/[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 172 by ph10, Tue Jun 5 10:40:13 2007 UTC revision 463 by ph10, Sun Oct 18 10:02:46 2009 UTC
# 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 or  PCRE_PARTIAL_HARD options, which can be set when calling \fBpcre_exec()\fP or
29  \fBpcre_dfa_exec()\fP. When this flag is set for \fBpcre_exec()\fP, the return  \fBpcre_dfa_exec()\fP. For backwards compatibility, PCRE_PARTIAL is a synonym
30  code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time  for PCRE_PARTIAL_SOFT. The essential difference between the two options is
31  during the matching process the last part of the subject string matched part of  whether or not a partial match is preferred to an alternative complete match,
32  the pattern. Unfortunately, for non-anchored matching, it is not possible to  though the details differ between the two matching functions. If both options
33  obtain the position of the start of the partial match. No captured data is set  are set, PCRE_PARTIAL_HARD takes precedence.
 when PCRE_ERROR_PARTIAL is returned.  
34  .P  .P
35  When PCRE_PARTIAL is set for \fBpcre_dfa_exec()\fP, the return code  Setting a partial matching option disables two of PCRE's optimizations. PCRE
36  PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of the  remembers the last literal byte in a pattern, and abandons matching immediately
37  subject is reached, there have been no complete matches, but there is still at  if such a byte is not present in the subject string. This optimization cannot
38  least one matching possibility. The portion of the string that provided the  be used for a subject string that might match only partially. If the pattern
39  partial match is set as the first matching string.  was studied, PCRE knows the minimum length of a matching string, and does not
40  .P  bother to run the matching function on shorter strings. This optimization is
41  Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers the  also disabled for partial matching.
 last literal byte in a pattern, and abandons matching immediately if such a  
 byte is not present in the subject string. This optimization cannot be used  
 for a subject string that might match only partially.  
42  .  .
43  .  .
44  .SH "RESTRICTED PATTERNS FOR PCRE_PARTIAL"  .SH "PARTIAL MATCHING USING pcre_exec()"
45  .rs  .rs
46  .sp  .sp
47  Because of the way certain internal optimizations are implemented in the  A partial match occurs during a call to \fBpcre_exec()\fP whenever the end of
48  \fBpcre_exec()\fP function, the PCRE_PARTIAL option cannot be used with all  the subject string is reached successfully, but matching cannot continue
49  patterns. These restrictions do not apply when \fBpcre_dfa_exec()\fP is used.  because more characters are needed. However, at least one character must have
50  For \fBpcre_exec()\fP, repeated single characters such as  been matched. (In other words, a partial match can never be an empty string.)
51    .P
52    If PCRE_PARTIAL_SOFT is set, the partial match is remembered, but matching
53    continues as normal, and other alternatives in the pattern are tried. If no
54    complete match can be found, \fBpcre_exec()\fP returns PCRE_ERROR_PARTIAL
55    instead of PCRE_ERROR_NOMATCH. If there are at least two slots in the offsets
56    vector, the first of them is set to the offset of the earliest character that
57    was inspected when the partial match was found. For convenience, the second
58    offset points to the end of the string so that a substring can easily be
59    identified.
60    .P
61    For the majority of patterns, the first offset identifies the start of the
62    partially matched string. However, for patterns that contain lookbehind
63    assertions, or \eK, or begin with \eb or \eB, earlier characters have been
64    inspected while carrying out the match. For example:
65    .sp
66      /(?<=abc)123/
67    .sp
68    This pattern matches "123", but only if it is preceded by "abc". If the subject
69    string is "xyzabc12", the offsets after a partial match are for the substring
70    "abc12", because all these characters are needed if another match is tried
71    with extra characters added.
72    .P
73    If there is more than one partial match, the first one that was found provides
74    the data that is returned. Consider this pattern:
75    .sp
76      /123\ew+X|dogY/
77  .sp  .sp
78    a{2,4}  If this is matched against the subject string "abc123dog", both
79    alternatives fail to match, but the end of the subject is reached during
80    matching, so PCRE_ERROR_PARTIAL is returned instead of PCRE_ERROR_NOMATCH. The
81    offsets are set to 3 and 9, identifying "123dog" as the first partial match
82    that was found. (In this example, there are two partial matches, because "dog"
83    on its own partially matches the second alternative.)
