Diff for /embedaddon/pcre/doc/pcrepartial.3 between versions 1.1.1.1 and 1.1.1.4

version 1.1.1.1, 2012/02/21 23:05:52 version 1.1.1.4, 2013/07/22 08:25:56
Line 1 Line 1
.TH PCREPARTIAL 3.TH PCREPARTIAL 3 "20 February 2013" "PCRE 8.33"
 .SH NAME  .SH NAME
 PCRE - Perl-compatible regular expressions  PCRE - Perl-compatible regular expressions
 .SH "PARTIAL MATCHING IN PCRE"  .SH "PARTIAL MATCHING IN PCRE"
 .rs  .rs
 .sp  .sp
In normal use of PCRE, if the subject string that is passed toIn normal use of PCRE, if the subject string that is passed to a matching
\fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP matches as far as it goes, but isfunction matches as far as it goes, but is too short to match the entire
too short to match the entire pattern, PCRE_ERROR_NOMATCH is returned. Therepattern, PCRE_ERROR_NOMATCH is returned. There are circumstances where it might
are circumstances where it might be helpful to distinguish this case from otherbe helpful to distinguish this case from other cases in which there is no
cases in which there is no match.match.
 .P  .P
 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
 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 25  entered. Partial matching can also be useful when the  Line 25  entered. Partial matching can also be useful when the 
 long and is not all available at once.  long and is not all available at once.
 .P  .P
 PCRE supports partial matching by means of the PCRE_PARTIAL_SOFT and  PCRE supports partial matching by means of the PCRE_PARTIAL_SOFT and
PCRE_PARTIAL_HARD options, which can be set when calling \fBpcre_exec()\fP orPCRE_PARTIAL_HARD options, which can be set when calling any of the matching
\fBpcre_dfa_exec()\fP. For backwards compatibility, PCRE_PARTIAL is a synonymfunctions. For backwards compatibility, PCRE_PARTIAL is a synonym for
for PCRE_PARTIAL_SOFT. The essential difference between the two options isPCRE_PARTIAL_SOFT. The essential difference between the two options is whether
whether or not a partial match is preferred to an alternative complete match,or not a partial match is preferred to an alternative complete match, though
though the details differ between the two matching functions. If both optionsthe details differ between the two types of matching function. If both options
 are set, PCRE_PARTIAL_HARD takes precedence.  are set, PCRE_PARTIAL_HARD takes precedence.
 .P  .P
Setting a partial matching option for \fBpcre_exec()\fP disables the use of anyIf you want to use partial matching with just-in-time optimized code, you must
just-in-time code that was set up by calling \fBpcre_study()\fP with thecall \fBpcre_study()\fP, \fBpcre16_study()\fP or  \fBpcre32_study()\fP with one
PCRE_STUDY_JIT_COMPILE option. It also disables two of PCRE's standardor both of these options:
optimizations. PCRE remembers the last literal byte in a pattern, and abandons.sp
matching immediately if such a byte is not present in the subject string. This  PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE
   PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE
 .sp
 PCRE_STUDY_JIT_COMPILE should also be set if you are going to run non-partial
 matches on the same pattern. If the appropriate JIT study mode has not been set
 for a match, the interpretive matching code is used.
 .P
 Setting a partial matching option disables two of PCRE's standard
 optimizations. PCRE remembers the last literal data unit in a pattern, and
 abandons matching immediately if it is not present in the subject string. This
 optimization cannot be used for a subject string that might match only  optimization cannot be used for a subject string that might match only
 partially. If the pattern was studied, PCRE knows the minimum length of a  partially. If the pattern was studied, PCRE knows the minimum length of a
 matching string, and does not bother to run the matching function on shorter  matching string, and does not bother to run the matching function on shorter
 strings. This optimization is also disabled for partial matching.  strings. This optimization is also disabled for partial matching.
 .  .
 .  .
