57 |
#undef min |
#undef min |
58 |
#undef max |
#undef max |
59 |
|
|
60 |
/* Values for setting in md->match_function_type to indicate two special types |
/* Values for setting in md->match_function_type to indicate two special types |
61 |
of call to match(). We do it this way to save on using another stack variable, |
of call to match(). We do it this way to save on using another stack variable, |
62 |
as stack usage is to be discouraged. */ |
as stack usage is to be discouraged. */ |
63 |
|
|
64 |
#define MATCH_CONDASSERT 1 /* Called to check a condition assertion */ |
#define MATCH_CONDASSERT 1 /* Called to check a condition assertion */ |
76 |
#define MATCH_ACCEPT (-999) |
#define MATCH_ACCEPT (-999) |
77 |
#define MATCH_COMMIT (-998) |
#define MATCH_COMMIT (-998) |
78 |
#define MATCH_KETRPOS (-997) |
#define MATCH_KETRPOS (-997) |
79 |
#define MATCH_PRUNE (-996) |
#define MATCH_ONCE (-996) |
80 |
#define MATCH_SKIP (-995) |
#define MATCH_PRUNE (-995) |
81 |
#define MATCH_SKIP_ARG (-994) |
#define MATCH_SKIP (-994) |
82 |
#define MATCH_THEN (-993) |
#define MATCH_SKIP_ARG (-993) |
83 |
|
#define MATCH_THEN (-992) |
84 |
|
|
85 |
/* This is a convenience macro for code that occurs many times. */ |
/* This is a convenience macro for code that occurs many times. */ |
86 |
|
|
138 |
|
|
139 |
/* Normally, if a back reference hasn't been set, the length that is passed is |
/* Normally, if a back reference hasn't been set, the length that is passed is |
140 |
negative, so the match always fails. However, in JavaScript compatibility mode, |
negative, so the match always fails. However, in JavaScript compatibility mode, |
141 |
the length passed is zero. Note that in caseless UTF-8 mode, the number of |
the length passed is zero. Note that in caseless UTF-8 mode, the number of |
142 |
subject bytes matched may be different to the number of reference bytes. |
subject bytes matched may be different to the number of reference bytes. |
143 |
|
|
144 |
Arguments: |
Arguments: |
185 |
#ifdef SUPPORT_UCP |
#ifdef SUPPORT_UCP |
186 |
if (md->utf8) |
if (md->utf8) |
187 |
{ |
{ |
188 |
/* Match characters up to the end of the reference. NOTE: the number of |
/* Match characters up to the end of the reference. NOTE: the number of |
189 |
bytes matched may differ, because there are some characters whose upper and |
bytes matched may differ, because there are some characters whose upper and |
190 |
lower case versions code as different numbers of bytes. For example, U+023A |
lower case versions code as different numbers of bytes. For example, U+023A |
191 |
(2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8); |
(2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8); |
192 |
a sequence of 3 of the former uses 6 bytes, as does a sequence of two of |
a sequence of 3 of the former uses 6 bytes, as does a sequence of two of |
193 |
the latter. It is important, therefore, to check the length along the |
the latter. It is important, therefore, to check the length along the |
194 |
reference, not along the subject (earlier code did this wrong). */ |
reference, not along the subject (earlier code did this wrong). */ |
195 |
|
|
196 |
USPTR endptr = p + length; |
USPTR endptr = p + length; |
197 |
while (p < endptr) |
while (p < endptr) |
198 |
{ |
{ |
210 |
/* The same code works when not in UTF-8 mode and in UTF-8 mode when there |
/* The same code works when not in UTF-8 mode and in UTF-8 mode when there |
211 |
is no UCP support. */ |
is no UCP support. */ |
212 |
{ |
{ |
213 |
if (eptr + length > md->end_subject) return -1; |
if (eptr + length > md->end_subject) return -1; |
214 |
while (length-- > 0) |
while (length-- > 0) |
215 |
{ if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; } |
{ if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; } |
216 |
} |
} |
217 |
} |
} |
218 |
|
|
219 |
/* In the caseful case, we can just compare the bytes, whether or not we |
/* In the caseful case, we can just compare the bytes, whether or not we |
220 |
are in UTF-8 mode. */ |
are in UTF-8 mode. */ |
221 |
|
|
222 |
else |
else |
223 |
{ |
{ |
224 |
if (eptr + length > md->end_subject) return -1; |
if (eptr + length > md->end_subject) return -1; |
225 |
while (length-- > 0) if (*p++ != *eptr++) return -1; |
while (length-- > 0) if (*p++ != *eptr++) return -1; |
226 |
} |
} |
227 |
|
|
228 |
return eptr - eptr_start; |
return eptr - eptr_start; |
277 |
RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, |
RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, |
278 |
RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, |
RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, |
279 |
RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, |
RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, |
280 |
RM61, RM62, RM63}; |
RM61, RM62, RM63 }; |
281 |
|
|
282 |
/* These versions of the macros use the stack, as normal. There are debugging |
/* These versions of the macros use the stack, as normal. There are debugging |
283 |
versions and production versions. Note that the "rw" argument of RMATCH isn't |
versions and production versions. Note that the "rw" argument of RMATCH isn't |
384 |
int Xprop_type; |
int Xprop_type; |
385 |
int Xprop_value; |
int Xprop_value; |
386 |
int Xprop_fail_result; |
int Xprop_fail_result; |
|
int Xprop_category; |
|
|
int Xprop_chartype; |
|
|
int Xprop_script; |
|
387 |
int Xoclength; |
int Xoclength; |
388 |
uschar Xocchars[8]; |
uschar Xocchars[8]; |
389 |
#endif |
#endif |
475 |
|
|
476 |
static int |
static int |
477 |
match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, |
match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, |
478 |
const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb, |
const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb, |
479 |
unsigned int rdepth) |
unsigned int rdepth) |
480 |
{ |
{ |
481 |
/* These variables do not need to be preserved over recursion in this function, |
/* These variables do not need to be preserved over recursion in this function, |
548 |
#define prop_type frame->Xprop_type |
#define prop_type frame->Xprop_type |
549 |
#define prop_value frame->Xprop_value |
#define prop_value frame->Xprop_value |
550 |
#define prop_fail_result frame->Xprop_fail_result |
#define prop_fail_result frame->Xprop_fail_result |
|
#define prop_category frame->Xprop_category |
|
|
#define prop_chartype frame->Xprop_chartype |
|
|
#define prop_script frame->Xprop_script |
|
551 |
#define oclength frame->Xoclength |
#define oclength frame->Xoclength |
552 |
#define occhars frame->Xocchars |
#define occhars frame->Xocchars |
553 |
#endif |
#endif |
585 |
below are for variables that do not have to be preserved over a recursive call |
below are for variables that do not have to be preserved over a recursive call |
586 |
to RMATCH(). */ |
to RMATCH(). */ |
587 |
|
|
588 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
589 |
const uschar *charptr; |
const uschar *charptr; |
590 |
#endif |
#endif |
591 |
const uschar *callpat; |
const uschar *callpat; |
592 |
const uschar *data; |
const uschar *data; |
593 |
const uschar *next; |
const uschar *next; |
594 |
USPTR pp; |
USPTR pp; |
595 |
const uschar *prev; |
const uschar *prev; |
596 |
USPTR saved_eptr; |
USPTR saved_eptr; |
597 |
|
|
598 |
recursion_info new_recursive; |
recursion_info new_recursive; |
599 |
|
|
600 |
BOOL cur_is_word; |
BOOL cur_is_word; |
601 |
BOOL condition; |
BOOL condition; |
602 |
BOOL prev_is_word; |
BOOL prev_is_word; |
603 |
|
|
605 |
int prop_type; |
int prop_type; |
606 |
int prop_value; |
int prop_value; |
607 |
int prop_fail_result; |
int prop_fail_result; |
|
int prop_category; |
|
|
int prop_chartype; |
|
|
int prop_script; |
|
608 |
int oclength; |
int oclength; |
609 |
uschar occhars[8]; |
uschar occhars[8]; |
610 |
#endif |
#endif |
624 |
eptrblock newptrb; |
eptrblock newptrb; |
625 |
#endif /* NO_RECURSE */ |
#endif /* NO_RECURSE */ |
626 |
|
|
627 |
/* To save space on the stack and in the heap frame, I have doubled up on some |
/* To save space on the stack and in the heap frame, I have doubled up on some |
628 |
of the local variables that are used only in localised parts of the code, but |
of the local variables that are used only in localised parts of the code, but |
629 |
still need to be preserved over recursive calls of match(). These macros define |
still need to be preserved over recursive calls of match(). These macros define |
630 |
the alternative names that are used. */ |
the alternative names that are used. */ |
631 |
|
|
632 |
#define allow_zero cur_is_word |
#define allow_zero cur_is_word |
672 |
if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT); |
if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT); |
673 |
|
|
674 |
/* At the start of a group with an unlimited repeat that may match an empty |
/* At the start of a group with an unlimited repeat that may match an empty |
675 |
string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is |
string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is |
676 |
done this way to save having to use another function argument, which would take |
done this way to save having to use another function argument, which would take |
677 |
up space on the stack. See also MATCH_CONDASSERT below. |
up space on the stack. See also MATCH_CONDASSERT below. |
678 |
|
|
679 |
When MATCH_CBEGROUP is set, add the current subject pointer to the chain of |
When MATCH_CBEGROUP is set, add the current subject pointer to the chain of |
697 |
{ |
{ |
698 |
minimize = possessive = FALSE; |
minimize = possessive = FALSE; |
699 |
op = *ecode; |
op = *ecode; |
700 |
|
|
701 |
switch(op) |
switch(op) |
702 |
{ |
{ |
703 |
case OP_MARK: |
case OP_MARK: |
800 |
subject position in the working slot at the top of the vector. We mustn't |
subject position in the working slot at the top of the vector. We mustn't |
801 |
change the current values of the data slot, because they may be set from a |
change the current values of the data slot, because they may be set from a |
802 |
previous iteration of this group, and be referred to by a reference inside |
previous iteration of this group, and be referred to by a reference inside |
803 |
the group. If we fail to match, we need to restore this value and also the |
the group. A failure to match might occur after the group has succeeded, |
804 |
values of the final offsets, in case they were set by a previous iteration |
if something later on doesn't match. For this reason, we need to restore |
805 |
of the same bracket. |
the working value and also the values of the final offsets, in case they |
806 |
|
were set by a previous iteration of the same bracket. |
807 |
|
|
808 |
If there isn't enough space in the offset vector, treat this as if it were |
If there isn't enough space in the offset vector, treat this as if it were |
809 |
a non-capturing bracket. Don't worry about setting the flag for the error |
a non-capturing bracket. Don't worry about setting the flag for the error |
813 |
case OP_SCBRA: |
case OP_SCBRA: |
814 |
number = GET2(ecode, 1+LINK_SIZE); |
number = GET2(ecode, 1+LINK_SIZE); |
815 |
offset = number << 1; |
offset = number << 1; |
816 |
|
|
817 |
#ifdef PCRE_DEBUG |
#ifdef PCRE_DEBUG |
818 |
printf("start bracket %d\n", number); |
printf("start bracket %d\n", number); |
819 |
printf("subject="); |
printf("subject="); |
834 |
|
|
835 |
for (;;) |
for (;;) |
836 |
{ |
{ |
