/[pcre]/code/trunk/doc/pcreperform.3
ViewVC logotype

Diff of /code/trunk/doc/pcreperform.3

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 74 by nigel, Sat Feb 24 21:40:03 2007 UTC revision 75 by nigel, Sat Feb 24 21:40:37 2007 UTC
# Line 1  Line 1 
1  .TH PCRE 3  .TH PCRE 3
2  .SH NAME  .SH NAME
3  PCRE - Perl-compatible regular expressions  PCRE - Perl-compatible regular expressions
4  .SH PCRE PERFORMANCE  .SH "PCRE PERFORMANCE"
5  .rs  .rs
6  .sp  .sp
7  Certain items that may appear in regular expression patterns are more efficient  Certain items that may appear in regular expression patterns are more efficient
8  than others. It is more efficient to use a character class like [aeiou] than a  than others. It is more efficient to use a character class like [aeiou] than a
9  set of alternatives such as (a|e|i|o|u). In general, the simplest construction  set of alternatives such as (a|e|i|o|u). In general, the simplest construction
10  that provides the required behaviour is usually the most efficient. Jeffrey  that provides the required behaviour is usually the most efficient. Jeffrey
11  Friedl's book contains a lot of discussion about optimizing regular expressions  Friedl's book contains a lot of useful general discussion about optimizing
12  for efficient performance.  regular expressions for efficient performance. This document contains a few
13    observations about PCRE.
14    .P
15    Using Unicode character properties (the \ep, \eP, and \eX escapes) is slow,
16    because PCRE has to scan a structure that contains data for over fifteen
17    thousand characters whenever it needs a character's property. If you can find
18    an alternative pattern that does not use character properties, it will probably
19    be faster.
20    .P
21  When a pattern begins with .* not in parentheses, or in parentheses that are  When a pattern begins with .* not in parentheses, or in parentheses that are
22  not the subject of a backreference, and the PCRE_DOTALL option is set, the  not the subject of a backreference, and the PCRE_DOTALL option is set, the
23  pattern is implicitly anchored by PCRE, since it can match only at the start of  pattern is implicitly anchored by PCRE, since it can match only at the start of
# Line 19  optimization, because the . metacharacte Line 26  optimization, because the . metacharacte
26  the subject string contains newlines, the pattern may match from the character  the subject string contains newlines, the pattern may match from the character
27  immediately following one of them instead of from the very start. For example,  immediately following one of them instead of from the very start. For example,
28  the pattern  the pattern
29    .sp
30    .*second    .*second
31    .sp
32  matches the subject "first\\nand second" (where \\n stands for a newline  matches the subject "first\enand second" (where \en stands for a newline
33  character), with the match starting at the seventh character. In order to do  character), with the match starting at the seventh character. In order to do
34  this, PCRE has to retry the match starting after every newline in the subject.  this, PCRE has to retry the match starting after every newline in the subject.
35    .P
36  If you are using such a pattern with subject strings that do not contain  If you are using such a pattern with subject strings that do not contain
37  newlines, the best performance is obtained by setting PCRE_DOTALL, or starting  newlines, the best performance is obtained by setting PCRE_DOTALL, or starting
38  the pattern with ^.* to indicate explicit anchoring. That saves PCRE from  the pattern with ^.* to indicate explicit anchoring. That saves PCRE from
39  having to scan along the subject looking for a newline to restart at.  having to scan along the subject looking for a newline to restart at.
40    .P
41  Beware of patterns that contain nested indefinite repeats. These can take a  Beware of patterns that contain nested indefinite repeats. These can take a
42  long time to run when applied to a string that does not match. Consider the  long time to run when applied to a string that does not match. Consider the
43  pattern fragment  pattern fragment
44    .sp
45    (a+)*    (a+)*
46    .sp
47  This can match "aaaa" in 33 different ways, and this number increases very  This can match "aaaa" in 33 different ways, and this number increases very
48  rapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4  rapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4
49  times, and for each of those cases other than 0, the + repeats can match  times, and for each of those cases other than 0, the + repeats can match
50  different numbers of times.) When the remainder of the pattern is such that the  different numbers of times.) When the remainder of the pattern is such that the
51  entire match is going to fail, PCRE has in principle to try every possible  entire match is going to fail, PCRE has in principle to try every possible
52  variation, and this can take an extremely long time.  variation, and this can take an extremely long time.
53    .P
54  An optimization catches some of the more simple cases such as  An optimization catches some of the more simple cases such as
55    .sp
56    (a+)*b    (a+)*b
57    .sp
58  where a literal character follows. Before embarking on the standard matching  where a literal character follows. Before embarking on the standard matching
59  procedure, PCRE checks that there is a "b" later in the subject string, and if  procedure, PCRE checks that there is a "b" later in the subject string, and if
60  there is not, it fails the match immediately. However, when there is no  there is not, it fails the match immediately. However, when there is no
61  following literal this optimization cannot be used. You can see the difference  following literal this optimization cannot be used. You can see the difference
62  by comparing the behaviour of  by comparing the behaviour of
63    .sp
64    (a+)*\\d    (a+)*\ed
65    .sp
66  with the pattern above. The former gives a failure almost instantly when  with the pattern above. The former gives a failure almost instantly when
67  applied to a whole line of "a" characters, whereas the latter takes an  applied to a whole line of "a" characters, whereas the latter takes an
68  appreciable time with strings longer than about 20 characters.  appreciable time with strings longer than about 20 characters.
69    .P
70    In many cases, the solution to this kind of performance issue is to use an
71    atomic group or a possessive quantifier.
72    .P
73  .in 0  .in 0
74  Last updated: 03 February 2003  Last updated: 09 September 2004
75  .br  .br
76  Copyright (c) 1997-2003 University of Cambridge.  Copyright (c) 1997-2004 University of Cambridge.

Legend:
Removed from v.74  
changed lines
  Added in v.75

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12