.SH "PARTIAL MATCHING USING pcre_exec()".SH "PARTIAL MATCHING USING pcre_exec() OR pcre[16|32]_exec()"
 .rs  .rs
 .sp  .sp
A partial match occurs during a call to \fBpcre_exec()\fP when the end of theA partial match occurs during a call to \fBpcre_exec()\fP or
subject string is reached successfully, but matching cannot continue because\fBpcre[16|32]_exec()\fP when the end of the subject string is reached
more characters are needed. However, at least one character in the subject mustsuccessfully, but matching cannot continue because more characters are needed.
have been inspected. This character need not form part of the final matchedHowever, at least one character in the subject must have been inspected. This
string; lookbehind assertions and the \eK escape sequence provide ways ofcharacter need not form part of the final matched string; lookbehind assertions
inspecting characters before the start of a matched substring. The requirementand the \eK escape sequence provide ways of inspecting characters before the
for inspecting at least one character exists because an empty string can alwaysstart of a matched substring. The requirement for inspecting at least one
be matched; without such a restriction there would always be a partial match ofcharacter exists because an empty string can always be matched; without such a
an empty string at the end of the subject.restriction there would always be a partial match of an empty string at the end
 of the subject.
 .P  .P
If there are at least two slots in the offsets vector when \fBpcre_exec()\fPIf there are at least two slots in the offsets vector when a partial match is
returns with a partial match, the first slot is set to the offset of thereturned, the first slot is set to the offset of the earliest character that
earliest character that was inspected when the partial match was found. Forwas inspected. For convenience, the second offset points to the end of the
convenience, the second offset points to the end of the subject so that asubject so that a substring can easily be identified. If there are at least
substring can easily be identified.three slots in the offsets vector, the third slot is set to the offset of the
 character where matching started.
 .P  .P
For the majority of patterns, the first offset identifies the start of theFor the majority of patterns, the contents of the first and third slots will be
partially matched string. However, for patterns that contain lookbehindthe same. However, for patterns that contain lookbehind assertions, or begin
assertions, or \eK, or begin with \eb or \eB, earlier characters have beenwith \eb or \eB, characters before the one where matching started may have been
inspected while carrying out the match. For example:inspected while carrying out the match. For example, consider this pattern:
 .sp  .sp
   /(?<=abc)123/    /(?<=abc)123/
 .sp  .sp
 This pattern matches "123", but only if it is preceded by "abc". If the subject  This pattern matches "123", but only if it is preceded by "abc". If the subject
string is "xyzabc12", the offsets after a partial match are for the substringstring is "xyzabc12", the first two offsets after a partial match are for the
"abc12", because all these characters are needed if another match is triedsubstring "abc12", because all these characters were inspected. However, the
with extra characters added to the subject.third offset is set to 6, because that is the offset where matching began.
 .P  .P
 What happens when a partial match is identified depends on which of the two  What happens when a partial match is identified depends on which of the two
 partial matching options are set.  partial matching options are set.
 .  .
 .  .
.SS "PCRE_PARTIAL_SOFT with pcre_exec()".SS "PCRE_PARTIAL_SOFT WITH pcre_exec() OR pcre[16|32]_exec()"
 .rs  .rs
 .sp  .sp
If PCRE_PARTIAL_SOFT is set when \fBpcre_exec()\fP identifies a partial match,If PCRE_PARTIAL_SOFT is set when \fBpcre_exec()\fP or \fBpcre[16|32]_exec()\fP
the partial match is remembered, but matching continues as normal, and otheridentifies a partial match, the partial match is remembered, but matching
alternatives in the pattern are tried. If no complete match can be found,continues as normal, and other alternatives in the pattern are tried. If no
\fBpcre_exec()\fP returns PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH.complete match can be found, PCRE_ERROR_PARTIAL is returned instead of
 PCRE_ERROR_NOMATCH.
 .P  .P
 This option is "soft" because it prefers a complete match over a partial match.  This option is "soft" because it prefers a complete match over a partial match.
 All the various matching items in a pattern behave as if the subject string is  All the various matching items in a pattern behave as if the subject string is
Line 105  example, there are two partial matches, because "dog"  Line 117  example, there are two partial matches, because "dog" 
 matches the second alternative.)  matches the second alternative.)
 .  .
 .  .