837 |
if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
838 |
RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
839 |
eptrb, RM1); |
eptrb, RM1); |
840 |
|
if (rrc == MATCH_ONCE) break; /* Backing up through an atomic group */ |
841 |
if (rrc != MATCH_NOMATCH && |
if (rrc != MATCH_NOMATCH && |
842 |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
843 |
RRETURN(rrc); |
RRETURN(rrc); |
844 |
md->capture_last = save_capture_last; |
md->capture_last = save_capture_last; |
845 |
ecode += GET(ecode, 1); |
ecode += GET(ecode, 1); |
846 |
if (*ecode != OP_ALT) break; |
if (*ecode != OP_ALT) break; |
847 |
} |
} |
848 |
|
|
849 |
DPRINTF(("bracket %d failed\n", number)); |
DPRINTF(("bracket %d failed\n", number)); |
|
|
|
850 |
md->offset_vector[offset] = save_offset1; |
md->offset_vector[offset] = save_offset1; |
851 |
md->offset_vector[offset+1] = save_offset2; |
md->offset_vector[offset+1] = save_offset2; |
852 |
md->offset_vector[md->offset_end - number] = save_offset3; |
md->offset_vector[md->offset_end - number] = save_offset3; |
853 |
|
|
854 |
|
/* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or |
855 |
|
MATCH_THEN. */ |
856 |
|
|
857 |
if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr; |
if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr; |
858 |
RRETURN(MATCH_NOMATCH); |
RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH)); |
859 |
} |
} |
860 |
|
|
861 |
/* FALL THROUGH ... Insufficient room for saving captured contents. Treat |
/* FALL THROUGH ... Insufficient room for saving captured contents. Treat |
869 |
/* VVVVVVVVVVVVVVVVVVVVVVVVV */ |
/* VVVVVVVVVVVVVVVVVVVVVVVVV */ |
870 |
/* VVVVVVVVVVVVVVVVVVVVVVVVV */ |
/* VVVVVVVVVVVVVVVVVVVVVVVVV */ |
871 |
|
|
872 |
/* Non-capturing bracket, except for possessive with unlimited repeat. Loop |
/* Non-capturing or atomic group, except for possessive with unlimited |
873 |
for all the alternatives. When we get to the final alternative within the |
repeat. Loop for all the alternatives. |
874 |
brackets, we used to return the result of a recursive call to match() |
|
875 |
whatever happened so it was possible to reduce stack usage by turning this |
When we get to the final alternative within the brackets, we used to return |
876 |
into a tail recursion, except in the case of a possibly empty group. |
the result of a recursive call to match() whatever happened so it was |
877 |
However, now that there is the possiblity of (*THEN) occurring in the final |
possible to reduce stack usage by turning this into a tail recursion, |
878 |
alternative, this optimization is no longer possible. */ |
except in the case of a possibly empty group. However, now that there is |
879 |
|
the possiblity of (*THEN) occurring in the final alternative, this |
880 |
|
optimization is no longer always possible. |
881 |
|
|
882 |
|
We can optimize if we know there are no (*THEN)s in the pattern; at present |
883 |
|
this is the best that can be done. |
884 |
|
|
885 |
|
MATCH_ONCE is returned when the end of an atomic group is successfully |
886 |
|
reached, but subsequent matching fails. It passes back up the tree (causing |
887 |
|
captured values to be reset) until the original atomic group level is |
888 |
|
reached. This is tested by comparing md->once_target with the start of the |
889 |
|
group. At this point, the return is converted into MATCH_NOMATCH so that |
890 |
|
previous backup points can be taken. */ |
891 |
|
|
892 |
|
case OP_ONCE: |
893 |
case OP_BRA: |
case OP_BRA: |
894 |
case OP_SBRA: |
case OP_SBRA: |
895 |
DPRINTF(("start non-capturing bracket\n")); |
DPRINTF(("start non-capturing bracket\n")); |
896 |
|
|
897 |
for (;;) |
for (;;) |
898 |
{ |
{ |
899 |
if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP; |
900 |
RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb, |
|
901 |
|
/* If this is not a possibly empty group, and there are no (*THEN)s in |
902 |
|
the pattern, and this is the final alternative, optimize as described |
903 |
|
above. */ |
904 |
|
|
905 |
|
else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT) |
906 |
|
{ |
907 |
|
ecode += _pcre_OP_lengths[*ecode]; |
908 |
|
goto TAIL_RECURSE; |
909 |
|
} |
910 |
|
|
911 |
|
/* In all other cases, we have to make another call to match(). */ |
912 |
|
|
913 |
|
RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb, |
914 |
RM2); |
RM2); |
915 |
if (rrc != MATCH_NOMATCH && |
if (rrc != MATCH_NOMATCH && |
916 |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
917 |
|
{ |
918 |
|
if (rrc == MATCH_ONCE) |
919 |
|
{ |
920 |
|
const uschar *scode = ecode; |
921 |
|
if (*scode != OP_ONCE) /* If not at start, find it */ |
922 |
|
{ |
923 |
|
while (*scode == OP_ALT) scode += GET(scode, 1); |
924 |
|
scode -= GET(scode, 1); |
925 |
|
} |
926 |
|
if (md->once_target == scode) rrc = MATCH_NOMATCH; |
927 |
|
} |
928 |
RRETURN(rrc); |
RRETURN(rrc); |
929 |
|
} |
930 |
ecode += GET(ecode, 1); |
ecode += GET(ecode, 1); |
931 |
if (*ecode != OP_ALT) break; |
if (*ecode != OP_ALT) break; |
932 |
} |
} |
|
|
|
933 |
if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr; |
if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr; |
934 |
RRETURN(MATCH_NOMATCH); |
RRETURN(MATCH_NOMATCH); |
935 |
|
|
936 |
/* Handle possessive capturing brackets with an unlimited repeat. We come |
/* Handle possessive capturing brackets with an unlimited repeat. We come |
937 |
here from BRAZERO with allow_zero set TRUE. The offset_vector values are |
here from BRAZERO with allow_zero set TRUE. The offset_vector values are |
938 |
handled similarly to the normal case above. However, the matching is |
handled similarly to the normal case above. However, the matching is |
939 |
different. The end of these brackets will always be OP_KETRPOS, which |
different. The end of these brackets will always be OP_KETRPOS, which |
940 |
returns MATCH_KETRPOS without going further in the pattern. By this means |
returns MATCH_KETRPOS without going further in the pattern. By this means |
941 |
we can handle the group by iteration rather than recursion, thereby |
we can handle the group by iteration rather than recursion, thereby |
942 |
reducing the amount of stack needed. */ |
reducing the amount of stack needed. */ |
943 |
|
|
944 |
case OP_CBRAPOS: |
case OP_CBRAPOS: |
945 |
case OP_SCBRAPOS: |
case OP_SCBRAPOS: |
946 |
allow_zero = FALSE; |
allow_zero = FALSE; |
947 |
|
|
948 |
POSSESSIVE_CAPTURE: |
POSSESSIVE_CAPTURE: |
949 |
number = GET2(ecode, 1+LINK_SIZE); |
number = GET2(ecode, 1+LINK_SIZE); |
950 |
offset = number << 1; |
offset = number << 1; |
959 |
if (offset < md->offset_max) |
if (offset < md->offset_max) |
960 |
{ |
{ |
961 |
matched_once = FALSE; |
matched_once = FALSE; |
962 |
code_offset = ecode - md->start_code; |
code_offset = ecode - md->start_code; |
963 |
|
|
964 |
save_offset1 = md->offset_vector[offset]; |
save_offset1 = md->offset_vector[offset]; |
965 |
save_offset2 = md->offset_vector[offset+1]; |
save_offset2 = md->offset_vector[offset+1]; |
967 |
save_capture_last = md->capture_last; |
save_capture_last = md->capture_last; |
968 |
|
|
969 |
DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); |
DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); |
970 |
|
|
971 |
/* Each time round the loop, save the current subject position for use |
/* Each time round the loop, save the current subject position for use |
972 |
when the group matches. For MATCH_MATCH, the group has matched, so we |
when the group matches. For MATCH_MATCH, the group has matched, so we |
973 |
restart it with a new subject starting position, remembering that we had |
restart it with a new subject starting position, remembering that we had |
974 |
at least one match. For MATCH_NOMATCH, carry on with the alternatives, as |
at least one match. For MATCH_NOMATCH, carry on with the alternatives, as |
975 |
usual. If we haven't matched any alternatives in any iteration, check to |
usual. If we haven't matched any alternatives in any iteration, check to |
976 |
see if a previous iteration matched. If so, the group has matched; |
see if a previous iteration matched. If so, the group has matched; |
977 |
continue from afterwards. Otherwise it has failed; restore the previous |
continue from afterwards. Otherwise it has failed; restore the previous |
978 |
capture values before returning NOMATCH. */ |
capture values before returning NOMATCH. */ |
979 |
|
|
980 |
for (;;) |
for (;;) |
981 |
{ |
{ |
982 |
md->offset_vector[md->offset_end - number] = |
md->offset_vector[md->offset_end - number] = |
983 |
(int)(eptr - md->start_subject); |
(int)(eptr - md->start_subject); |
984 |
if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
985 |
RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
986 |
eptrb, RM63); |
eptrb, RM63); |
987 |
if (rrc == MATCH_KETRPOS) |
if (rrc == MATCH_KETRPOS) |
988 |
{ |
{ |
989 |
offset_top = md->end_offset_top; |
offset_top = md->end_offset_top; |
990 |
eptr = md->end_match_ptr; |
eptr = md->end_match_ptr; |
991 |
ecode = md->start_code + code_offset; |
ecode = md->start_code + code_offset; |
992 |
save_capture_last = md->capture_last; |
save_capture_last = md->capture_last; |
993 |
matched_once = TRUE; |
matched_once = TRUE; |
994 |
continue; |
continue; |
995 |
} |
} |
996 |
if (rrc != MATCH_NOMATCH && |
if (rrc != MATCH_NOMATCH && |
997 |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
998 |
RRETURN(rrc); |
RRETURN(rrc); |
999 |
md->capture_last = save_capture_last; |
md->capture_last = save_capture_last; |
1000 |
ecode += GET(ecode, 1); |
ecode += GET(ecode, 1); |
1001 |
if (*ecode != OP_ALT) break; |
if (*ecode != OP_ALT) break; |
1002 |
} |
} |
1003 |
|
|
1004 |
if (!matched_once) |
if (!matched_once) |
1005 |
{ |
{ |
1006 |
md->offset_vector[offset] = save_offset1; |
md->offset_vector[offset] = save_offset1; |
1007 |
md->offset_vector[offset+1] = save_offset2; |
md->offset_vector[offset+1] = save_offset2; |
1008 |
md->offset_vector[md->offset_end - number] = save_offset3; |
md->offset_vector[md->offset_end - number] = save_offset3; |
1009 |
} |
} |
1010 |
|
|
1011 |
if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr; |
if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr; |
1012 |
if (allow_zero || matched_once) |
if (allow_zero || matched_once) |
1013 |
{ |
{ |
1014 |
ecode += 1 + LINK_SIZE; |
ecode += 1 + LINK_SIZE; |
1015 |
break; |
break; |
1016 |
} |
} |
1017 |
|
|
1018 |
RRETURN(MATCH_NOMATCH); |
RRETURN(MATCH_NOMATCH); |
1019 |
} |
} |
1020 |
|
|
1021 |
/* FALL THROUGH ... Insufficient room for saving captured contents. Treat |
/* FALL THROUGH ... Insufficient room for saving captured contents. Treat |
1022 |
as a non-capturing bracket. */ |
as a non-capturing bracket. */ |
1023 |
|
|
1029 |
/* VVVVVVVVVVVVVVVVVVVVVVVVV */ |
/* VVVVVVVVVVVVVVVVVVVVVVVVV */ |
1030 |
/* VVVVVVVVVVVVVVVVVVVVVVVVV */ |
/* VVVVVVVVVVVVVVVVVVVVVVVVV */ |
1031 |
|
|
1032 |
/* Non-capturing possessive bracket with unlimited repeat. We come here |
/* Non-capturing possessive bracket with unlimited repeat. We come here |
1033 |
from BRAZERO with allow_zero = TRUE. The code is similar to the above, |
from BRAZERO with allow_zero = TRUE. The code is similar to the above, |
1034 |
without the capturing complication. It is written out separately for speed |
without the capturing complication. It is written out separately for speed |
1035 |
and cleanliness. */ |
and cleanliness. */ |
1036 |
|
|
1037 |
