doc/html/pcrematching.html

<html>
<head>
<title>pcrematching specification</title>
</head>
<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
<h1>pcrematching 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>
<ul>
<li><a name="TOC1" href="#SEC1">PCRE MATCHING ALGORITHMS</a>
<li><a name="TOC2" href="#SEC2">REGULAR EXPRESSIONS AS TREES</a>
<li><a name="TOC3" href="#SEC3">THE STANDARD MATCHING ALGORITHM</a>
<li><a name="TOC4" href="#SEC4">THE ALTERNATIVE MATCHING ALGORITHM</a>
<li><a name="TOC5" href="#SEC5">ADVANTAGES OF THE ALTERNATIVE ALGORITHM</a>
<li><a name="TOC6" href="#SEC6">DISADVANTAGES OF THE ALTERNATIVE ALGORITHM</a>
<li><a name="TOC7" href="#SEC7">AUTHOR</a>
<li><a name="TOC8" href="#SEC8">REVISION</a>
</ul>
<br><a name="SEC1" href="#TOC1">PCRE MATCHING ALGORITHMS</a><br>
<P>
This document describes the two different algorithms that are available in PCRE
for matching a compiled regular expression against a given subject string. The
"standard" algorithm is the one provided by the <b>pcre_exec()</b> and
<b>pcre16_exec()</b> functions. These work in the same was as Perl's matching
function, and provide a Perl-compatible matching operation. The just-in-time 
(JIT) optimization that is described in the
<a href="pcrejit.html"><b>pcrejit</b></a>
documentation is compatible with these functions.
</P>
<P>
An alternative algorithm is provided by the <b>pcre_dfa_exec()</b> and 
<b>pcre16_dfa_exec()</b> functions; they operate in a different way, and are not
Perl-compatible. This alternative has advantages and disadvantages compared
with the standard algorithm, and these are described below.
</P>
<P>
When there is only one possible way in which a given subject string can match a
pattern, the two algorithms give the same answer. A difference arises, however,
when there are multiple possibilities. For example, if the pattern
<pre>
  ^&#60;.*&#62;
</pre>
is matched against the string
<pre>
  &#60;something&#62; &#60;something else&#62; &#60;something further&#62;
</pre>
there are three possible answers. The standard algorithm finds only one of
them, whereas the alternative algorithm finds all three.
</P>
<br><a name="SEC2" href="#TOC1">REGULAR EXPRESSIONS AS TREES</a><br>
<P>
The set of strings that are matched by a regular expression can be represented
as a tree structure. An unlimited repetition in the pattern makes the tree of
infinite size, but it is still a tree. Matching the pattern to a given subject
string (from a given starting point) can be thought of as a search of the tree.
There are two ways to search a tree: depth-first and breadth-first, and these
correspond to the two matching algorithms provided by PCRE.
</P>
<br><a name="SEC3" href="#TOC1">THE STANDARD MATCHING ALGORITHM</a><br>
<P>
In the terminology of Jeffrey Friedl's book "Mastering Regular
Expressions", the standard algorithm is an "NFA algorithm". It conducts a
depth-first search of the pattern tree. That is, it proceeds along a single
path through the tree, checking that the subject matches what is required. When
there is a mismatch, the algorithm tries any alternatives at the current point,
and if they all fail, it backs up to the previous branch point in the tree, and
tries the next alternative branch at that level. This often involves backing up
(moving to the left) in the subject string as well. The order in which
repetition branches are tried is controlled by the greedy or ungreedy nature of
the quantifier.
</P>
<P>
If a leaf node is reached, a matching string has been found, and at that point
the algorithm stops. Thus, if there is more than one possible match, this
algorithm returns the first one that it finds. Whether this is the shortest,
the longest, or some intermediate length depends on the way the greedy and
ungreedy repetition quantifiers are specified in the pattern.
</P>
<P>
Because it ends up with a single path through the tree, it is relatively
straightforward for this algorithm to keep track of the substrings that are
matched by portions of the pattern in parentheses. This provides support for
capturing parentheses and back references.
