1: <html>
2: <head>
3: <title>pcrepartial specification</title>
4: </head>
5: <body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6: <h1>pcrepartial 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">PARTIAL MATCHING IN PCRE</a>
17: <li><a name="TOC2" href="#SEC2">PARTIAL MATCHING USING pcre_exec() OR pcre[16|32]_exec()</a>
18: <li><a name="TOC3" href="#SEC3">PARTIAL MATCHING USING pcre_dfa_exec() OR pcre[16|32]_dfa_exec()</a>
19: <li><a name="TOC4" href="#SEC4">PARTIAL MATCHING AND WORD BOUNDARIES</a>
20: <li><a name="TOC5" href="#SEC5">FORMERLY RESTRICTED PATTERNS</a>
21: <li><a name="TOC6" href="#SEC6">EXAMPLE OF PARTIAL MATCHING USING PCRETEST</a>
22: <li><a name="TOC7" href="#SEC7">MULTI-SEGMENT MATCHING WITH pcre_dfa_exec() OR pcre[16|32]_dfa_exec()</a>
23: <li><a name="TOC8" href="#SEC8">MULTI-SEGMENT MATCHING WITH pcre_exec() OR pcre[16|32]_exec()</a>
24: <li><a name="TOC9" href="#SEC9">ISSUES WITH MULTI-SEGMENT MATCHING</a>
25: <li><a name="TOC10" href="#SEC10">AUTHOR</a>
26: <li><a name="TOC11" href="#SEC11">REVISION</a>
27: </ul>
28: <br><a name="SEC1" href="#TOC1">PARTIAL MATCHING IN PCRE</a><br>
29: <P>
30: In normal use of PCRE, if the subject string that is passed to a matching
31: function matches as far as it goes, but is too short to match the entire
32: pattern, PCRE_ERROR_NOMATCH is returned. There are circumstances where it might
33: be helpful to distinguish this case from other cases in which there is no
34: match.
35: </P>
36: <P>
37: Consider, for example, an application where a human is required to type in data
38: for a field with specific formatting requirements. An example might be a date
39: in the form <i>ddmmmyy</i>, defined by this pattern:
40: <pre>
41: ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
42: </pre>
43: If the application sees the user's keystrokes one by one, and can check that
44: what has been typed so far is potentially valid, it is able to raise an error
45: as soon as a mistake is made, by beeping and not reflecting the character that
46: has been typed, for example. This immediate feedback is likely to be a better
47: user interface than a check that is delayed until the entire string has been
48: entered. Partial matching can also be useful when the subject string is very
49: long and is not all available at once.
50: </P>
51: <P>
52: PCRE supports partial matching by means of the PCRE_PARTIAL_SOFT and
53: PCRE_PARTIAL_HARD options, which can be set when calling any of the matching
54: functions. For backwards compatibility, PCRE_PARTIAL is a synonym for
55: PCRE_PARTIAL_SOFT. The essential difference between the two options is whether
56: or not a partial match is preferred to an alternative complete match, though
57: the details differ between the two types of matching function. If both options
58: are set, PCRE_PARTIAL_HARD takes precedence.
59: </P>
60: <P>
61: If you want to use partial matching with just-in-time optimized code, you must
62: call <b>pcre_study()</b>, <b>pcre16_study()</b> or <b>pcre32_study()</b> with one
63: or both of these options:
64: <pre>
65: PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE
66: PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE
67: </pre>
68: PCRE_STUDY_JIT_COMPILE should also be set if you are going to run non-partial
69: matches on the same pattern. If the appropriate JIT study mode has not been set
70: for a match, the interpretive matching code is used.
71: </P>
72: <P>
73: Setting a partial matching option disables two of PCRE's standard
74: optimizations. PCRE remembers the last literal data unit in a pattern, and
75: abandons matching immediately if it is not present in the subject string. This
76: optimization cannot be used for a subject string that might match only
77: partially. If the pattern was studied, PCRE knows the minimum length of a
78: matching string, and does not bother to run the matching function on shorter
79: strings. This optimization is also disabled for partial matching.
