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

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

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

revision 75 by nigel, Sat Feb 24 21:40:37 2007 UTC revision 435 by ph10, Sat Sep 5 10:20:44 2009 UTC
# Line 1  Line 1 
1  .TH PCRE 3  .TH PCREPARTIAL 3
2  .SH NAME  .SH NAME
3  PCRE - Perl-compatible regular expressions  PCRE - Perl-compatible regular expressions
4  .SH "PARTIAL MATCHING IN PCRE"  .SH "PARTIAL MATCHING IN PCRE"
5  .rs  .rs
6  .sp  .sp
7  In normal use of PCRE, if the subject string that is passed to  In normal use of PCRE, if the subject string that is passed to
8  \fBpcre_exec()\fP matches as far as it goes, but is too short to match the  \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP matches as far as it goes, but is
9  entire pattern, PCRE_ERROR_NOMATCH is returned. There are circumstances where  too short to match the entire pattern, PCRE_ERROR_NOMATCH is returned. There
10  it might be helpful to distinguish this case from other cases in which there is  are circumstances where it might be helpful to distinguish this case from other
11  no match.  cases in which there is no match.
12  .P  .P
13  Consider, for example, an application where a human is required to type in data  Consider, for example, an application where a human is required to type in data
14  for a field with specific formatting requirements. An example might be a date  for a field with specific formatting requirements. An example might be a date
# Line 18  in the form \fIddmmmyy\fP, defined by th Line 18  in the form \fIddmmmyy\fP, defined by th
18  .sp  .sp
19  If the application sees the user's keystrokes one by one, and can check that  If the application sees the user's keystrokes one by one, and can check that
20  what has been typed so far is potentially valid, it is able to raise an error  what has been typed so far is potentially valid, it is able to raise an error
21  as soon as a mistake is made, possibly beeping and not reflecting the  as soon as a mistake is made, by beeping and not reflecting the character that
22  character that has been typed. This immediate feedback is likely to be a better  has been typed, for example. This immediate feedback is likely to be a better
23  user interface than a check that is delayed until the entire string has been  user interface than a check that is delayed until the entire string has been
24  entered.  entered. Partial matching can also sometimes be useful when the subject string
25    is very long and is not all available at once.
26  .P  .P
27  PCRE supports the concept of partial matching by means of the PCRE_PARTIAL  PCRE supports partial matching by means of the PCRE_PARTIAL_SOFT and
28  option, which can be set when calling \fBpcre_exec()\fP. When this is done, the  PCRE_PARTIAL_HARD options, which can be set when calling \fBpcre_exec()\fP or
29  return code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any  \fBpcre_dfa_exec()\fP. For backwards compatibility, PCRE_PARTIAL is a synonym
30  time during the matching process the entire subject string matched part of the  for PCRE_PARTIAL_SOFT. The essential difference between the two options is
31  pattern. No captured data is set when this occurs.  whether or not a partial match is preferred to an alternative complete match,
32    though the details differ between the two matching functions. If both options
33    are set, PCRE_PARTIAL_HARD takes precedence.
34  .P  .P
35  Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers the  Setting a partial matching option disables one of PCRE's optimizations. PCRE
36  last literal byte in a pattern, and abandons matching immediately if such a  remembers the last literal byte in a pattern, and abandons matching immediately
37  byte is not present in the subject string. This optimization cannot be used  if such a byte is not present in the subject string. This optimization cannot
38  for a subject string that might match only partially.  be used for a subject string that might match only partially.
39  .  .
40  .  .
41  .SH "RESTRICTED PATTERNS FOR PCRE_PARTIAL"  .SH "PARTIAL MATCHING USING pcre_exec()"
42  .rs  .rs
43  .sp  .sp
44  Because of the way certain internal optimizations are implemented in PCRE, the  A partial match occurs during a call to \fBpcre_exec()\fP whenever the end of
45  PCRE_PARTIAL option cannot be used with all patterns. Repeated single  the subject string is reached successfully, but matching cannot continue
46  characters such as  because more characters are needed. However, at least one character must have
47    been matched. (In other words, a partial match can never be an empty string.)