84    .P
85    If PCRE_PARTIAL_HARD is set for \fBpcre_exec()\fP, it returns
86    PCRE_ERROR_PARTIAL as soon as a partial match is found, without continuing to
87    search for possible complete matches. The difference between the two options
88    can be illustrated by a pattern such as:
89    .sp
90      /dog(sbody)?/
91    .sp
92    This matches either "dog" or "dogsbody", greedily (that is, it prefers the
93    longer string if possible). If it is matched against the string "dog" with
94    PCRE_PARTIAL_SOFT, it yields a complete match for "dog". However, if
95    PCRE_PARTIAL_HARD is set, the result is PCRE_ERROR_PARTIAL. On the other hand,
96    if the pattern is made ungreedy the result is different:
97    .sp
98      /dog(sbody)??/
99    .sp
100    In this case the result is always a complete match because \fBpcre_exec()\fP
101    finds that first, and it never continues after finding a match. It might be
102    easier to follow this explanation by thinking of the two patterns like this:
103  .sp  .sp
104  and repeated single metasequences such as    /dog(sbody)?/    is the same as  /dogsbody|dog/
105      /dog(sbody)??/   is the same as  /dog|dogsbody/
106  .sp  .sp
107    \ed+  The second pattern will never match "dogsbody" when \fBpcre_exec()\fP is
108    used, because it will always find the shorter match first.
109    .
110    .
111    .SH "PARTIAL MATCHING USING pcre_dfa_exec()"
112    .rs
113  .sp  .sp
114  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
115  Optional items such as \ed? (where the maximum is one) are permitted.  character, without backtracking, searching for all possible matches
116  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
117  examples above can be coded thus:  pattern, there is the possibility of a partial match, again provided that at
118    least one character has matched.
119    .P
120    When PCRE_PARTIAL_SOFT is set, PCRE_ERROR_PARTIAL is returned only if there
121    have been no complete matches. Otherwise, the complete matches are returned.
122    However, if PCRE_PARTIAL_HARD is set, a partial match takes precedence over any
123    complete matches. The portion of the string that was inspected when the longest
124    partial match was found is set as the first matching string, provided there are
125    at least two slots in the offsets vector.
126    .P
127    Because \fBpcre_dfa_exec()\fP always searches for all possible matches, and
128    there is no difference between greedy and ungreedy repetition, its behaviour is
129    different from \fBpcre_exec\fP when PCRE_PARTIAL_HARD is set. Consider the
130    string "dog" matched against the ungreedy pattern shown above:
131    .sp
132      /dog(sbody)??/
133    .sp
134    Whereas \fBpcre_exec()\fP stops as soon as it finds the complete match for
135    "dog", \fBpcre_dfa_exec()\fP also finds the partial match for "dogsbody", and
136    so returns that when PCRE_PARTIAL_HARD is set.
137    .
138    .
139    .SH "PARTIAL MATCHING AND WORD BOUNDARIES"
140    .rs
141  .sp  .sp
142    (a){2,4}  If a pattern ends with one of sequences \ew or \eW, which test for word
143    (\ed)+  boundaries, partial matching with PCRE_PARTIAL_SOFT can give counter-intuitive
144    results. Consider this pattern:
145    .sp
146      /\ebcat\eb/
147    .sp
148    This matches "cat", provided there is a word boundary at either end. If the
149    subject string is "the cat", the comparison of the final "t" with a following
150    character cannot take place, so a partial match is found. However,
151    \fBpcre_exec()\fP carries on with normal matching, which matches \eb at the end
152    of the subject when the last character is a letter, thus finding a complete
153    match. The result, therefore, is \fInot\fP PCRE_ERROR_PARTIAL. The same thing
154    happens with \fBpcre_dfa_exec()\fP, because it also finds the complete match.
155    .P
156    Using PCRE_PARTIAL_HARD in this case does yield PCRE_ERROR_PARTIAL, because
157    then the partial match takes precedence.
158    .
159    .
160    .SH "FORMERLY RESTRICTED PATTERNS"
161    .rs
162  .sp  .sp
163  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
164  envisaged for this facility, this is not felt to be a major restriction.  optimizations were implemented in the \fBpcre_exec()\fP function, the
165    PCRE_PARTIAL option (predecessor of PCRE_PARTIAL_SOFT) could not be used with
166    all patterns. From release 8.00 onwards, the restrictions no longer apply, and
167    partial matching with \fBpcre_exec()\fP can be requested for any pattern.