case OP_BRAPOS: |
case OP_BRAPOS: |
1038 |
case OP_SBRAPOS: |
case OP_SBRAPOS: |
1039 |
allow_zero = FALSE; |
allow_zero = FALSE; |
1040 |
|
|
1041 |
POSSESSIVE_NON_CAPTURE: |
POSSESSIVE_NON_CAPTURE: |
1042 |
matched_once = FALSE; |
matched_once = FALSE; |
1043 |
code_offset = ecode - md->start_code; |
code_offset = ecode - md->start_code; |
1044 |
|
|
1045 |
for (;;) |
for (;;) |
1046 |
{ |
{ |
1047 |
if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
1048 |
RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
1049 |
eptrb, RM48); |
eptrb, RM48); |
1050 |
if (rrc == MATCH_KETRPOS) |
if (rrc == MATCH_KETRPOS) |
1051 |
{ |
{ |
1052 |
offset_top = md->end_offset_top; |
offset_top = md->end_offset_top; |
1053 |
eptr = md->end_match_ptr; |
eptr = md->end_match_ptr; |
1054 |
ecode = md->start_code + code_offset; |
ecode = md->start_code + code_offset; |
1055 |
matched_once = TRUE; |
matched_once = TRUE; |
1056 |
continue; |
continue; |
1057 |
} |
} |
1058 |
if (rrc != MATCH_NOMATCH && |
if (rrc != MATCH_NOMATCH && |
1059 |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
1060 |
RRETURN(rrc); |
RRETURN(rrc); |
1061 |
ecode += GET(ecode, 1); |
ecode += GET(ecode, 1); |
1062 |
if (*ecode != OP_ALT) break; |
if (*ecode != OP_ALT) break; |
1063 |
} |
} |
1064 |
|
|
1065 |
if (matched_once || allow_zero) |
if (matched_once || allow_zero) |
1066 |
{ |
{ |
1067 |
ecode += 1 + LINK_SIZE; |
ecode += 1 + LINK_SIZE; |
1068 |
break; |
break; |
1069 |
} |
} |
1070 |
RRETURN(MATCH_NOMATCH); |
RRETURN(MATCH_NOMATCH); |
1071 |
|
|
1072 |
/* Control never reaches here. */ |
/* Control never reaches here. */ |
1088 |
if (pcre_callout != NULL) |
if (pcre_callout != NULL) |
1089 |
{ |
{ |
1090 |
pcre_callout_block cb; |
pcre_callout_block cb; |
1091 |
cb.version = 1; /* Version 1 of the callout block */ |
cb.version = 2; /* Version 1 of the callout block */ |
1092 |
cb.callout_number = ecode[LINK_SIZE+2]; |
cb.callout_number = ecode[LINK_SIZE+2]; |
1093 |
cb.offset_vector = md->offset_vector; |
cb.offset_vector = md->offset_vector; |
1094 |
cb.subject = (PCRE_SPTR)md->start_subject; |
cb.subject = (PCRE_SPTR)md->start_subject; |
1100 |
cb.capture_top = offset_top/2; |
cb.capture_top = offset_top/2; |
1101 |
cb.capture_last = md->capture_last; |
cb.capture_last = md->capture_last; |
1102 |
cb.callout_data = md->callout_data; |
cb.callout_data = md->callout_data; |
1103 |
|
cb.mark = markptr; |
1104 |
if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
1105 |
if (rrc < 0) RRETURN(rrc); |
if (rrc < 0) RRETURN(rrc); |
1106 |
} |
} |
1259 |
|
|
1260 |
else |
else |
1261 |
{ |
{ |
1262 |
md->match_function_type = MATCH_CONDASSERT; |
md->match_function_type = MATCH_CONDASSERT; |
1263 |
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3); |
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3); |
1264 |
if (rrc == MATCH_MATCH) |
if (rrc == MATCH_MATCH) |
1265 |
{ |
{ |
1266 |
|
if (md->end_offset_top > offset_top) |
1267 |
|
offset_top = md->end_offset_top; /* Captures may have happened */ |
1268 |
condition = TRUE; |
condition = TRUE; |
1269 |
ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); |
ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); |
1270 |
while (*ecode == OP_ALT) ecode += GET(ecode, 1); |
while (*ecode == OP_ALT) ecode += GET(ecode, 1); |
1282 |
} |
} |
1283 |
|
|
1284 |
/* We are now at the branch that is to be obeyed. As there is only one, |
/* We are now at the branch that is to be obeyed. As there is only one, |
1285 |
we used to use tail recursion to avoid using another stack frame, except |
we used always to use tail recursion to avoid using another stack frame, |
1286 |
when there was unlimited repeat of a possibly empty group. However, that |
except when there was unlimited repeat of a possibly empty group. However, |
1287 |
strategy no longer works because of the possibilty of (*THEN) being |
that strategy no longer works because of the possibilty of (*THEN) being |
1288 |
encountered in the branch. A recursive call to match() is always required, |
encountered in the branch. However, we can still use tail recursion if |
1289 |
unless the second alternative doesn't exist, in which case we can just |
there are no (*THEN)s in the pattern. Otherwise, a recursive call to |
1290 |
plough on. */ |
match() is always required, unless the second alternative doesn't exist, in |
1291 |
|
which case we can just plough on. */ |
1292 |
|
|
1293 |
if (condition || *ecode == OP_ALT) |
if (condition || *ecode == OP_ALT) |
1294 |
{ |
{ |
1295 |
if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP; |
if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP; |
1296 |
|
else if (!md->hasthen) |
1297 |
|
{ |
1298 |
|
ecode += 1 + LINK_SIZE; |
1299 |
|
goto TAIL_RECURSE; |
1300 |
|
} |
1301 |
|
|
1302 |
|
/* A call to match() is required. */ |
1303 |
|
|
1304 |
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49); |
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49); |
1305 |
if (rrc == MATCH_THEN && md->start_match_ptr == ecode) |
|
1306 |
rrc = MATCH_NOMATCH; |
/* If the result is THEN from within the "true" branch of the condition, |
1307 |
|
md->start_match_ptr will point to the original OP_COND, not to the start |
1308 |
|
of the branch, so we have do work to see if it matches. If THEN comes |
1309 |
|
from the "false" branch, md->start_match_ptr does point to OP_ALT. */ |
1310 |
|
|
1311 |
|
if (rrc == MATCH_THEN) |
1312 |
|
{ |
1313 |
|
if (*ecode != OP_ALT) |
1314 |
|
{ |
1315 |
|
do ecode += GET(ecode, 1); while (*ecode == OP_ALT); |
1316 |
|
ecode -= GET(ecode, 1); |
1317 |
|
} |
1318 |
|
if (md->start_match_ptr == ecode) rrc = MATCH_NOMATCH; |
1319 |
|
} |
1320 |
RRETURN(rrc); |
RRETURN(rrc); |
1321 |
} |
} |
1322 |
else /* Condition false & no alternative */ |
|
1323 |
|
/* Condition false & no alternative; continue after the group. */ |
1324 |
|
|
1325 |
|
else |
1326 |
{ |
{ |
1327 |
ecode += 1 + LINK_SIZE; |
ecode += 1 + LINK_SIZE; |
1328 |
} |
} |
1353 |
break; |
break; |
1354 |
|
|
1355 |
|
|
1356 |
/* End of the pattern, either real or forced. If we are in a recursion, we |
/* End of the pattern, either real or forced. */ |
|
should restore the offsets appropriately, and if it's a top-level |
|
|
recursion, continue from after the call. */ |
|
1357 |
|
|
|
case OP_ACCEPT: |
|
|
case OP_ASSERT_ACCEPT: |
|
1358 |
case OP_END: |
case OP_END: |
1359 |
if (md->recursive != NULL) |
case OP_ACCEPT: |
1360 |
{ |
case OP_ASSERT_ACCEPT: |
|
recursion_info *rec = md->recursive; |
|
|
md->recursive = rec->prevrec; |
|
|
memmove(md->offset_vector, rec->offset_save, |
|
|
rec->saved_max * sizeof(int)); |
|
|
offset_top = rec->save_offset_top; |
|
|
if (rec->group_num == 0) |
|
|
{ |
|
|
ecode = rec->after_call; |
|
|
break; |
|
|
} |
|
|
} |
|
1361 |
|
|
1362 |
/* Otherwise, if we have matched an empty string, fail if not in an |
/* If we have matched an empty string, fail if not in an assertion and not |
1363 |
assertion and if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART |
in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART |
1364 |
is set and we have matched at the start of the subject. In both cases, |
is set and we have matched at the start of the subject. In both cases, |
1365 |
backtracking will then try other alternatives, if any. */ |
backtracking will then try other alternatives, if any. */ |
1366 |
|
|
1367 |
else if (eptr == mstart && op != OP_ASSERT_ACCEPT && |
if (eptr == mstart && op != OP_ASSERT_ACCEPT && |
1368 |
(md->notempty || |
md->recursive == NULL && |
1369 |
(md->notempty_atstart && |
(md->notempty || |
1370 |
mstart == md->start_subject + md->start_offset))) |
(md->notempty_atstart && |
1371 |
|
mstart == md->start_subject + md->start_offset))) |
1372 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
1373 |
|
|
1374 |
/* Otherwise, we have a match. */ |
/* Otherwise, we have a match. */ |
1375 |
|
|
1376 |
md->end_match_ptr = eptr; /* Record where we ended */ |
md->end_match_ptr = eptr; /* Record where we ended */ |
1377 |
md->end_offset_top = offset_top; /* and how many extracts were taken */ |
md->end_offset_top = offset_top; /* and how many extracts were taken */ |
1378 |
md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
1387 |
matching won't pass the KET for an assertion. If any one branch matches, |
matching won't pass the KET for an assertion. If any one branch matches, |
1388 |
the assertion is true. Lookbehind assertions have an OP_REVERSE item at the |
the assertion is true. Lookbehind assertions have an OP_REVERSE item at the |
1389 |
start of each branch to move the current point backwards, so the code at |
start of each branch to move the current point backwards, so the code at |
1390 |
this level is identical to the lookahead case. When the assertion is part |
this level is identical to the lookahead case. When the assertion is part |
1391 |
of a condition, we want to return immediately afterwards. The caller of |
of a condition, we want to return immediately afterwards. The caller of |
1392 |
this incarnation of the match() function will have set MATCH_CONDASSERT in |
this incarnation of the match() function will have set MATCH_CONDASSERT in |
1393 |
md->match_function type, and one of these opcodes will be the first opcode |
md->match_function type, and one of these opcodes will be the first opcode |
1394 |
that is processed. We use a local variable that is preserved over calls to |
that is processed. We use a local variable that is preserved over calls to |
1395 |
match() to remember this case. */ |
match() to remember this case. */ |
1396 |
|
|
1397 |
case OP_ASSERT: |
case OP_ASSERT: |
1401 |
condassert = TRUE; |
condassert = TRUE; |
1402 |
md->match_function_type = 0; |
md->match_function_type = 0; |
1403 |
} |
} |
1404 |
else condassert = FALSE; |
else condassert = FALSE; |
1405 |
|
|
1406 |
do |
do |
1407 |
{ |
{ |
1408 |
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4); |
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4); |
1409 |
if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
1410 |
{ |
{ |
1411 |
mstart = md->start_match_ptr; /* In case \K reset it */ |
mstart = md->start_match_ptr; /* In case \K reset it */ |
1412 |
|
markptr = md->mark; |
1413 |
break; |
break; |
1414 |
} |
} |
1415 |
if (rrc != MATCH_NOMATCH && |
if (rrc != MATCH_NOMATCH && |
1418 |
ecode += GET(ecode, 1); |
ecode += GET(ecode, 1); |
1419 |
} |
} |
1420 |
while (*ecode == OP_ALT); |
while (*ecode == OP_ALT); |
1421 |
|
|
1422 |
if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); |
if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); |
1423 |
|
|
1424 |
/* If checking an assertion for a condition, return MATCH_MATCH. */ |
/* If checking an assertion for a condition, return MATCH_MATCH. */ |
1444 |
condassert = TRUE; |
condassert = TRUE; |
1445 |
md->match_function_type = 0; |
md->match_function_type = 0; |
1446 |
} |
} |
1447 |
else condassert = FALSE; |
else condassert = FALSE; |
1448 |
|
|
1449 |
do |
do |
1450 |
{ |
{ |
1463 |
while (*ecode == OP_ALT); |
while (*ecode == OP_ALT); |
1464 |
|
|
1465 |
if (condassert) RRETURN(MATCH_MATCH); /* Condition assertion */ |
if (condassert) RRETURN(MATCH_MATCH); /* Condition assertion */ |
1466 |
|
|
1467 |
ecode += 1 + LINK_SIZE; |
ecode += 1 + LINK_SIZE; |
1468 |
continue; |
continue; |
1469 |
|
|
1508 |
if (pcre_callout != NULL) |
if (pcre_callout != NULL) |
1509 |
{ |
{ |
1510 |
pcre_callout_block cb; |