80: </P>
81: <br><a name="SEC2" href="#TOC1">PARTIAL MATCHING USING pcre_exec() OR pcre[16|32]_exec()</a><br>
82: <P>
83: A partial match occurs during a call to <b>pcre_exec()</b> or
84: <b>pcre[16|32]_exec()</b> when the end of the subject string is reached
85: successfully, but matching cannot continue because more characters are needed.
86: However, at least one character in the subject must have been inspected. This
87: character need not form part of the final matched string; lookbehind assertions
88: and the \K escape sequence provide ways of inspecting characters before the
89: start of a matched substring. The requirement for inspecting at least one
90: character exists because an empty string can always be matched; without such a
91: restriction there would always be a partial match of an empty string at the end
92: of the subject.
93: </P>
94: <P>
95: If there are at least two slots in the offsets vector when a partial match is
96: returned, the first slot is set to the offset of the earliest character that
97: was inspected. For convenience, the second offset points to the end of the
98: subject so that a substring can easily be identified. If there are at least
99: three slots in the offsets vector, the third slot is set to the offset of the
100: character where matching started.
101: </P>
102: <P>
103: For the majority of patterns, the contents of the first and third slots will be
104: the same. However, for patterns that contain lookbehind assertions, or begin
105: with \b or \B, characters before the one where matching started may have been
106: inspected while carrying out the match. For example, consider this pattern:
107: <pre>
108: /(?<=abc)123/
109: </pre>
110: This pattern matches "123", but only if it is preceded by "abc". If the subject
111: string is "xyzabc12", the first two offsets after a partial match are for the
112: substring "abc12", because all these characters were inspected. However, the
113: third offset is set to 6, because that is the offset where matching began.
114: </P>
115: <P>
116: What happens when a partial match is identified depends on which of the two
117: partial matching options are set.
118: </P>
119: <br><b>
120: PCRE_PARTIAL_SOFT WITH pcre_exec() OR pcre[16|32]_exec()
121: </b><br>
122: <P>
123: If PCRE_PARTIAL_SOFT is set when <b>pcre_exec()</b> or <b>pcre[16|32]_exec()</b>
124: identifies a partial match, the partial match is remembered, but matching
125: continues as normal, and other alternatives in the pattern are tried. If no
126: complete match can be found, PCRE_ERROR_PARTIAL is returned instead of
127: PCRE_ERROR_NOMATCH.
128: </P>
129: <P>
130: This option is "soft" because it prefers a complete match over a partial match.
131: All the various matching items in a pattern behave as if the subject string is
132: potentially complete. For example, \z, \Z, and $ match at the end of the
133: subject, as normal, and for \b and \B the end of the subject is treated as a
134: non-alphanumeric.
135: </P>
136: <P>
137: If there is more than one partial match, the first one that was found provides
138: the data that is returned. Consider this pattern:
139: <pre>
140: /123\w+X|dogY/
141: </pre>
142: If this is matched against the subject string "abc123dog", both
143: alternatives fail to match, but the end of the subject is reached during
144: matching, so PCRE_ERROR_PARTIAL is returned. The offsets are set to 3 and 9,
145: identifying "123dog" as the first partial match that was found. (In this
146: example, there are two partial matches, because "dog" on its own partially
147: matches the second alternative.)
148: </P>
149: <br><b>
150: PCRE_PARTIAL_HARD WITH pcre_exec() OR pcre[16|32]_exec()
151: </b><br>
152: <P>
153: If PCRE_PARTIAL_HARD is set for <b>pcre_exec()</b> or <b>pcre[16|32]_exec()</b>,
154: PCRE_ERROR_PARTIAL is returned as soon as a partial match is found, without
155: continuing to search for possible complete matches. This option is "hard"
156: because it prefers an earlier partial match over a later complete match. For
157: this reason, the assumption is made that the end of the supplied subject string
158: may not be the true end of the available data, and so, if \z, \Z, \b, \B,
159: or $ are encountered at the end of the subject, the result is
160: PCRE_ERROR_PARTIAL, provided that at least one character in the subject has
161: been inspected.