168  .P  .P
169  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
170  \fBpcre_exec()\fP returns the error code PCRE_ERROR_BADPARTIAL (-13).  repeated metasequences. If PCRE_PARTIAL was set for a pattern that did not
171  You can use the PCRE_INFO_OKPARTIAL call to \fBpcre_fullinfo()\fP to find out  conform to the restrictions, \fBpcre_exec()\fP returned the error code
172  if a compiled pattern can be used for partial matching.  PCRE_ERROR_BADPARTIAL (-13). This error code is no longer in use. The
173    PCRE_INFO_OKPARTIAL call to \fBpcre_fullinfo()\fP to find out if a compiled
174    pattern can be used for partial matching now always returns 1.
175  .  .
176  .  .
177  .SH "EXAMPLE OF PARTIAL MATCHING USING PCRETEST"  .SH "EXAMPLE OF PARTIAL MATCHING USING PCRETEST"
178  .rs  .rs
179  .sp  .sp
180  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
181  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
182  uses the date example quoted above:  that uses the date example quoted above:
183  .sp  .sp
184      re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/      re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/
185    data> 25jun04\eP    data> 25jun04\eP
186     0: 25jun04     0: 25jun04
187     1: jun     1: jun
188    data> 25dec3\eP    data> 25dec3\eP
189    Partial match    Partial match: 23dec3
190    data> 3ju\eP    data> 3ju\eP
191    Partial match    Partial match: 3ju
192    data> 3juj\eP    data> 3juj\eP
193    No match    No match
194    data> j\eP    data> j\eP
# Line 97  uses the date example quoted above: Line 196  uses the date example quoted above:
196  .sp  .sp
197  The first data string is matched completely, so \fBpcretest\fP shows the  The first data string is matched completely, so \fBpcretest\fP shows the
198  matched substrings. The remaining four strings do not match the complete  matched substrings. The remaining four strings do not match the complete
199  pattern, but the first two are partial matches. The same test, using  pattern, but the first two are partial matches. Similar output is obtained
200  \fBpcre_dfa_exec()\fP matching (by means of the \eD escape sequence), produces  when \fBpcre_dfa_exec()\fP is used.
201  the following output:  .P
202  .sp  If the escape sequence \eP is present more than once in a \fBpcretest\fP data
203      re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/  line, the PCRE_PARTIAL_HARD option is set for the match.
   data> 25jun04\eP\eD  
    0: 25jun04  
   data> 23dec3\eP\eD  
   Partial match: 23dec3  
   data> 3ju\eP\eD  
   Partial match: 3ju  
   data> 3juj\eP\eD  
   No match  
   data> j\eP\eD  
   No match  
 .sp  
 Notice that in this case the portion of the string that was matched is made  
 available.  
204  .  .
205  .  .
206  .SH "MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()"  .SH "MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()"
# Line 123  available. Line 209  available.
209  When a partial match has been found using \fBpcre_dfa_exec()\fP, it is possible  When a partial match has been found using \fBpcre_dfa_exec()\fP, it is possible
210  to continue the match by providing additional subject data and calling  to continue the match by providing additional subject data and calling
211  \fBpcre_dfa_exec()\fP again with the same compiled regular expression, this  \fBpcre_dfa_exec()\fP again with the same compiled regular expression, this
212  time setting the PCRE_DFA_RESTART option. You must also pass the same working  time setting the PCRE_DFA_RESTART option. You must pass the same working
213  space as before, because this is where details of the previous partial match  space as before, because this is where details of the previous partial match
214  are stored. Here is an example using \fBpcretest\fP, using the \eR escape  are stored. Here is an example using \fBpcretest\fP, using the \eR escape
215  sequence to set the PCRE_DFA_RESTART option (\eP and \eD are as above):  sequence to set the PCRE_DFA_RESTART option (\eD specifies the use of
216    \fBpcre_dfa_exec()\fP):
217  .sp  .sp
218      re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/      re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/
219    data> 23ja\eP\eD    data> 23ja\eP\eD
# Line 140  Notice that when the match is complete, Line 227  Notice that when the match is complete,
227  not retain the previously partially-matched string. It is up to the calling  not retain the previously partially-matched string. It is up to the calling
228  program to do that if it needs to.  program to do that if it needs to.