pcre_callout_block cb; |
1511 |
cb.version = 1; /* Version 1 of the callout block */ |
cb.version = 2; /* Version 1 of the callout block */ |
1512 |
cb.callout_number = ecode[1]; |
cb.callout_number = ecode[1]; |
1513 |
cb.offset_vector = md->offset_vector; |
cb.offset_vector = md->offset_vector; |
1514 |
cb.subject = (PCRE_SPTR)md->start_subject; |
cb.subject = (PCRE_SPTR)md->start_subject; |
1520 |
cb.capture_top = offset_top/2; |
cb.capture_top = offset_top/2; |
1521 |
cb.capture_last = md->capture_last; |
cb.capture_last = md->capture_last; |
1522 |
cb.callout_data = md->callout_data; |
cb.callout_data = md->callout_data; |
1523 |
|
cb.mark = markptr; |
1524 |
if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
1525 |
if (rrc < 0) RRETURN(rrc); |
if (rrc < 0) RRETURN(rrc); |
1526 |
} |
} |
1531 |
offset data is the offset to the starting bracket from the start of the |
offset data is the offset to the starting bracket from the start of the |
1532 |
whole pattern. (This is so that it works from duplicated subpatterns.) |
whole pattern. (This is so that it works from duplicated subpatterns.) |
1533 |
|
|
1534 |
If there are any capturing brackets started but not finished, we have to |
The state of the capturing groups is preserved over recursion, and |
1535 |
save their starting points and reinstate them after the recursion. However, |
re-instated afterwards. We don't know how many are started and not yet |
1536 |
we don't know how many such there are (offset_top records the completed |
finished (offset_top records the completed total) so we just have to save |
1537 |
total) so we just have to save all the potential data. There may be up to |
all the potential data. There may be up to 65535 such values, which is too |
1538 |
65535 such values, which is too large to put on the stack, but using malloc |
large to put on the stack, but using malloc for small numbers seems |
1539 |
for small numbers seems expensive. As a compromise, the stack is used when |
expensive. As a compromise, the stack is used when there are no more than |
1540 |
there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc |
REC_STACK_SAVE_MAX values to store; otherwise malloc is used. |
|
is used. A problem is what to do if the malloc fails ... there is no way of |
|
|
returning to the top level with an error. Save the top REC_STACK_SAVE_MAX |
|
|
values on the stack, and accept that the rest may be wrong. |
|
1541 |
|
|
1542 |
There are also other values that have to be saved. We use a chained |
There are also other values that have to be saved. We use a chained |
1543 |
sequence of blocks that actually live on the stack. Thanks to Robin Houston |
sequence of blocks that actually live on the stack. Thanks to Robin Houston |
1544 |
for the original version of this logic. */ |
for the original version of this logic. It has, however, been hacked around |
1545 |
|
a lot, so he is not to blame for the current way it works. */ |
1546 |
|
|
1547 |
case OP_RECURSE: |
case OP_RECURSE: |
1548 |
{ |
{ |
1549 |
|
recursion_info *ri; |
1550 |
|
int recno; |
1551 |
|
|
1552 |
callpat = md->start_code + GET(ecode, 1); |
callpat = md->start_code + GET(ecode, 1); |
1553 |
new_recursive.group_num = (callpat == md->start_code)? 0 : |
recno = (callpat == md->start_code)? 0 : |
1554 |
GET2(callpat, 1 + LINK_SIZE); |
GET2(callpat, 1 + LINK_SIZE); |
1555 |
|
|
1556 |
|
/* Check for repeating a recursion without advancing the subject pointer. |
1557 |
|
This should catch convoluted mutual recursions. (Some simple cases are |
1558 |
|
caught at compile time.) */ |
1559 |
|
|
1560 |
|
for (ri = md->recursive; ri != NULL; ri = ri->prevrec) |
1561 |
|
if (recno == ri->group_num && eptr == ri->subject_position) |
1562 |
|
RRETURN(PCRE_ERROR_RECURSELOOP); |
1563 |
|
|
1564 |
/* Add to "recursing stack" */ |
/* Add to "recursing stack" */ |
1565 |
|
|
1566 |
|
new_recursive.group_num = recno; |
1567 |
|
new_recursive.subject_position = eptr; |
1568 |
new_recursive.prevrec = md->recursive; |
new_recursive.prevrec = md->recursive; |
1569 |
md->recursive = &new_recursive; |
md->recursive = &new_recursive; |
1570 |
|
|
1571 |
/* Find where to continue from afterwards */ |
/* Where to continue from afterwards */ |
1572 |
|
|
1573 |
ecode += 1 + LINK_SIZE; |
ecode += 1 + LINK_SIZE; |
|
new_recursive.after_call = ecode; |
|
1574 |
|
|
1575 |
/* Now save the offset data. */ |
/* Now save the offset data */ |
1576 |
|
|
1577 |
new_recursive.saved_max = md->offset_end; |
new_recursive.saved_max = md->offset_end; |
1578 |
if (new_recursive.saved_max <= REC_STACK_SAVE_MAX) |
if (new_recursive.saved_max <= REC_STACK_SAVE_MAX) |
1583 |
(int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int)); |
(int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int)); |
1584 |
if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); |
if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); |
1585 |
} |
} |
|
|
|
1586 |
memcpy(new_recursive.offset_save, md->offset_vector, |
memcpy(new_recursive.offset_save, md->offset_vector, |
1587 |
new_recursive.saved_max * sizeof(int)); |
new_recursive.saved_max * sizeof(int)); |
1588 |
new_recursive.save_offset_top = offset_top; |
|
1589 |
|
/* OK, now we can do the recursion. After processing each alternative, |
1590 |
/* OK, now we can do the recursion. For each top-level alternative we |
restore the offset data. If there were nested recursions, md->recursive |
1591 |
restore the offset and recursion data. */ |
might be changed, so reset it before looping. */ |
1592 |
|
|
1593 |
DPRINTF(("Recursing into group %d\n", new_recursive.group_num)); |
DPRINTF(("Recursing into group %d\n", new_recursive.group_num)); |
1594 |
cbegroup = (*callpat >= OP_SBRA); |
cbegroup = (*callpat >= OP_SBRA); |
1597 |
if (cbegroup) md->match_function_type = MATCH_CBEGROUP; |
if (cbegroup) md->match_function_type = MATCH_CBEGROUP; |
1598 |
RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, |
RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, |
1599 |
md, eptrb, RM6); |
md, eptrb, RM6); |
1600 |
|
memcpy(md->offset_vector, new_recursive.offset_save, |
1601 |
|
new_recursive.saved_max * sizeof(int)); |
1602 |
|
md->recursive = new_recursive.prevrec; |
1603 |
if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
1604 |
{ |
{ |
1605 |
DPRINTF(("Recursion matched\n")); |
DPRINTF(("Recursion matched\n")); |
|
md->recursive = new_recursive.prevrec; |
|
1606 |
if (new_recursive.offset_save != stacksave) |
if (new_recursive.offset_save != stacksave) |
1607 |
(pcre_free)(new_recursive.offset_save); |
(pcre_free)(new_recursive.offset_save); |
1608 |
MRRETURN(MATCH_MATCH); |
|
1609 |
|
/* Set where we got to in the subject, and reset the start in case |
1610 |
|
it was changed by \K. This *is* propagated back out of a recursion, |
1611 |
|
for Perl compatibility. */ |
1612 |
|
|
1613 |
|
eptr = md->end_match_ptr; |
1614 |
|
mstart = md->start_match_ptr; |
1615 |
|
goto RECURSION_MATCHED; /* Exit loop; end processing */ |
1616 |
} |
} |
1617 |
else if (rrc != MATCH_NOMATCH && |
else if (rrc != MATCH_NOMATCH && |
1618 |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
1624 |
} |
} |
1625 |
|
|
1626 |
md->recursive = &new_recursive; |
md->recursive = &new_recursive; |
|
memcpy(md->offset_vector, new_recursive.offset_save, |
|
|
new_recursive.saved_max * sizeof(int)); |
|
1627 |
callpat += GET(callpat, 1); |
callpat += GET(callpat, 1); |
1628 |
} |
} |
1629 |
while (*callpat == OP_ALT); |
while (*callpat == OP_ALT); |
1634 |
(pcre_free)(new_recursive.offset_save); |
(pcre_free)(new_recursive.offset_save); |
1635 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
1636 |
} |
} |
|
/* Control never reaches here */ |
|
|
|
|
|
/* "Once" brackets are like assertion brackets except that after a match, |
|
|
the point in the subject string is not moved back. Thus there can never be |
|
|
a move back into the brackets. Friedl calls these "atomic" subpatterns. |
|
|
Check the alternative branches in turn - the matching won't pass the KET |
|
|
for this kind of subpattern. If any one branch matches, we carry on as at |
|
|
the end of a normal bracket, leaving the subject pointer, but resetting |
|
|
the start-of-match value in case it was changed by \K. */ |
|
|
|
|
|
case OP_ONCE: |
|
|
prev = ecode; |
|
|
saved_eptr = eptr; |
|
|
|
|
|
do |
|
|
{ |
|
|
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7); |
|
|
if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ |
|
|
{ |
|
|
mstart = md->start_match_ptr; |
|
|
break; |
|
|
} |
|
|
if (rrc != MATCH_NOMATCH && |
|
|
(rrc != MATCH_THEN || md->start_match_ptr != ecode)) |
|
|
RRETURN(rrc); |
|
|
ecode += GET(ecode,1); |
|
|
} |
|
|
while (*ecode == OP_ALT); |
|
|
|
|
|
/* If hit the end of the group (which could be repeated), fail */ |
|
|
|
|
|
if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH); |
|
|
|
|
|
/* Continue after the group, updating the offsets high water mark, since |
|
|
extracts may have been taken. */ |
|
|
|
|
|
do ecode += GET(ecode, 1); while (*ecode == OP_ALT); |
|
|
|
|
|
offset_top = md->end_offset_top; |
|
|
eptr = md->end_match_ptr; |
|
|
|
|
|
/* For a non-repeating ket, just continue at this level. This also |
|
|
happens for a repeating ket if no characters were matched in the group. |
|
|
This is the forcible breaking of infinite loops as implemented in Perl |
|
|
5.005. If there is an options reset, it will get obeyed in the normal |
|
|
course of events. */ |
|
|
|
|
|
if (*ecode == OP_KET || eptr == saved_eptr) |
|
|
{ |
|
|
ecode += 1+LINK_SIZE; |
|
|
break; |
|
|
} |
|
1637 |
|
|
1638 |
/* The repeating kets try the rest of the pattern or restart from the |
RECURSION_MATCHED: |
1639 |
preceding bracket, in the appropriate order. The second "call" of match() |
break; |
|
uses tail recursion, to avoid using another stack frame. */ |
|
|
|
|
|
if (*ecode == OP_KETRMIN) |
|
|
{ |
|
|
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM8); |
|
|
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
|
|
ecode = prev; |
|
|
goto TAIL_RECURSE; |
|
|
} |
|
|
else /* OP_KETRMAX */ |
|
|
{ |
|
|
md->match_function_type = MATCH_CBEGROUP; |
|
|
RMATCH(eptr, prev, offset_top, md, eptrb, RM9); |
|
|
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
|
|
ecode += 1 + LINK_SIZE; |
|
|
goto TAIL_RECURSE; |
|
|
} |
|
|
/* Control never gets here */ |
|
1640 |
|
|
1641 |
/* An alternation is the end of a branch; scan along to find the end of the |
/* An alternation is the end of a branch; scan along to find the end of the |
1642 |
bracketed group and go to there. */ |
bracketed group and go to there. */ |
1650 |
at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets |
at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets |
1651 |
with fixed upper repeat limits are compiled as a number of copies, with the |
with fixed upper repeat limits are compiled as a number of copies, with the |
1652 |
optional ones preceded by BRAZERO or BRAMINZERO. */ |
optional ones preceded by BRAZERO or BRAMINZERO. */ |
1653 |
|
|
1654 |
case OP_BRAZERO: |
case OP_BRAZERO: |
1655 |
next = ecode + 1; |
next = ecode + 1; |
1656 |
RMATCH(eptr, next, offset_top, md, eptrb, RM10); |
RMATCH(eptr, next, offset_top, md, eptrb, RM10); |
1658 |
do next += GET(next, 1); while (*next == OP_ALT); |
do next += GET(next, 1); while (*next == OP_ALT); |
1659 |
ecode = next + 1 + LINK_SIZE; |
ecode = next + 1 + LINK_SIZE; |
1660 |
break; |
break; |
