7 |
|
|
8 |
This program is free software; you can redistribute it and/or modify |
This program is free software; you can redistribute it and/or modify |
9 |
it under the terms of the GNU General Public License as published by |
it under the terms of the GNU General Public License as published by |
10 |
the Free Software Foundation; either version 2, or (at your option) |
the Free Software Foundation; either version 2 of the License, or |
11 |
any later version. |
(at your option) any later version. |
12 |
|
|
13 |
This program is distributed in the hope that it will be useful, |
This program is distributed in the hope that it will be useful, |
14 |
but WITHOUT ANY WARRANTY; without even the implied warranty of |
but WITHOUT ANY WARRANTY; without even the implied warranty of |
17 |
|
|
18 |
You should have received a copy of the GNU General Public License |
You should have received a copy of the GNU General Public License |
19 |
along with this program; if not, write to the Free Software |
along with this program; if not, write to the Free Software |
20 |
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ |
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA |
21 |
|
02110-1301 USA |
22 |
|
*/ |
23 |
|
|
24 |
/* AIX requires this to be the first thing in the file. */ |
/* AIX requires this to be the first thing in the file. */ |
25 |
#if defined (_AIX) && !defined (REGEX_MALLOC) |
#if defined (_AIX) && !defined (REGEX_MALLOC) |
79 |
|
|
80 |
/* This must be nonzero for the wordchar and notwordchar pattern |
/* This must be nonzero for the wordchar and notwordchar pattern |
81 |
commands in re_match_2. */ |
commands in re_match_2. */ |
82 |
#ifndef Sword |
#ifndef Sword |
83 |
#define Sword 1 |
#define Sword 1 |
84 |
#endif |
#endif |
85 |
|
|
190 |
use `alloca' instead of `malloc'. This is because using malloc in |
use `alloca' instead of `malloc'. This is because using malloc in |
191 |
re_search* or re_match* could cause memory leaks when C-g is used in |
re_search* or re_match* could cause memory leaks when C-g is used in |
192 |
Emacs; also, malloc is slower and causes storage fragmentation. On |
Emacs; also, malloc is slower and causes storage fragmentation. On |
193 |
the other hand, malloc is more portable, and easier to debug. |
the other hand, malloc is more portable, and easier to debug. |
194 |
|
|
195 |
Because we sometimes use alloca, some routines have to be macros, |
Because we sometimes use alloca, some routines have to be macros, |
196 |
not functions -- `alloca'-allocated space disappears at the end of the |
not functions -- `alloca'-allocated space disappears at the end of the |
197 |
function it is called in. */ |
function it is called in. */ |
216 |
#ifndef _AIX /* Already did AIX, up at the top. */ |
#ifndef _AIX /* Already did AIX, up at the top. */ |
217 |
char *alloca (); |
char *alloca (); |
218 |
#endif /* not _AIX */ |
#endif /* not _AIX */ |
219 |
#endif /* not HAVE_ALLOCA_H */ |
#endif /* not HAVE_ALLOCA_H */ |
220 |
#endif /* not __GNUC__ */ |
#endif /* not __GNUC__ */ |
221 |
|
|
222 |
#endif /* not alloca */ |
#endif /* not alloca */ |
321 |
|
|
322 |
/* Analogously, for end of buffer/string. */ |
/* Analogously, for end of buffer/string. */ |
323 |
endbuf, |
endbuf, |
324 |
|
|
325 |
/* Followed by two byte relative address to which to jump. */ |
/* Followed by two byte relative address to which to jump. */ |
326 |
jump, |
jump, |
327 |
|
|
328 |
/* Same as jump, but marks the end of an alternative. */ |
/* Same as jump, but marks the end of an alternative. */ |
329 |
jump_past_alt, |
jump_past_alt, |
331 |
/* Followed by two-byte relative address of place to resume at |
/* Followed by two-byte relative address of place to resume at |
332 |
in case of failure. */ |
in case of failure. */ |
333 |
on_failure_jump, |
on_failure_jump, |
334 |
|
|
335 |
/* Like on_failure_jump, but pushes a placeholder instead of the |
/* Like on_failure_jump, but pushes a placeholder instead of the |
336 |
current string position when executed. */ |
current string position when executed. */ |
337 |
on_failure_keep_string_jump, |
on_failure_keep_string_jump, |
338 |
|
|
339 |
/* Throw away latest failure point and then jump to following |
/* Throw away latest failure point and then jump to following |
340 |
two-byte relative address. */ |
two-byte relative address. */ |
341 |
pop_failure_jump, |
pop_failure_jump, |
431 |
int *dest; |
int *dest; |
432 |
unsigned char *source; |
unsigned char *source; |
433 |
{ |
{ |
434 |
int temp = SIGN_EXTEND_CHAR (*(source + 1)); |
int temp = SIGN_EXTEND_CHAR (*(source + 1)); |
435 |
*dest = *source & 0377; |
*dest = *source & 0377; |
436 |
*dest += temp << 8; |
*dest += temp << 8; |
437 |
} |
} |
457 |
extract_number_and_incr (destination, source) |
extract_number_and_incr (destination, source) |
458 |
int *destination; |
int *destination; |
459 |
unsigned char **source; |
unsigned char **source; |
460 |
{ |
{ |
461 |
extract_number (destination, *source); |
extract_number (destination, *source); |
462 |
*source += 2; |
*source += 2; |
463 |
} |
} |
506 |
char *fastmap; |
char *fastmap; |
507 |
{ |
{ |
508 |
unsigned was_a_range = 0; |
unsigned was_a_range = 0; |
509 |
unsigned i = 0; |
unsigned i = 0; |
510 |
|
|
511 |
while (i < (1 << BYTEWIDTH)) |
while (i < (1 << BYTEWIDTH)) |
512 |
{ |
{ |
513 |
if (fastmap[i++]) |
if (fastmap[i++]) |
526 |
} |
} |
527 |
} |
} |
528 |
} |
} |
529 |
putchar ('\n'); |
putchar ('\n'); |
530 |
} |
} |
531 |
|
|
532 |
|
|
547 |
printf ("(null)\n"); |
printf ("(null)\n"); |
548 |
return; |
return; |
549 |
} |
} |
550 |
|
|
551 |
/* Loop over pattern commands. */ |
/* Loop over pattern commands. */ |
552 |
while (p < pend) |
while (p < pend) |
553 |
{ |
{ |
596 |
|
|
597 |
printf ("/charset [%s", |
printf ("/charset [%s", |
598 |
(re_opcode_t) *(p - 1) == charset_not ? "^" : ""); |
(re_opcode_t) *(p - 1) == charset_not ? "^" : ""); |
599 |
|
|
600 |
assert (p + *p < pend); |
assert (p + *p < pend); |
601 |
|
|
602 |
for (c = 0; c < 256; c++) |
for (c = 0; c < 256; c++) |
615 |
printchar (last); |
printchar (last); |
616 |
in_range = 0; |
in_range = 0; |
617 |
} |
} |
618 |
|
|
619 |
if (! in_range) |
if (! in_range) |
620 |
printchar (c); |
printchar (c); |
621 |
|
|
657 |
case push_dummy_failure: |
case push_dummy_failure: |
658 |
printf ("/push_dummy_failure"); |
printf ("/push_dummy_failure"); |
659 |
break; |
break; |
660 |
|
|
661 |
case maybe_pop_jump: |
case maybe_pop_jump: |
662 |
extract_number_and_incr (&mcnt, &p); |
extract_number_and_incr (&mcnt, &p); |
663 |
printf ("/maybe_pop_jump to %d", p + mcnt - start); |
printf ("/maybe_pop_jump to %d", p + mcnt - start); |
666 |
case pop_failure_jump: |
case pop_failure_jump: |
667 |
extract_number_and_incr (&mcnt, &p); |
extract_number_and_incr (&mcnt, &p); |
668 |
printf ("/pop_failure_jump to %d", p + mcnt - start); |
printf ("/pop_failure_jump to %d", p + mcnt - start); |
669 |
break; |
break; |
670 |
|
|
671 |
case jump_past_alt: |
case jump_past_alt: |
672 |
extract_number_and_incr (&mcnt, &p); |
extract_number_and_incr (&mcnt, &p); |
673 |
printf ("/jump_past_alt to %d", p + mcnt - start); |
printf ("/jump_past_alt to %d", p + mcnt - start); |
674 |
break; |
break; |
675 |
|
|
676 |
case jump: |
case jump: |
677 |
extract_number_and_incr (&mcnt, &p); |
extract_number_and_incr (&mcnt, &p); |
678 |
printf ("/jump to %d", p + mcnt - start); |
printf ("/jump to %d", p + mcnt - start); |
679 |
break; |
break; |
680 |
|
|
681 |
case succeed_n: |
case succeed_n: |
682 |
extract_number_and_incr (&mcnt, &p); |
extract_number_and_incr (&mcnt, &p); |
683 |
extract_number_and_incr (&mcnt2, &p); |
extract_number_and_incr (&mcnt2, &p); |
684 |
printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2); |
printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2); |
685 |
break; |
break; |
686 |
|
|
687 |
case jump_n: |
case jump_n: |
688 |
extract_number_and_incr (&mcnt, &p); |
extract_number_and_incr (&mcnt, &p); |
689 |
extract_number_and_incr (&mcnt2, &p); |
extract_number_and_incr (&mcnt2, &p); |
690 |
printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2); |
printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2); |
691 |
break; |
break; |
692 |
|
|
693 |
case set_number_at: |
case set_number_at: |
694 |
extract_number_and_incr (&mcnt, &p); |
extract_number_and_incr (&mcnt, &p); |
695 |
extract_number_and_incr (&mcnt2, &p); |
extract_number_and_incr (&mcnt2, &p); |
696 |
printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2); |
printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2); |
697 |
break; |
break; |
698 |
|
|
699 |
case wordbound: |
case wordbound: |
700 |
printf ("/wordbound"); |
printf ("/wordbound"); |
701 |
break; |
break; |
707 |
case wordbeg: |
case wordbeg: |
708 |
printf ("/wordbeg"); |
printf ("/wordbeg"); |
709 |
break; |
break; |
710 |
|
|
711 |
case wordend: |
case wordend: |
712 |
printf ("/wordend"); |
printf ("/wordend"); |
713 |
|
|
714 |
#ifdef emacs |
#ifdef emacs |
715 |
case before_dot: |
case before_dot: |
716 |
printf ("/before_dot"); |
printf ("/before_dot"); |
729 |
mcnt = *p++; |
mcnt = *p++; |
730 |
printf ("/%d", mcnt); |
printf ("/%d", mcnt); |
731 |
break; |
break; |
732 |
|
|
733 |
case notsyntaxspec: |
case notsyntaxspec: |
734 |
printf ("/notsyntaxspec"); |
printf ("/notsyntaxspec"); |
735 |
mcnt = *p++; |
mcnt = *p++; |
740 |
case wordchar: |
case wordchar: |
741 |
printf ("/wordchar"); |
printf ("/wordchar"); |
742 |
break; |
break; |
743 |
|
|
744 |
case notwordchar: |
case notwordchar: |
745 |
printf ("/notwordchar"); |
printf ("/notwordchar"); |
746 |
break; |
break; |
800 |
int size2; |
int size2; |