.SS "PCRE_PARTIAL_HARD with pcre_exec()".SS "PCRE_PARTIAL_HARD WITH pcre_exec() OR pcre[16|32]_exec()"
 .rs  .rs
 .sp  .sp
If PCRE_PARTIAL_HARD is set for \fBpcre_exec()\fP, it returnsIf PCRE_PARTIAL_HARD is set for \fBpcre_exec()\fP or \fBpcre[16|32]_exec()\fP,
PCRE_ERROR_PARTIAL as soon as a partial match is found, without continuing toPCRE_ERROR_PARTIAL is returned as soon as a partial match is found, without
search for possible complete matches. This option is "hard" because it preferscontinuing to search for possible complete matches. This option is "hard"
an earlier partial match over a later complete match. For this reason, thebecause it prefers an earlier partial match over a later complete match. For
assumption is made that the end of the supplied subject string may not be thethis reason, the assumption is made that the end of the supplied subject string
true end of the available data, and so, if \ez, \eZ, \eb, \eB, or $ aremay not be the true end of the available data, and so, if \ez, \eZ, \eb, \eB,
encountered at the end of the subject, the result is PCRE_ERROR_PARTIAL.or $ are encountered at the end of the subject, the result is
 PCRE_ERROR_PARTIAL, provided that at least one character in the subject has
 been inspected.
 .P  .P
Setting PCRE_PARTIAL_HARD also affects the way \fBpcre_exec()\fP checks UTF-8Setting PCRE_PARTIAL_HARD also affects the way UTF-8 and UTF-16
subject strings for validity. Normally, an invalid UTF-8 sequence causes thesubject strings are checked for validity. Normally, an invalid sequence
error PCRE_ERROR_BADUTF8. However, in the special case of a truncated UTF-8causes the error PCRE_ERROR_BADUTF8 or PCRE_ERROR_BADUTF16. However, in the
character at the end of the subject, PCRE_ERROR_SHORTUTF8 is returned whenspecial case of a truncated character at the end of the subject,
 PCRE_ERROR_SHORTUTF8 or PCRE_ERROR_SHORTUTF16 is returned when
 PCRE_PARTIAL_HARD is set.  PCRE_PARTIAL_HARD is set.
 .  .
 .  .
Line 139  if the pattern is made ungreedy the result is differen Line 154  if the pattern is made ungreedy the result is differen
 .sp  .sp
   /dog(sbody)??/    /dog(sbody)??/
 .sp  .sp
In this case the result is always a complete match because \fBpcre_exec()\fPIn this case the result is always a complete match because that is found first,
finds that first, and it never continues after finding a match. It might beand matching never continues after finding a complete match. It might be easier
easier to follow this explanation by thinking of the two patterns like this:to follow this explanation by thinking of the two patterns like this:
 .sp  .sp
   /dog(sbody)?/    is the same as  /dogsbody|dog/    /dog(sbody)?/    is the same as  /dogsbody|dog/
   /dog(sbody)??/   is the same as  /dog|dogsbody/    /dog(sbody)??/   is the same as  /dog|dogsbody/
 .sp  .sp
The second pattern will never match "dogsbody" when \fBpcre_exec()\fP isThe second pattern will never match "dogsbody", because it will always find the
used, because it will always find the shorter match first.shorter match first.
 .  .
 .  .
.SH "PARTIAL MATCHING USING pcre_dfa_exec()".SH "PARTIAL MATCHING USING pcre_dfa_exec() OR pcre[16|32]_dfa_exec()"
 .rs  .rs
 .sp  .sp
The \fBpcre_dfa_exec()\fP function moves along the subject string character byThe DFA functions move along the subject string character by character, without
character, without backtracking, searching for all possible matchesbacktracking, searching for all possible matches simultaneously. If the end of
simultaneously. If the end of the subject is reached before the end of thethe subject is reached before the end of the pattern, there is the possibility
pattern, there is the possibility of a partial match, again provided that atof a partial match, again provided that at least one character has been
least one character has been inspected.inspected.
 .P  .P
 When PCRE_PARTIAL_SOFT is set, PCRE_ERROR_PARTIAL is returned only if there  When PCRE_PARTIAL_SOFT is set, PCRE_ERROR_PARTIAL is returned only if there
 have been no complete matches. Otherwise, the complete matches are returned.  have been no complete matches. Otherwise, the complete matches are returned.