1661 |
|
|
1662 |
case OP_BRAMINZERO: |
case OP_BRAMINZERO: |
1663 |
next = ecode + 1; |
next = ecode + 1; |
1664 |
do next += GET(next, 1); while (*next == OP_ALT); |
do next += GET(next, 1); while (*next == OP_ALT); |
1672 |
do next += GET(next,1); while (*next == OP_ALT); |
do next += GET(next,1); while (*next == OP_ALT); |
1673 |
ecode = next + 1 + LINK_SIZE; |
ecode = next + 1 + LINK_SIZE; |
1674 |
break; |
break; |
1675 |
|
|
1676 |
/* BRAPOSZERO occurs before a possessive bracket group. Don't do anything |
/* BRAPOSZERO occurs before a possessive bracket group. Don't do anything |
1677 |
here; just jump to the group, with allow_zero set TRUE. */ |
here; just jump to the group, with allow_zero set TRUE. */ |
1678 |
|
|
1679 |
case OP_BRAPOSZERO: |
case OP_BRAPOSZERO: |
1680 |
op = *(++ecode); |
op = *(++ecode); |
1681 |
allow_zero = TRUE; |
allow_zero = TRUE; |
1682 |
if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE; |
if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE; |
1683 |
goto POSSESSIVE_NON_CAPTURE; |
goto POSSESSIVE_NON_CAPTURE; |
1687 |
case OP_KET: |
case OP_KET: |
1688 |
case OP_KETRMIN: |
case OP_KETRMIN: |
1689 |
case OP_KETRMAX: |
case OP_KETRMAX: |
1690 |
case OP_KETRPOS: |
case OP_KETRPOS: |
1691 |
prev = ecode - GET(ecode, 1); |
prev = ecode - GET(ecode, 1); |
1692 |
|
|
1693 |
/* If this was a group that remembered the subject start, in order to break |
/* If this was a group that remembered the subject start, in order to break |
1694 |
infinite repeats of empty string matches, retrieve the subject start from |
infinite repeats of empty string matches, retrieve the subject start from |
1695 |
the chain. Otherwise, set it NULL. */ |
the chain. Otherwise, set it NULL. */ |
1696 |
|
|
1697 |
if (*prev >= OP_SBRA) |
if (*prev >= OP_SBRA || *prev == OP_ONCE) |
1698 |
{ |
{ |
1699 |
saved_eptr = eptrb->epb_saved_eptr; /* Value at start of group */ |
saved_eptr = eptrb->epb_saved_eptr; /* Value at start of group */ |
1700 |
eptrb = eptrb->epb_prev; /* Backup to previous group */ |
eptrb = eptrb->epb_prev; /* Backup to previous group */ |
1701 |
} |
} |
1702 |
else saved_eptr = NULL; |
else saved_eptr = NULL; |
1703 |
|
|
1704 |
/* If we are at the end of an assertion group or an atomic group, stop |
/* If we are at the end of an assertion group, stop matching and return |
1705 |
matching and return MATCH_MATCH, but record the current high water mark for |
MATCH_MATCH, but record the current high water mark for use by positive |
1706 |
use by positive assertions. We also need to record the match start in case |
assertions. We also need to record the match start in case it was changed |
1707 |
it was changed by \K. */ |
by \K. */ |
1708 |
|
|
1709 |
if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || |
if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || |
1710 |
*prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || |
*prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT) |
|
*prev == OP_ONCE) |
|
1711 |
{ |
{ |
1712 |
md->end_match_ptr = eptr; /* For ONCE */ |
md->end_match_ptr = eptr; /* For ONCE */ |
1713 |
md->end_offset_top = offset_top; |
md->end_offset_top = offset_top; |
1714 |
md->start_match_ptr = mstart; |
md->start_match_ptr = mstart; |
1715 |
MRRETURN(MATCH_MATCH); |
MRRETURN(MATCH_MATCH); /* Sets md->mark */ |
1716 |
} |
} |
1717 |
|
|
1718 |
/* For capturing groups we have to check the group number back at the start |
/* For capturing groups we have to check the group number back at the start |
1719 |
and if necessary complete handling an extraction by setting the offsets and |
and if necessary complete handling an extraction by setting the offsets and |
1720 |
bumping the high water mark. Note that whole-pattern recursion is coded as |
bumping the high water mark. Whole-pattern recursion is coded as a recurse |
1721 |
a recurse into group 0, so it won't be picked up here. Instead, we catch it |
into group 0, so it won't be picked up here. Instead, we catch it when the |
1722 |
when the OP_END is reached. Other recursion is handled here. */ |
OP_END is reached. Other recursion is handled here. We just have to record |
1723 |
|
the current subject position and start match pointer and give a MATCH |
1724 |
|
return. */ |
1725 |
|
|
1726 |
if (*prev == OP_CBRA || *prev == OP_SCBRA || |
if (*prev == OP_CBRA || *prev == OP_SCBRA || |
1727 |
*prev == OP_CBRAPOS || *prev == OP_SCBRAPOS) |
*prev == OP_CBRAPOS || *prev == OP_SCBRAPOS) |
1734 |
printf("\n"); |
printf("\n"); |
1735 |
#endif |
#endif |
1736 |
|
|
1737 |
|
/* Handle a recursively called group. */ |
1738 |
|
|
1739 |
|
if (md->recursive != NULL && md->recursive->group_num == number) |
1740 |
|
{ |
1741 |
|
md->end_match_ptr = eptr; |
1742 |
|
md->start_match_ptr = mstart; |
1743 |
|
RRETURN(MATCH_MATCH); |
1744 |
|
} |
1745 |
|
|
1746 |
|
/* Deal with capturing */ |
1747 |
|
|
1748 |
md->capture_last = number; |
md->capture_last = number; |
1749 |
if (offset >= md->offset_max) md->offset_overflow = TRUE; else |
if (offset >= md->offset_max) md->offset_overflow = TRUE; else |
1750 |
{ |
{ |
1751 |
/* If offset is greater than offset_top, it means that we are |
/* If offset is greater than offset_top, it means that we are |
1752 |
"skipping" a capturing group, and that group's offsets must be marked |
"skipping" a capturing group, and that group's offsets must be marked |
1753 |
unset. In earlier versions of PCRE, all the offsets were unset at the |
unset. In earlier versions of PCRE, all the offsets were unset at the |
1754 |
start of matching, but this doesn't work because atomic groups and |
start of matching, but this doesn't work because atomic groups and |
1755 |
assertions can cause a value to be set that should later be unset. |
assertions can cause a value to be set that should later be unset. |
1756 |
Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as |
Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as |
1757 |
part of the atomic group, but this is not on the final matching path, |
part of the atomic group, but this is not on the final matching path, |
1758 |
so must be unset when 2 is set. (If there is no group 2, there is no |
so must be unset when 2 is set. (If there is no group 2, there is no |
1759 |
problem, because offset_top will then be 2, indicating no capture.) */ |
problem, because offset_top will then be 2, indicating no capture.) */ |
1760 |
|
|
1761 |
if (offset > offset_top) |
if (offset > offset_top) |
1762 |
{ |
{ |
1763 |
register int *iptr = md->offset_vector + offset_top; |
register int *iptr = md->offset_vector + offset_top; |
1764 |
register int *iend = md->offset_vector + offset; |
register int *iend = md->offset_vector + offset; |
1765 |
while (iptr < iend) *iptr++ = -1; |
while (iptr < iend) *iptr++ = -1; |
1766 |
} |
} |
1767 |
|
|
1768 |
/* Now make the extraction */ |
/* Now make the extraction */ |
1769 |
|
|
1770 |
md->offset_vector[offset] = |
md->offset_vector[offset] = |
1772 |
md->offset_vector[offset+1] = (int)(eptr - md->start_subject); |
md->offset_vector[offset+1] = (int)(eptr - md->start_subject); |
1773 |
if (offset_top <= offset) offset_top = offset + 2; |
if (offset_top <= offset) offset_top = offset + 2; |
1774 |
} |
} |
|
|
|
|
/* Handle a recursively called group. Restore the offsets |
|
|
appropriately and continue from after the call. */ |
|
|
|
|
|
if (md->recursive != NULL && md->recursive->group_num == number) |
|
|
{ |
|
|
recursion_info *rec = md->recursive; |
|
|
DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); |
|
|
md->recursive = rec->prevrec; |
|
|
memcpy(md->offset_vector, rec->offset_save, |
|
|
rec->saved_max * sizeof(int)); |
|
|
offset_top = rec->save_offset_top; |
|
|
ecode = rec->after_call; |
|
|
break; |
|
|
} |
|
1775 |
} |
} |
1776 |
|
|
1777 |
/* For a non-repeating ket, just continue at this level. This also |
/* For an ordinary non-repeating ket, just continue at this level. This |
1778 |
happens for a repeating ket if no characters were matched in the group. |
also happens for a repeating ket if no characters were matched in the |
1779 |
This is the forcible breaking of infinite loops as implemented in Perl |
group. This is the forcible breaking of infinite loops as implemented in |
1780 |
5.005. If there is an options reset, it will get obeyed in the normal |
Perl 5.005. For a non-repeating atomic group, establish a backup point by |
1781 |
course of events. */ |
processing the rest of the pattern at a lower level. If this results in a |
1782 |
|
NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby |
1783 |
|
bypassing intermediate backup points, but resetting any captures that |
1784 |
|
happened along the way. */ |
1785 |
|
|
1786 |
if (*ecode == OP_KET || eptr == saved_eptr) |
if (*ecode == OP_KET || eptr == saved_eptr) |
1787 |
{ |
{ |
1788 |
ecode += 1 + LINK_SIZE; |
if (*prev == OP_ONCE) |
1789 |
|
{ |
1790 |
|
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12); |
1791 |
|
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
1792 |
|
md->once_target = prev; /* Level at which to change to MATCH_NOMATCH */ |
1793 |
|
RRETURN(MATCH_ONCE); |
1794 |
|
} |
1795 |
|
ecode += 1 + LINK_SIZE; /* Carry on at this level */ |
1796 |
break; |
break; |
1797 |
} |
} |
1798 |
|
|
1799 |
/* OP_KETRPOS is a possessive repeating ket. Remember the current position, |
/* OP_KETRPOS is a possessive repeating ket. Remember the current position, |
1800 |
and return the MATCH_KETRPOS. This makes it possible to do the repeats one |
and return the MATCH_KETRPOS. This makes it possible to do the repeats one |
1801 |
at a time from the outer level, thus saving stack. */ |
at a time from the outer level, thus saving stack. */ |
1802 |
|
|
1803 |
if (*ecode == OP_KETRPOS) |
if (*ecode == OP_KETRPOS) |
1804 |
{ |
{ |
1805 |
md->end_match_ptr = eptr; |
md->end_match_ptr = eptr; |
1806 |
md->end_offset_top = offset_top; |
md->end_offset_top = offset_top; |
1807 |
RRETURN(MATCH_KETRPOS); |
RRETURN(MATCH_KETRPOS); |
1808 |
} |
} |
1809 |
|
|
1810 |
/* The normal repeating kets try the rest of the pattern or restart from |
/* The normal repeating kets try the rest of the pattern or restart from |
1811 |
the preceding bracket, in the appropriate order. In the second case, we can |
the preceding bracket, in the appropriate order. In the second case, we can |
1812 |
use tail recursion to avoid using another stack frame, unless we have an |
use tail recursion to avoid using another stack frame, unless we have an |
1813 |
unlimited repeat of a group that can match an empty string. */ |
an atomic group or an unlimited repeat of a group that can match an empty |
1814 |
|
string. */ |
1815 |
|
|
1816 |
if (*ecode == OP_KETRMIN) |
if (*ecode == OP_KETRMIN) |
1817 |
{ |
{ |
1818 |
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12); |
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7); |
1819 |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
1820 |
|
if (*prev == OP_ONCE) |
1821 |
|
{ |
1822 |
|
RMATCH(eptr, prev, offset_top, md, eptrb, RM8); |
1823 |
|
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
1824 |
|
md->once_target = prev; /* Level at which to change to MATCH_NOMATCH */ |
1825 |
|
RRETURN(MATCH_ONCE); |
1826 |
|
} |
1827 |
if (*prev >= OP_SBRA) /* Could match an empty string */ |
if (*prev >= OP_SBRA) /* Could match an empty string */ |
1828 |
{ |
{ |
1829 |
md->match_function_type = MATCH_CBEGROUP; |