162: </P>
163: <P>
164: Setting PCRE_PARTIAL_HARD also affects the way UTF-8 and UTF-16
165: subject strings are checked for validity. Normally, an invalid sequence
166: causes the error PCRE_ERROR_BADUTF8 or PCRE_ERROR_BADUTF16. However, in the
167: special case of a truncated character at the end of the subject,
168: PCRE_ERROR_SHORTUTF8 or PCRE_ERROR_SHORTUTF16 is returned when
169: PCRE_PARTIAL_HARD is set.
170: </P>
171: <br><b>
172: Comparing hard and soft partial matching
173: </b><br>
174: <P>
175: The difference between the two partial matching options can be illustrated by a
176: pattern such as:
177: <pre>
178: /dog(sbody)?/
179: </pre>
180: This matches either "dog" or "dogsbody", greedily (that is, it prefers the
181: longer string if possible). If it is matched against the string "dog" with
182: PCRE_PARTIAL_SOFT, it yields a complete match for "dog". However, if
183: PCRE_PARTIAL_HARD is set, the result is PCRE_ERROR_PARTIAL. On the other hand,
184: if the pattern is made ungreedy the result is different:
185: <pre>
186: /dog(sbody)??/
187: </pre>
188: In this case the result is always a complete match because that is found first,
189: and matching never continues after finding a complete match. It might be easier
190: to follow this explanation by thinking of the two patterns like this:
191: <pre>
192: /dog(sbody)?/ is the same as /dogsbody|dog/
193: /dog(sbody)??/ is the same as /dog|dogsbody/
194: </pre>
195: The second pattern will never match "dogsbody", because it will always find the
196: shorter match first.
197: </P>
198: <br><a name="SEC3" href="#TOC1">PARTIAL MATCHING USING pcre_dfa_exec() OR pcre[16|32]_dfa_exec()</a><br>
199: <P>
200: The DFA functions move along the subject string character by character, without
201: backtracking, searching for all possible matches simultaneously. If the end of
202: the subject is reached before the end of the pattern, there is the possibility
203: of a partial match, again provided that at least one character has been
204: inspected.
205: </P>
206: <P>
207: When PCRE_PARTIAL_SOFT is set, PCRE_ERROR_PARTIAL is returned only if there
208: have been no complete matches. Otherwise, the complete matches are returned.
209: However, if PCRE_PARTIAL_HARD is set, a partial match takes precedence over any
210: complete matches. The portion of the string that was inspected when the longest
211: partial match was found is set as the first matching string, provided there are
212: at least two slots in the offsets vector.
213: </P>
214: <P>
215: Because the DFA functions always search for all possible matches, and there is
216: no difference between greedy and ungreedy repetition, their behaviour is
217: different from the standard functions when PCRE_PARTIAL_HARD is set. Consider
218: the string "dog" matched against the ungreedy pattern shown above:
219: <pre>
220: /dog(sbody)??/
221: </pre>
222: Whereas the standard functions stop as soon as they find the complete match for
223: "dog", the DFA functions also find the partial match for "dogsbody", and so
224: return that when PCRE_PARTIAL_HARD is set.
225: </P>
226: <br><a name="SEC4" href="#TOC1">PARTIAL MATCHING AND WORD BOUNDARIES</a><br>
227: <P>
228: If a pattern ends with one of sequences \b or \B, which test for word
229: boundaries, partial matching with PCRE_PARTIAL_SOFT can give counter-intuitive
230: results. Consider this pattern:
231: <pre>
232: /\bcat\b/
233: </pre>
234: This matches "cat", provided there is a word boundary at either end. If the
235: subject string is "the cat", the comparison of the final "t" with a following
236: character cannot take place, so a partial match is found. However, normal
237: matching carries on, and \b matches at the end of the subject when the last
238: character is a letter, so a complete match is found. The result, therefore, is
239: <i>not</i> PCRE_ERROR_PARTIAL. Using PCRE_PARTIAL_HARD in this case does yield
240: PCRE_ERROR_PARTIAL, because then the partial match takes precedence.