Line 166  complete matches. The portion of the string that was i Line 181  complete matches. The portion of the string that was i
 partial match was found is set as the first matching string, provided there are  partial match was found is set as the first matching string, provided there are
 at least two slots in the offsets vector.  at least two slots in the offsets vector.
 .P  .P
Because \fBpcre_dfa_exec()\fP always searches for all possible matches, andBecause the DFA functions always search for all possible matches, and there is
there is no difference between greedy and ungreedy repetition, its behaviour isno difference between greedy and ungreedy repetition, their behaviour is
different from \fBpcre_exec\fP when PCRE_PARTIAL_HARD is set. Consider thedifferent from the standard functions when PCRE_PARTIAL_HARD is set. Consider
string "dog" matched against the ungreedy pattern shown above:the string "dog" matched against the ungreedy pattern shown above:
 .sp  .sp
   /dog(sbody)??/    /dog(sbody)??/
 .sp  .sp
Whereas \fBpcre_exec()\fP stops as soon as it finds the complete match forWhereas the standard functions stop as soon as they find the complete match for
"dog", \fBpcre_dfa_exec()\fP also finds the partial match for "dogsbody", and"dog", the DFA functions also find the partial match for "dogsbody", and so
so returns that when PCRE_PARTIAL_HARD is set.return that when PCRE_PARTIAL_HARD is set.
 .  .
 .  .
 .SH "PARTIAL MATCHING AND WORD BOUNDARIES"  .SH "PARTIAL MATCHING AND WORD BOUNDARIES"
Line 189  results. Consider this pattern: Line 204  results. Consider this pattern:
 .sp  .sp
 This matches "cat", provided there is a word boundary at either end. If the  This matches "cat", provided there is a word boundary at either end. If the
 subject string is "the cat", the comparison of the final "t" with a following  subject string is "the cat", the comparison of the final "t" with a following
character cannot take place, so a partial match is found. However,character cannot take place, so a partial match is found. However, normal
\fBpcre_exec()\fP carries on with normal matching, which matches \eb at the endmatching carries on, and \eb matches at the end of the subject when the last
of the subject when the last character is a letter, thus finding a completecharacter is a letter, so a complete match is found. The result, therefore, is
match. The result, therefore, is \fInot\fP PCRE_ERROR_PARTIAL. The same thing\fInot\fP PCRE_ERROR_PARTIAL. Using PCRE_PARTIAL_HARD in this case does yield
happens with \fBpcre_dfa_exec()\fP, because it also finds the complete match.PCRE_ERROR_PARTIAL, because then the partial match takes precedence.
.P 
Using PCRE_PARTIAL_HARD in this case does yield PCRE_ERROR_PARTIAL, because 
then the partial match takes precedence. 
 .  .
 .  .
 .SH "FORMERLY RESTRICTED PATTERNS"  .SH "FORMERLY RESTRICTED PATTERNS"
Line 206  For releases of PCRE prior to 8.00, because of the way Line 218  For releases of PCRE prior to 8.00, because of the way
 optimizations were implemented in the \fBpcre_exec()\fP function, the  optimizations were implemented in the \fBpcre_exec()\fP function, the
 PCRE_PARTIAL option (predecessor of PCRE_PARTIAL_SOFT) could not be used with  PCRE_PARTIAL option (predecessor of PCRE_PARTIAL_SOFT) could not be used with
 all patterns. From release 8.00 onwards, the restrictions no longer apply, and  all patterns. From release 8.00 onwards, the restrictions no longer apply, and
partial matching with \fBpcre_exec()\fP can be requested for any pattern.partial matching with can be requested for any pattern.
 .P  .P
 Items that were formerly restricted were repeated single characters and  Items that were formerly restricted were repeated single characters and
 repeated metasequences. If PCRE_PARTIAL was set for a pattern that did not  repeated metasequences. If PCRE_PARTIAL was set for a pattern that did not
Line 239  that uses the date example quoted above: Line 251  that uses the date example quoted above:
 The first data string is matched completely, so \fBpcretest\fP shows the  The first data string is matched completely, so \fBpcretest\fP shows the
 matched substrings. The remaining four strings do not match the complete  matched substrings. The remaining four strings do not match the complete
 pattern, but the first two are partial matches. Similar output is obtained  pattern, but the first two are partial matches. Similar output is obtained
when \fBpcre_dfa_exec()\fP is used.if DFA matching is used.