md->match_function_type = MATCH_CBEGROUP; |
1830 |
RMATCH(eptr, prev, offset_top, md, eptrb, RM50); |
RMATCH(eptr, prev, offset_top, md, eptrb, RM50); |
1831 |
RRETURN(rrc); |
RRETURN(rrc); |
1832 |
} |
} |
1835 |
} |
} |
1836 |
else /* OP_KETRMAX */ |
else /* OP_KETRMAX */ |
1837 |
{ |
{ |
1838 |
if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
if (*prev >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
1839 |
RMATCH(eptr, prev, offset_top, md, eptrb, RM13); |
RMATCH(eptr, prev, offset_top, md, eptrb, RM13); |
1840 |
|
if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH; |
1841 |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
1842 |
|
if (*prev == OP_ONCE) |
1843 |
|
{ |
1844 |
|
RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9); |
1845 |
|
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
1846 |
|
md->once_target = prev; |
1847 |
|
RRETURN(MATCH_ONCE); |
1848 |
|
} |
1849 |
ecode += 1 + LINK_SIZE; |
ecode += 1 + LINK_SIZE; |
1850 |
goto TAIL_RECURSE; |
goto TAIL_RECURSE; |
1851 |
} |
} |
1855 |
|
|
1856 |
case OP_CIRC: |
case OP_CIRC: |
1857 |
if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); |
if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); |
1858 |
|
|
1859 |
/* Start of subject assertion */ |
/* Start of subject assertion */ |
1860 |
|
|
1861 |
case OP_SOD: |
case OP_SOD: |
1862 |
if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); |
if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); |
1863 |
ecode++; |
ecode++; |
1864 |
break; |
break; |
1865 |
|
|
1866 |
/* Multiline mode: start of subject unless notbol, or after any newline. */ |
/* Multiline mode: start of subject unless notbol, or after any newline. */ |
1867 |
|
|
1868 |
case OP_CIRCM: |
case OP_CIRCM: |
1901 |
ecode++; |
ecode++; |
1902 |
break; |
break; |
1903 |
|
|
1904 |
/* Not multiline mode: assert before a terminating newline or before end of |
/* Not multiline mode: assert before a terminating newline or before end of |
1905 |
subject unless noteol is set. */ |
subject unless noteol is set. */ |
1906 |
|
|
1907 |
case OP_DOLL: |
case OP_DOLL: |
2057 |
/* Fall through */ |
/* Fall through */ |
2058 |
|
|
2059 |
case OP_ALLANY: |
case OP_ALLANY: |
2060 |
if (eptr++ >= md->end_subject) |
if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ |
2061 |
{ |
{ /* not be updated before SCHECK_PARTIAL. */ |
2062 |
SCHECK_PARTIAL(); |
SCHECK_PARTIAL(); |
2063 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
2064 |
} |
} |
2065 |
|
eptr++; |
2066 |
if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
2067 |
ecode++; |
ecode++; |
2068 |
break; |
break; |
2071 |
any byte, even newline, independent of the setting of PCRE_DOTALL. */ |
any byte, even newline, independent of the setting of PCRE_DOTALL. */ |
2072 |
|
|
2073 |
case OP_ANYBYTE: |
case OP_ANYBYTE: |
2074 |
if (eptr++ >= md->end_subject) |
if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ |
2075 |
{ |
{ /* not be updated before SCHECK_PARTIAL. */ |
2076 |
SCHECK_PARTIAL(); |
SCHECK_PARTIAL(); |
2077 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
2078 |
} |
} |
2079 |
|
eptr++; |
2080 |
ecode++; |
ecode++; |
2081 |
break; |
break; |
2082 |
|
|
2192 |
switch(c) |
switch(c) |
2193 |
{ |
{ |
2194 |
default: MRRETURN(MATCH_NOMATCH); |
default: MRRETURN(MATCH_NOMATCH); |
2195 |
|
|
2196 |
case 0x000d: |
case 0x000d: |
2197 |
if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
2198 |
break; |
break; |
2416 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
2417 |
} |
} |
2418 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
2419 |
|
if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); |
2420 |
|
while (eptr < md->end_subject) |
2421 |
{ |
{ |
2422 |
int category = UCD_CATEGORY(c); |
int len = 1; |
2423 |
if (category == ucp_M) MRRETURN(MATCH_NOMATCH); |
if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
2424 |
while (eptr < md->end_subject) |
if (UCD_CATEGORY(c) != ucp_M) break; |
2425 |
{ |
eptr += len; |
|
int len = 1; |
|
|
if (!utf8) c = *eptr; else |
|
|
{ |
|
|
GETCHARLEN(c, eptr, len); |
|
|
} |
|
|
category = UCD_CATEGORY(c); |
|
|
if (category != ucp_M) break; |
|
|
eptr += len; |
|
|
} |
|
2426 |
} |
} |
2427 |
ecode++; |
ecode++; |
2428 |
break; |
break; |
2438 |
loops). */ |
loops). */ |
2439 |
|
|
2440 |
case OP_REF: |
case OP_REF: |
2441 |
case OP_REFI: |
case OP_REFI: |
2442 |
caseless = op == OP_REFI; |
caseless = op == OP_REFI; |
2443 |
offset = GET2(ecode, 1) << 1; /* Doubled ref number */ |
offset = GET2(ecode, 1) << 1; /* Doubled ref number */ |
2444 |
ecode += 3; |
ecode += 3; |
2445 |
|
|
2507 |
|
|
2508 |
for (i = 1; i <= min; i++) |
for (i = 1; i <= min; i++) |
2509 |
{ |
{ |
2510 |
int slength; |
int slength; |
2511 |
if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
2512 |
{ |
{ |
2513 |
CHECK_PARTIAL(); |
CHECK_PARTIAL(); |
2527 |
{ |
{ |
2528 |
for (fi = min;; fi++) |
for (fi = min;; fi++) |
2529 |
{ |
{ |
2530 |
int slength; |
int slength; |
2531 |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); |
2532 |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
2533 |
if (fi >= max) MRRETURN(MATCH_NOMATCH); |
if (fi >= max) MRRETURN(MATCH_NOMATCH); |
2548 |
pp = eptr; |
pp = eptr; |
2549 |
for (i = min; i < max; i++) |
for (i = min; i < max; i++) |
2550 |
{ |
{ |
2551 |
int slength; |
int slength; |
2552 |
if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
2553 |
{ |
{ |
2554 |
CHECK_PARTIAL(); |
CHECK_PARTIAL(); |
3281 |
checking can be multibyte. */ |
checking can be multibyte. */ |
3282 |
|
|
3283 |
case OP_NOT: |
case OP_NOT: |
3284 |
case OP_NOTI: |
case OP_NOTI: |
3285 |
if (eptr >= md->end_subject) |
if (eptr >= md->end_subject) |
3286 |
{ |
{ |
3287 |
SCHECK_PARTIAL(); |
SCHECK_PARTIAL(); |
3776 |
case PT_LAMP: |
case PT_LAMP: |
3777 |
for (i = 1; i <= min; i++) |
for (i = 1; i <= min; i++) |
3778 |
{ |
{ |
3779 |
|
int chartype; |
3780 |
if (eptr >= md->end_subject) |
if (eptr >= md->end_subject) |
3781 |
{ |
{ |
3782 |
SCHECK_PARTIAL(); |
SCHECK_PARTIAL(); |
3783 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3784 |
} |
} |
3785 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
3786 |
prop_chartype = UCD_CHARTYPE(c); |
chartype = UCD_CHARTYPE(c); |
3787 |
if ((prop_chartype == ucp_Lu || |
if ((chartype == ucp_Lu || |
3788 |
prop_chartype == ucp_Ll || |
chartype == ucp_Ll || |
3789 |
prop_chartype == ucp_Lt) == prop_fail_result) |
chartype == ucp_Lt) == prop_fail_result) |
3790 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3791 |
} |
} |
3792 |
break; |
break; |
3800 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3801 |
} |
} |
3802 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
3803 |
prop_category = UCD_CATEGORY(c); |
if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) |
|
if ((prop_category == prop_value) == prop_fail_result) |
|
3804 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3805 |
} |
} |
3806 |
break; |
break; |
3814 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3815 |
} |
} |
3816 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
3817 |
prop_chartype = UCD_CHARTYPE(c); |
if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) |
|
if ((prop_chartype == prop_value) == prop_fail_result) |
|
3818 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3819 |
} |
} |
3820 |
break; |
break; |
3828 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3829 |
} |
} |
3830 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
3831 |
prop_script = UCD_SCRIPT(c); |
if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) |
|
if ((prop_script == prop_value) == prop_fail_result) |
|
3832 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3833 |
} |
} |
3834 |
break; |
break; |
3836 |
case PT_ALNUM: |
case PT_ALNUM: |
3837 |
for (i = 1; i <= min; i++) |
for (i = 1; i <= min; i++) |
3838 |
{ |
{ |
3839 |
|
int category; |
3840 |
if (eptr >= md->end_subject) |
if (eptr >= md->end_subject) |
3841 |
{ |
{ |
3842 |
SCHECK_PARTIAL(); |
SCHECK_PARTIAL(); |
3843 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3844 |
} |
} |
3845 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
3846 |
prop_category = UCD_CATEGORY(c); |
category = UCD_CATEGORY(c); |
3847 |
if ((prop_category == ucp_L || prop_category == ucp_N) |
if ((category == ucp_L || category == ucp_N) == prop_fail_result) |
|
== prop_fail_result) |
|
3848 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3849 |
} |
} |
3850 |
break; |
break; |
3858 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3859 |
} |
} |
3860 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
3861 |
prop_category = UCD_CATEGORY(c); |
if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
3862 |
c == CHAR_FF || c == CHAR_CR) |
c == CHAR_FF || c == CHAR_CR) |
3863 |
== prop_fail_result) |
== prop_fail_result) |
3864 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3874 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3875 |
} |
} |
3876 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
3877 |
prop_category = UCD_CATEGORY(c); |
if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
3878 |
c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
3879 |
== prop_fail_result) |
== prop_fail_result) |
3880 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3884 |
case PT_WORD: |
case PT_WORD: |
3885 |
for (i = 1; i <= min; i++) |
for (i = 1; i <= min; i++) |
3886 |
{ |
{ |
3887 |
|
int category; |
3888 |
if (eptr >= md->end_subject) |
if (eptr >= md->end_subject) |
3889 |
{ |
{ |
3890 |
SCHECK_PARTIAL(); |
SCHECK_PARTIAL(); |
3891 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3892 |
} |
} |
3893 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
3894 |
prop_category = UCD_CATEGORY(c); |
category = UCD_CATEGORY(c); |
3895 |
if ((prop_category == ucp_L || prop_category == ucp_N || |
if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) |
|
c == CHAR_UNDERSCORE) |
|
3896 |
== prop_fail_result) |
== prop_fail_result) |
3897 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3898 |
} |
} |
3918 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
3919 |
} |
} |
3920 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
3921 |
prop_category = UCD_CATEGORY(c); |
if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); |
|
if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
|
3922 |
while (eptr < md->end_subject) |
while (eptr < md->end_subject) |
3923 |
{ |
{ |
3924 |
int len = 1; |
int len = 1; |
3925 |
if (!utf8) c = *eptr; |
if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
3926 |
else { GETCHARLEN(c, eptr, len); } |
if (UCD_CATEGORY(c) != ucp_M) break; |
|
prop_category = UCD_CATEGORY(c); |
|
|
if (prop_category != ucp_M) break; |
|
3927 |
eptr += len; |
eptr += len; |
3928 |
} |
} |
3929 |
} |
} |
3981 |
switch(c) |
switch(c) |
3982 |
{ |
{ |
3983 |
default: MRRETURN(MATCH_NOMATCH); |
default: MRRETURN(MATCH_NOMATCH); |
3984 |
|
|
3985 |
case 0x000d: |
case 0x000d: |
3986 |
if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
3987 |
break; |
break; |
4258 |
switch(*eptr++) |
switch(*eptr++) |
4259 |
{ |
{ |
4260 |
default: MRRETURN(MATCH_NOMATCH); |
default: MRRETURN(MATCH_NOMATCH); |
4261 |
|
|
4262 |
case 0x000d: |
case 0x000d: |
4263 |
if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
4264 |
break; |
break; |
4265 |
|
|
4266 |
case 0x000a: |
case 0x000a: |
4267 |
break; |
break; |
4268 |
|
|
4468 |
case PT_LAMP: |
case PT_LAMP: |
4469 |
for (fi = min;; fi++) |
