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. However,
18 there is one case where the memory usage of a compiled pattern can be
19 unexpectedly large. If a parenthesized subpattern has a quantifier with
20 a minimum greater than 1 and/or a limited maximum, the whole subpattern
21 is repeated in the compiled code. For example, the pattern
22 (abc|def){2,4}
24 is compiled as if it were
26 (abc|def)(abc|def)((abc|def)(abc|def)?)?
28 (Technical aside: It is done this way so that backtrack points within
30 each of the repetitions can be independently maintained.)
31 For regular expressions whose quantifiers use only small numbers, this
33 is not usually a problem. However, if the numbers are large, and par‐
34 ticularly if such repetitions are nested, the memory usage can become
35 an embarrassment. For example, the very simple pattern
36 ((ab){1,1000}c){1,3}
38 uses 51K bytes when compiled using the 8-bit library. When PCRE2 is
40 compiled with its default internal pointer size of two bytes, the size
41 limit on a compiled pattern is 64K code units in the 8-bit and 16-bit
42 libraries, and this is reached with the above pattern if the outer rep‐
43 etition is increased from 3 to 4. PCRE2 can be compiled to use larger
44 internal pointers and thus handle larger compiled patterns, but it is
45 better to try to rewrite your pattern to use less memory if you can.
46 One way of reducing the memory usage for such patterns is to make use
48 of PCRE2's "subroutine" facility. Re-writing the above pattern as
49 ((ab)(?2){0,999}c)(?1){0,2}
51 reduces the memory requirements to 18K, and indeed it remains under 20K
53 even with the outer repetition increased to 100. However, this pattern
54 is not exactly equivalent, because the "subroutine" calls are treated
55 as atomic groups into which there can be no backtracking if there is a
56 subsequent matching failure. Therefore, PCRE2 cannot do this kind of
57 rewriting automatically. Furthermore, there is a noticeable loss of
58 speed when executing the modified pattern. Nevertheless, if the atomic
59 grouping is not a problem and the loss of speed is acceptable, this
60 kind of rewriting will allow you to process patterns that PCRE2 cannot
61 otherwise handle.
62
64 When pcre2_match() is used for matching, certain kinds of pattern can
66 cause it to use large amounts of the process stack. In some environ‐
67 ments the default process stack is quite small, and if it runs out the
68 result is often SIGSEGV. Rewriting your pattern can often help. The
69 pcre2stack documentation discusses this issue in detail.
70
72 Certain items in regular expression patterns are processed more effi‐
74 ciently than others. It is more efficient to use a character class like
75 [aeiou] than a set of single-character alternatives such as
76 (a|e|i|o|u). In general, the simplest construction that provides the
77 required behaviour is usually the most efficient. Jeffrey Friedl's book
78 contains a lot of useful general discussion about optimizing regular
79 expressions for efficient performance. This document contains a few
80 observations about PCRE2.
81 Using Unicode character properties (the \p, \P, and \X escapes) is
83 slow, because PCRE2 has to use a multi-stage table lookup whenever it
84 needs a character's property. If you can find an alternative pattern
85 that does not use character properties, it will probably be faster.
86 By default, the escape sequences \b, \d, \s, and \w, and the POSIX
88 character classes such as [:alpha:] do not use Unicode properties,
89 partly for backwards compatibility, and partly for performance reasons.
90 However, you can set the PCRE2_UCP option or start the pattern with
91 (*UCP) if you want Unicode character properties to be used. This can
92 double the matching time for items such as \d, when matched with
93 pcre2_match(); the performance loss is less with a DFA matching func‐
94 tion, and in both cases there is not much difference for \b.
95 When a pattern begins with .* not in atomic parentheses, nor in paren‐
97 theses that are the subject of a backreference, and the PCRE2_DOTALL
98 option is set, the pattern is implicitly anchored by PCRE2, since it
99 can match only at the start of a subject string. If the pattern has
100 multiple top-level branches, they must all be anchorable. The optimiza‐
101 tion can be disabled by the PCRE2_NO_DOTSTAR_ANCHOR option, and is
102 automatically disabled if the pattern contains (*PRUNE) or (*SKIP).
103 If PCRE2_DOTALL is not set, PCRE2 cannot make this optimization,
105 because the dot metacharacter does not then match a newline, and if the
106 subject string contains newlines, the pattern may match from the char‐
107 acter immediately following one of them instead of from the very start.
108 For example, the pattern
109 .*second
111 matches the subject "first\nand second" (where \n stands for a newline
113 character), with the match starting at the seventh character. In order
114 to do this, PCRE2 has to retry the match starting after every newline
115 in the subject.
116 If you are using such a pattern with subject strings that do not con‐
118 tain newlines, the best performance is obtained by setting
119 PCRE2_DOTALL, or starting the pattern with ^.* or ^.*? to indicate
120 explicit anchoring. That saves PCRE2 from having to scan along the sub‐
121 ject looking for a newline to restart at.
122 Beware of patterns that contain nested indefinite repeats. These can
124 take a long time to run when applied to a string that does not match.
125 Consider the pattern fragment
126 ^(a+)*
128 This can match "aaaa" in 16 different ways, and this number increases
130 very rapidly as the string gets longer. (The * repeat can match 0, 1,
131 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
132 repeats can match different numbers of times.) When the remainder of
133 the pattern is such that the entire match is going to fail, PCRE2 has
134 in principle to try every possible variation, and this can take an
135 extremely long time, even for relatively short strings.
136 An optimization catches some of the more simple cases such as
138 (a+)*b
140 where a literal character follows. Before embarking on the standard
142 matching procedure, PCRE2 checks that there is a "b" later in the sub‐
143 ject string, and if there is not, it fails the match immediately. How‐
144 ever, when there is no following literal this optimization cannot be
145 used. You can see the difference by comparing the behaviour of
146 (a+)*\d
148 with the pattern above. The former gives a failure almost instantly
150 when applied to a whole line of "a" characters, whereas the latter
151 takes an appreciable time with strings longer than about 20 characters.
152 In many cases, the solution to this kind of performance issue is to use
154 an atomic group or a possessive quantifier.
155
157 Philip Hazel
159 University Computing Service
160 Cambridge, England.
161
163 Last updated: 02 January 2015
165 Copyright (c) 1997-2015 University of Cambridge.
166PCRE2 10.00 02 January 2015 PCRE2PERFORM(3)