</P>
<br><a name="SEC4" href="#TOC1">THE ALTERNATIVE MATCHING ALGORITHM</a><br>
<P>
This algorithm conducts a breadth-first search of the tree. Starting from the
first matching point in the subject, it scans the subject string from left to
right, once, character by character, and as it does this, it remembers all the
paths through the tree that represent valid matches. In Friedl's terminology,
this is a kind of "DFA algorithm", though it is not implemented as a
traditional finite state machine (it keeps multiple states active
simultaneously).
</P>
<P>
Although the general principle of this matching algorithm is that it scans the
subject string only once, without backtracking, there is one exception: when a
lookaround assertion is encountered, the characters following or preceding the
current point have to be independently inspected.
</P>
<P>
The scan continues until either the end of the subject is reached, or there are
no more unterminated paths. At this point, terminated paths represent the
different matching possibilities (if there are none, the match has failed).
Thus, if there is more than one possible match, this algorithm finds all of
them, and in particular, it finds the longest. The matches are returned in
decreasing order of length. There is an option to stop the algorithm after the
first match (which is necessarily the shortest) is found.
</P>
<P>
Note that all the matches that are found start at the same point in the
subject. If the pattern
<pre>
  cat(er(pillar)?)?
</pre>
is matched against the string "the caterpillar catchment", the result will be
the three strings "caterpillar", "cater", and "cat" that start at the fifth
character of the subject. The algorithm does not automatically move on to find
matches that start at later positions.
</P>
<P>
There are a number of features of PCRE regular expressions that are not
supported by the alternative matching algorithm. They are as follows:
</P>
<P>
1. Because the algorithm finds all possible matches, the greedy or ungreedy
nature of repetition quantifiers is not relevant. Greedy and ungreedy
quantifiers are treated in exactly the same way. However, possessive
quantifiers can make a difference when what follows could also match what is
quantified, for example in a pattern like this:
<pre>
  ^a++\w!
</pre>
This pattern matches "aaab!" but not "aaa!", which would be matched by a
non-possessive quantifier. Similarly, if an atomic group is present, it is
matched as if it were a standalone pattern at the current point, and the
longest match is then "locked in" for the rest of the overall pattern.
</P>
<P>
2. When dealing with multiple paths through the tree simultaneously, it is not
straightforward to keep track of captured substrings for the different matching
possibilities, and PCRE's implementation of this algorithm does not attempt to
do this. This means that no captured substrings are available.
</P>
<P>
3. Because no substrings are captured, back references within the pattern are
not supported, and cause errors if encountered.
</P>
<P>
4. For the same reason, conditional expressions that use a backreference as the
condition or test for a specific group recursion are not supported.
</P>
<P>
5. Because many paths through the tree may be active, the \K escape sequence,
which resets the start of the match when encountered (but may be on some paths
and not on others), is not supported. It causes an error if encountered.
</P>
<P>
6. Callouts are supported, but the value of the <i>capture_top</i> field is
always 1, and the value of the <i>capture_last</i> field is always -1.
</P>
<P>
7. The \C escape sequence, which (in the standard algorithm) always matches a
single data unit, even in UTF-8 or UTF-16 modes, is not supported in these
modes, because the alternative algorithm moves through the subject string one
character (not data unit) at a time, for all active paths through the tree.
</P>
<P>
8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE) are not
supported. (*FAIL) is supported, and behaves like a failing negative assertion.
</P>
<br><a name="SEC5" href="#TOC1">ADVANTAGES OF THE ALTERNATIVE ALGORITHM</a><br>
<P>
Using the alternative matching algorithm provides the following advantages:
</P>
<P>
1. All possible matches (at a single point in the subject) are automatically
found, and in particular, the longest match is found. To find more than one
match using the standard algorithm, you have to do kludgy things with
callouts.