801 |
{ |
{ |
802 |
unsigned this_char; |
unsigned this_char; |
803 |
|
|
804 |
if (where == NULL) |
if (where == NULL) |
805 |
printf ("(null)"); |
printf ("(null)"); |
806 |
else |
else |
810 |
for (this_char = where - string1; this_char < size1; this_char++) |
for (this_char = where - string1; this_char < size1; this_char++) |
811 |
printchar (string1[this_char]); |
printchar (string1[this_char]); |
812 |
|
|
813 |
where = string2; |
where = string2; |
814 |
} |
} |
815 |
|
|
816 |
for (this_char = where - string2; this_char < size2; this_char++) |
for (this_char = where - string2; this_char < size2; this_char++) |
851 |
reg_syntax_t syntax; |
reg_syntax_t syntax; |
852 |
{ |
{ |
853 |
reg_syntax_t ret = re_syntax_options; |
reg_syntax_t ret = re_syntax_options; |
854 |
|
|
855 |
re_syntax_options = syntax; |
re_syntax_options = syntax; |
856 |
return ret; |
return ret; |
857 |
} |
} |
889 |
ralloc heap) shift the data out from underneath the regexp |
ralloc heap) shift the data out from underneath the regexp |
890 |
routines. |
routines. |
891 |
|
|
892 |
Here's another reason to avoid allocation: Emacs |
Here's another reason to avoid allocation: Emacs |
893 |
processes input from X in a signal handler; processing X input may |
processes input from X in a signal handler; processing X input may |
894 |
call malloc; if input arrives while a matching routine is calling |
call malloc; if input arrives while a matching routine is calling |
895 |
malloc, then we're scrod. But Emacs can't just block input while |
malloc, then we're scrod. But Emacs can't just block input while |
911 |
/* Failure stack declarations and macros; both re_compile_fastmap and |
/* Failure stack declarations and macros; both re_compile_fastmap and |
912 |
re_match_2 use a failure stack. These have to be macros because of |
re_match_2 use a failure stack. These have to be macros because of |
913 |
REGEX_ALLOCATE. */ |
REGEX_ALLOCATE. */ |
914 |
|
|
915 |
|
|
916 |
/* Number of failure points for which to initially allocate space |
/* Number of failure points for which to initially allocate space |
917 |
when matching. If this number is exceeded, we allocate more |
when matching. If this number is exceeded, we allocate more |
966 |
/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. |
/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. |
967 |
|
|
968 |
Return 1 if succeeds, and 0 if either ran out of memory |
Return 1 if succeeds, and 0 if either ran out of memory |
969 |
allocating space for it or it was already too large. |
allocating space for it or it was already too large. |
970 |
|
|
971 |
REGEX_REALLOCATE requires `destination' be declared. */ |
REGEX_REALLOCATE requires `destination' be declared. */ |
972 |
|
|
973 |
#define DOUBLE_FAIL_STACK(fail_stack) \ |
#define DOUBLE_FAIL_STACK(fail_stack) \ |
984 |
1))) |
1))) |
985 |
|
|
986 |
|
|
987 |
/* Push PATTERN_OP on FAIL_STACK. |
/* Push PATTERN_OP on FAIL_STACK. |
988 |
|
|
989 |
Return 1 if was able to do so and 0 if ran out of memory allocating |
Return 1 if was able to do so and 0 if ran out of memory allocating |
990 |
space to do so. */ |
space to do so. */ |
1015 |
|
|
1016 |
|
|
1017 |
/* Push the information about the state we will need |
/* Push the information about the state we will need |
1018 |
if we ever fail back to it. |
if we ever fail back to it. |
1019 |
|
|
1020 |
Requires variables fail_stack, regstart, regend, reg_info, and |
Requires variables fail_stack, regstart, regend, reg_info, and |
1021 |
num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be |
num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be |
1022 |
declared. |
declared. |
1023 |
|
|
1024 |
Does `return FAILURE_CODE' if runs out of memory. */ |
Does `return FAILURE_CODE' if runs out of memory. */ |
1025 |
|
|
1026 |
#define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \ |
#define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \ |
1128 |
LOW_REG, HIGH_REG -- the highest and lowest active registers. |
LOW_REG, HIGH_REG -- the highest and lowest active registers. |
1129 |
REGSTART, REGEND -- arrays of string positions. |
REGSTART, REGEND -- arrays of string positions. |
1130 |
REG_INFO -- array of information about each subexpression. |
REG_INFO -- array of information about each subexpression. |
1131 |
|
|
1132 |
Also assumes the variables `fail_stack' and (if debugging), `bufp', |
Also assumes the variables `fail_stack' and (if debugging), `bufp', |
1133 |
`pend', `string1', `size1', `string2', and `size2'. */ |
`pend', `string1', `size1', `string2', and `size2'. */ |
1134 |
|
|
1196 |
onto the failure stack. Other register information, such as the |
onto the failure stack. Other register information, such as the |
1197 |
starting and ending positions (which are addresses), and the list of |
starting and ending positions (which are addresses), and the list of |
1198 |
inner groups (which is a bits list) are maintained in separate |
inner groups (which is a bits list) are maintained in separate |
1199 |
variables. |
variables. |
1200 |
|
|
1201 |
We are making a (strictly speaking) nonportable assumption here: that |
We are making a (strictly speaking) nonportable assumption here: that |
1202 |
the compiler will pack our bit fields into something that fits into |
the compiler will pack our bit fields into something that fits into |
1203 |
the type of `word', i.e., is something that fits into one item on the |
the type of `word', i.e., is something that fits into one item on the |
1255 |
static const char ** regstart, ** regend; |
static const char ** regstart, ** regend; |
1256 |
static const char ** old_regstart, ** old_regend; |
static const char ** old_regstart, ** old_regend; |
1257 |
static const char **best_regstart, **best_regend; |
static const char **best_regstart, **best_regend; |
1258 |
static register_info_type *reg_info; |
static register_info_type *reg_info; |
1259 |
static const char **reg_dummy; |
static const char **reg_dummy; |
1260 |
static register_info_type *reg_info_dummy; |
static register_info_type *reg_info_dummy; |
1261 |
#endif |
#endif |
1269 |
static boolean group_in_compile_stack (); |
static boolean group_in_compile_stack (); |
1270 |
static reg_errcode_t compile_range (); |
static reg_errcode_t compile_range (); |
1271 |
|
|
1272 |
/* Fetch the next character in the uncompiled pattern---translating it |
/* Fetch the next character in the uncompiled pattern---translating it |
1273 |
if necessary. Also cast from a signed character in the constant |
if necessary. Also cast from a signed character in the constant |
1274 |
string passed to us by the user to an unsigned char that we can use |
string passed to us by the user to an unsigned char that we can use |
1275 |
as an array index (in, e.g., `translate'). */ |
as an array index (in, e.g., `translate'). */ |
1409 |
pattern_offset_t begalt_offset; |
pattern_offset_t begalt_offset; |
1410 |
pattern_offset_t fixup_alt_jump; |
pattern_offset_t fixup_alt_jump; |
1411 |
pattern_offset_t inner_group_offset; |
pattern_offset_t inner_group_offset; |
1412 |
pattern_offset_t laststart_offset; |
pattern_offset_t laststart_offset; |
1413 |
regnum_t regnum; |
regnum_t regnum; |
1414 |
} compile_stack_elt_t; |
} compile_stack_elt_t; |
1415 |
|
|
1452 |
PATFETCH (c); \ |
PATFETCH (c); \ |
1453 |
} \ |
} \ |
1454 |
} \ |
} \ |
1455 |
} |
} |
1456 |
|
|
1457 |
#define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ |
#define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ |
1458 |
|
|
1478 |
`fastmap_accurate' is zero; |
`fastmap_accurate' is zero; |
1479 |
`re_nsub' is the number of subexpressions in PATTERN; |
`re_nsub' is the number of subexpressions in PATTERN; |
1480 |
`not_bol' and `not_eol' are zero; |
`not_bol' and `not_eol' are zero; |
1481 |
|
|
1482 |
The `fastmap' and `newline_anchor' fields are neither |
The `fastmap' and `newline_anchor' fields are neither |
1483 |
examined nor set. */ |
examined nor set. */ |
1484 |
|
|
1497 |
`char *' (i.e., signed), we declare these variables as unsigned, so |
`char *' (i.e., signed), we declare these variables as unsigned, so |
1498 |
they can be reliably used as array indices. */ |
they can be reliably used as array indices. */ |
1499 |
register unsigned char c, c1; |
register unsigned char c, c1; |
1500 |
|
|
1501 |
/* A random temporary spot in PATTERN. */ |
/* A random temporary spot in PATTERN. */ |
1502 |
const char *p1; |
const char *p1; |
1503 |
|
|
1504 |
/* Points to the end of the buffer, where we should append. */ |
/* Points to the end of the buffer, where we should append. */ |
1505 |
register unsigned char *b; |
register unsigned char *b; |
1506 |
|
|
1507 |
/* Keeps track of unclosed groups. */ |
/* Keeps track of unclosed groups. */ |
1508 |
compile_stack_type compile_stack; |
compile_stack_type compile_stack; |
1509 |
|
|
1510 |
/* Points to the current (ending) position in the pattern. */ |
/* Points to the current (ending) position in the pattern. */ |
1511 |
const char *p = pattern; |
const char *p = pattern; |
1512 |
const char *pend = pattern + size; |
const char *pend = pattern + size; |
1513 |
|
|
1514 |
/* How to translate the characters in the pattern. */ |
/* How to translate the characters in the pattern. */ |
1515 |
char *translate = bufp->translate; |
char *translate = bufp->translate; |
1516 |
|
|
1531 |
/* Place in the uncompiled pattern (i.e., the {) to |
/* Place in the uncompiled pattern (i.e., the {) to |
1532 |
which to go back if the interval is invalid. */ |
which to go back if the interval is invalid. */ |
1533 |
const char *beg_interval; |
const char *beg_interval; |
1534 |
|
|
1535 |