229  .P  .P
230  You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial matching  You can set the PCRE_PARTIAL_SOFT or PCRE_PARTIAL_HARD options with
231  over multiple segments. This facility can be used to pass very long subject  PCRE_DFA_RESTART to continue partial matching over multiple segments. This
232  strings to \fBpcre_dfa_exec()\fP. However, some care is needed for certain  facility can be used to pass very long subject strings to
233  types of pattern.  \fBpcre_dfa_exec()\fP.
234    .
235    .
236    .SH "MULTI-SEGMENT MATCHING WITH pcre_exec()"
237    .rs
238    .sp
239    From release 8.00, \fBpcre_exec()\fP can also be used to do multi-segment
240    matching. Unlike \fBpcre_dfa_exec()\fP, it is not possible to restart the
241    previous match with a new segment of data. Instead, new data must be added to
242    the previous subject string, and the entire match re-run, starting from the
243    point where the partial match occurred. Earlier data can be discarded.
244    Consider an unanchored pattern that matches dates:
245    .sp
246        re> /\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed/
247      data> The date is 23ja\eP
248      Partial match: 23ja
249    .sp
250    The this stage, an application could discard the text preceding "23ja", add on
251    text from the next segment, and call \fBpcre_exec()\fP again. Unlike
252    \fBpcre_dfa_exec()\fP, the entire matching string must always be available, and
253    the complete matching process occurs for each call, so more memory and more
254    processing time is needed.
255    .P
256    \fBNote:\fP If the pattern contains lookbehind assertions, or \eK, or starts
257    with \eb or \eB, the string that is returned for a partial match will include
258    characters that precede the partially matched string itself, because these must
259    be retained when adding on more characters for a subsequent matching attempt.
260    .
261    .
262    .SH "ISSUES WITH MULTI-SEGMENT MATCHING"
263    .rs
264    .sp
265    Certain types of pattern may give problems with multi-segment matching,
266    whichever matching function is used.
267  .P  .P
268  1. If the pattern contains tests for the beginning or end of a line, you need  1. If the pattern contains tests for the beginning or end of a line, you need
269  to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropriate, when the  to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropriate, when the
270  subject string for any call does not contain the beginning or end of a line.  subject string for any call does not contain the beginning or end of a line.
271  .P  .P
272  2. If the pattern contains backward assertions (including \eb or \eB), you need  2. Lookbehind assertions at the start of a pattern are catered for in the
273  to arrange for some overlap in the subject strings to allow for this. For  offsets that are returned for a partial match. However, in theory, a lookbehind
274  example, you could pass the subject in chunks that are 500 bytes long, but in  assertion later in the pattern could require even earlier characters to be
275  a buffer of 700 bytes, with the starting offset set to 200 and the previous 200  inspected, and it might not have been reached when a partial match occurs. This
276  bytes at the start of the buffer.  is probably an extremely unlikely case; you could guard against it to a certain
277  .P  extent by always including extra characters at the start.
278  3. Matching a subject string that is split into multiple segments does not  .P
279  always produce exactly the same result as matching over one single long string.  3. Matching a subject string that is split into multiple segments may not
280  The difference arises when there are multiple matching possibilities, because a  always produce exactly the same result as matching over one single long string,
281  partial match result is given only when there are no completed matches in a  especially when PCRE_PARTIAL_SOFT is used. The section "Partial Matching and
282  call to \fBpcre_dfa_exec()\fP. This means that as soon as the shortest match has  Word Boundaries" above describes an issue that arises if the pattern ends with
283    \eb or \eB. Another kind of difference may occur when there are multiple
284    matching possibilities, because a partial match result is given only when there
285    are no completed matches. This means that as soon as the shortest match has
286  been found, continuation to a new subject segment is no longer possible.  been found, continuation to a new subject segment is no longer possible.