</P>
<P>
2. Because the alternative algorithm scans the subject string just once, and
never needs to backtrack (except for lookbehinds), it is possible to pass very
long subject strings to the matching function in several pieces, checking for
partial matching each time. Although it is possible to do multi-segment
matching using the standard algorithm by retaining partially matched
substrings, it is more complicated. The
<a href="pcrepartial.html"><b>pcrepartial</b></a>
documentation gives details of partial matching and discusses multi-segment
matching.
</P>
<br><a name="SEC6" href="#TOC1">DISADVANTAGES OF THE ALTERNATIVE ALGORITHM</a><br>
<P>
The alternative algorithm suffers from a number of disadvantages:
</P>
<P>
1. It is substantially slower than the standard algorithm. This is partly
because it has to search for all possible matches, but is also because it is
less susceptible to optimization.
</P>
<P>
2. Capturing parentheses and back references are not supported.
</P>
<P>
3. Although atomic groups are supported, their use does not provide the
performance advantage that it does for the standard algorithm.
</P>
<br><a name="SEC7" href="#TOC1">AUTHOR</a><br>
<P>
Philip Hazel
<br>
University Computing Service
<br>
Cambridge CB2 3QH, England.
<br>
</P>
<br><a name="SEC8" href="#TOC1">REVISION</a><br>
<P>
Last updated: 08 January 2012
<br>
Copyright &copy; 1997-2012 University of Cambridge.
<br>
<p>
Return to the <a href="index.html">PCRE index page</a>.
</p>
1	<html>
2	<head>
3	<title>pcrematching specification</title>
4	</head>
5	<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6	<h1>pcrematching 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	<ul>
16	<li><a name="TOC1" href="#SEC1">PCRE MATCHING ALGORITHMS</a>
17	<li><a name="TOC2" href="#SEC2">REGULAR EXPRESSIONS AS TREES</a>
18	<li><a name="TOC3" href="#SEC3">THE STANDARD MATCHING ALGORITHM</a>
19	<li><a name="TOC4" href="#SEC4">THE ALTERNATIVE MATCHING ALGORITHM</a>
20	<li><a name="TOC5" href="#SEC5">ADVANTAGES OF THE ALTERNATIVE ALGORITHM</a>
21	<li><a name="TOC6" href="#SEC6">DISADVANTAGES OF THE ALTERNATIVE ALGORITHM</a>
22	<li><a name="TOC7" href="#SEC7">AUTHOR</a>
23	<li><a name="TOC8" href="#SEC8">REVISION</a>
24	</ul>
25	<br><a name="SEC1" href="#TOC1">PCRE MATCHING ALGORITHMS</a><br>
26	<P>
27	This document describes the two different algorithms that are available in PCRE
28	for matching a compiled regular expression against a given subject string. The
29	"standard" algorithm is the one provided by the <b>pcre_exec()</b> and
30	<b>pcre16_exec()</b> functions. These work in the same was as Perl's matching
31	function, and provide a Perl-compatible matching operation. The just-in-time
32	(JIT) optimization that is described in the
33	<a href="pcrejit.html"><b>pcrejit</b></a>
34	documentation is compatible with these functions.
35	</P>
36	<P>
37	An alternative algorithm is provided by the <b>pcre_dfa_exec()</b> and
38	<b>pcre16_dfa_exec()</b> functions; they operate in a different way, and are not
39	Perl-compatible. This alternative has advantages and disadvantages compared
40	with the standard algorithm, and these are described below.
41	</P>
42	<P>
43	When there is only one possible way in which a given subject string can match a
44	pattern, the two algorithms give the same answer. A difference arises, however,
45	when there are multiple possibilities. For example, if the pattern
46	<pre>
47	^<.*>
48	</pre>
49	is matched against the string
50	<pre>
51	<something> <something else> <something further>
52	</pre>
53	there are three possible answers. The standard algorithm finds only one of
54	them, whereas the alternative algorithm finds all three.