 .P  .P
 If the escape sequence \eP is present more than once in a \fBpcretest\fP data  If the escape sequence \eP is present more than once in a \fBpcretest\fP data
 line, the PCRE_PARTIAL_HARD option is set for the match.  line, the PCRE_PARTIAL_HARD option is set for the match.
 .  .
 .  .
.SH "MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()".SH "MULTI-SEGMENT MATCHING WITH pcre_dfa_exec() OR pcre[16|32]_dfa_exec()"
 .rs  .rs
 .sp  .sp
When a partial match has been found using \fBpcre_dfa_exec()\fP, it is possibleWhen a partial match has been found using a DFA matching function, it is
to continue the match by providing additional subject data and callingpossible to continue the match by providing additional subject data and calling
\fBpcre_dfa_exec()\fP again with the same compiled regular expression, thisthe function again with the same compiled regular expression, this time setting
time setting the PCRE_DFA_RESTART option. You must pass the same workingthe PCRE_DFA_RESTART option. You must pass the same working space as before,
space as before, because this is where details of the previous partial matchbecause this is where details of the previous partial match are stored. Here is
are stored. Here is an example using \fBpcretest\fP, using the \eR escapean example using \fBpcretest\fP, using the \eR escape sequence to set the
sequence to set the PCRE_DFA_RESTART option (\eD specifies the use ofPCRE_DFA_RESTART option (\eD specifies the use of the DFA matching function):
\fBpcre_dfa_exec()\fP): 
 .sp  .sp
     re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/      re> /^\ed?\ed(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\ed\ed$/
   data> 23ja\eP\eD    data> 23ja\eP\eD
Line 271  program to do that if it needs to. Line 282  program to do that if it needs to.
 .P  .P
 You can set the PCRE_PARTIAL_SOFT or PCRE_PARTIAL_HARD options with  You can set the PCRE_PARTIAL_SOFT or PCRE_PARTIAL_HARD options with
 PCRE_DFA_RESTART to continue partial matching over multiple segments. This  PCRE_DFA_RESTART to continue partial matching over multiple segments. This
facility can be used to pass very long subject strings tofacility can be used to pass very long subject strings to the DFA matching
\fBpcre_dfa_exec()\fP.functions.
 .  .
 .  .
.SH "MULTI-SEGMENT MATCHING WITH pcre_exec()".SH "MULTI-SEGMENT MATCHING WITH pcre_exec() OR pcre[16|32]_exec()"
 .rs  .rs
 .sp  .sp
From release 8.00, \fBpcre_exec()\fP can also be used to do multi-segmentFrom release 8.00, the standard matching functions can also be used to do
matching. Unlike \fBpcre_dfa_exec()\fP, it is not possible to restart themulti-segment matching. Unlike the DFA functions, it is not possible to
previous match with a new segment of data. Instead, new data must be added torestart the previous match with a new segment of data. Instead, new data must
the previous subject string, and the entire match re-run, starting from thebe added to the previous subject string, and the entire match re-run, starting
point where the partial match occurred. Earlier data can be discarded. It isfrom the point where the partial match occurred. Earlier data can be discarded.
best to use PCRE_PARTIAL_HARD in this situation, because it does not treat the.P
end of a segment as the end of the subject when matching \ez, \eZ, \eb, \eB,It is best to use PCRE_PARTIAL_HARD in this situation, because it does not
and $. Consider an unanchored pattern that matches dates:treat the end of a segment as the end of the subject when matching \ez, \eZ,
 \eb, \eB, and $. Consider an unanchored pattern that matches dates:
 .sp  .sp
     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/
   data> The date is 23ja\eP\eP    data> The date is 23ja\eP\eP
   Partial match: 23ja    Partial match: 23ja
 .sp  .sp
 At this stage, an application could discard the text preceding "23ja", add on  At this stage, an application could discard the text preceding "23ja", add on
text from the next segment, and call \fBpcre_exec()\fP again. Unliketext from the next segment, and call the matching function again. Unlike the
\fBpcre_dfa_exec()\fP, the entire matching string must always be available, andDFA matching functions, the entire matching string must always be available,
the complete matching process occurs for each call, so more memory and moreand the complete matching process occurs for each call, so more memory and more
 processing time is needed.  processing time is needed.
