1PCRE2PERFORM(3) Library Functions Manual PCRE2PERFORM(3)
2
6 PCRE2 - Perl-compatible regular expressions (revised API)
7
9 Two aspects of performance are discussed below: memory usage and pro‐
11 cessing time. The way you express your pattern as a regular expression
12 can affect both of them.
13
15 Patterns are compiled by PCRE2 into a reasonably efficient interpretive
17 code, so that most simple patterns do not use much memory for storing
18 the compiled version. However, there is one case where the memory usage
19 of a compiled pattern can be unexpectedly large. If a parenthesized
20 subpattern has a quantifier with a minimum greater than 1 and/or a lim‐
21 ited maximum, the whole subpattern is repeated in the compiled code.
22 For example, the pattern
23 (abc|def){2,4}
25 is compiled as if it were
27 (abc|def)(abc|def)((abc|def)(abc|def)?)?
29 (Technical aside: It is done this way so that backtrack points within
31 each of the repetitions can be independently maintained.)
32 For regular expressions whose quantifiers use only small numbers, this
34 is not usually a problem. However, if the numbers are large, and par‐
35 ticularly if such repetitions are nested, the memory usage can become
36 an embarrassment. For example, the very simple pattern
37 ((ab){1,1000}c){1,3}
39 uses over 50KiB when compiled using the 8-bit library. When PCRE2 is
41 compiled with its default internal pointer size of two bytes, the size
42 limit on a compiled pattern is 65535 code units in the 8-bit and 16-bit
43 libraries, and this is reached with the above pattern if the outer rep‐
44 etition is increased from 3 to 4. PCRE2 can be compiled to use larger
45 internal pointers and thus handle larger compiled patterns, but it is
46 better to try to rewrite your pattern to use less memory if you can.
47 One way of reducing the memory usage for such patterns is to make use
49 of PCRE2's "subroutine" facility. Re-writing the above pattern as
50 ((ab)(?2){0,999}c)(?1){0,2}
52 reduces the memory requirements to around 16KiB, and indeed it remains
54 under 20KiB even with the outer repetition increased to 100. However,
55 this kind of pattern is not always exactly equivalent, because any cap‐
56 tures within subroutine calls are lost when the subroutine completes.
57 If this is not a problem, this kind of rewriting will allow you to
58 process patterns that PCRE2 cannot otherwise handle. The matching per‐
59 formance of the two different versions of the pattern are roughly the
60 same. (This applies from release 10.30 - things were different in ear‐
61 lier releases.)
62
64 From release 10.30, the interpretive (non-JIT) version of pcre2_match()
66 uses very little system stack at run time. In earlier releases recur‐
67 sive function calls could use a great deal of stack, and this could
68 cause problems, but this usage has been eliminated. Backtracking posi‐
69 tions are now explicitly remembered in memory frames controlled by the
70 code. An initial 20KiB vector of frames is allocated on the system
71 stack (enough for about 100 frames for small patterns), but if this is
72 insufficient, heap memory is used. The amount of heap memory can be
73 limited; if the limit is set to zero, only the initial stack vector is
74 used. Rewriting patterns to be time-efficient, as described below, may
75 also reduce the memory requirements.
76 In contrast to pcre2_match(), pcre2_dfa_match() does use recursive
78 function calls, but only for processing atomic groups, lookaround
79 assertions, and recursion within the pattern. The original version of
80 the code used to allocate quite large internal workspace vectors on the
81 stack, which caused some problems for some patterns in environments
82 with small stacks. From release 10.32 the code for pcre2_dfa_match()
83 has been re-factored to use heap memory when necessary for internal
84 workspace when recursing, though recursive function calls are still
85 used.
86 The "match depth" parameter can be used to limit the depth of function
88 recursion, and the "match heap" parameter to limit heap memory in
89 pcre2_dfa_match().
90
92 Certain items in regular expression patterns are processed more effi‐
94 ciently than others. It is more efficient to use a character class like
95 [aeiou] than a set of single-character alternatives such as
96 (a|e|i|o|u). In general, the simplest construction that provides the
97 required behaviour is usually the most efficient. Jeffrey Friedl's book
98 contains a lot of useful general discussion about optimizing regular
99 expressions for efficient performance. This document contains a few
100 observations about PCRE2.
101 Using Unicode character properties (the \p, \P, and \X escapes) is
103 slow, because PCRE2 has to use a multi-stage table lookup whenever it
104 needs a character's property. If you can find an alternative pattern
105 that does not use character properties, it will probably be faster.
106 By default, the escape sequences \b, \d, \s, and \w, and the POSIX
108 character classes such as [:alpha:] do not use Unicode properties,
109 partly for backwards compatibility, and partly for performance reasons.