for (fi = min;; fi++) |
4470 |
{ |
{ |
4471 |
|
int chartype; |
4472 |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); |
4473 |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
4474 |
if (fi >= max) MRRETURN(MATCH_NOMATCH); |
if (fi >= max) MRRETURN(MATCH_NOMATCH); |
4478 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4479 |
} |
} |
4480 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
4481 |
prop_chartype = UCD_CHARTYPE(c); |
chartype = UCD_CHARTYPE(c); |
4482 |
if ((prop_chartype == ucp_Lu || |
if ((chartype == ucp_Lu || |
4483 |
prop_chartype == ucp_Ll || |
chartype == ucp_Ll || |
4484 |
prop_chartype == ucp_Lt) == prop_fail_result) |
chartype == ucp_Lt) == prop_fail_result) |
4485 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4486 |
} |
} |
4487 |
/* Control never gets here */ |
/* Control never gets here */ |
4498 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4499 |
} |
} |
4500 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
4501 |
prop_category = UCD_CATEGORY(c); |
if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) |
|
if ((prop_category == prop_value) == prop_fail_result) |
|
4502 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4503 |
} |
} |
4504 |
/* Control never gets here */ |
/* Control never gets here */ |
4515 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4516 |
} |
} |
4517 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
4518 |
prop_chartype = UCD_CHARTYPE(c); |
if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) |
|
if ((prop_chartype == prop_value) == prop_fail_result) |
|
4519 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4520 |
} |
} |
4521 |
/* Control never gets here */ |
/* Control never gets here */ |
4532 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4533 |
} |
} |
4534 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
4535 |
prop_script = UCD_SCRIPT(c); |
if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) |
|
if ((prop_script == prop_value) == prop_fail_result) |
|
4536 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4537 |
} |
} |
4538 |
/* Control never gets here */ |
/* Control never gets here */ |
4540 |
case PT_ALNUM: |
case PT_ALNUM: |
4541 |
for (fi = min;; fi++) |
for (fi = min;; fi++) |
4542 |
{ |
{ |
4543 |
|
int category; |
4544 |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM59); |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM59); |
4545 |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
4546 |
if (fi >= max) MRRETURN(MATCH_NOMATCH); |
if (fi >= max) MRRETURN(MATCH_NOMATCH); |
4550 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4551 |
} |
} |
4552 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
4553 |
prop_category = UCD_CATEGORY(c); |
category = UCD_CATEGORY(c); |
4554 |
if ((prop_category == ucp_L || prop_category == ucp_N) |
if ((category == ucp_L || category == ucp_N) == prop_fail_result) |
|
== prop_fail_result) |
|
4555 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4556 |
} |
} |
4557 |
/* Control never gets here */ |
/* Control never gets here */ |
4568 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4569 |
} |
} |
4570 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
4571 |
prop_category = UCD_CATEGORY(c); |
if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
4572 |
c == CHAR_FF || c == CHAR_CR) |
c == CHAR_FF || c == CHAR_CR) |
4573 |
== prop_fail_result) |
== prop_fail_result) |
4574 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4587 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4588 |
} |
} |
4589 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
4590 |
prop_category = UCD_CATEGORY(c); |
if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
4591 |
c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
4592 |
== prop_fail_result) |
== prop_fail_result) |
4593 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4597 |
case PT_WORD: |
case PT_WORD: |
4598 |
for (fi = min;; fi++) |
for (fi = min;; fi++) |
4599 |
{ |
{ |
4600 |
|
int category; |
4601 |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM62); |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM62); |
4602 |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
4603 |
if (fi >= max) MRRETURN(MATCH_NOMATCH); |
if (fi >= max) MRRETURN(MATCH_NOMATCH); |
4607 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4608 |
} |
} |
4609 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
4610 |
prop_category = UCD_CATEGORY(c); |
category = UCD_CATEGORY(c); |
4611 |
if ((prop_category == ucp_L || |
if ((category == ucp_L || |
4612 |
prop_category == ucp_N || |
category == ucp_N || |
4613 |
c == CHAR_UNDERSCORE) |
c == CHAR_UNDERSCORE) |
4614 |
== prop_fail_result) |
== prop_fail_result) |
4615 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4639 |
MRRETURN(MATCH_NOMATCH); |
MRRETURN(MATCH_NOMATCH); |
4640 |
} |
} |
4641 |
GETCHARINCTEST(c, eptr); |
GETCHARINCTEST(c, eptr); |
4642 |
prop_category = UCD_CATEGORY(c); |
if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); |
|
if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
|
4643 |
while (eptr < md->end_subject) |
while (eptr < md->end_subject) |
4644 |
{ |
{ |
4645 |
int len = 1; |
int len = 1; |
4646 |
if (!utf8) c = *eptr; |
if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
4647 |
else { GETCHARLEN(c, eptr, len); } |
if (UCD_CATEGORY(c) != ucp_M) break; |
|
prop_category = UCD_CATEGORY(c); |
|
|
if (prop_category != ucp_M) break; |
|
4648 |
eptr += len; |
eptr += len; |
4649 |
} |
} |
4650 |
} |
} |
4651 |
} |
} |
|
|
|
4652 |
else |
else |
4653 |
#endif /* SUPPORT_UCP */ |
#endif /* SUPPORT_UCP */ |
4654 |
|
|
4969 |
case PT_LAMP: |
case PT_LAMP: |
4970 |
for (i = min; i < max; i++) |
for (i = min; i < max; i++) |
4971 |
{ |
{ |
4972 |
|
int chartype; |
4973 |
int len = 1; |
int len = 1; |
4974 |
if (eptr >= md->end_subject) |
if (eptr >= md->end_subject) |
4975 |
{ |
{ |
4977 |
break; |
break; |
4978 |
} |
} |
4979 |
GETCHARLENTEST(c, eptr, len); |
GETCHARLENTEST(c, eptr, len); |
4980 |
prop_chartype = UCD_CHARTYPE(c); |
chartype = UCD_CHARTYPE(c); |
4981 |
if ((prop_chartype == ucp_Lu || |
if ((chartype == ucp_Lu || |
4982 |
prop_chartype == ucp_Ll || |
chartype == ucp_Ll || |
4983 |
prop_chartype == ucp_Lt) == prop_fail_result) |
chartype == ucp_Lt) == prop_fail_result) |
4984 |
break; |
break; |
4985 |
eptr+= len; |
eptr+= len; |
4986 |
} |
} |
4996 |
break; |
break; |
4997 |
} |
} |
4998 |
GETCHARLENTEST(c, eptr, len); |
GETCHARLENTEST(c, eptr, len); |
4999 |
prop_category = UCD_CATEGORY(c); |
if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break; |
|
if ((prop_category == prop_value) == prop_fail_result) |
|
|
break; |
|
5000 |
eptr+= len; |
eptr+= len; |
5001 |
} |
} |
5002 |
break; |
break; |
5011 |
break; |
break; |
5012 |
} |
} |
5013 |
GETCHARLENTEST(c, eptr, len); |
GETCHARLENTEST(c, eptr, len); |
5014 |
prop_chartype = UCD_CHARTYPE(c); |
if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break; |
|
if ((prop_chartype == prop_value) == prop_fail_result) |
|
|
break; |
|
5015 |
eptr+= len; |
eptr+= len; |
5016 |
} |
} |
5017 |
break; |
break; |
5026 |
break; |
break; |
5027 |
} |
} |
5028 |
GETCHARLENTEST(c, eptr, len); |
GETCHARLENTEST(c, eptr, len); |
5029 |
prop_script = UCD_SCRIPT(c); |
if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break; |
|
if ((prop_script == prop_value) == prop_fail_result) |
|
|
break; |
|
5030 |
eptr+= len; |
eptr+= len; |
5031 |
} |
} |
5032 |
break; |
break; |
5034 |
case PT_ALNUM: |
case PT_ALNUM: |
5035 |
for (i = min; i < max; i++) |
for (i = min; i < max; i++) |
5036 |
{ |
{ |
5037 |
|
int category; |
5038 |
int len = 1; |
int len = 1; |
5039 |
if (eptr >= md->end_subject) |
if (eptr >= md->end_subject) |
5040 |
{ |
{ |
5042 |
break; |
break; |
5043 |
} |
} |
5044 |
GETCHARLENTEST(c, eptr, len); |
GETCHARLENTEST(c, eptr, len); |
5045 |
prop_category = UCD_CATEGORY(c); |
category = UCD_CATEGORY(c); |
5046 |
if ((prop_category == ucp_L || prop_category == ucp_N) |
if ((category == ucp_L || category == ucp_N) == prop_fail_result) |
|
== prop_fail_result) |
|
5047 |
break; |
break; |
5048 |
eptr+= len; |
eptr+= len; |
5049 |
} |
} |
5059 |
break; |
break; |
5060 |
} |
} |
5061 |
GETCHARLENTEST(c, eptr, len); |
GETCHARLENTEST(c, eptr, len); |
5062 |
prop_category = UCD_CATEGORY(c); |
if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
5063 |
c == CHAR_FF || c == CHAR_CR) |
c == CHAR_FF || c == CHAR_CR) |
5064 |
== prop_fail_result) |
== prop_fail_result) |
5065 |
break; |
break; |
5077 |
break; |
break; |
5078 |
} |
} |
5079 |
GETCHARLENTEST(c, eptr, len); |
GETCHARLENTEST(c, eptr, len); |
5080 |
prop_category = UCD_CATEGORY(c); |
if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
|
5081 |
c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
5082 |
== prop_fail_result) |
== prop_fail_result) |
5083 |
break; |
break; |
5088 |
case PT_WORD: |
case PT_WORD: |
5089 |
for (i = min; i < max; i++) |
for (i = min; i < max; i++) |
5090 |
{ |
{ |
5091 |
|
int category; |
5092 |
int len = 1; |
int len = 1; |
5093 |
if (eptr >= md->end_subject) |
if (eptr >= md->end_subject) |
5094 |
{ |
{ |
5096 |
break; |
break; |
5097 |
} |
} |
5098 |
GETCHARLENTEST(c, eptr, len); |
GETCHARLENTEST(c, eptr, len); |
5099 |
prop_category = UCD_CATEGORY(c); |
category = UCD_CATEGORY(c); |
5100 |
if ((prop_category == ucp_L || prop_category == ucp_N || |
if ((category == ucp_L || category == ucp_N || |
5101 |
c == CHAR_UNDERSCORE) == prop_fail_result) |
c == CHAR_UNDERSCORE) == prop_fail_result) |
5102 |
break; |
break; |
5103 |
eptr+= len; |
eptr+= len; |
5127 |
{ |
{ |
5128 |
for (i = min; i < max; i++) |
for (i = min; i < max; i++) |
5129 |
{ |
{ |
5130 |
|
int len = 1; |
5131 |
if (eptr >= md->end_subject) |
if (eptr >= md->end_subject) |
5132 |
{ |
{ |
5133 |
SCHECK_PARTIAL(); |
SCHECK_PARTIAL(); |
5134 |
break; |
break; |
5135 |
} |
} |
5136 |
GETCHARINCTEST(c, eptr); |
if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
5137 |
prop_category = UCD_CATEGORY(c); |
if (UCD_CATEGORY(c) == ucp_M) break; |
5138 |
if (prop_category == ucp_M) break; |
eptr += len; |
5139 |
while (eptr < md->end_subject) |
while (eptr < md->end_subject) |
5140 |
{ |
{ |
5141 |
int len = 1; |
len = 1; |
5142 |
if (!utf8) c = *eptr; else |
if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
5143 |
{ |
if (UCD_CATEGORY(c) != ucp_M) break; |
|
GETCHARLEN(c, eptr, len); |
|
|
} |
|
|
prop_category = UCD_CATEGORY(c); |
|
|
if (prop_category != ucp_M) break; |
|
5144 |
eptr += len; |
eptr += len; |
5145 |
} |
} |
5146 |
} |
} |
5156 |
if (eptr-- == pp) break; /* Stop if tried at original pos */ |
if (eptr-- == pp) break; /* Stop if tried at original pos */ |
5157 |
for (;;) /* Move back over one extended */ |
for (;;) /* Move back over one extended */ |
5158 |
{ |
{ |
|
int len = 1; |
|
5159 |
if (!utf8) c = *eptr; else |
if (!utf8) c = *eptr; else |
5160 |
{ |
{ |
5161 |
BACKCHAR(eptr); |
BACKCHAR(eptr); |
5162 |
GETCHARLEN(c, eptr, len); |
GETCHAR(c, eptr); |
5163 |
} |
} |
5164 |
prop_category = UCD_CATEGORY(c); |
if (UCD_CATEGORY(c) != ucp_M) break; |
|
if (prop_category != ucp_M) break; |
|
5165 |
eptr--; |
eptr--; |
5166 |
} |
} |
5167 |
} |
} |
5225 |
while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
5226 |
} |
} |
5227 |
} |
} |
5228 |
else eptr = md->end_subject; /* Unlimited UTF-8 repeat */ |
else |
5229 |
|
{ |
5230 |
|
eptr = md->end_subject; /* Unlimited UTF-8 repeat */ |