55	</P>
56	<br><a name="SEC2" href="#TOC1">REGULAR EXPRESSIONS AS TREES</a><br>
57	<P>
58	The set of strings that are matched by a regular expression can be represented
59	as a tree structure. An unlimited repetition in the pattern makes the tree of
60	infinite size, but it is still a tree. Matching the pattern to a given subject
61	string (from a given starting point) can be thought of as a search of the tree.
62	There are two ways to search a tree: depth-first and breadth-first, and these
63	correspond to the two matching algorithms provided by PCRE.
64	</P>
65	<br><a name="SEC3" href="#TOC1">THE STANDARD MATCHING ALGORITHM</a><br>
66	<P>
67	In the terminology of Jeffrey Friedl's book "Mastering Regular
68	Expressions", the standard algorithm is an "NFA algorithm". It conducts a
69	depth-first search of the pattern tree. That is, it proceeds along a single
70	path through the tree, checking that the subject matches what is required. When
71	there is a mismatch, the algorithm tries any alternatives at the current point,
72	and if they all fail, it backs up to the previous branch point in the tree, and
73	tries the next alternative branch at that level. This often involves backing up
74	(moving to the left) in the subject string as well. The order in which
75	repetition branches are tried is controlled by the greedy or ungreedy nature of
76	the quantifier.
77	</P>
78	<P>
79	If a leaf node is reached, a matching string has been found, and at that point
80	the algorithm stops. Thus, if there is more than one possible match, this
81	algorithm returns the first one that it finds. Whether this is the shortest,
82	the longest, or some intermediate length depends on the way the greedy and
83	ungreedy repetition quantifiers are specified in the pattern.
84	</P>
85	<P>
86	Because it ends up with a single path through the tree, it is relatively
87	straightforward for this algorithm to keep track of the substrings that are
88	matched by portions of the pattern in parentheses. This provides support for
89	capturing parentheses and back references.
90	</P>
91	<br><a name="SEC4" href="#TOC1">THE ALTERNATIVE MATCHING ALGORITHM</a><br>
92	<P>
93	This algorithm conducts a breadth-first search of the tree. Starting from the
94	first matching point in the subject, it scans the subject string from left to
95	right, once, character by character, and as it does this, it remembers all the
96	paths through the tree that represent valid matches. In Friedl's terminology,
97	this is a kind of "DFA algorithm", though it is not implemented as a
98	traditional finite state machine (it keeps multiple states active
99	simultaneously).
100	</P>
101	<P>
102	Although the general principle of this matching algorithm is that it scans the
103	subject string only once, without backtracking, there is one exception: when a
104	lookaround assertion is encountered, the characters following or preceding the
105	current point have to be independently inspected.
106	</P>
107	<P>
108	The scan continues until either the end of the subject is reached, or there are
109	no more unterminated paths. At this point, terminated paths represent the
110	different matching possibilities (if there are none, the match has failed).
111	Thus, if there is more than one possible match, this algorithm finds all of
112	them, and in particular, it finds the longest. The matches are returned in
113	decreasing order of length. There is an option to stop the algorithm after the
114	first match (which is necessarily the shortest) is found.
115	</P>
116	<P>
117	Note that all the matches that are found start at the same point in the
118	subject. If the pattern
119	<pre>
120	cat(er(pillar)?)?
121	</pre>
122	is matched against the string "the caterpillar catchment", the result will be
123	the three strings "caterpillar", "cater", and "cat" that start at the fifth
124	character of the subject. The algorithm does not automatically move on to find
125	matches that start at later positions.
126	</P>
127	<P>
128	There are a number of features of PCRE regular expressions that are not
129	supported by the alternative matching algorithm. They are as follows:
130	</P>
131	<P>
132	1. Because the algorithm finds all possible matches, the greedy or ungreedy
133	nature of repetition quantifiers is not relevant. Greedy and ungreedy
134	quantifiers are treated in exactly the same way. However, possessive
135	quantifiers can make a difference when what follows could also match what is
136	quantified, for example in a pattern like this:
137	<pre>
138	^a++\w!