110 However, you can set the PCRE2_UCP option or start the pattern with
111 (*UCP) if you want Unicode character properties to be used. This can
112 double the matching time for items such as \d, when matched with
113 pcre2_match(); the performance loss is less with a DFA matching func‐
114 tion, and in both cases there is not much difference for \b.
115 When a pattern begins with .* not in atomic parentheses, nor in paren‐
117 theses that are the subject of a backreference, and the PCRE2_DOTALL
118 option is set, the pattern is implicitly anchored by PCRE2, since it
119 can match only at the start of a subject string. If the pattern has
120 multiple top-level branches, they must all be anchorable. The optimiza‐
121 tion can be disabled by the PCRE2_NO_DOTSTAR_ANCHOR option, and is
122 automatically disabled if the pattern contains (*PRUNE) or (*SKIP).
123 If PCRE2_DOTALL is not set, PCRE2 cannot make this optimization,
125 because the dot metacharacter does not then match a newline, and if the
126 subject string contains newlines, the pattern may match from the char‐
127 acter immediately following one of them instead of from the very start.
128 For example, the pattern
129 .*second
131 matches the subject "first\nand second" (where \n stands for a newline
133 character), with the match starting at the seventh character. In order
134 to do this, PCRE2 has to retry the match starting after every newline
135 in the subject.
136 If you are using such a pattern with subject strings that do not con‐
138 tain newlines, the best performance is obtained by setting
139 PCRE2_DOTALL, or starting the pattern with ^.* or ^.*? to indicate
140 explicit anchoring. That saves PCRE2 from having to scan along the sub‐
141 ject looking for a newline to restart at.
142 Beware of patterns that contain nested indefinite repeats. These can
144 take a long time to run when applied to a string that does not match.
145 Consider the pattern fragment
146 ^(a+)*
148 This can match "aaaa" in 16 different ways, and this number increases
150 very rapidly as the string gets longer. (The * repeat can match 0, 1,
151 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
152 repeats can match different numbers of times.) When the remainder of
153 the pattern is such that the entire match is going to fail, PCRE2 has
154 in principle to try every possible variation, and this can take an
155 extremely long time, even for relatively short strings.
156 An optimization catches some of the more simple cases such as
158 (a+)*b
160 where a literal character follows. Before embarking on the standard
162 matching procedure, PCRE2 checks that there is a "b" later in the sub‐
163 ject string, and if there is not, it fails the match immediately. How‐
164 ever, when there is no following literal this optimization cannot be
165 used. You can see the difference by comparing the behaviour of
166 (a+)*\d
168 with the pattern above. The former gives a failure almost instantly
170 when applied to a whole line of "a" characters, whereas the latter
171 takes an appreciable time with strings longer than about 20 characters.
172 In many cases, the solution to this kind of performance issue is to use
174 an atomic group or a possessive quantifier. This can often reduce mem‐
175 ory requirements as well. As another example, consider this pattern:
176 ([^<]|<(?!inet))+
178 It matches from wherever it starts until it encounters "<inet" or the
180 end of the data, and is the kind of pattern that might be used when
181 processing an XML file. Each iteration of the outer parentheses matches
182 either one character that is not "<" or a "<" that is not followed by
183 "inet". However, each time a parenthesis is processed, a backtracking
184 position is passed, so this formulation uses a memory frame for each
185 matched character. For a long string, a lot of memory is required. Con‐
186 sider now this rewritten pattern, which matches exactly the same
187 strings:
188 ([^<]++|<(?!inet))+
190 This runs much faster, because sequences of characters that do not con‐
192 tain "<" are "swallowed" in one item inside the parentheses, and a pos‐
193 sessive quantifier is used to stop any backtracking into the runs of
194 non-"<" characters. This version also uses a lot less memory because
195 entry to a new set of parentheses happens only when a "<" character
196 that is not followed by "inet" is encountered (and we assume this is
197 relatively rare).
198 This example shows that one way of optimizing performance when matching
200 long subject strings is to write repeated parenthesized subpatterns to
201 match more than one character whenever possible.
202 SETTING RESOURCE LIMITS
204 You can set limits on the amount of processing that takes place when
206 matching, and on the amount of heap memory that is used. The default
207 values of the limits are very large, and unlikely ever to operate. They
208 can be changed when PCRE2 is built, and they can also be set when
209 pcre2_match() or pcre2_dfa_match() is called. For details of these
210 interfaces, see the pcre2build documentation and the section entitled
211 "The match context" in the pcre2api documentation.
212 The pcre2test test program has a modifier called "find_limits" which,
214 if applied to a subject line, causes it to find the smallest limits
215 that allow a pattern to match. This is done by repeatedly matching with
216 different limits.
217
219 Philip Hazel
221 University Computing Service
222 Cambridge, England.
223
225 Last updated: 25 April 2018
227 Copyright (c) 1997-2018 University of Cambridge.
228PCRE2 10.32 25 April 2018 PCRE2PERFORM(3)