/* Address of the place where a forward jump should go to the end of |
/* Address of the place where a forward jump should go to the end of |
1536 |
the containing expression. Each alternative of an `or' -- except the |
the containing expression. Each alternative of an `or' -- except the |
1537 |
last -- ends with a forward jump of this sort. */ |
last -- ends with a forward jump of this sort. */ |
1547 |
if (debug) |
if (debug) |
1548 |
{ |
{ |
1549 |
unsigned debug_count; |
unsigned debug_count; |
1550 |
|
|
1551 |
for (debug_count = 0; debug_count < size; debug_count++) |
for (debug_count = 0; debug_count < size; debug_count++) |
1552 |
printchar (pattern[debug_count]); |
printchar (pattern[debug_count]); |
1553 |
putchar ('\n'); |
putchar ('\n'); |
1571 |
printer (for debugging) will think there's no pattern. We reset it |
printer (for debugging) will think there's no pattern. We reset it |
1572 |
at the end. */ |
at the end. */ |
1573 |
bufp->used = 0; |
bufp->used = 0; |
1574 |
|
|
1575 |
/* Always count groups, whether or not bufp->no_sub is set. */ |
/* Always count groups, whether or not bufp->no_sub is set. */ |
1576 |
bufp->re_nsub = 0; |
bufp->re_nsub = 0; |
1577 |
|
|
1578 |
#if !defined (emacs) && !defined (SYNTAX_TABLE) |
#if !defined (emacs) && !defined (SYNTAX_TABLE) |
1579 |
/* Initialize the syntax table. */ |
/* Initialize the syntax table. */ |
1624 |
case '$': |
case '$': |
1625 |
{ |
{ |
1626 |
if ( /* If at end of pattern, it's an operator. */ |
if ( /* If at end of pattern, it's an operator. */ |
1627 |
p == pend |
p == pend |
1628 |
/* If context independent, it's an operator. */ |
/* If context independent, it's an operator. */ |
1629 |
|| syntax & RE_CONTEXT_INDEP_ANCHORS |
|| syntax & RE_CONTEXT_INDEP_ANCHORS |
1630 |
/* Otherwise, depends on what's next. */ |
/* Otherwise, depends on what's next. */ |
1655 |
{ |
{ |
1656 |
/* Are we optimizing this jump? */ |
/* Are we optimizing this jump? */ |
1657 |
boolean keep_string_p = false; |
boolean keep_string_p = false; |
1658 |
|
|
1659 |
/* 1 means zero (many) matches is allowed. */ |
/* 1 means zero (many) matches is allowed. */ |
1660 |
char zero_times_ok = 0, many_times_ok = 0; |
char zero_times_ok = 0, many_times_ok = 0; |
1661 |
|
|
1703 |
|
|
1704 |
/* Star, etc. applied to an empty pattern is equivalent |
/* Star, etc. applied to an empty pattern is equivalent |
1705 |
to an empty pattern. */ |
to an empty pattern. */ |
1706 |
if (!laststart) |
if (!laststart) |
1707 |
break; |
break; |
1708 |
|
|
1709 |
/* Now we know whether or not zero matches is allowed |
/* Now we know whether or not zero matches is allowed |
1712 |
{ /* More than one repetition is allowed, so put in at the |
{ /* More than one repetition is allowed, so put in at the |
1713 |
end a backward relative jump from `b' to before the next |
end a backward relative jump from `b' to before the next |
1714 |
jump we're going to put in below (which jumps from |
jump we're going to put in below (which jumps from |
1715 |
laststart to after this jump). |
laststart to after this jump). |
1716 |
|
|
1717 |
But if we are at the `*' in the exact sequence `.*\n', |
But if we are at the `*' in the exact sequence `.*\n', |
1718 |
insert an unconditional jump backwards to the ., |
insert an unconditional jump backwards to the ., |
1789 |
|
|
1790 |
/* We test `*p == '^' twice, instead of using an if |
/* We test `*p == '^' twice, instead of using an if |
1791 |
statement, so we only need one BUF_PUSH. */ |
statement, so we only need one BUF_PUSH. */ |
1792 |
BUF_PUSH (*p == '^' ? charset_not : charset); |
BUF_PUSH (*p == '^' ? charset_not : charset); |
1793 |
if (*p == '^') |
if (*p == '^') |
1794 |
p++; |
p++; |
1795 |
|
|
1839 |
was a character: if this is a hyphen not at the |
was a character: if this is a hyphen not at the |
1840 |
beginning or the end of a list, then it's the range |
beginning or the end of a list, then it's the range |
1841 |
operator. */ |
operator. */ |
1842 |
if (c == '-' |
if (c == '-' |
1843 |
&& !(p - 2 >= pattern && p[-2] == '[') |
&& !(p - 2 >= pattern && p[-2] == '[') |
1844 |
&& !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') |
&& !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') |
1845 |
&& *p != ']') |
&& *p != ']') |
1846 |
{ |
{ |
1855 |
|
|
1856 |
/* Move past the `-'. */ |
/* Move past the `-'. */ |
1857 |
PATFETCH (c1); |
PATFETCH (c1); |
1858 |
|
|
1859 |
ret = compile_range (&p, pend, translate, syntax, b); |
ret = compile_range (&p, pend, translate, syntax, b); |
1860 |
if (ret != REG_NOERROR) FREE_STACK_RETURN (ret); |
if (ret != REG_NOERROR) FREE_STACK_RETURN (ret); |
1861 |
} |
} |
1884 |
str[c1] = '\0'; |
str[c1] = '\0'; |
1885 |
|
|
1886 |
/* If isn't a word bracketed by `[:' and:`]': |
/* If isn't a word bracketed by `[:' and:`]': |
1887 |
undo the ending character, the letters, and leave |
undo the ending character, the letters, and leave |
1888 |
the leading `:' and `[' (but set bits for them). */ |
the leading `:' and `[' (but set bits for them). */ |
1889 |
if (c == ':' && *p == ']') |
if (c == ':' && *p == ']') |
1890 |
{ |
{ |
1901 |
boolean is_space = STREQ (str, "space"); |
boolean is_space = STREQ (str, "space"); |
1902 |
boolean is_upper = STREQ (str, "upper"); |
boolean is_upper = STREQ (str, "upper"); |
1903 |
boolean is_xdigit = STREQ (str, "xdigit"); |
boolean is_xdigit = STREQ (str, "xdigit"); |
1904 |
|
|
1905 |
if (!IS_CHAR_CLASS (str)) |
if (!IS_CHAR_CLASS (str)) |
1906 |
FREE_STACK_RETURN (REG_ECTYPE); |
FREE_STACK_RETURN (REG_ECTYPE); |
1907 |
|
|
1908 |
/* Throw away the ] at the end of the character |
/* Throw away the ] at the end of the character |
1909 |
class. */ |
class. */ |
1910 |
PATFETCH (c); |
PATFETCH (c); |
1911 |
|
|
1912 |
if (p == pend) FREE_STACK_RETURN (REG_EBRACK); |
if (p == pend) FREE_STACK_RETURN (REG_EBRACK); |
1913 |
|
|
1936 |
else |
else |
1937 |
{ |
{ |
1938 |
c1++; |
c1++; |
1939 |
while (c1--) |
while (c1--) |
1940 |
PATUNFETCH; |
PATUNFETCH; |
1941 |
SET_LIST_BIT ('['); |
SET_LIST_BIT ('['); |
1942 |
SET_LIST_BIT (':'); |
SET_LIST_BIT (':'); |
1952 |
|
|
1953 |
/* Discard any (non)matching list bytes that are all 0 at the |
/* Discard any (non)matching list bytes that are all 0 at the |
1954 |
end of the map. Decrease the map-length byte too. */ |
end of the map. Decrease the map-length byte too. */ |
1955 |
while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) |
while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) |
1956 |
b[-1]--; |
b[-1]--; |
1957 |
b += b[-1]; |
b += b[-1]; |
1958 |
} |
} |
1959 |
break; |
break; |
2013 |
regnum++; |
regnum++; |
2014 |
|
|
2015 |
if (COMPILE_STACK_FULL) |
if (COMPILE_STACK_FULL) |
2016 |
{ |
{ |
2017 |
RETALLOC (compile_stack.stack, compile_stack.size << 1, |
RETALLOC (compile_stack.stack, compile_stack.size << 1, |
2018 |
compile_stack_elt_t); |
compile_stack_elt_t); |
2019 |
if (compile_stack.stack == NULL) return REG_ESPACE; |
if (compile_stack.stack == NULL) return REG_ESPACE; |
2026 |
whole pattern moves because of realloc, they will still |
whole pattern moves because of realloc, they will still |
2027 |
be valid. */ |
be valid. */ |
2028 |
COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; |
COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; |
2029 |
COMPILE_STACK_TOP.fixup_alt_jump |
COMPILE_STACK_TOP.fixup_alt_jump |
2030 |
= fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; |
= fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; |
2031 |
COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer; |
COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer; |
2032 |
COMPILE_STACK_TOP.regnum = regnum; |
COMPILE_STACK_TOP.regnum = regnum; |
2040 |
COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2; |
COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2; |
2041 |
BUF_PUSH_3 (start_memory, regnum, 0); |
BUF_PUSH_3 (start_memory, regnum, 0); |
2042 |
} |
} |
2043 |
|
|
2044 |
compile_stack.avail++; |
compile_stack.avail++; |
2045 |
|
|
2046 |
fixup_alt_jump = 0; |
fixup_alt_jump = 0; |
2069 |
`pop_failure_jump' to pop. See comments at |
`pop_failure_jump' to pop. See comments at |
2070 |
`push_dummy_failure' in `re_match_2'. */ |
`push_dummy_failure' in `re_match_2'. */ |
2071 |
BUF_PUSH (push_dummy_failure); |
BUF_PUSH (push_dummy_failure); |
2072 |
|
|
2073 |
/* We allocated space for this jump when we assigned |
/* We allocated space for this jump when we assigned |
2074 |
to `fixup_alt_jump', in the `handle_alt' case below. */ |
to `fixup_alt_jump', in the `handle_alt' case below. */ |
2075 |
STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1); |
STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1); |
2091 |
as in `(ab)c(de)' -- the second group is #2. */ |
as in `(ab)c(de)' -- the second group is #2. */ |
2092 |
regnum_t this_group_regnum; |
regnum_t this_group_regnum; |
2093 |
|
|
2094 |
compile_stack.avail--; |
compile_stack.avail--; |
2095 |
begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; |
begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; |
2096 |
fixup_alt_jump |
fixup_alt_jump |
2097 |
= COMPILE_STACK_TOP.fixup_alt_jump |
= COMPILE_STACK_TOP.fixup_alt_jump |
2098 |
? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 |
? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 |
2099 |
: 0; |
: 0; |
2100 |
laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; |
laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; |
2101 |
this_group_regnum = COMPILE_STACK_TOP.regnum; |
this_group_regnum = COMPILE_STACK_TOP.regnum; |
2110 |
{ |
{ |
2111 |
unsigned char *inner_group_loc |
unsigned char *inner_group_loc |
2112 |
= bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; |
= bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; |
2113 |
|
|
2114 |
*inner_group_loc = regnum - this_group_regnum; |
*inner_group_loc = regnum - this_group_regnum; |
2115 |
BUF_PUSH_3 (stop_memory, this_group_regnum, |
BUF_PUSH_3 (stop_memory, this_group_regnum, |
2116 |
regnum - this_group_regnum); |
regnum - this_group_regnum); |
2139 |
jump (put in below, which in turn will jump to the next |
jump (put in below, which in turn will jump to the next |
2140 |
(if any) alternative's such jump, etc.). The last such |
(if any) alternative's such jump, etc.). The last such |
2141 |
jump jumps to the correct final destination. A picture: |
jump jumps to the correct final destination. A picture: |
2142 |
_____ _____ |
_____ _____ |
2143 |
| | | | |
| | | | |
2144 |
| v | v |
| v | v |
2145 |
a | b | c |
a | b | c |
2146 |
|
|
2147 |
If we are at `b', then fixup_alt_jump right now points to a |
If we are at `b', then fixup_alt_jump right now points to a |
2148 |
three-byte space after `a'. We'll put in the jump, set |
three-byte space after `a'. We'll put in the jump, set |
2164 |
break; |
break; |
2165 |
|
|
2166 |
|
|
2167 |
case '{': |
case '{': |
2168 |
/* If \{ is a literal. */ |
/* If \{ is a literal. */ |
2169 |
if (!(syntax & RE_INTERVALS) |
if (!(syntax & RE_INTERVALS) |
2170 |
/* If we're at `\{' and it's not the open-interval |
/* If we're at `\{' and it's not the open-interval |
2171 |
operator. */ |
operator. */ |
2172 |
|| ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) |
|| ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) |
2173 |
|| (p - 2 == pattern && p == pend)) |
|| (p - 2 == pattern && p == pend)) |
2206 |
{ |
{ |
2207 |
if (syntax & RE_NO_BK_BRACES) |
if (syntax & RE_NO_BK_BRACES) |
2208 |
goto unfetch_interval; |
goto unfetch_interval; |
2209 |
else |
else |
2210 |
FREE_STACK_RETURN (REG_BADBR); |
FREE_STACK_RETURN (REG_BADBR); |
2211 |
} |
} |
2212 |
|
|
2213 |
if (!(syntax & RE_NO_BK_BRACES)) |
if (!(syntax & RE_NO_BK_BRACES)) |
2214 |
{ |
{ |
2215 |
if (c != '\\') FREE_STACK_RETURN (REG_EBRACE); |
if (c != '\\') FREE_STACK_RETURN (REG_EBRACE); |
2216 |
|
|
2221 |
{ |
{ |
2222 |
if (syntax & RE_NO_BK_BRACES) |
if (syntax & RE_NO_BK_BRACES) |
2223 |
goto unfetch_interval; |
goto unfetch_interval; |
2224 |
else |
else |
2225 |
FREE_STACK_RETURN (REG_BADBR); |
FREE_STACK_RETURN (REG_BADBR); |
2226 |
} |
} |
2227 |
|
|
2257 |
jump_n <succeed_n addr> <jump count> |
jump_n <succeed_n addr> <jump count> |
2258 |
(The upper bound and `jump_n' are omitted if |
(The upper bound and `jump_n' are omitted if |
2259 |
`upper_bound' is 1, though.) */ |
`upper_bound' is 1, though.) */ |
2260 |
else |
else |
2261 |
{ /* If the upper bound is > 1, we need to insert |
{ /* If the upper bound is > 1, we need to insert |
2262 |
more at the end of the loop. */ |
more at the end of the loop. */ |
2263 |
unsigned nbytes = 10 + (upper_bound > 1) * 10; |
unsigned nbytes = 10 + (upper_bound > 1) * 10; |
2274 |
lower_bound); |
lower_bound); |
2275 |
b += 5; |
b += 5; |
2276 |
|
|
2277 |
/* Code to initialize the lower bound. Insert |
/* Code to initialize the lower bound. Insert |
2278 |
before the `succeed_n'. The `5' is the last two |
before the `succeed_n'. The `5' is the last two |
2279 |
bytes of this `set_number_at', plus 3 bytes of |
bytes of this `set_number_at', plus 3 bytes of |
2280 |
the following `succeed_n'. */ |
the following `succeed_n'. */ |
2285 |
{ /* More than one repetition is allowed, so |
{ /* More than one repetition is allowed, so |
2286 |
append a backward jump to the `succeed_n' |
append a backward jump to the `succeed_n' |
2287 |
that starts this interval. |
that starts this interval. |
2288 |
|
|
2289 |
When we've reached this during matching, |
When we've reached this during matching, |
2290 |
we'll have matched the interval once, so |
we'll have matched the interval once, so |
2291 |
jump back only `upper_bound - 1' times. */ |
jump back only `upper_bound - 1' times. */ |
2303 |
so everything is getting moved up by 5. |
so everything is getting moved up by 5. |
2304 |
Conclusion: (b - 2) - (laststart + 3) + 5, |
Conclusion: (b - 2) - (laststart + 3) + 5, |
2305 |
i.e., b - laststart. |
i.e., b - laststart. |
2306 |
|
|
2307 |
We insert this at the beginning of the loop |
We insert this at the beginning of the loop |
2308 |
so that if we fail during matching, we'll |
so that if we fail during matching, we'll |
2309 |
reinitialize the bounds. */ |
reinitialize the bounds. */ |
2324 |
beg_interval = NULL; |
beg_interval = NULL; |
2325 |
|
|
2326 |
/* normal_char and normal_backslash need `c'. */ |
/* normal_char and normal_backslash need `c'. */ |
2327 |
PATFETCH (c); |
PATFETCH (c); |
2328 |
|
|
2329 |
if (!(syntax & RE_NO_BK_BRACES)) |
if (!(syntax & RE_NO_BK_BRACES)) |
2330 |
{ |
{ |
2340 |
BUF_PUSH (at_dot); |
BUF_PUSH (at_dot); |
2341 |
break; |
break; |
2342 |
|
|
2343 |
case 's': |
case 's': |
2344 |
laststart = b; |
laststart = b; |
2345 |
PATFETCH (c); |
PATFETCH (c); |
2346 |
BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); |
BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); |
2431 |
/* Expects the character in `c'. */ |
/* Expects the character in `c'. */ |
2432 |
normal_char: |
normal_char: |
2433 |
/* If no exactn currently being built. */ |
/* If no exactn currently being built. */ |
2434 |
if (!pending_exact |
if (!pending_exact |
2435 |
|
|
2436 |
/* If last exactn not at current position. */ |
/* If last exactn not at current position. */ |
2437 |
|| pending_exact + *pending_exact + 1 != b |
|| pending_exact + *pending_exact + 1 != b |
2438 |
|
|
2439 |
/* We have only one byte following the exactn for the count. */ |
/* We have only one byte following the exactn for the count. */ |
2440 |
|| *pending_exact == (1 << BYTEWIDTH) - 1 |
|| *pending_exact == (1 << BYTEWIDTH) - 1 |
2441 |
|
|
2450 |
: (p[0] == '\\' && p[1] == '{')))) |
: (p[0] == '\\' && p[1] == '{')))) |
2451 |
{ |
{ |
2452 |
/* Start building a new exactn. */ |
/* Start building a new exactn. */ |
2453 |
|
|
2454 |
laststart = b; |
laststart = b; |
2455 |
|
|
2456 |
BUF_PUSH_2 (exactn, 0); |
BUF_PUSH_2 (exactn, 0); |
2457 |
pending_exact = b - 1; |
pending_exact = b - 1; |
2458 |
} |
} |
2459 |
|
|
2460 |
BUF_PUSH (c); |
BUF_PUSH (c); |
2461 |
(*pending_exact)++; |
(*pending_exact)++; |
2462 |
break; |
break; |
2463 |
} /* switch (c) */ |
} /* switch (c) */ |
2464 |
} /* while p != pend */ |
} /* while p != pend */ |
2465 |
|
|
2466 |
|
|
2467 |
/* Through the pattern now. */ |
/* Through the pattern now. */ |
2468 |
|
|
2469 |
if (fixup_alt_jump) |
if (fixup_alt_jump) |
2470 |
STORE_JUMP (jump_past_alt, fixup_alt_jump, b); |
STORE_JUMP (jump_past_alt, fixup_alt_jump, b); |
2471 |
|
|
2472 |
if (!COMPILE_STACK_EMPTY) |
if (!COMPILE_STACK_EMPTY) |
2473 |
FREE_STACK_RETURN (REG_EPAREN); |
FREE_STACK_RETURN (REG_EPAREN); |
2474 |
|
|
2475 |
free (compile_stack.stack); |
free (compile_stack.stack); |
2502 |
#ifdef emacs |
#ifdef emacs |
2503 |
if (! fail_stack.stack) |
if (! fail_stack.stack) |
2504 |
fail_stack.stack |
fail_stack.stack |
2505 |
= (fail_stack_elt_t *) xmalloc (fail_stack.size |
= (fail_stack_elt_t *) xmalloc (fail_stack.size |
2506 |
* sizeof (fail_stack_elt_t)); |
* sizeof (fail_stack_elt_t)); |
2507 |
else |
else |
2508 |
fail_stack.stack |
fail_stack.stack |
2512 |
#else /* not emacs */ |
#else /* not emacs */ |
2513 |
if (! fail_stack.stack) |
if (! fail_stack.stack) |
2514 |
fail_stack.stack |
fail_stack.stack |
2515 |
= (fail_stack_elt_t *) malloc (fail_stack.size |
= (fail_stack_elt_t *) malloc (fail_stack.size |
2516 |
* sizeof (fail_stack_elt_t)); |
* sizeof (fail_stack_elt_t)); |
2517 |
else |
else |
2518 |
fail_stack.stack |
fail_stack.stack |
2575 |
re_opcode_t op; |
re_opcode_t op; |
2576 |
unsigned char *loc; |
unsigned char *loc; |
2577 |
int arg; |
int arg; |
2578 |
unsigned char *end; |
unsigned char *end; |
2579 |
{ |
{ |
2580 |
register unsigned char *pfrom = end; |
register unsigned char *pfrom = end; |
2581 |
register unsigned char *pto = end + 3; |
register unsigned char *pto = end + 3; |
2582 |
|
|
2583 |
while (pfrom != loc) |
while (pfrom != loc) |
2584 |
*--pto = *--pfrom; |
*--pto = *--pfrom; |
2585 |
|
|
2586 |
store_op1 (op, loc, arg); |
store_op1 (op, loc, arg); |
2587 |
} |
} |
2588 |
|
|
2594 |
re_opcode_t op; |
re_opcode_t op; |
2595 |
unsigned char *loc; |
unsigned char *loc; |
2596 |
int arg1, arg2; |
int arg1, arg2; |
2597 |
unsigned char *end; |
unsigned char *end; |
2598 |
{ |
{ |
2599 |
register unsigned char *pfrom = end; |
register unsigned char *pfrom = end; |
2600 |
register unsigned char *pto = end + 5; |
register unsigned char *pto = end + 5; |
2601 |
|
|
2602 |
while (pfrom != loc) |
while (pfrom != loc) |
2603 |
*--pto = *--pfrom; |
*--pto = *--pfrom; |
2604 |
|
|
2605 |
store_op2 (op, loc, arg1, arg2); |
store_op2 (op, loc, arg1, arg2); |
2606 |
} |
} |
2607 |
|
|
2617 |
{ |
{ |
2618 |
const char *prev = p - 2; |
const char *prev = p - 2; |
2619 |
boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\'; |
boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\'; |
2620 |
|
|
2621 |
return |
return |
2622 |
/* After a subexpression? */ |
/* After a subexpression? */ |
2623 |
(*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) |
(*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) |
2637 |
const char *next = p; |
const char *next = p; |
2638 |
boolean next_backslash = *next == '\\'; |
boolean next_backslash = *next == '\\'; |
2639 |
const char *next_next = p + 1 < pend ? p + 1 : NULL; |
const char *next_next = p + 1 < pend ? p + 1 : NULL; |
2640 |
|
|
2641 |
return |