48    .P
49    If PCRE_PARTIAL_SOFT is set, the partial match is remembered, but matching
50    continues as normal, and other alternatives in the pattern are tried. If no
51    complete match can be found, \fBpcre_exec()\fP returns PCRE_ERROR_PARTIAL
52    instead of PCRE_ERROR_NOMATCH. If there are at least two slots in the offsets
53    vector, the first of them is set to the offset of the earliest character that
54    was inspected when the partial match was found. For convenience, the second
55    offset points to the end of the string so that a substring can easily be
56    extracted.
57    .P
58    For the majority of patterns, the first offset identifies the start of the
59    partially matched string. However, for patterns that contain lookbehind
60    assertions, or \eK, or begin with \eb or \eB, earlier characters have been
61    inspected while carrying out the match. For example:
62    .sp
63      /(?<=abc)123/
64    .sp
65    This pattern matches "123", but only if it is preceded by "abc". If the subject
66    string is "xyzabc12", the offsets after a partial match are for the substring
67    "abc12", because all these characters are needed if another match is tried
68    with extra characters added.
69    .P
70    If there is more than one partial match, the first one that was found provides
71    the data that is returned. Consider this pattern:
72  .sp  .sp
73    a{2,4}    /123\ew+X|dogY/
74  .sp  .sp
75  and repeated single metasequences such as  If this is matched against the subject string "abc123dog", both
76    alternatives fail to match, but the end of the subject is reached during
77    matching, so PCRE_ERROR_PARTIAL is returned instead of PCRE_ERROR_NOMATCH. The
78    offsets are set to 3 and 9, identifying "123dog" as the first partial match
79    that was found. (In this example, there are two partial matches, because "dog"
80    on its own partially matches the second alternative.)
81    .P
82    If PCRE_PARTIAL_HARD is set for \fBpcre_exec()\fP, it returns
83    PCRE_ERROR_PARTIAL as soon as a partial match is found, without continuing to
84    search for possible complete matches. The difference between the two options
85    can be illustrated by a pattern such as:
86    .sp
87      /dog(sbody)?/
88    .sp
89    This matches either "dog" or "dogsbody", greedily (that is, it prefers the
90    longer string if possible). If it is matched against the string "dog" with
91    PCRE_PARTIAL_SOFT, it yields a complete match for "dog". However, if
92    PCRE_PARTIAL_HARD is set, the result is PCRE_ERROR_PARTIAL. On the other hand,
93    if the pattern is made ungreedy the result is different:
94    .sp
95      /dog(sbody)??/
96    .sp
97    In this case the result is always a complete match because \fBpcre_exec()\fP
98    finds that first, and it never continues after finding a match. It might be
99    easier to follow this explanation by thinking of the two patterns like this:
100    .sp
101      /dog(sbody)?/    is the same as  /dogsbody|dog/
102      /dog(sbody)??/   is the same as  /dog|dogsbody/
103  .sp  .sp
104    \ed+  The second pattern will never match "dogsbody" when \fBpcre_exec()\fP is
105    used, because it will always find the shorter match first.
106    .
107    .
108    .SH "PARTIAL MATCHING USING pcre_dfa_exec()"
109    .rs
110  .sp  .sp
111  are not permitted if the maximum number of occurrences is greater than one.  The \fBpcre_dfa_exec()\fP function moves along the subject string character by
112  Optional items such as \ed? (where the maximum is one) are permitted.  character, without backtracking, searching for all possible matches
113  Quantifiers with any values are permitted after parentheses, so the invalid  simultaneously. If the end of the subject is reached before the end of the
114  examples above can be coded thus:  pattern, there is the possibility of a partial match, again provided that at
115    least one character has matched.
116    .P
117    When PCRE_PARTIAL_SOFT is set, PCRE_ERROR_PARTIAL is returned only if there
118    have been no complete matches. Otherwise, the complete matches are returned.
