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

Contents of /code/trunk/doc/html/pcreperform.html

Parent Directory Parent Directory | Revision Log Revision Log


Revision 63 - (show annotations)
Sat Feb 24 21:40:03 2007 UTC (8 years, 6 months ago) by nigel
File MIME type: text/html
File size: 3459 byte(s)
Error occurred while calculating annotation data.
Load pcre-4.0 into code/trunk.
1 <html>
2 <head>
3 <title>pcreperform specification</title>
4 </head>
5 <body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6 This HTML document has been generated automatically from the original man page.
7 If there is any nonsense in it, please consult the man page, in case the
8 conversion went wrong.<br>
9 <ul>
10 <li><a name="TOC1" href="#SEC1">PCRE PERFORMANCE</a>
11 </ul>
12 <br><a name="SEC1" href="#TOC1">PCRE PERFORMANCE</a><br>
13 <P>
14 Certain items that may appear in regular expression patterns are more efficient
15 than others. It is more efficient to use a character class like [aeiou] than a
16 set of alternatives such as (a|e|i|o|u). In general, the simplest construction
17 that provides the required behaviour is usually the most efficient. Jeffrey
18 Friedl's book contains a lot of discussion about optimizing regular expressions
19 for efficient performance.
20 </P>
21 <P>
22 When a pattern begins with .* not in parentheses, or in parentheses that are
23 not the subject of a backreference, and the PCRE_DOTALL option is set, the
24 pattern is implicitly anchored by PCRE, since it can match only at the start of
25 a subject string. However, if PCRE_DOTALL is not set, PCRE cannot make this
26 optimization, because the . metacharacter does not then match a newline, and if
27 the subject string contains newlines, the pattern may match from the character
28 immediately following one of them instead of from the very start. For example,
29 the pattern
30 </P>
31 <P>
32 <pre>
33 .*second
34 </PRE>
35 </P>
36 <P>
37 matches the subject "first\nand second" (where \n stands for a newline
38 character), with the match starting at the seventh character. In order to do
39 this, PCRE has to retry the match starting after every newline in the subject.
40 </P>
41 <P>
42 If you are using such a pattern with subject strings that do not contain
43 newlines, the best performance is obtained by setting PCRE_DOTALL, or starting
44 the pattern with ^.* to indicate explicit anchoring. That saves PCRE from
45 having to scan along the subject looking for a newline to restart at.
46 </P>
47 <P>
48 Beware of patterns that contain nested indefinite repeats. These can take a
49 long time to run when applied to a string that does not match. Consider the
50 pattern fragment
51 </P>
52 <P>
53 <pre>
54 (a+)*
55 </PRE>
56 </P>
57 <P>
58 This can match "aaaa" in 33 different ways, and this number increases very
59 rapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4
60 times, and for each of those cases other than 0, the + repeats can match
61 different numbers of times.) When the remainder of the pattern is such that the
62 entire match is going to fail, PCRE has in principle to try every possible
63 variation, and this can take an extremely long time.
64 </P>
65 <P>
66 An optimization catches some of the more simple cases such as
67 </P>
68 <P>
69 <pre>
70 (a+)*b
71 </PRE>
72 </P>
73 <P>
74 where a literal character follows. Before embarking on the standard matching
75 procedure, PCRE checks that there is a "b" later in the subject string, and if
76 there is not, it fails the match immediately. However, when there is no
77 following literal this optimization cannot be used. You can see the difference
78 by comparing the behaviour of
79 </P>
80 <P>
81 <pre>
82 (a+)*\d
83 </PRE>
84 </P>
85 <P>
86 with the pattern above. The former gives a failure almost instantly when
87 applied to a whole line of "a" characters, whereas the latter takes an
88 appreciable time with strings longer than about 20 characters.
89 </P>
90 <P>
91 Last updated: 03 February 2003
92 <br>
93 Copyright &copy; 1997-2003 University of Cambridge.

  ViewVC Help
Powered by ViewVC 1.1.5