return |
2642 |
/* Before a subexpression? */ |
/* Before a subexpression? */ |
2643 |
(syntax & RE_NO_BK_PARENS ? *next == ')' |
(syntax & RE_NO_BK_PARENS ? *next == ')' |
2648 |
} |
} |
2649 |
|
|
2650 |
|
|
2651 |
/* Returns true if REGNUM is in one of COMPILE_STACK's elements and |
/* Returns true if REGNUM is in one of COMPILE_STACK's elements and |
2652 |
false if it's not. */ |
false if it's not. */ |
2653 |
|
|
2654 |
static boolean |
static boolean |
2658 |
{ |
{ |
2659 |
int this_element; |
int this_element; |
2660 |
|
|
2661 |
for (this_element = compile_stack.avail - 1; |
for (this_element = compile_stack.avail - 1; |
2662 |
this_element >= 0; |
this_element >= 0; |
2663 |
this_element--) |
this_element--) |
2664 |
if (compile_stack.stack[this_element].regnum == regnum) |
if (compile_stack.stack[this_element].regnum == regnum) |
2665 |
return true; |
return true; |
2673 |
starting character is in `P[-2]'. (`P[-1]' is the character `-'.) |
starting character is in `P[-2]'. (`P[-1]' is the character `-'.) |
2674 |
Then we set the translation of all bits between the starting and |
Then we set the translation of all bits between the starting and |
2675 |
ending characters (inclusive) in the compiled pattern B. |
ending characters (inclusive) in the compiled pattern B. |
2676 |
|
|
2677 |
Return an error code. |
Return an error code. |
2678 |
|
|
2679 |
We use these short variable names so we can use the same macros as |
We use these short variable names so we can use the same macros as |
2680 |
`regex_compile' itself. */ |
`regex_compile' itself. */ |
2681 |
|
|
2690 |
|
|
2691 |
const char *p = *p_ptr; |
const char *p = *p_ptr; |
2692 |
int range_start, range_end; |
int range_start, range_end; |
2693 |
|
|
2694 |
if (p == pend) |
if (p == pend) |
2695 |
return REG_ERANGE; |
return REG_ERANGE; |
2696 |
|
|
2699 |
is set, the range endpoints will be negative if we fetch using a |
is set, the range endpoints will be negative if we fetch using a |
2700 |
signed char *. |
signed char *. |
2701 |
|
|
2702 |
We also want to fetch the endpoints without translating them; the |
We also want to fetch the endpoints without translating them; the |
2703 |
appropriate translation is done in the bit-setting loop below. */ |
appropriate translation is done in the bit-setting loop below. */ |
2704 |
/* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */ |
/* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */ |
2705 |
range_start = ((const unsigned char *) p)[-2]; |
range_start = ((const unsigned char *) p)[-2]; |
2721 |
{ |
{ |
2722 |
SET_LIST_BIT (TRANSLATE (this_char)); |
SET_LIST_BIT (TRANSLATE (this_char)); |
2723 |
} |
} |
2724 |
|
|
2725 |
return REG_NOERROR; |
return REG_NOERROR; |
2726 |
} |
} |
2727 |
|
|
2732 |
|
|
2733 |
The caller must supply the address of a (1 << BYTEWIDTH)-byte data |
The caller must supply the address of a (1 << BYTEWIDTH)-byte data |
2734 |
area as BUFP->fastmap. |
area as BUFP->fastmap. |
2735 |
|
|
2736 |
We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in |
We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in |
2737 |
the pattern buffer. |
the pattern buffer. |
2738 |
|
|
2751 |
#endif |
#endif |
2752 |
/* We don't push any register information onto the failure stack. */ |
/* We don't push any register information onto the failure stack. */ |
2753 |
unsigned num_regs = 0; |
unsigned num_regs = 0; |
2754 |
|
|
2755 |
register char *fastmap = bufp->fastmap; |
register char *fastmap = bufp->fastmap; |
2756 |
unsigned char *pattern = bufp->buffer; |
unsigned char *pattern = bufp->buffer; |
2757 |
unsigned long size = bufp->used; |
unsigned long size = bufp->used; |
2768 |
boolean succeed_n_p = false; |
boolean succeed_n_p = false; |
2769 |
|
|
2770 |
assert (fastmap != NULL && p != NULL); |
assert (fastmap != NULL && p != NULL); |
2771 |
|
|
2772 |
INIT_FAIL_STACK (); |
INIT_FAIL_STACK (); |
2773 |
bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */ |
bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */ |
2774 |
bufp->fastmap_accurate = 1; /* It will be when we're done. */ |
bufp->fastmap_accurate = 1; /* It will be when we're done. */ |
2775 |
bufp->can_be_null = 0; |
bufp->can_be_null = 0; |
2776 |
|
|
2777 |
while (p != pend || !FAIL_STACK_EMPTY ()) |
while (p != pend || !FAIL_STACK_EMPTY ()) |
2778 |
{ |
{ |
2779 |
if (p == pend) |
if (p == pend) |
2780 |
{ |
{ |
2781 |
bufp->can_be_null |= path_can_be_null; |
bufp->can_be_null |= path_can_be_null; |
2782 |
|
|
2783 |
/* Reset for next path. */ |
/* Reset for next path. */ |
2784 |
path_can_be_null = true; |
path_can_be_null = true; |
2785 |
|
|
2786 |
p = fail_stack.stack[--fail_stack.avail]; |
p = fail_stack.stack[--fail_stack.avail]; |
2787 |
} |
} |
2788 |
|
|
2789 |
/* We should never be about to go beyond the end of the pattern. */ |
/* We should never be about to go beyond the end of the pattern. */ |
2790 |
assert (p < pend); |
assert (p < pend); |
2791 |
|
|
2792 |
#ifdef SWITCH_ENUM_BUG |
#ifdef SWITCH_ENUM_BUG |
2793 |
switch ((int) ((re_opcode_t) *p++)) |
switch ((int) ((re_opcode_t) *p++)) |
2794 |
#else |
#else |
2915 |
case jump_past_alt: |
case jump_past_alt: |
2916 |
case dummy_failure_jump: |
case dummy_failure_jump: |
2917 |
EXTRACT_NUMBER_AND_INCR (j, p); |
EXTRACT_NUMBER_AND_INCR (j, p); |
2918 |
p += j; |
p += j; |
2919 |
if (j > 0) |
if (j > 0) |
2920 |
continue; |
continue; |
2921 |
|
|
2922 |
/* Jump backward implies we just went through the body of a |
/* Jump backward implies we just went through the body of a |
2923 |
loop and matched nothing. Opcode jumped to should be |
loop and matched nothing. Opcode jumped to should be |
2924 |
`on_failure_jump' or `succeed_n'. Just treat it like an |
`on_failure_jump' or `succeed_n'. Just treat it like an |
2930 |
|
|
2931 |
p++; |
p++; |
2932 |
EXTRACT_NUMBER_AND_INCR (j, p); |
EXTRACT_NUMBER_AND_INCR (j, p); |
2933 |
p += j; |
p += j; |
2934 |
|
|
2935 |
/* If what's on the stack is where we are now, pop it. */ |
/* If what's on the stack is where we are now, pop it. */ |
2936 |
if (!FAIL_STACK_EMPTY () |
if (!FAIL_STACK_EMPTY () |
2937 |
&& fail_stack.stack[fail_stack.avail - 1] == p) |
&& fail_stack.stack[fail_stack.avail - 1] == p) |
2938 |
fail_stack.avail--; |
fail_stack.avail--; |
2939 |
|
|
2971 |
|
|
2972 |
case succeed_n: |
case succeed_n: |
2973 |
/* Get to the number of times to succeed. */ |
/* Get to the number of times to succeed. */ |
2974 |
p += 2; |
p += 2; |
2975 |
|
|
2976 |
/* Increment p past the n for when k != 0. */ |
/* Increment p past the n for when k != 0. */ |
2977 |
EXTRACT_NUMBER_AND_INCR (k, p); |
EXTRACT_NUMBER_AND_INCR (k, p); |
3062 |
int size, startpos, range; |
int size, startpos, range; |
3063 |
struct re_registers *regs; |
struct re_registers *regs; |
3064 |
{ |
{ |
3065 |
return re_search_2 (bufp, NULL, 0, string, size, startpos, range, |
return re_search_2 (bufp, NULL, 0, string, size, startpos, range, |
3066 |
regs, size); |
regs, size); |
3067 |
} |
} |
3068 |
|
|
3070 |
/* Using the compiled pattern in BUFP->buffer, first tries to match the |
/* Using the compiled pattern in BUFP->buffer, first tries to match the |
3071 |
virtual concatenation of STRING1 and STRING2, starting first at index |
virtual concatenation of STRING1 and STRING2, starting first at index |
3072 |
STARTPOS, then at STARTPOS + 1, and so on. |
STARTPOS, then at STARTPOS + 1, and so on. |
3073 |
|
|
3074 |
STRING1 and STRING2 have length SIZE1 and SIZE2, respectively. |
STRING1 and STRING2 have length SIZE1 and SIZE2, respectively. |
3075 |
|
|
3076 |
RANGE is how far to scan while trying to match. RANGE = 0 means try |
RANGE is how far to scan while trying to match. RANGE = 0 means try |
3077 |
only at STARTPOS; in general, the last start tried is STARTPOS + |
only at STARTPOS; in general, the last start tried is STARTPOS + |
3078 |
RANGE. |
RANGE. |
3079 |
|
|
3080 |
In REGS, return the indices of the virtual concatenation of STRING1 |
In REGS, return the indices of the virtual concatenation of STRING1 |
3081 |
and STRING2 that matched the entire BUFP->buffer and its contained |
and STRING2 that matched the entire BUFP->buffer and its contained |
3082 |
subexpressions. |
subexpressions. |
3083 |
|
|
3084 |
Do not consider matching one past the index STOP in the virtual |
Do not consider matching one past the index STOP in the virtual |
3085 |
concatenation of STRING1 and STRING2. |
concatenation of STRING1 and STRING2. |
3086 |
|
|
3107 |
/* Check for out-of-range STARTPOS. */ |
/* Check for out-of-range STARTPOS. */ |
3108 |
if (startpos < 0 || startpos > total_size) |
if (startpos < 0 || startpos > total_size) |
3109 |
return -1; |
return -1; |
3110 |
|
|
3111 |
/* Fix up RANGE if it might eventually take us outside |
/* Fix up RANGE if it might eventually take us outside |
3112 |
the virtual concatenation of STRING1 and STRING2. */ |
the virtual concatenation of STRING1 and STRING2. */ |
3113 |
if (endpos < -1) |
if (endpos < -1) |
3129 |
if (fastmap && !bufp->fastmap_accurate) |
if (fastmap && !bufp->fastmap_accurate) |
3130 |
if (re_compile_fastmap (bufp) == -2) |
if (re_compile_fastmap (bufp) == -2) |
3131 |
return -2; |
return -2; |
3132 |
|
|
3133 |
/* Loop through the string, looking for a place to start matching. */ |
/* Loop through the string, looking for a place to start matching. */ |
3134 |
for (;;) |
for (;;) |
3135 |
{ |
{ |
3136 |
/* If a fastmap is supplied, skip quickly over characters that |
/* If a fastmap is supplied, skip quickly over characters that |
3137 |
cannot be the start of a match. If the pattern can match the |
cannot be the start of a match. If the pattern can match the |
3138 |