119    However, if PCRE_PARTIAL_HARD is set, a partial match takes precedence over any
120    complete matches. The portion of the string that was inspected when the longest
121    partial match was found is set as the first matching string, provided there are
122    at least two slots in the offsets vector.
123    .P
124    Because \fBpcre_dfa_exec()\fP always searches for all possible matches, and
125    there is no difference between greedy and ungreedy repetition, its behaviour is
126    different from \fBpcre_exec\fP when PCRE_PARTIAL_HARD is set. Consider the
127    string "dog" matched against the ungreedy pattern shown above:
128    .sp
129      /dog(sbody)??/
130    .sp
131    Whereas \fBpcre_exec()\fP stops as soon as it finds the complete match for
132    "dog", \fBpcre_dfa_exec()\fP also finds the partial match for "dogsbody", and
133    so returns that when PCRE_PARTIAL_HARD is set.
134    .
135    .
136    .SH "PARTIAL MATCHING AND WORD BOUNDARIES"
137    .rs
138  .sp  .sp
139    (a){2,4}  If a pattern ends with one of sequences \ew or \eW, which test for word
140    (\ed)+  boundaries, partial matching with PCRE_PARTIAL_SOFT can give counter-intuitive
141    results. Consider this pattern:
142    .sp
143      /\ebcat\eb/
144    .sp
145    This matches "cat", provided there is a word boundary at either end. If the
146    subject string is "the cat", the comparison of the final "t" with a following
147    character cannot take place, so a partial match is found. However,
148    \fBpcre_exec()\fP carries on with normal matching, which matches \eb at the end
149    of the subject when the last character is a letter, thus finding a complete
150    match. The result, therefore, is \fInot\fP PCRE_ERROR_PARTIAL. The same thing
151    happens with \fBpcre_dfa_exec()\fP, because it also finds the complete match.
152    .P
153    Using PCRE_PARTIAL_HARD in this case does yield PCRE_ERROR_PARTIAL, because
154    then the partial match takes precedence.
155    .
156    .
157    .SH "FORMERLY RESTRICTED PATTERNS"
158    .rs
159  .sp  .sp
160  These constructions run more slowly, but for the kinds of application that are  For releases of PCRE prior to 8.00, because of the way certain internal
161  envisaged for this facility, this is not felt to be a major restriction.  optimizations were implemented in the \fBpcre_exec()\fP function, the
162    PCRE_PARTIAL option (predecessor of PCRE_PARTIAL_SOFT) could not be used with
163    all patterns. From release 8.00 onwards, the restrictions no longer apply, and
164    partial matching with \fBpcre_exec()\fP can be requested for any pattern.
165  .P  .P
166  If PCRE_PARTIAL is set for a pattern that does not conform to the restrictions,  Items that were formerly restricted were repeated single characters and
167  \fBpcre_exec()\fP returns the error code PCRE_ERROR_BADPARTIAL (-13).  repeated metasequences. If PCRE_PARTIAL was set for a pattern that did not
168    conform to the restrictions, \fBpcre_exec()\fP returned the error code
169    PCRE_ERROR_BADPARTIAL (-13). This error code is no longer in use. The
170    PCRE_INFO_OKPARTIAL call to \fBpcre_fullinfo()\fP to find out if a compiled
171    pattern can be used for partial matching now always returns 1.
172  .  .
173  .  .