241: </P>
242: <br><a name="SEC5" href="#TOC1">FORMERLY RESTRICTED PATTERNS</a><br>
243: <P>
244: For releases of PCRE prior to 8.00, because of the way certain internal
245: optimizations were implemented in the <b>pcre_exec()</b> function, the
246: PCRE_PARTIAL option (predecessor of PCRE_PARTIAL_SOFT) could not be used with
247: all patterns. From release 8.00 onwards, the restrictions no longer apply, and
248: partial matching with can be requested for any pattern.
249: </P>
250: <P>
251: Items that were formerly restricted were repeated single characters and
252: repeated metasequences. If PCRE_PARTIAL was set for a pattern that did not
253: conform to the restrictions, <b>pcre_exec()</b> returned the error code
254: PCRE_ERROR_BADPARTIAL (-13). This error code is no longer in use. The
255: PCRE_INFO_OKPARTIAL call to <b>pcre_fullinfo()</b> to find out if a compiled
256: pattern can be used for partial matching now always returns 1.
257: </P>
258: <br><a name="SEC6" href="#TOC1">EXAMPLE OF PARTIAL MATCHING USING PCRETEST</a><br>
259: <P>
260: If the escape sequence \P is present in a <b>pcretest</b> data line, the
261: PCRE_PARTIAL_SOFT option is used for the match. Here is a run of <b>pcretest</b>
262: that uses the date example quoted above:
263: <pre>
264: re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
265: data> 25jun04\P
266: 0: 25jun04
267: 1: jun
268: data> 25dec3\P
269: Partial match: 23dec3
270: data> 3ju\P
271: Partial match: 3ju
272: data> 3juj\P
273: No match
274: data> j\P
275: No match
276: </pre>
277: The first data string is matched completely, so <b>pcretest</b> shows the
278: matched substrings. The remaining four strings do not match the complete
279: pattern, but the first two are partial matches. Similar output is obtained
280: if DFA matching is used.
281: </P>
282: <P>
283: If the escape sequence \P is present more than once in a <b>pcretest</b> data
284: line, the PCRE_PARTIAL_HARD option is set for the match.
285: </P>
286: <br><a name="SEC7" href="#TOC1">MULTI-SEGMENT MATCHING WITH pcre_dfa_exec() OR pcre[16|32]_dfa_exec()</a><br>
287: <P>
288: When a partial match has been found using a DFA matching function, it is
289: possible to continue the match by providing additional subject data and calling
290: the function again with the same compiled regular expression, this time setting
291: the PCRE_DFA_RESTART option. You must pass the same working space as before,
292: because this is where details of the previous partial match are stored. Here is
293: an example using <b>pcretest</b>, using the \R escape sequence to set the
294: PCRE_DFA_RESTART option (\D specifies the use of the DFA matching function):
295: <pre>
296: re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
297: data> 23ja\P\D
298: Partial match: 23ja
299: data> n05\R\D
300: 0: n05
301: </pre>
302: The first call has "23ja" as the subject, and requests partial matching; the
303: second call has "n05" as the subject for the continued (restarted) match.
304: Notice that when the match is complete, only the last part is shown; PCRE does
305: not retain the previously partially-matched string. It is up to the calling
306: program to do that if it needs to.
307: </P>
308: <P>
309: You can set the PCRE_PARTIAL_SOFT or PCRE_PARTIAL_HARD options with
310: PCRE_DFA_RESTART to continue partial matching over multiple segments. This
311: facility can be used to pass very long subject strings to the DFA matching
312: functions.
313: </P>
314: <br><a name="SEC8" href="#TOC1">MULTI-SEGMENT MATCHING WITH pcre_exec() OR pcre[16|32]_exec()</a><br>
315: <P>
316: From release 8.00, the standard matching functions can also be used to do
317: multi-segment matching. Unlike the DFA functions, it is not possible to
318: restart the previous match with a new segment of data. Instead, new data must
319: be added to the previous subject string, and the entire match re-run, starting
320: from the point where the partial match occurred. Earlier data can be discarded.