5231 |
|
SCHECK_PARTIAL(); |
5232 |
|
} |
5233 |
break; |
break; |
5234 |
|
|
5235 |
/* The byte case is the same as non-UTF8 */ |
/* The byte case is the same as non-UTF8 */ |
5450 |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
5451 |
if (eptr-- == pp) break; /* Stop if tried at original pos */ |
if (eptr-- == pp) break; /* Stop if tried at original pos */ |
5452 |
BACKCHAR(eptr); |
BACKCHAR(eptr); |
5453 |
if (ctype == OP_ANYNL && eptr > pp && *eptr == '\n' && |
if (ctype == OP_ANYNL && eptr > pp && *eptr == '\n' && |
5454 |
eptr[-1] == '\r') eptr--; |
eptr[-1] == '\r') eptr--; |
5455 |
} |
} |
5456 |
} |
} |
5662 |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM47); |
RMATCH(eptr, ecode, offset_top, md, eptrb, RM47); |
5663 |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
5664 |
eptr--; |
eptr--; |
5665 |
if (ctype == OP_ANYNL && eptr > pp && *eptr == '\n' && |
if (ctype == OP_ANYNL && eptr > pp && *eptr == '\n' && |
5666 |
eptr[-1] == '\r') eptr--; |
eptr[-1] == '\r') eptr--; |
5667 |
} |
} |
5668 |
} |
} |
5804 |
PCRE_SPTR subject, int length, int start_offset, int options, int *offsets, |
PCRE_SPTR subject, int length, int start_offset, int options, int *offsets, |
5805 |
int offsetcount) |
int offsetcount) |
5806 |
{ |
{ |
5807 |
int rc, ocount; |
int rc, ocount, arg_offset_max; |
5808 |
int first_byte = -1; |
int first_byte = -1; |
5809 |
int req_byte = -1; |
int req_byte = -1; |
5810 |
int req_byte2 = -1; |
int req_byte2 = -1; |
5840 |
if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; |
if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; |
5841 |
if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET; |
if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET; |
5842 |
|
|
5843 |
/* This information is for finding all the numbers associated with a given |
/* These two settings are used in the code for checking a UTF-8 string that |
5844 |
name, for condition testing. */ |
follows immediately afterwards. Other values in the md block are used only |
5845 |
|
during "normal" pcre_exec() processing, not when the JIT support is in use, |
5846 |
|
so they are set up later. */ |
5847 |
|
|
5848 |
|
utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0; |
5849 |
|
md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 : |
5850 |
|
((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0; |
5851 |
|
|
5852 |
|
/* Check a UTF-8 string if required. Pass back the character offset and error |
5853 |
|
code for an invalid string if a results vector is available. */ |
5854 |
|
|
5855 |
|
#ifdef SUPPORT_UTF8 |
5856 |
|
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0) |
5857 |
|
{ |
5858 |
|
int erroroffset; |
5859 |
|
int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset); |
5860 |
|
if (errorcode != 0) |
5861 |
|
{ |
5862 |
|
if (offsetcount >= 2) |
5863 |
|
{ |
5864 |
|
offsets[0] = erroroffset; |
5865 |
|
offsets[1] = errorcode; |
5866 |
|
} |
5867 |
|
return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)? |
5868 |
|
PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8; |
5869 |
|
} |
5870 |
|
|
5871 |
|
/* Check that a start_offset points to the start of a UTF-8 character. */ |
5872 |
|
if (start_offset > 0 && start_offset < length && |
5873 |
|
(((USPTR)subject)[start_offset] & 0xc0) == 0x80) |
5874 |
|
return PCRE_ERROR_BADUTF8_OFFSET; |
5875 |
|
} |
5876 |
|
#endif |
5877 |
|
|
5878 |
|
/* If the pattern was successfully studied with JIT support, run the JIT |
5879 |
|
executable instead of the rest of this function. Most options must be set at |
5880 |
|
compile time for the JIT code to be usable. Fallback to the normal code path if |
5881 |
|
an unsupported flag is set. In particular, JIT does not support partial |
5882 |
|
matching. */ |
5883 |
|
|
5884 |
|
#ifdef SUPPORT_JIT |
5885 |
|
if (extra_data != NULL |
5886 |
|
&& (extra_data->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0 |
5887 |
|
&& extra_data->executable_jit != NULL |
5888 |
|
&& (options & ~(PCRE_NO_UTF8_CHECK | PCRE_NOTBOL | PCRE_NOTEOL | |
5889 |
|
PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART)) == 0) |
5890 |
|
return _pcre_jit_exec(re, extra_data->executable_jit, subject, length, |
5891 |
|
start_offset, options, ((extra_data->flags & PCRE_EXTRA_MATCH_LIMIT) == 0) |
5892 |
|
? MATCH_LIMIT : extra_data->match_limit, offsets, offsetcount); |
5893 |
|
#endif |
5894 |
|
|
5895 |
|
/* Carry on with non-JIT matching. This information is for finding all the |
5896 |
|
numbers associated with a given name, for condition testing. */ |
5897 |
|
|
5898 |
md->name_table = (uschar *)re + re->name_table_offset; |
md->name_table = (uschar *)re + re->name_table_offset; |
5899 |
md->name_count = re->name_count; |
md->name_count = re->name_count; |
5960 |
end_subject = md->end_subject; |
end_subject = md->end_subject; |
5961 |
|
|
5962 |
md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0; |
md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0; |
|
utf8 = md->utf8 = (re->options & PCRE_UTF8) != 0; |
|
5963 |
md->use_ucp = (re->options & PCRE_UCP) != 0; |
md->use_ucp = (re->options & PCRE_UCP) != 0; |
5964 |
md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0; |
md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0; |
5965 |
|
|
5970 |
md->noteol = (options & PCRE_NOTEOL) != 0; |
md->noteol = (options & PCRE_NOTEOL) != 0; |
5971 |
md->notempty = (options & PCRE_NOTEMPTY) != 0; |
md->notempty = (options & PCRE_NOTEMPTY) != 0; |
5972 |
md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0; |
md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0; |
5973 |
md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 : |
|
|
((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0; |
|
|
|
|
|
|
|
5974 |
md->hitend = FALSE; |
md->hitend = FALSE; |
5975 |
md->mark = NULL; /* In case never set */ |
md->mark = NULL; /* In case never set */ |
5976 |
|
|
5977 |
md->recursive = NULL; /* No recursion at top level */ |
md->recursive = NULL; /* No recursion at top level */ |
5978 |
|
md->hasthen = (re->flags & PCRE_HASTHEN) != 0; |
5979 |
|
|
5980 |
md->lcc = tables + lcc_offset; |
md->lcc = tables + lcc_offset; |
5981 |
md->ctypes = tables + ctypes_offset; |
md->ctypes = tables + ctypes_offset; |
6053 |
if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0) |
if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0) |
6054 |
return PCRE_ERROR_BADPARTIAL; |
return PCRE_ERROR_BADPARTIAL; |
6055 |
|
|
|
/* Check a UTF-8 string if required. Pass back the character offset and error |
|
|
code for an invalid string if a results vector is available. */ |
|
|
|
|
|
#ifdef SUPPORT_UTF8 |
|
|
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0) |
|
|
{ |
|
|
int erroroffset; |
|
|
int errorcode = _pcre_valid_utf8((USPTR)subject, length, &erroroffset); |
|
|
if (errorcode != 0) |
|
|
{ |
|
|
if (offsetcount >= 2) |
|
|
{ |
|
|
offsets[0] = erroroffset; |
|
|
offsets[1] = errorcode; |
|
|
} |
|
|
return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)? |
|
|
PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8; |
|
|
} |
|
|
|
|
|
/* Check that a start_offset points to the start of a UTF-8 character. */ |
|
|
|
|
|
if (start_offset > 0 && start_offset < length && |
|
|
(((USPTR)subject)[start_offset] & 0xc0) == 0x80) |
|
|
return PCRE_ERROR_BADUTF8_OFFSET; |
|
|
} |
|
|
#endif |
|
|
|
|
6056 |
/* If the expression has got more back references than the offsets supplied can |
/* If the expression has got more back references than the offsets supplied can |
6057 |
hold, we get a temporary chunk of working store to use during the matching. |
hold, we get a temporary chunk of working store to use during the matching. |
6058 |
Otherwise, we can use the vector supplied, rounding down its size to a multiple |
Otherwise, we can use the vector supplied, rounding down its size to a multiple |
6059 |
of 3. */ |
of 3. */ |
6060 |
|
|
6061 |
ocount = offsetcount - (offsetcount % 3); |
ocount = offsetcount - (offsetcount % 3); |
6062 |
|
arg_offset_max = (2*ocount)/3; |
6063 |
|
|
6064 |
if (re->top_backref > 0 && re->top_backref >= ocount/3) |
if (re->top_backref > 0 && re->top_backref >= ocount/3) |
6065 |
{ |
{ |
6322 |
md->start_match_ptr = start_match; |
md->start_match_ptr = start_match; |
6323 |
md->start_used_ptr = start_match; |
md->start_used_ptr = start_match; |
6324 |
md->match_call_count = 0; |
md->match_call_count = 0; |
6325 |
md->match_function_type = 0; |
md->match_function_type = 0; |
6326 |
md->end_offset_top = 0; |
md->end_offset_top = 0; |
6327 |
rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0); |
rc = match(start_match, md->start_code, start_match, NULL, 2, md, NULL, 0); |
6328 |
if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr; |
if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr; |
6434 |
{ |
{ |
6435 |
if (using_temporary_offsets) |
if (using_temporary_offsets) |
6436 |
{ |
{ |
6437 |
if (offsetcount >= 4) |
if (arg_offset_max >= 4) |
6438 |
{ |
{ |
6439 |
memcpy(offsets + 2, md->offset_vector + 2, |
memcpy(offsets + 2, md->offset_vector + 2, |
6440 |
(offsetcount - 2) * sizeof(int)); |
(arg_offset_max - 2) * sizeof(int)); |
6441 |
DPRINTF(("Copied offsets from temporary memory\n")); |
DPRINTF(("Copied offsets from temporary memory\n")); |
6442 |
} |
} |
6443 |
if (md->end_offset_top > offsetcount) md->offset_overflow = TRUE; |
if (md->end_offset_top > arg_offset_max) md->offset_overflow = TRUE; |
6444 |
DPRINTF(("Freeing temporary memory\n")); |
DPRINTF(("Freeing temporary memory\n")); |
6445 |
(pcre_free)(md->offset_vector); |
(pcre_free)(md->offset_vector); |
6446 |
} |
} |
6447 |
|
|
6448 |
/* Set the return code to the number of captured strings, or 0 if there are |
/* Set the return code to the number of captured strings, or 0 if there were |
6449 |
too many to fit into the vector. */ |
too many to fit into the vector. */ |
6450 |
|
|
6451 |
rc = md->offset_overflow? 0 : md->end_offset_top/2; |
rc = (md->offset_overflow && md->end_offset_top >= arg_offset_max)? |
6452 |
|
0 : md->end_offset_top/2; |
6453 |
|
|
6454 |
|
/* If there is space in the offset vector, set any unused pairs at the end of |
6455 |
|
the pattern to -1 for backwards compatibility. It is documented that this |
6456 |
|
happens. In earlier versions, the whole set of potential capturing offsets |
6457 |
|
was set to -1 each time round the loop, but this is handled differently now. |
6458 |
|
"Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only |
6459 |
|
those at the end that need unsetting here. We can't just unset them all at |
6460 |
|
the start of the whole thing because they may get set in one branch that is |
6461 |
|
not the final matching branch. */ |
6462 |
|
|
6463 |
|
if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL) |
6464 |
|
{ |
6465 |
|
register int *iptr, *iend; |
6466 |
|
int resetcount = 2 + re->top_bracket * 2; |
6467 |
|
if (resetcount > offsetcount) resetcount = ocount; |
6468 |
|
iptr = offsets + md->end_offset_top; |
6469 |
|
iend = offsets + resetcount; |
6470 |
|
while (iptr < iend) *iptr++ = -1; |
6471 |
|
} |
6472 |
|
|
6473 |
/* If there is space, set up the whole thing as substring 0. The value of |
/* If there is space, set up the whole thing as substring 0. The value of |
6474 |
md->start_match_ptr might be modified if \K was encountered on the success |
md->start_match_ptr might be modified if \K was encountered on the success |