174  .SH "EXAMPLE OF PARTIAL MATCHING USING PCRETEST"  .SH "EXAMPLE OF PARTIAL MATCHING USING PCRETEST"
175  .rs  .rs
176  .sp  .sp
177  If the escape sequence \eP is present in a \fBpcretest\fP data line, the  If the escape sequence \eP is present in a \fBpcretest\fP data line, the
178  PCRE_PARTIAL flag is used for the match. Here is a run of \fBpcretest\fP that  PCRE_PARTIAL_SOFT option is used for the match. Here is a run of \fBpcretest\fP
179  uses the date example quoted above:  that uses the date example quoted above:
180  .sp  .sp
181      re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/      re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/
182    data> 25jun04\P    data> 25jun04\eP
183     0: 25jun04     0: 25jun04
184     1: jun     1: jun
185    data> 25dec3\P    data> 25dec3\eP
186    Partial match    Partial match: 23dec3
187    data> 3ju\P    data> 3ju\eP
188    Partial match    Partial match: 3ju
189    data> 3juj\P    data> 3juj\eP
190    No match    No match
191    data> j\P    data> j\eP
192    No match    No match
193  .sp  .sp
194  The first data string is matched completely, so \fBpcretest\fP shows the  The first data string is matched completely, so \fBpcretest\fP shows the
195  matched substrings. The remaining four strings do not match the complete  matched substrings. The remaining four strings do not match the complete
196  pattern, but the first two are partial matches.  pattern, but the first two are partial matches. Similar output is obtained
197    when \fBpcre_dfa_exec()\fP is used.
198    .P
199    If the escape sequence \eP is present more than once in a \fBpcretest\fP data
200    line, the PCRE_PARTIAL_HARD option is set for the match.
201  .  .
202  .  .
203    .SH "MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()"
204    .rs
205    .sp
206    When a partial match has been found using \fBpcre_dfa_exec()\fP, it is possible
207    to continue the match by providing additional subject data and calling
208    \fBpcre_dfa_exec()\fP again with the same compiled regular expression, this
209    time setting the PCRE_DFA_RESTART option. You must pass the same working
210    space as before, because this is where details of the previous partial match
211    are stored. Here is an example using \fBpcretest\fP, using the \eR escape
212    sequence to set the PCRE_DFA_RESTART option (\eD specifies the use of
213    \fBpcre_dfa_exec()\fP):
214    .sp
215        re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/
216      data> 23ja\eP\eD
217      Partial match: 23ja
218      data> n05\eR\eD
219       0: n05
220    .sp
221    The first call has "23ja" as the subject, and requests partial matching; the
222    second call has "n05" as the subject for the continued (restarted) match.
223    Notice that when the match is complete, only the last part is shown; PCRE does
224    not retain the previously partially-matched string. It is up to the calling
225    program to do that if it needs to.
226  .P  .P
227  .in 0  You can set the PCRE_PARTIAL_SOFT or PCRE_PARTIAL_HARD options with
228  Last updated: 08 September 2004  PCRE_DFA_RESTART to continue partial matching over multiple segments. This
229  .br  facility can be used to pass very long subject strings to
230  Copyright (c) 1997-2004 University of Cambridge.  \fBpcre_dfa_exec()\fP.
231    .
232    .
233    .SH "MULTI-SEGMENT MATCHING WITH pcre_exec()"
234    .rs
235    .sp
236    From release 8.00, \fBpcre_exec()\fP can also be used to do multi-segment
237    matching. Unlike \fBpcre_dfa_exec()\fP, it is not possible to restart the
238    previous match with a new segment of data. Instead, new data must be added to
239    the previous subject string, and the entire match re-run, starting from the
240    point where the partial match occurred. Earlier data can be discarded.
241    Consider an unanchored pattern that matches dates:
242    .sp
243        re> /\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed/
244      data> The date is 23ja\eP
245      Partial match: 23ja
246    .sp
247    The this stage, an application could discard the text preceding "23ja", add on
248    text from the next segment, and call \fBpcre_exec()\fP again. Unlike
249    \fBpcre_dfa_exec()\fP, the entire matching string must always be available, and
250    the complete matching process occurs for each call, so more memory and more
251    processing time is needed.
252    .P
253    \fBNote:\fP If the pattern contains lookbehind assertions, or \eK, or starts
254    with \eb or \eB, the string that is returned for a partial match will include
255    characters that precede the partially matched string itself, because these must
256    be retained when adding on more characters for a subsequent matching attempt.
257    .
258    .
259    .SH "ISSUES WITH MULTI-SEGMENT MATCHING"
260    .rs
261    .sp
262    Certain types of pattern may give problems with multi-segment matching,
263    whichever matching function is used.