 .P  .P
 \fBNote:\fP If the pattern contains lookbehind assertions, or \eK, or starts  \fBNote:\fP If the pattern contains lookbehind assertions, or \eK, or starts
with \eb or \eB, the string that is returned for a partial match will includewith \eb or \eB, the string that is returned for a partial match includes
characters that precede the partially matched string itself, because these mustcharacters that precede the start of what would be returned for a complete
be retained when adding on more characters for a subsequent matching attempt.match, because it contains all the characters that were inspected during the
 partial match.
 .  .
 .  .
 .SH "ISSUES WITH MULTI-SEGMENT MATCHING"  .SH "ISSUES WITH MULTI-SEGMENT MATCHING"
Line 315  beginning of a line. There is also a PCRE_NOTEOL optio Line 328  beginning of a line. There is also a PCRE_NOTEOL optio
 doing multi-segment matching you should be using PCRE_PARTIAL_HARD, which  doing multi-segment matching you should be using PCRE_PARTIAL_HARD, which
 includes the effect of PCRE_NOTEOL.  includes the effect of PCRE_NOTEOL.
 .P  .P
2. Lookbehind assertions at the start of a pattern are catered for in the2. Lookbehind assertions that have already been obeyed are catered for in the
offsets that are returned for a partial match. However, in theory, a lookbehindoffsets that are returned for a partial match. However a lookbehind assertion
assertion later in the pattern could require even earlier characters to belater in the pattern could require even earlier characters to be inspected. You
inspected, and it might not have been reached when a partial match occurs. Thiscan handle this case by using the PCRE_INFO_MAXLOOKBEHIND option of the
is probably an extremely unlikely case; you could guard against it to a certain\fBpcre_fullinfo()\fP or \fBpcre[16|32]_fullinfo()\fP functions to obtain the
extent by always including extra characters at the start.length of the longest lookbehind in the pattern. This length is given in
 characters, not bytes. If you always retain at least that many characters
 before the partially matched string, all should be well. (Of course, near the
 start of the subject, fewer characters may be present; in that case all
 characters should be retained.)
 .P  .P
3. Matching a subject string that is split into multiple segments may notFrom release 8.33, there is a more accurate way of deciding which characters to
 retain. Instead of subtracting the length of the longest lookbehind from the
 earliest inspected character (\fIoffsets[0]\fP), the match start position
 (\fIoffsets[2]\fP) should be used, and the next match attempt started at the
 \fIoffsets[2]\fP character by setting the \fIstartoffset\fP argument of
 \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP.
 .P
 For example, if the pattern "(?<=123)abc" is partially
 matched against the string "xx123a", the three offset values returned are 2, 6,
 and 5. This indicates that the matching process that gave a partial match
 started at offset 5, but the characters "123a" were all inspected. The maximum
 lookbehind for that pattern is 3, so taking that away from 5 shows that we need
 only keep "123a", and the next match attempt can be started at offset 3 (that
 is, at "a") when further characters have been added. When the match start is
 not the earliest inspected character, \fBpcretest\fP shows it explicitly:
 .sp
     re> "(?<=123)abc"
   data> xx123a\eP\eP
   Partial match at offset 5: 123a
 .P
 3. Because a partial match must always contain at least one character, what
 might be considered a partial match of an empty string actually gives a "no
 match" result. For example:
 .sp
     re> /c(?<=abc)x/
   data> ab\eP
   No match
 .sp
 If the next segment begins "cx", a match should be found, but this will only
 happen if characters from the previous segment are retained. For this reason, a
 "no match" result should be interpreted as "partial match of an empty string"
 when the pattern contains lookbehinds.
