doc/html/pcreperform.html

<html>
<head>
<title>pcreperform specification</title>
</head>
<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
<h1>pcreperform man page</h1>
<p>
Return to the <a href="index.html">PCRE index page</a>.
</p>
<p>
This page is part of the PCRE HTML documentation. It was generated automatically
from the original man page. If there is any nonsense in it, please consult the
man page, in case the conversion went wrong.
<br>
<br><b>
PCRE PERFORMANCE
</b><br>
<P>
Two aspects of performance are discussed below: memory usage and processing
time. The way you express your pattern as a regular expression can affect both
of them.
</P>
<br><b>
MEMORY USAGE
</b><br>
<P>
Patterns are compiled by PCRE into a reasonably efficient byte code, so that
most simple patterns do not use much memory. However, there is one case where
memory usage can be unexpectedly large. When a parenthesized subpattern has a
quantifier with a minimum greater than 1 and/or a limited maximum, the whole
subpattern is repeated in the compiled code. For example, the pattern
<pre>
  (abc|def){2,4}
</pre>
is compiled as if it were
<pre>
  (abc|def)(abc|def)((abc|def)(abc|def)?)?
</pre>
(Technical aside: It is done this way so that backtrack points within each of
the repetitions can be independently maintained.)
</P>
<P>
For regular expressions whose quantifiers use only small numbers, this is not
usually a problem. However, if the numbers are large, and particularly if such
repetitions are nested, the memory usage can become an embarrassment. For
example, the very simple pattern
<pre>
  ((ab){1,1000}c){1,3}
</pre>
uses 51K bytes when compiled. When PCRE is compiled with its default internal
pointer size of two bytes, the size limit on a compiled pattern is 64K, and
this is reached with the above pattern if the outer repetition is increased
from 3 to 4. PCRE can be compiled to use larger internal pointers and thus
handle larger compiled patterns, but it is better to try to rewrite your
pattern to use less memory if you can.
</P>
<P>
One way of reducing the memory usage for such patterns is to make use of PCRE's
<a href="pcrepattern.html#subpatternsassubroutines">"subroutine"</a>
facility. Re-writing the above pattern as
<pre>
  ((ab)(?2){0,999}c)(?1){0,2}
</pre>
reduces the memory requirements to 18K, and indeed it remains under 20K even
with the outer repetition increased to 100. However, this pattern is not
exactly equivalent, because the "subroutine" calls are treated as
<a href="pcrepattern.html#atomicgroup">atomic groups</a>
into which there can be no backtracking if there is a subsequent matching
failure. Therefore, PCRE cannot do this kind of rewriting automatically.
Furthermore, there is a noticeable loss of speed when executing the modified
pattern. Nevertheless, if the atomic grouping is not a problem and the loss of
speed is acceptable, this kind of rewriting will allow you to process patterns
that PCRE cannot otherwise handle.
</P>
<br><b>
PROCESSING TIME
</b><br>
<P>
Certain items in regular expression patterns are processed more efficiently
than others. It is more efficient to use a character class like [aeiou] than a
set of single-character alternatives such as (a|e|i|o|u). In general, the
simplest construction that provides the required behaviour is usually the most
efficient. Jeffrey Friedl's book contains a lot of useful general discussion
about optimizing regular expressions for efficient performance. This document
contains a few observations about PCRE.
</P>
<P>
Using Unicode character properties (the \p, \P, and \X escapes) is slow,
because PCRE has to scan a structure that contains data for over fifteen
thousand characters whenever it needs a character's property. If you can find
an alternative pattern that does not use character properties, it will probably
be faster.
</P>
<P>
When a pattern begins with .* not in parentheses, or in parentheses that are
not the subject of a backreference, and the PCRE_DOTALL option is set, the
pattern is implicitly anchored by PCRE, since it can match only at the start of
a subject string. However, if PCRE_DOTALL is not set, PCRE cannot make this
optimization, because the . metacharacter does not then match a newline, and if
the subject string contains newlines, the pattern may match from the character
immediately following one of them instead of from the very start. For example,
the pattern
<pre>
  .*second
</pre>
matches the subject "first\nand second" (where \n stands for a newline
character), with the match starting at the seventh character. In order to do
this, PCRE has to retry the match starting after every newline in the subject.
</P>
<P>
If you are using such a pattern with subject strings that do not contain
newlines, the best performance is obtained by setting PCRE_DOTALL, or starting
the pattern with ^.* or ^.*? to indicate explicit anchoring. That saves PCRE
from having to scan along the subject looking for a newline to restart at.
</P>
<P>
Beware of patterns that contain nested indefinite repeats. These can take a
long time to run when applied to a string that does not match. Consider the
pattern fragment
<pre>
  ^(a+)*
</pre>
This can match "aaaa" in 16 different ways, and this number increases very
rapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4
times, and for each of those cases other than 0 or 4, the + repeats can match
different numbers of times.) When the remainder of the pattern is such that the
entire match is going to fail, PCRE has in principle to try every possible
variation, and this can take an extremely long time, even for relatively short
strings.
</P>
<P>
An optimization catches some of the more simple cases such as
<pre>
  (a+)*b
</pre>
where a literal character follows. Before embarking on the standard matching
procedure, PCRE checks that there is a "b" later in the subject string, and if
there is not, it fails the match immediately. However, when there is no
following literal this optimization cannot be used. You can see the difference
by comparing the behaviour of
<pre>
  (a+)*\d
</pre>
with the pattern above. The former gives a failure almost instantly when
applied to a whole line of "a" characters, whereas the latter takes an
appreciable time with strings longer than about 20 characters.
</P>
<P>
In many cases, the solution to this kind of performance issue is to use an
atomic group or a possessive quantifier.
</P>
<br><b>
AUTHOR
</b><br>
<P>
Philip Hazel
<br>
University Computing Service
<br>
Cambridge CB2 3QH, England.
<br>
</P>
<br><b>
REVISION
</b><br>
<P>
Last updated: 06 March 2007
<br>
Copyright &copy; 1997-2007 University of Cambridge.
<br>
<p>
Return to the <a href="index.html">PCRE index page</a>.
</p>
1	<html>
2	<head>
3	<title>pcreperform specification</title>
4	</head>
5	<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6	<h1>pcreperform man page</h1>
7	<p>
8	Return to the <a href="index.html">PCRE index page</a>.
9	</p>
10	<p>
11	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	<br>
15	<br><b>
16	PCRE PERFORMANCE
17	</b><br>
18	<P>
19	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	than others. It is more efficient to use a character class like [aeiou] than a
81	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	</P>
87	<P>
88	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	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	</pre>
106	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	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	</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	^(a+)*
122	</pre>
123	This can match "aaaa" in 16 different ways, and this number increases very
124	rapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4
125	times, and for each of those cases other than 0 or 4, the + repeats can match
126	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	variation, and this can take an extremely long time, even for relatively short
129	strings.
130	</P>
131	<P>
132	An optimization catches some of the more simple cases such as
133	<pre>
134	(a+)*b
135	</pre>
136	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	</pre>
144	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	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	<br><b>
153	AUTHOR
154	</b><br>
155	<P>
156	Philip Hazel
157	<br>
158	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 © 1997-2007 University of Cambridge.
170	<br>
171	<p>
172	Return to the <a href="index.html">PCRE index page</a>.
173	</p>
Name	Value
svn:eol-style	native
svn:keywords	"Author Date Id Revision Url"