321: </P>
322: <P>
323: It is best to use PCRE_PARTIAL_HARD in this situation, because it does not
324: treat the end of a segment as the end of the subject when matching \z, \Z,
325: \b, \B, and $. Consider an unanchored pattern that matches dates:
326: <pre>
327: re> /\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d/
328: data> The date is 23ja\P\P
329: Partial match: 23ja
330: </pre>
331: At this stage, an application could discard the text preceding "23ja", add on
332: text from the next segment, and call the matching function again. Unlike the
333: DFA matching functions, the entire matching string must always be available,
334: and the complete matching process occurs for each call, so more memory and more
335: processing time is needed.
336: </P>
337: <P>
338: <b>Note:</b> If the pattern contains lookbehind assertions, or \K, or starts
339: with \b or \B, the string that is returned for a partial match includes
340: characters that precede the start of what would be returned for a complete
341: match, because it contains all the characters that were inspected during the
342: partial match.
343: </P>
344: <br><a name="SEC9" href="#TOC1">ISSUES WITH MULTI-SEGMENT MATCHING</a><br>
345: <P>
346: Certain types of pattern may give problems with multi-segment matching,
347: whichever matching function is used.
348: </P>
349: <P>
350: 1. If the pattern contains a test for the beginning of a line, you need to pass
351: the PCRE_NOTBOL option when the subject string for any call does start at the
352: beginning of a line. There is also a PCRE_NOTEOL option, but in practice when
353: doing multi-segment matching you should be using PCRE_PARTIAL_HARD, which
354: includes the effect of PCRE_NOTEOL.
355: </P>
356: <P>
357: 2. Lookbehind assertions that have already been obeyed are catered for in the
358: offsets that are returned for a partial match. However a lookbehind assertion
359: later in the pattern could require even earlier characters to be inspected. You
360: can handle this case by using the PCRE_INFO_MAXLOOKBEHIND option of the
361: <b>pcre_fullinfo()</b> or <b>pcre[16|32]_fullinfo()</b> functions to obtain the
362: length of the longest lookbehind in the pattern. This length is given in
363: characters, not bytes. If you always retain at least that many characters
364: before the partially matched string, all should be well. (Of course, near the
365: start of the subject, fewer characters may be present; in that case all
366: characters should be retained.)
367: </P>
368: <P>
369: From release 8.33, there is a more accurate way of deciding which characters to
370: retain. Instead of subtracting the length of the longest lookbehind from the
371: earliest inspected character (<i>offsets[0]</i>), the match start position
372: (<i>offsets[2]</i>) should be used, and the next match attempt started at the
373: <i>offsets[2]</i> character by setting the <i>startoffset</i> argument of
374: <b>pcre_exec()</b> or <b>pcre_dfa_exec()</b>.
375: </P>
376: <P>
377: For example, if the pattern "(?<=123)abc" is partially
378: matched against the string "xx123a", the three offset values returned are 2, 6,
379: and 5. This indicates that the matching process that gave a partial match
380: started at offset 5, but the characters "123a" were all inspected. The maximum
381: lookbehind for that pattern is 3, so taking that away from 5 shows that we need
382: only keep "123a", and the next match attempt can be started at offset 3 (that
383: is, at "a") when further characters have been added. When the match start is
384: not the earliest inspected character, <b>pcretest</b> shows it explicitly:
385: <pre>
386: re> "(?<=123)abc"
387: data> xx123a\P\P
388: Partial match at offset 5: 123a
389: </PRE>
390: </P>
391: <P>
392: 3. Because a partial match must always contain at least one character, what
393: might be considered a partial match of an empty string actually gives a "no
394: match" result. For example:
395: <pre>
396: re> /c(?<=abc)x/
397: data> ab\P
398: No match
399: </pre>
400: If the next segment begins "cx", a match should be found, but this will only
401: happen if characters from the previous segment are retained. For this reason, a
402: "no match" result should be interpreted as "partial match of an empty string"
403: when the pattern contains lookbehinds.