null string, however, we don't need to skip characters; we want |
null string, however, we don't need to skip characters; we want |
3149 |
lim = range - (size1 - startpos); |
lim = range - (size1 - startpos); |
3150 |
|
|
3151 |
d = (startpos >= size1 ? string2 - size1 : string1) + startpos; |
d = (startpos >= size1 ? string2 - size1 : string1) + startpos; |
3152 |
|
|
3153 |
/* Written out as an if-else to avoid testing `translate' |
/* Written out as an if-else to avoid testing `translate' |
3154 |
inside the loop. */ |
inside the loop. */ |
3155 |
if (translate) |
if (translate) |
3166 |
else /* Searching backwards. */ |
else /* Searching backwards. */ |
3167 |
{ |
{ |
3168 |
register char c = (size1 == 0 || startpos >= size1 |
register char c = (size1 == 0 || startpos >= size1 |
3169 |
? string2[startpos - size1] |
? string2[startpos - size1] |
3170 |
: string1[startpos]); |
: string1[startpos]); |
3171 |
|
|
3172 |
if (!fastmap[(unsigned char) TRANSLATE (c)]) |
if (!fastmap[(unsigned char) TRANSLATE (c)]) |
3189 |
|
|
3190 |
if (val >= 0) |
if (val >= 0) |
3191 |
return startpos; |
return startpos; |
3192 |
|
|
3193 |
if (val == -2) |
if (val == -2) |
3194 |
return -2; |
return -2; |
3195 |
|
|
3196 |
advance: |
advance: |
3197 |
if (!range) |
if (!range) |
3198 |
break; |
break; |
3199 |
else if (range > 0) |
else if (range > 0) |
3200 |
{ |
{ |
3201 |
range--; |
range--; |
3202 |
startpos++; |
startpos++; |
3203 |
} |
} |
3204 |
else |
else |
3205 |
{ |
{ |
3206 |
range++; |
range++; |
3207 |
startpos--; |
startpos--; |
3208 |
} |
} |
3209 |
} |
} |
3245 |
/* Test if at very beginning or at very end of the virtual concatenation |
/* Test if at very beginning or at very end of the virtual concatenation |
3246 |
of `string1' and `string2'. If only one string, it's `string2'. */ |
of `string1' and `string2'. If only one string, it's `string2'. */ |
3247 |
#define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) |
#define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) |
3248 |
#define AT_STRINGS_END(d) ((d) == end2) |
#define AT_STRINGS_END(d) ((d) == end2) |
3249 |
|
|
3250 |
|
|
3251 |
/* Test if D points to a character which is word-constituent. We have |
/* Test if D points to a character which is word-constituent. We have |
3323 |
the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 |
the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 |
3324 |
and SIZE2, respectively). We start matching at POS, and stop |
and SIZE2, respectively). We start matching at POS, and stop |
3325 |
matching at STOP. |
matching at STOP. |
3326 |
|
|
3327 |
If REGS is non-null and the `no_sub' field of BUFP is nonzero, we |
If REGS is non-null and the `no_sub' field of BUFP is nonzero, we |
3328 |
store offsets for the substring each group matched in REGS. See the |
store offsets for the substring each group matched in REGS. See the |
3329 |
documentation for exactly how many groups we fill. |
documentation for exactly how many groups we fill. |
3371 |
|
|
3372 |
/* Where we are in the data, and the end of the current string. */ |
/* Where we are in the data, and the end of the current string. */ |
3373 |
const char *d, *dend; |
const char *d, *dend; |
3374 |
|
|
3375 |
/* Where we are in the pattern, and the end of the pattern. */ |
/* Where we are in the pattern, and the end of the pattern. */ |
3376 |
unsigned char *p = bufp->buffer; |
unsigned char *p = bufp->buffer; |
3377 |
register unsigned char *pend = p + bufp->used; |
register unsigned char *pend = p + bufp->used; |
3404 |
return, for use in backreferences. The number here includes |
return, for use in backreferences. The number here includes |
3405 |
an element for register zero. */ |
an element for register zero. */ |
3406 |
unsigned num_regs = bufp->re_nsub + 1; |
unsigned num_regs = bufp->re_nsub + 1; |
3407 |
|
|
3408 |
/* The currently active registers. */ |
/* The currently active registers. */ |
3409 |
unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
3410 |
unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
3436 |
subexpression. These two fields get reset each time through any |
subexpression. These two fields get reset each time through any |
3437 |
loop their register is in. */ |
loop their register is in. */ |
3438 |
#ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ |
#ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ |
3439 |
register_info_type *reg_info; |
register_info_type *reg_info; |
3440 |
#endif |
#endif |
3441 |
|
|
3442 |
/* The following record the register info as found in the above |
/* The following record the register info as found in the above |
3443 |
variables when we find a match better than any we've seen before. |
variables when we find a match better than any we've seen before. |
3444 |
This happens as we backtrack through the failure points, which in |
This happens as we backtrack through the failure points, which in |
3445 |
turn happens only if we have not yet matched the entire string. */ |
turn happens only if we have not yet matched the entire string. */ |
3446 |
unsigned best_regs_set = false; |
unsigned best_regs_set = false; |
3447 |
#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ |
#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ |
3448 |
const char **best_regstart, **best_regend; |
const char **best_regstart, **best_regend; |
3449 |
#endif |
#endif |
3450 |
|
|
3451 |
/* Logically, this is `best_regend[0]'. But we don't want to have to |
/* Logically, this is `best_regend[0]'. But we don't want to have to |
3452 |
allocate space for that if we're not allocating space for anything |
allocate space for that if we're not allocating space for anything |
3453 |
else (see below). Also, we never need info about register 0 for |
else (see below). Also, we never need info about register 0 for |
3466 |
|
|
3467 |
#ifdef DEBUG |
#ifdef DEBUG |
3468 |
/* Counts the total number of registers pushed. */ |
/* Counts the total number of registers pushed. */ |
3469 |
unsigned num_regs_pushed = 0; |
unsigned num_regs_pushed = 0; |
3470 |
#endif |
#endif |
3471 |
|
|
3472 |
DEBUG_PRINT1 ("\n\nEntering re_match_2.\n"); |
DEBUG_PRINT1 ("\n\nEntering re_match_2.\n"); |
3473 |
|
|
3474 |
INIT_FAIL_STACK (); |
INIT_FAIL_STACK (); |
3475 |
|
|
3476 |
#ifdef MATCH_MAY_ALLOCATE |
#ifdef MATCH_MAY_ALLOCATE |
3477 |
/* Do not bother to initialize all the register variables if there are |
/* Do not bother to initialize all the register variables if there are |
3478 |
no groups in the pattern, as it takes a fair amount of time. If |
no groups in the pattern, as it takes a fair amount of time. If |
3491 |
reg_dummy = REGEX_TALLOC (num_regs, const char *); |
reg_dummy = REGEX_TALLOC (num_regs, const char *); |
3492 |
reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); |
reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); |
3493 |
|
|
3494 |
if (!(regstart && regend && old_regstart && old_regend && reg_info |
if (!(regstart && regend && old_regstart && old_regend && reg_info |
3495 |
&& best_regstart && best_regend && reg_dummy && reg_info_dummy)) |
&& best_regstart && best_regend && reg_dummy && reg_info_dummy)) |
3496 |
{ |
{ |
3497 |
FREE_VARIABLES (); |
FREE_VARIABLES (); |
3498 |
return -2; |
return -2; |
3516 |
FREE_VARIABLES (); |
FREE_VARIABLES (); |
3517 |
return -1; |
return -1; |
3518 |
} |
} |
3519 |
|
|
3520 |
/* Initialize subexpression text positions to -1 to mark ones that no |
/* Initialize subexpression text positions to -1 to mark ones that no |
3521 |
start_memory/stop_memory has been seen for. Also initialize the |
start_memory/stop_memory has been seen for. Also initialize the |
3522 |
register information struct. */ |
register information struct. */ |
3523 |
for (mcnt = 1; mcnt < num_regs; mcnt++) |
for (mcnt = 1; mcnt < num_regs; mcnt++) |
3524 |
{ |
{ |
3525 |
regstart[mcnt] = regend[mcnt] |
regstart[mcnt] = regend[mcnt] |
3526 |
= old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; |
= old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; |
3527 |
|
|
3528 |
REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; |
REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; |
3529 |
IS_ACTIVE (reg_info[mcnt]) = 0; |
IS_ACTIVE (reg_info[mcnt]) = 0; |
3530 |
MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
3531 |
EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
3532 |
} |
} |
3533 |
|
|
3534 |
/* We move `string1' into `string2' if the latter's empty -- but not if |
/* We move `string1' into `string2' if the latter's empty -- but not if |
3535 |
`string1' is null. */ |
`string1' is null. */ |
3536 |
if (size2 == 0 && string1 != NULL) |
if (size2 == 0 && string1 != NULL) |
3555 |
end_match_2 = string2 + stop - size1; |
end_match_2 = string2 + stop - size1; |
3556 |
} |
} |
3557 |
|
|
3558 |
/* `p' scans through the pattern as `d' scans through the data. |
/* `p' scans through the pattern as `d' scans through the data. |
3559 |
`dend' is the end of the input string that `d' points within. `d' |
`dend' is the end of the input string that `d' points within. `d' |
3560 |
is advanced into the following input string whenever necessary, but |
is advanced into the following input string whenever necessary, but |
3561 |
this happens before fetching; therefore, at the beginning of the |
this happens before fetching; therefore, at the beginning of the |
3577 |
DEBUG_PRINT1 ("The string to match is: `"); |
DEBUG_PRINT1 ("The string to match is: `"); |
3578 |
DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); |
DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); |
3579 |
DEBUG_PRINT1 ("'\n"); |
DEBUG_PRINT1 ("'\n"); |
3580 |
|
|
3581 |
