/[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 111 - (hide annotations) (download) (as text)
Thu Mar 8 16:53:09 2007 UTC (7 years, 4 months ago) by ph10
File MIME type: text/html
File size: 6744 byte(s)
Create the PrepareRelease script to process the documentation and create the 
.generic files for distribution, also to remove trailing spaces. Update a lot 
more of the build-time documentation. Arrange for PrepareRelease and its 
sub-scripts to be distributed.

1 nigel 63 <html>
2     <head>
3     <title>pcreperform specification</title>
4     </head>
5     <body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6 nigel 75 <h1>pcreperform man page</h1>
7     <p>
8     Return to the <a href="index.html">PCRE index page</a>.
9     </p>
10 ph10 111 <p>
11 nigel 75 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 <br>
15 nigel 75 <br><b>
16     PCRE PERFORMANCE
17     </b><br>
18 nigel 63 <P>
19 nigel 93 Two aspects of performance are discussed below: memory usage and processing
20     time. The way you express your pattern as a regular expression can affect both
21     of them.
22     </P>
23     <br><b>
24     MEMORY USAGE
25     </b><br>
26     <P>
27     Patterns are compiled by PCRE into a reasonably efficient byte code, so that
28     most simple patterns do not use much memory. However, there is one case where
29     memory usage can be unexpectedly large. When a parenthesized subpattern has a
30     quantifier with a minimum greater than 1 and/or a limited maximum, the whole
31     subpattern is repeated in the compiled code. For example, the pattern
32     <pre>
33     (abc|def){2,4}
34     </pre>
35     is compiled as if it were
36     <pre>
37     (abc|def)(abc|def)((abc|def)(abc|def)?)?
38     </pre>
39     (Technical aside: It is done this way so that backtrack points within each of
40     the repetitions can be independently maintained.)
41     </P>
42     <P>
43     For regular expressions whose quantifiers use only small numbers, this is not
44     usually a problem. However, if the numbers are large, and particularly if such
45     repetitions are nested, the memory usage can become an embarrassment. For
46     example, the very simple pattern
47     <pre>
48     ((ab){1,1000}c){1,3}
49     </pre>
50     uses 51K bytes when compiled. When PCRE is compiled with its default internal
51     pointer size of two bytes, the size limit on a compiled pattern is 64K, and
52     this is reached with the above pattern if the outer repetition is increased
53     from 3 to 4. PCRE can be compiled to use larger internal pointers and thus
54     handle larger compiled patterns, but it is better to try to rewrite your
55     pattern to use less memory if you can.
56     </P>
57     <P>
58     One way of reducing the memory usage for such patterns is to make use of PCRE's
59     <a href="pcrepattern.html#subpatternsassubroutines">"subroutine"</a>
60     facility. Re-writing the above pattern as
61     <pre>
62     ((ab)(?2){0,999}c)(?1){0,2}
63     </pre>
64     reduces the memory requirements to 18K, and indeed it remains under 20K even
65     with the outer repetition increased to 100. However, this pattern is not
66     exactly equivalent, because the "subroutine" calls are treated as
67     <a href="pcrepattern.html#atomicgroup">atomic groups</a>
68     into which there can be no backtracking if there is a subsequent matching
69     failure. Therefore, PCRE cannot do this kind of rewriting automatically.
70     Furthermore, there is a noticeable loss of speed when executing the modified
71     pattern. Nevertheless, if the atomic grouping is not a problem and the loss of
72     speed is acceptable, this kind of rewriting will allow you to process patterns
73     that PCRE cannot otherwise handle.
74     </P>
75     <br><b>
76     PROCESSING TIME
77     </b><br>
78     <P>
79     Certain items in regular expression patterns are processed more efficiently
80 nigel 63 than others. It is more efficient to use a character class like [aeiou] than a
81 nigel 93 set of single-character alternatives such as (a|e|i|o|u). In general, the
82     simplest construction that provides the required behaviour is usually the most
83     efficient. Jeffrey Friedl's book contains a lot of useful general discussion