287  Consider this \fBpcretest\fP example:  Consider again this \fBpcretest\fP example:
288  .sp  .sp
289      re> /dog(sbody)?/      re> /dog(sbody)?/
290      data> dogsb\eP
291       0: dog
292    data> do\eP\eD    data> do\eP\eD
293    Partial match: do    Partial match: do
294    data> gsb\eR\eP\eD    data> gsb\eR\eP\eD
# Line 172  Consider this \fBpcretest\fP example: Line 297  Consider this \fBpcretest\fP example:
297     0: dogsbody     0: dogsbody
298     1: dog     1: dog
299  .sp  .sp
300  The pattern matches the words "dog" or "dogsbody". When the subject is  The first data line passes the string "dogsb" to \fBpcre_exec()\fP, setting the
301  presented in several parts ("do" and "gsb" being the first two) the match stops  PCRE_PARTIAL_SOFT option. Although the string is a partial match for
302  when "dog" has been found, and it is not possible to continue. On the other  "dogsbody", the result is not PCRE_ERROR_PARTIAL, because the shorter string
303  hand, if "dogsbody" is presented as a single string, both matches are found.  "dog" is a complete match. Similarly, when the subject is presented to
304    \fBpcre_dfa_exec()\fP in several parts ("do" and "gsb" being the first two) the
305    match stops when "dog" has been found, and it is not possible to continue. On
306    the other hand, if "dogsbody" is presented as a single string,
307    \fBpcre_dfa_exec()\fP finds both matches.
308  .P  .P
309  Because of this phenomenon, it does not usually make sense to end a pattern  Because of these problems, it is probably best to use PCRE_PARTIAL_HARD when
310  that is going to be matched in this way with a variable repeat.  matching multi-segment data. The example above then behaves differently:
311    .sp
312        re> /dog(sbody)?/
313      data> dogsb\eP\eP
314      Partial match: dogsb
315      data> do\eP\eD
316      Partial match: do
317      data> gsb\eR\eP\eP\eD
318      Partial match: gsb
319    .sp
320  .P  .P
321  4. Patterns that contain alternatives at the top level which do not all  4. Patterns that contain alternatives at the top level which do not all
322  start with the same pattern item may not work as expected. For example,  start with the same pattern item may not work as expected when
323  consider this pattern:  PCRE_DFA_RESTART is used with \fBpcre_dfa_exec()\fP. For example, consider this
324    pattern:
325  .sp  .sp
326    1234|3789    1234|3789
327  .sp  .sp
328  If the first part of the subject is "ABC123", a partial match of the first  If the first part of the subject is "ABC123", a partial match of the first
329  alternative is found at offset 3. There is no partial match for the second  alternative is found at offset 3. There is no partial match for the second
330  alternative, because such a match does not start at the same point in the  alternative, because such a match does not start at the same point in the
331  subject string. Attempting to continue with the string "789" does not yield a  subject string. Attempting to continue with the string "7890" does not yield a
332  match because only those alternatives that match at one point in the subject  match because only those alternatives that match at one point in the subject
333  are remembered. The problem arises because the start of the second alternative  are remembered. The problem arises because the start of the second alternative
334  matches within the first alternative. There is no problem with anchored  matches within the first alternative. There is no problem with anchored
# Line 197  patterns or patterns such as: Line 336  patterns or patterns such as:
336  .sp  .sp
337    1234|ABCD    1234|ABCD
338  .sp  .sp
339  where no string can be a partial match for both alternatives.  where no string can be a partial match for both alternatives. This is not a
340    problem if \fPpcre_exec()\fP is used, because the entire match has to be rerun
341    each time:
342    .sp
343        re> /1234|3789/
344      data> ABC123\eP
345      Partial match: 123
346      data> 1237890
347       0: 3789
348    .sp
349    Of course, instead of using PCRE_DFA_PARTIAL, the same technique of re-running
350    the entire match can also be used with \fBpcre_dfa_exec()\fP.
351  .  .
352  .  .
353  .SH AUTHOR  .SH AUTHOR
# Line 214  Cambridge CB2 3QH, England. Line 364  Cambridge CB2 3QH, England.
364  .rs  .rs
365  .sp  .sp
366  .nf  .nf
367  Last updated: 04 June 2007  Last updated: 18 October 2009
368  Copyright (c) 1997-2007 University of Cambridge.  Copyright (c) 1997-2009 University of Cambridge.
369  .fi  .fi

Legend:
Removed from v.172  
changed lines
  Added in v.463

  ViewVC Help
Powered by ViewVC 1.1.5