 .P
 4. Matching a subject string that is split into multiple segments may not
 always produce exactly the same result as matching over one single long string,  always produce exactly the same result as matching over one single long string,
 especially when PCRE_PARTIAL_SOFT is used. The section "Partial Matching and  especially when PCRE_PARTIAL_SOFT is used. The section "Partial Matching and
 Word Boundaries" above describes an issue that arises if the pattern ends with  Word Boundaries" above describes an issue that arises if the pattern ends with
Line 343  longer possible. Consider again this \fBpcretest\fP ex Line 393  longer possible. Consider again this \fBpcretest\fP ex
    0: dogsbody     0: dogsbody
    1: dog     1: dog
 .sp  .sp
The first data line passes the string "dogsb" to \fBpcre_exec()\fP, setting theThe first data line passes the string "dogsb" to a standard matching function,
PCRE_PARTIAL_SOFT option. Although the string is a partial match forsetting the PCRE_PARTIAL_SOFT option. Although the string is a partial match
"dogsbody", the result is not PCRE_ERROR_PARTIAL, because the shorter stringfor "dogsbody", the result is not PCRE_ERROR_PARTIAL, because the shorter
"dog" is a complete match. Similarly, when the subject is presented tostring "dog" is a complete match. Similarly, when the subject is presented to
\fBpcre_dfa_exec()\fP in several parts ("do" and "gsb" being the first two) thea DFA matching function in several parts ("do" and "gsb" being the first two)
match stops when "dog" has been found, and it is not possible to continue. Onthe match stops when "dog" has been found, and it is not possible to continue.
the other hand, if "dogsbody" is presented as a single string,On the other hand, if "dogsbody" is presented as a single string, a DFA
\fBpcre_dfa_exec()\fP finds both matches.matching function finds both matches.
 .P  .P
 Because of these problems, it is best to use PCRE_PARTIAL_HARD when matching  Because of these problems, it is best to use PCRE_PARTIAL_HARD when matching
 multi-segment data. The example above then behaves differently:  multi-segment data. The example above then behaves differently:
Line 363  multi-segment data. The example above then behaves dif Line 413  multi-segment data. The example above then behaves dif
   data> gsb\eR\eP\eP\eD    data> gsb\eR\eP\eP\eD
   Partial match: gsb    Partial match: gsb
 .sp  .sp
4. Patterns that contain alternatives at the top level which do not all5. Patterns that contain alternatives at the top level which do not all start
start with the same pattern item may not work as expected whenwith the same pattern item may not work as expected when PCRE_DFA_RESTART is
PCRE_DFA_RESTART is used with \fBpcre_dfa_exec()\fP. For example, consider thisused. For example, consider this pattern:
pattern: 
 .sp  .sp
   1234|3789    1234|3789
 .sp  .sp
Line 382  patterns or patterns such as: Line 431  patterns or patterns such as:
   1234|ABCD    1234|ABCD
 .sp  .sp
 where no string can be a partial match for both alternatives. This is not a  where no string can be a partial match for both alternatives. This is not a
problem if \fBpcre_exec()\fP is used, because the entire match has to be rerunproblem if a standard matching function is used, because the entire match has
each time:to be rerun each time:
 .sp  .sp
     re> /1234|3789/      re> /1234|3789/
   data> ABC123\eP\eP    data> ABC123\eP\eP
Line 392  each time: Line 441  each time:
    0: 3789     0: 3789
 .sp  .sp
 Of course, instead of using PCRE_DFA_RESTART, the same technique of re-running  Of course, instead of using PCRE_DFA_RESTART, the same technique of re-running
the entire match can also be used with \fBpcre_dfa_exec()\fP. Anotherthe entire match can also be used with the DFA matching functions. Another
 possibility is to work with two buffers. If a partial match at offset \fIn\fP  possibility is to work with two buffers. If a partial match at offset \fIn\fP
 in the first buffer is followed by "no match" when PCRE_DFA_RESTART is used on  in the first buffer is followed by "no match" when PCRE_DFA_RESTART is used on
 the second buffer, you can then try a new match starting at offset \fIn+1\fP in  the second buffer, you can then try a new match starting at offset \fIn+1\fP in
Line 413  Cambridge CB2 3QH, England. Line 462  Cambridge CB2 3QH, England.
 .rs  .rs
 .sp  .sp
 .nf  .nf
Last updated: 26 August 2011Last updated: 20 February 2013
Copyright (c) 1997-2011 University of Cambridge.Copyright (c) 1997-2013 University of Cambridge.
 .fi  .fi

Removed from v.1.1.1.1  
changed lines
  Added in v.1.1.1.4


FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>