/* This loops over pattern commands. It exits by returning from the |
/* This loops over pattern commands. It exits by returning from the |
3582 |
function if the match is complete, or it drops through if the match |
function if the match is complete, or it drops through if the match |
3583 |
fails at this starting point in the input data. */ |
fails at this starting point in the input data. */ |
3588 |
if (p == pend) |
if (p == pend) |
3589 |
{ /* End of pattern means we might have succeeded. */ |
{ /* End of pattern means we might have succeeded. */ |
3590 |
DEBUG_PRINT1 ("end of pattern ... "); |
DEBUG_PRINT1 ("end of pattern ... "); |
3591 |
|
|
3592 |
/* If we haven't matched the entire string, and we want the |
/* If we haven't matched the entire string, and we want the |
3593 |
longest match, try backtracking. */ |
longest match, try backtracking. */ |
3594 |
if (d != end_match_2) |
if (d != end_match_2) |
3595 |
{ |
{ |
3596 |
/* 1 if this match ends in the same string (string1 or string2) |
/* 1 if this match ends in the same string (string1 or string2) |
3597 |
as the best previous match. */ |
as the best previous match. */ |
3598 |
boolean same_str_p = (FIRST_STRING_P (match_end) |
boolean same_str_p = (FIRST_STRING_P (match_end) |
3599 |
== MATCHING_IN_FIRST_STRING); |
== MATCHING_IN_FIRST_STRING); |
3600 |
/* 1 if this match is the best seen so far. */ |
/* 1 if this match is the best seen so far. */ |
3601 |
boolean best_match_p; |
boolean best_match_p; |
3608 |
best_match_p = !MATCHING_IN_FIRST_STRING; |
best_match_p = !MATCHING_IN_FIRST_STRING; |
3609 |
|
|
3610 |
DEBUG_PRINT1 ("backtracking.\n"); |
DEBUG_PRINT1 ("backtracking.\n"); |
3611 |
|
|
3612 |
if (!FAIL_STACK_EMPTY ()) |
if (!FAIL_STACK_EMPTY ()) |
3613 |
{ /* More failure points to try. */ |
{ /* More failure points to try. */ |
3614 |
|
|
3617 |
{ |
{ |
3618 |
best_regs_set = true; |
best_regs_set = true; |
3619 |
match_end = d; |
match_end = d; |
3620 |
|
|
3621 |
DEBUG_PRINT1 ("\nSAVING match as best so far.\n"); |
DEBUG_PRINT1 ("\nSAVING match as best so far.\n"); |
3622 |
|
|
3623 |
for (mcnt = 1; mcnt < num_regs; mcnt++) |
for (mcnt = 1; mcnt < num_regs; mcnt++) |
3624 |
{ |
{ |
3625 |
best_regstart[mcnt] = regstart[mcnt]; |
best_regstart[mcnt] = regstart[mcnt]; |
3626 |
best_regend[mcnt] = regend[mcnt]; |
best_regend[mcnt] = regend[mcnt]; |
3627 |
} |
} |
3628 |
} |
} |
3629 |
goto fail; |
goto fail; |
3630 |
} |
} |
3631 |
|
|
3632 |
/* If no failure points, don't restore garbage. And if |
/* If no failure points, don't restore garbage. And if |
3641 |
strings `x-' and `y-z-', if the two strings are |
strings `x-' and `y-z-', if the two strings are |
3642 |
not consecutive in memory. */ |
not consecutive in memory. */ |
3643 |
DEBUG_PRINT1 ("Restoring best registers.\n"); |
DEBUG_PRINT1 ("Restoring best registers.\n"); |
3644 |
|
|
3645 |
d = match_end; |
d = match_end; |
3646 |
dend = ((d >= string1 && d <= end1) |
dend = ((d >= string1 && d <= end1) |
3647 |
? end_match_1 : end_match_2); |
? end_match_1 : end_match_2); |
3701 |
? ((regoff_t) (d - string1)) |
? ((regoff_t) (d - string1)) |
3702 |
: ((regoff_t) (d - string2 + size1))); |
: ((regoff_t) (d - string2 + size1))); |
3703 |
} |
} |
3704 |
|
|
3705 |
/* Go through the first `min (num_regs, regs->num_regs)' |
/* Go through the first `min (num_regs, regs->num_regs)' |
3706 |
registers, since that is all we initialized. */ |
registers, since that is all we initialized. */ |
3707 |
for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++) |
for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++) |
3716 |
= (regoff_t) POINTER_TO_OFFSET (regend[mcnt]); |
= (regoff_t) POINTER_TO_OFFSET (regend[mcnt]); |
3717 |
} |
} |
3718 |
} |
} |
3719 |
|
|
3720 |
/* If the regs structure we return has more elements than |
/* If the regs structure we return has more elements than |
3721 |
were in the pattern, set the extra elements to -1. If |
were in the pattern, set the extra elements to -1. If |
3722 |
we (re)allocated the registers, this is the case, |
we (re)allocated the registers, this is the case, |
3732 |
nfailure_points_pushed - nfailure_points_popped); |
nfailure_points_pushed - nfailure_points_popped); |
3733 |
DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed); |
DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed); |
3734 |
|
|
3735 |
mcnt = d - pos - (MATCHING_IN_FIRST_STRING |
mcnt = d - pos - (MATCHING_IN_FIRST_STRING |
3736 |
? string1 |
? string1 |
3737 |
: string2 - size1); |
: string2 - size1); |
3738 |
|
|
3739 |
DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt); |
DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt); |
3823 |
p += 1 + *p; |
p += 1 + *p; |
3824 |
|
|
3825 |
if (!not) goto fail; |
if (!not) goto fail; |
3826 |
|
|
3827 |
SET_REGS_MATCHED (); |
SET_REGS_MATCHED (); |
3828 |
d++; |
d++; |
3829 |
break; |
break; |
3840 |
|
|
3841 |
/* Find out if this group can match the empty string. */ |
/* Find out if this group can match the empty string. */ |
3842 |
p1 = p; /* To send to group_match_null_string_p. */ |
p1 = p; /* To send to group_match_null_string_p. */ |
3843 |
|
|
3844 |
if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) |
if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) |
3845 |
REG_MATCH_NULL_STRING_P (reg_info[*p]) |
REG_MATCH_NULL_STRING_P (reg_info[*p]) |
3846 |
= group_match_null_string_p (&p1, pend, reg_info); |
= group_match_null_string_p (&p1, pend, reg_info); |
3847 |
|
|
3848 |
/* Save the position in the string where we were the last time |
/* Save the position in the string where we were the last time |
3853 |
old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
3854 |
? REG_UNSET (regstart[*p]) ? d : regstart[*p] |
? REG_UNSET (regstart[*p]) ? d : regstart[*p] |
3855 |
: regstart[*p]; |
: regstart[*p]; |
3856 |
DEBUG_PRINT2 (" old_regstart: %d\n", |
DEBUG_PRINT2 (" old_regstart: %d\n", |
3857 |
POINTER_TO_OFFSET (old_regstart[*p])); |
POINTER_TO_OFFSET (old_regstart[*p])); |
3858 |
|
|
3859 |
regstart[*p] = d; |
regstart[*p] = d; |
3861 |
|
|
3862 |
IS_ACTIVE (reg_info[*p]) = 1; |
IS_ACTIVE (reg_info[*p]) = 1; |
3863 |
MATCHED_SOMETHING (reg_info[*p]) = 0; |
MATCHED_SOMETHING (reg_info[*p]) = 0; |
3864 |
|
|
3865 |
/* This is the new highest active register. */ |
/* This is the new highest active register. */ |
3866 |
highest_active_reg = *p; |
highest_active_reg = *p; |
3867 |
|
|
3868 |
/* If nothing was active before, this is the new lowest active |
/* If nothing was active before, this is the new lowest active |
3869 |
register. */ |
register. */ |
3870 |
if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) |
if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) |
3881 |
number, and the number of inner groups. */ |
number, and the number of inner groups. */ |
3882 |
case stop_memory: |
case stop_memory: |
3883 |
DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); |
DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); |
3884 |
|
|
3885 |
/* We need to save the string position the last time we were at |
/* We need to save the string position the last time we were at |
3886 |
this close-group operator in case the group is operated |
this close-group operator in case the group is operated |
3887 |
upon by a repetition operator, e.g., with `((a*)*(b*)*)*' |
upon by a repetition operator, e.g., with `((a*)*(b*)*)*' |
3890 |
old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
3891 |
? REG_UNSET (regend[*p]) ? d : regend[*p] |
? REG_UNSET (regend[*p]) ? d : regend[*p] |
3892 |
: regend[*p]; |
: regend[*p]; |
3893 |
DEBUG_PRINT2 (" old_regend: %d\n", |
DEBUG_PRINT2 (" old_regend: %d\n", |
3894 |
POINTER_TO_OFFSET (old_regend[*p])); |
POINTER_TO_OFFSET (old_regend[*p])); |
3895 |
|
|
3896 |
regend[*p] = d; |
regend[*p] = d; |
3898 |
|
|
3899 |
/* This register isn't active anymore. */ |
/* This register isn't active anymore. */ |
3900 |
IS_ACTIVE (reg_info[*p]) = 0; |
IS_ACTIVE (reg_info[*p]) = 0; |
3901 |
|
|
3902 |
/* If this was the only register active, nothing is active |
/* If this was the only register active, nothing is active |
3903 |
anymore. */ |
anymore. */ |
3904 |
if (lowest_active_reg == highest_active_reg) |
if (lowest_active_reg == highest_active_reg) |
3914 |
unsigned char r = *p - 1; |
unsigned char r = *p - 1; |
3915 |
while (r > 0 && !IS_ACTIVE (reg_info[r])) |
while (r > 0 && !IS_ACTIVE (reg_info[r])) |
3916 |
r--; |
r--; |
3917 |
|
|
3918 |
/* If we end up at register zero, that means that we saved |
/* If we end up at register zero, that means that we saved |
3919 |
the registers as the result of an `on_failure_jump', not |
the registers as the result of an `on_failure_jump', not |
3920 |
a `start_memory', and we jumped to past the innermost |
a `start_memory', and we jumped to past the innermost |
3930 |
else |
else |
3931 |
highest_active_reg = r; |
highest_active_reg = r; |
3932 |
} |
} |
3933 |
|
|
3934 |
/* If just failed to match something this time around with a |
/* If just failed to match something this time around with a |
3935 |
group that's operated on by a repetition operator, try to |
group that's operated on by a repetition operator, try to |
3936 |
force exit from the ``loop'', and restore the register |
force exit from the ``loop'', and restore the register |
3938 |
last match. */ |
last match. */ |
3939 |
if ((!MATCHED_SOMETHING (reg_info[*p]) |
if ((!MATCHED_SOMETHING (reg_info[*p]) |
3940 |
|| just_past_start_mem == p - 1) |
|| just_past_start_mem == p - 1) |
3941 |
&& (p + 2) < pend) |
&& (p + 2) < pend) |
3942 |
{ |
{ |
3943 |
boolean is_a_jump_n = false; |
boolean is_a_jump_n = false; |
3944 |
|
|
3945 |
p1 = p + 2; |
p1 = p + 2; |
3946 |