264    .P
265    1. If the pattern contains tests for the beginning or end of a line, you need
266    to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropriate, when the
267    subject string for any call does not contain the beginning or end of a line.
268    .P
269    2. Lookbehind assertions at the start of a pattern are catered for in the
270    offsets that are returned for a partial match. However, in theory, a lookbehind
271    assertion later in the pattern could require even earlier characters to be
272    inspected, and it might not have been reached when a partial match occurs. This
273    is probably an extremely unlikely case; you could guard against it to a certain
274    extent by always including extra characters at the start.
275    .P
276    3. Matching a subject string that is split into multiple segments may not
277    always produce exactly the same result as matching over one single long string,
278    especially when PCRE_PARTIAL_SOFT is used. The section "Partial Matching and
279    Word Boundaries" above describes an issue that arises if the pattern ends with
280    \eb or \eB. Another kind of difference may occur when there are multiple
281    matching possibilities, because a partial match result is given only when there
282    are no completed matches. This means that as soon as the shortest match has
283    been found, continuation to a new subject segment is no longer possible.
284    Consider again this \fBpcretest\fP example:
285    .sp
286        re> /dog(sbody)?/
287      data> dogsb\eP
288       0: dog
289      data> do\eP\eD
290      Partial match: do
291      data> gsb\eR\eP\eD
292       0: g
293      data> dogsbody\eD
294       0: dogsbody
295       1: dog
296    .sp
297    The first data line passes the string "dogsb" to \fBpcre_exec()\fP, setting the
298    PCRE_PARTIAL_SOFT option. Although the string is a partial match for
299    "dogsbody", the result is not PCRE_ERROR_PARTIAL, because the shorter string
300    "dog" is a complete match. Similarly, when the subject is presented to
301    \fBpcre_dfa_exec()\fP in several parts ("do" and "gsb" being the first two) the
302    match stops when "dog" has been found, and it is not possible to continue. On
303    the other hand, if "dogsbody" is presented as a single string,
304    \fBpcre_dfa_exec()\fP finds both matches.
305    .P
306    Because of these problems, it is probably best to use PCRE_PARTIAL_HARD when
307    matching multi-segment data. The example above then behaves differently:
308    .sp
309        re> /dog(sbody)?/
310      data> dogsb\eP\eP
311      Partial match: dogsb
312      data> do\eP\eD
313      Partial match: do
314      data> gsb\eR\eP\eP\eD
315      Partial match: gsb
316    .sp
317    .P
318    4. Patterns that contain alternatives at the top level which do not all
319    start with the same pattern item may not work as expected when
320    \fBpcre_dfa_exec()\fP is used. For example, consider this pattern:
321    .sp
322      1234|3789
323    .sp
324    If the first part of the subject is "ABC123", a partial match of the first
325    alternative is found at offset 3. There is no partial match for the second
326    alternative, because such a match does not start at the same point in the
327    subject string. Attempting to continue with the string "7890" does not yield a
328    match because only those alternatives that match at one point in the subject
329    are remembered. The problem arises because the start of the second alternative
330    matches within the first alternative. There is no problem with anchored
331    patterns or patterns such as:
332    .sp
333      1234|ABCD
334    .sp
335    where no string can be a partial match for both alternatives. This is not a
336    problem if \fPpcre_exec()\fP is used, because the entire match has to be rerun
337    each time:
338    .sp
339        re> /1234|3789/
340      data> ABC123\eP
341      Partial match: 123
342      data> 1237890
343       0: 3789
344    .sp
345    .
346    .
347    .SH AUTHOR
348    .rs
349    .sp
350    .nf
351    Philip Hazel
352    University Computing Service
353    Cambridge CB2 3QH, England.
354    .fi
355    .
356    .
357    .SH REVISION
358    .rs
359    .sp
360    .nf
361    Last updated: 05 September 2009
362    Copyright (c) 1997-2009 University of Cambridge.
363    .fi

Legend:
Removed from v.75  
changed lines
  Added in v.435

  ViewVC Help
Powered by ViewVC 1.1.5