequence, which (in the standard algorithm) matches a single 161 nigel 93 byte, even in UTF-8 mode, is not supported because the alternative algorithm 162 moves through the subject string one character at a time, for all active paths 163 nigel 77 through the tree. 164

165 ph10 211

166 ph10 345 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE) are not 167 supported. (*FAIL) is supported, and behaves like a failing negative assertion. 168 ph10 211

169 nigel 93
170 nigel 77

171 nigel 93 Using the alternative matching algorithm provides the following advantages: 172 nigel 77

173

174 1. All possible matches (at a single point in the subject) are automatically 175 found, and in particular, the longest match is found. To find more than one 176 match using the standard algorithm, you have to do kludgy things with 177 callouts. 178

179

180 ph10 429 2. Because the alternative algorithm scans the subject string just once, and 181 nigel 93 never needs to backtrack, it is possible to pass very long subject strings to 182 the matching function in several pieces, checking for partial matching each 183 time. 184 nigel 77

185 nigel 93
186 nigel 77

187 nigel 93 The alternative algorithm suffers from a number of disadvantages: 188 nigel 77

189

190 1. It is substantially slower than the standard algorithm. This is partly 191 because it has to search for all possible matches, but is also because it is 192 less susceptible to optimization. 193

194

195 2. Capturing parentheses and back references are not supported. 196

197

198 nigel 93 3. Although atomic groups are supported, their use does not provide the 199 performance advantage that it does for the standard algorithm. 200 nigel 77

201 ph10 99
AUTHOR
202 nigel 77

203 ph10 99 Philip Hazel 204 nigel 77
205 ph10 99 University Computing Service 206
207 Cambridge CB2 3QH, England. 208
209

210
REVISION
211

212 ph10 429 Last updated: 25 August 2009 213 ph10 99
214 ph10 429 Copyright © 1997-2009 University of Cambridge. 215 ph10 99
216 nigel 77

## Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"