139	</pre>
140	This pattern matches "aaab!" but not "aaa!", which would be matched by a
141	non-possessive quantifier. Similarly, if an atomic group is present, it is
142	matched as if it were a standalone pattern at the current point, and the
143	longest match is then "locked in" for the rest of the overall pattern.
144	</P>
145	<P>
146	2. When dealing with multiple paths through the tree simultaneously, it is not
147	straightforward to keep track of captured substrings for the different matching
148	possibilities, and PCRE's implementation of this algorithm does not attempt to
149	do this. This means that no captured substrings are available.
150	</P>
151	<P>
152	3. Because no substrings are captured, back references within the pattern are
153	not supported, and cause errors if encountered.
154	</P>
155	<P>
156	4. For the same reason, conditional expressions that use a backreference as the
157	condition or test for a specific group recursion are not supported.
158	</P>
159	<P>
160	5. Because many paths through the tree may be active, the \K escape sequence,
161	which resets the start of the match when encountered (but may be on some paths
162	and not on others), is not supported. It causes an error if encountered.
163	</P>
164	<P>
165	6. Callouts are supported, but the value of the <i>capture_top</i> field is
166	always 1, and the value of the <i>capture_last</i> field is always -1.
167	</P>
168	<P>
169	7. The \C escape sequence, which (in the standard algorithm) always matches a
170	single data unit, even in UTF-8 or UTF-16 modes, is not supported in these
171	modes, because the alternative algorithm moves through the subject string one
172	character (not data unit) at a time, for all active paths through the tree.
173	</P>
174	<P>
175	8. Except for (FAIL), the backtracking control verbs such as (PRUNE) are not
176	supported. (*FAIL) is supported, and behaves like a failing negative assertion.
177	</P>
178	<br><a name="SEC5" href="#TOC1">ADVANTAGES OF THE ALTERNATIVE ALGORITHM</a><br>
179	<P>
180	Using the alternative matching algorithm provides the following advantages:
181	</P>
182	<P>
183	1. All possible matches (at a single point in the subject) are automatically
184	found, and in particular, the longest match is found. To find more than one
185	match using the standard algorithm, you have to do kludgy things with
186	callouts.
187	</P>
188	<P>
189	2. Because the alternative algorithm scans the subject string just once, and
190	never needs to backtrack (except for lookbehinds), it is possible to pass very
191	long subject strings to the matching function in several pieces, checking for
192	partial matching each time. Although it is possible to do multi-segment
193	matching using the standard algorithm by retaining partially matched
194	substrings, it is more complicated. The
195	<a href="pcrepartial.html"><b>pcrepartial</b></a>
196	documentation gives details of partial matching and discusses multi-segment
197	matching.
198	</P>
199	<br><a name="SEC6" href="#TOC1">DISADVANTAGES OF THE ALTERNATIVE ALGORITHM</a><br>
200	<P>
201	The alternative algorithm suffers from a number of disadvantages:
202	</P>
203	<P>
204	1. It is substantially slower than the standard algorithm. This is partly
205	because it has to search for all possible matches, but is also because it is
206	less susceptible to optimization.
207	</P>
208	<P>
209	2. Capturing parentheses and back references are not supported.
210	</P>
211	<P>
212	3. Although atomic groups are supported, their use does not provide the
213	performance advantage that it does for the standard algorithm.
214	</P>
215	<br><a name="SEC7" href="#TOC1">AUTHOR</a><br>
216	<P>
217	Philip Hazel
218	<br>
219	University Computing Service
220	<br>
221	Cambridge CB2 3QH, England.
222	<br>
223	</P>
224	<br><a name="SEC8" href="#TOC1">REVISION</a><br>
225	<P>
226	Last updated: 08 January 2012
227	<br>
228	Copyright © 1997-2012 University of Cambridge.
229	<br>
230	<p>
231	Return to the <a href="index.html">PCRE index page</a>.
232	</p>
Name	Value
svn:eol-style	native
svn:keywords	"Author Date Id Revision Url"