404: </P>
405: <P>
406: 4. Matching a subject string that is split into multiple segments may not
407: always produce exactly the same result as matching over one single long string,
408: especially when PCRE_PARTIAL_SOFT is used. The section "Partial Matching and
409: Word Boundaries" above describes an issue that arises if the pattern ends with
410: \b or \B. Another kind of difference may occur when there are multiple
411: matching possibilities, because (for PCRE_PARTIAL_SOFT) a partial match result
412: is given only when there are no completed matches. This means that as soon as
413: the shortest match has been found, continuation to a new subject segment is no
414: longer possible. Consider again this <b>pcretest</b> example:
415: <pre>
416: re> /dog(sbody)?/
417: data> dogsb\P
418: 0: dog
419: data> do\P\D
420: Partial match: do
421: data> gsb\R\P\D
422: 0: g
423: data> dogsbody\D
424: 0: dogsbody
425: 1: dog
426: </pre>
427: The first data line passes the string "dogsb" to a standard matching function,
428: setting the PCRE_PARTIAL_SOFT option. Although the string is a partial match
429: for "dogsbody", the result is not PCRE_ERROR_PARTIAL, because the shorter
430: string "dog" is a complete match. Similarly, when the subject is presented to
431: a DFA matching function in several parts ("do" and "gsb" being the first two)
432: the match stops when "dog" has been found, and it is not possible to continue.
433: On the other hand, if "dogsbody" is presented as a single string, a DFA
434: matching function finds both matches.
435: </P>
436: <P>
437: Because of these problems, it is best to use PCRE_PARTIAL_HARD when matching
438: multi-segment data. The example above then behaves differently:
439: <pre>
440: re> /dog(sbody)?/
441: data> dogsb\P\P
442: Partial match: dogsb
443: data> do\P\D
444: Partial match: do
445: data> gsb\R\P\P\D
446: Partial match: gsb
447: </pre>
448: 5. Patterns that contain alternatives at the top level which do not all start
449: with the same pattern item may not work as expected when PCRE_DFA_RESTART is
450: used. For example, consider this pattern:
451: <pre>
452: 1234|3789
453: </pre>
454: If the first part of the subject is "ABC123", a partial match of the first
455: alternative is found at offset 3. There is no partial match for the second
456: alternative, because such a match does not start at the same point in the
457: subject string. Attempting to continue with the string "7890" does not yield a
458: match because only those alternatives that match at one point in the subject
459: are remembered. The problem arises because the start of the second alternative
460: matches within the first alternative. There is no problem with anchored
461: patterns or patterns such as:
462: <pre>
463: 1234|ABCD
464: </pre>
465: where no string can be a partial match for both alternatives. This is not a
466: problem if a standard matching function is used, because the entire match has
467: to be rerun each time:
468: <pre>
469: re> /1234|3789/
470: data> ABC123\P\P
471: Partial match: 123
472: data> 1237890
473: 0: 3789
474: </pre>
475: Of course, instead of using PCRE_DFA_RESTART, the same technique of re-running
476: the entire match can also be used with the DFA matching functions. Another
477: possibility is to work with two buffers. If a partial match at offset <i>n</i>
478: in the first buffer is followed by "no match" when PCRE_DFA_RESTART is used on
479: the second buffer, you can then try a new match starting at offset <i>n+1</i> in
480: the first buffer.
481: </P>
482: <br><a name="SEC10" href="#TOC1">AUTHOR</a><br>
483: <P>
484: Philip Hazel
485: <br>
486: University Computing Service
487: <br>
488: Cambridge CB2 3QH, England.
489: <br>
490: </P>
491: <br><a name="SEC11" href="#TOC1">REVISION</a><br>
492: <P>
493: Last updated: 20 February 2013
494: <br>
495: Copyright © 1997-2013 University of Cambridge.
496: <br>
497: <p>
498: Return to the <a href="index.html">PCRE index page</a>.
499: </p>
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>