84     about optimizing regular expressions for efficient performance. This document
85     contains a few observations about PCRE.
86 nigel 63 </P>
87     <P>
88 nigel 75 Using Unicode character properties (the \p, \P, and \X escapes) is slow,
89     because PCRE has to scan a structure that contains data for over fifteen
90     thousand characters whenever it needs a character's property. If you can find
91     an alternative pattern that does not use character properties, it will probably
92     be faster.
93     </P>
94     <P>
95 nigel 63 When a pattern begins with .* not in parentheses, or in parentheses that are
96     not the subject of a backreference, and the PCRE_DOTALL option is set, the
97     pattern is implicitly anchored by PCRE, since it can match only at the start of
98     a subject string. However, if PCRE_DOTALL is not set, PCRE cannot make this
99     optimization, because the . metacharacter does not then match a newline, and if
100     the subject string contains newlines, the pattern may match from the character
101     immediately following one of them instead of from the very start. For example,
102     the pattern
103     <pre>
104     .*second
105 nigel 75 </pre>
106 nigel 63 matches the subject "first\nand second" (where \n stands for a newline
107     character), with the match starting at the seventh character. In order to do
108     this, PCRE has to retry the match starting after every newline in the subject.
109     </P>
110     <P>
111     If you are using such a pattern with subject strings that do not contain
112     newlines, the best performance is obtained by setting PCRE_DOTALL, or starting
113 nigel 77 the pattern with ^.* or ^.*? to indicate explicit anchoring. That saves PCRE
114     from having to scan along the subject looking for a newline to restart at.
115 nigel 63 </P>
116     <P>
117     Beware of patterns that contain nested indefinite repeats. These can take a
118     long time to run when applied to a string that does not match. Consider the
119     pattern fragment
120     <pre>
121 nigel 93 ^(a+)*
122 nigel 75 </pre>
123 nigel 93 This can match "aaaa" in 16 different ways, and this number increases very
124 nigel 63 rapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4
125 nigel 93 times, and for each of those cases other than 0 or 4, the + repeats can match
126 nigel 63 different numbers of times.) When the remainder of the pattern is such that the
127     entire match is going to fail, PCRE has in principle to try every possible
128 nigel 93 variation, and this can take an extremely long time, even for relatively short
129     strings.
130 nigel 63 </P>
131     <P>
132     An optimization catches some of the more simple cases such as
133     <pre>
134     (a+)*b
135 nigel 75 </pre>
136 nigel 63 where a literal character follows. Before embarking on the standard matching
137     procedure, PCRE checks that there is a "b" later in the subject string, and if
138     there is not, it fails the match immediately. However, when there is no
139     following literal this optimization cannot be used. You can see the difference
140     by comparing the behaviour of
141     <pre>
142     (a+)*\d
143 nigel 75 </pre>
144 nigel 63 with the pattern above. The former gives a failure almost instantly when
145     applied to a whole line of "a" characters, whereas the latter takes an
146     appreciable time with strings longer than about 20 characters.
147     </P>
148     <P>
149 nigel 75 In many cases, the solution to this kind of performance issue is to use an
150     atomic group or a possessive quantifier.
151     </P>
152 ph10 99 <br><b>
153     AUTHOR
154     </b><br>
155 nigel 75 <P>
156 ph10 99 Philip Hazel
157 nigel 63 <br>
158 ph10 99 University Computing Service
159     <br>
160     Cambridge CB2 3QH, England.
161     <br>
162     </P>
163     <br><b>
164     REVISION
165     </b><br>
166     <P>
167     Last updated: 06 March 2007
168     <br>
169     Copyright &copy; 1997-2007 University of Cambridge.
170     <br>
171 nigel 75 <p>
172     Return to the <a href="index.html">PCRE index page</a>.
173     </p>

Properties

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

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12