mcnt = 0; |
mcnt = 0; |
3947 |
switch ((re_opcode_t) *p1++) |
switch ((re_opcode_t) *p1++) |
3956 |
if (is_a_jump_n) |
if (is_a_jump_n) |
3957 |
p1 += 2; |
p1 += 2; |
3958 |
break; |
break; |
3959 |
|
|
3960 |
default: |
default: |
3961 |
/* do nothing */ ; |
/* do nothing */ ; |
3962 |
} |
} |
3963 |
p1 += mcnt; |
p1 += mcnt; |
3964 |
|
|
3965 |
/* If the next operation is a jump backwards in the pattern |
/* If the next operation is a jump backwards in the pattern |
3966 |
to an on_failure_jump right before the start_memory |
to an on_failure_jump right before the start_memory |
3967 |
corresponding to this stop_memory, exit from the loop |
corresponding to this stop_memory, exit from the loop |
3975 |
failed match, e.g., with `(a*)*b' against `ab' for |
failed match, e.g., with `(a*)*b' against `ab' for |
3976 |
regstart[1], and, e.g., with `((a*)*(b*)*)*' |
regstart[1], and, e.g., with `((a*)*(b*)*)*' |
3977 |
against `aba' for regend[3]. |
against `aba' for regend[3]. |
3978 |
|
|
3979 |
Also restore the registers for inner groups for, |
Also restore the registers for inner groups for, |
3980 |
e.g., `((a*)(b*))*' against `aba' (register 3 would |
e.g., `((a*)(b*))*' against `aba' (register 3 would |
3981 |
otherwise get trashed). */ |
otherwise get trashed). */ |
3982 |
|
|
3983 |
if (EVER_MATCHED_SOMETHING (reg_info[*p])) |
if (EVER_MATCHED_SOMETHING (reg_info[*p])) |
3984 |
{ |
{ |
3985 |
unsigned r; |
unsigned r; |
3986 |
|
|
3987 |
EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; |
EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; |
3988 |
|
|
3989 |
/* Restore this and inner groups' (if any) registers. */ |
/* Restore this and inner groups' (if any) registers. */ |
3990 |
for (r = *p; r < *p + *(p + 1); r++) |
for (r = *p; r < *p + *(p + 1); r++) |
3991 |
{ |
{ |
3994 |
/* xx why this test? */ |
/* xx why this test? */ |
3995 |
if ((int) old_regend[r] >= (int) regstart[r]) |
if ((int) old_regend[r] >= (int) regstart[r]) |
3996 |
regend[r] = old_regend[r]; |
regend[r] = old_regend[r]; |
3997 |
} |
} |
3998 |
} |
} |
3999 |
p1++; |
p1++; |
4000 |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4003 |
goto fail; |
goto fail; |
4004 |
} |
} |
4005 |
} |
} |
4006 |
|
|
4007 |
/* Move past the register number and the inner group count. */ |
/* Move past the register number and the inner group count. */ |
4008 |
p += 2; |
p += 2; |
4009 |
break; |
break; |
4020 |
/* Can't back reference a group which we've never matched. */ |
/* Can't back reference a group which we've never matched. */ |
4021 |
if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) |
if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) |
4022 |
goto fail; |
goto fail; |
4023 |
|
|
4024 |
/* Where in input to try to start matching. */ |
/* Where in input to try to start matching. */ |
4025 |
d2 = regstart[regno]; |
d2 = regstart[regno]; |
4026 |
|
|
4027 |
/* Where to stop matching; if both the place to start and |
/* Where to stop matching; if both the place to start and |
4028 |
the place to stop matching are in the same string, then |
the place to stop matching are in the same string, then |
4029 |
set to the place to stop, otherwise, for now have to use |
set to the place to stop, otherwise, for now have to use |
4030 |
the end of the first string. */ |
the end of the first string. */ |
4031 |
|
|
4032 |
dend2 = ((FIRST_STRING_P (regstart[regno]) |
dend2 = ((FIRST_STRING_P (regstart[regno]) |
4033 |
== FIRST_STRING_P (regend[regno])) |
== FIRST_STRING_P (regend[regno])) |
4034 |
? regend[regno] : end_match_1); |
? regend[regno] : end_match_1); |
4035 |
for (;;) |
for (;;) |
4053 |
|
|
4054 |
/* How many characters left in this segment to match. */ |
/* How many characters left in this segment to match. */ |
4055 |
mcnt = dend - d; |
mcnt = dend - d; |
4056 |
|
|
4057 |
/* Want how many consecutive characters we can match in |
/* Want how many consecutive characters we can match in |
4058 |
one shot, so, if necessary, adjust the count. */ |
one shot, so, if necessary, adjust the count. */ |
4059 |
if (mcnt > dend2 - d2) |
if (mcnt > dend2 - d2) |
4060 |
mcnt = dend2 - d2; |
mcnt = dend2 - d2; |
4061 |
|
|
4062 |
/* Compare that many; failure if mismatch, else move |
/* Compare that many; failure if mismatch, else move |
4063 |
past them. */ |
past them. */ |
4064 |
if (translate |
if (translate |
4065 |
? bcmp_translate (d, d2, mcnt, translate) |
? bcmp_translate (d, d2, mcnt, translate) |
4066 |
: bcmp (d, d2, mcnt)) |
: bcmp (d, d2, mcnt)) |
4067 |
goto fail; |
goto fail; |
4068 |
d += mcnt, d2 += mcnt; |
d += mcnt, d2 += mcnt; |
4076 |
`newline_anchor' is set, after newlines. */ |
`newline_anchor' is set, after newlines. */ |
4077 |
case begline: |
case begline: |
4078 |
DEBUG_PRINT1 ("EXECUTING begline.\n"); |
DEBUG_PRINT1 ("EXECUTING begline.\n"); |
4079 |
|
|
4080 |
if (AT_STRINGS_BEG (d)) |
if (AT_STRINGS_BEG (d)) |
4081 |
{ |
{ |
4082 |
if (!bufp->not_bol) break; |
if (!bufp->not_bol) break; |
4097 |
{ |
{ |
4098 |
if (!bufp->not_eol) break; |
if (!bufp->not_eol) break; |
4099 |
} |
} |
4100 |
|
|
4101 |
/* We have to ``prefetch'' the next character. */ |
/* We have to ``prefetch'' the next character. */ |
4102 |
else if ((d == end1 ? *string2 : *d) == '\n' |
else if ((d == end1 ? *string2 : *d) == '\n' |
4103 |
&& bufp->newline_anchor) |
&& bufp->newline_anchor) |
4131 |
then the . fails against the \n. But the next thing we want |
then the . fails against the \n. But the next thing we want |
4132 |
to do is match the \n against the \n; if we restored the |
to do is match the \n against the \n; if we restored the |
4133 |
string value, we would be back at the foo. |
string value, we would be back at the foo. |
4134 |
|
|
4135 |
Because this is used only in specific cases, we don't need to |
Because this is used only in specific cases, we don't need to |
4136 |
check all the things that `on_failure_jump' does, to make |
check all the things that `on_failure_jump' does, to make |
4137 |
sure the right things get saved on the stack. Hence we don't |
sure the right things get saved on the stack. Hence we don't |
4141 |
case; that seems worse than this. */ |
case; that seems worse than this. */ |
4142 |
case on_failure_keep_string_jump: |
case on_failure_keep_string_jump: |
4143 |
DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump"); |
DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump"); |
4144 |
|
|
4145 |
EXTRACT_NUMBER_AND_INCR (mcnt, p); |
EXTRACT_NUMBER_AND_INCR (mcnt, p); |
4146 |
DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt); |
DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt); |
4147 |
|
|
4150 |
|
|
4151 |
|
|
4152 |
/* Uses of on_failure_jump: |
/* Uses of on_failure_jump: |
4153 |
|
|
4154 |
Each alternative starts with an on_failure_jump that points |
Each alternative starts with an on_failure_jump that points |
4155 |
to the beginning of the next alternative. Each alternative |
to the beginning of the next alternative. Each alternative |
4156 |
except the last ends with a jump that in effect jumps past |
except the last ends with a jump that in effect jumps past |
4216 |
would have to backtrack because of (as in, e.g., `a*a') |
would have to backtrack because of (as in, e.g., `a*a') |
4217 |
then we can change to pop_failure_jump, because we'll |
then we can change to pop_failure_jump, because we'll |
4218 |
never have to backtrack. |
never have to backtrack. |
4219 |
|
|
4220 |
This is not true in the case of alternatives: in |
This is not true in the case of alternatives: in |
4221 |
`(a|ab)*' we do need to backtrack to the `ab' alternative |
`(a|ab)*' we do need to backtrack to the `ab' alternative |
4222 |
(e.g., if the string was `ab'). But instead of trying to |
(e.g., if the string was `ab'). But instead of trying to |
4242 |
|
|
4243 |
p1 = p + mcnt; |
p1 = p + mcnt; |
4244 |
/* p1[0] ... p1[2] are the `on_failure_jump' corresponding |
/* p1[0] ... p1[2] are the `on_failure_jump' corresponding |
4245 |
to the `maybe_finalize_jump' of this case. Examine what |
to the `maybe_finalize_jump' of this case. Examine what |
4246 |
follows. */ |
follows. */ |
4247 |
|
|
4248 |
/* If we're at the end of the pattern, we can change. */ |
/* If we're at the end of the pattern, we can change. */ |
4268 |
DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", |
DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", |
4269 |
c, p1[5]); |
c, p1[5]); |
4270 |
} |
} |
4271 |
|
|
4272 |
else if ((re_opcode_t) p1[3] == charset |
else if ((re_opcode_t) p1[3] == charset |
4273 |
|| (re_opcode_t) p1[3] == charset_not) |
|| (re_opcode_t) p1[3] == charset_not) |
4274 |
{ |
{ |
4275 |
int not = (re_opcode_t) p1[3] == charset_not; |
int not = (re_opcode_t) p1[3] == charset_not; |
4276 |
|
|
4277 |
if (c < (unsigned char) (p1[4] * BYTEWIDTH) |
if (c < (unsigned char) (p1[4] * BYTEWIDTH) |
4278 |
&& p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) |
&& p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) |
4279 |
not = !not; |
not = !not; |
4303 |
DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", |
DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", |
4304 |
c, p1[5]); |
c, p1[5]); |
4305 |
} |
} |
4306 |
|
|
4307 |
else if ((re_opcode_t) p1[3] == charset_not) |
else if ((re_opcode_t) p1[3] == charset_not) |
4308 |
{ |
{ |
4309 |
int idx; |
int idx; |
4374 |
} |
} |
4375 |
/* Note fall through. */ |
/* Note fall through. */ |
4376 |
|
|
4377 |
|
|
4378 |
/* Unconditionally jump (without popping any failure points). */ |
/* Unconditionally jump (without popping any failure points). */ |
4379 |
case jump: |
case jump: |
4380 |
unconditional_jump: |
unconditional_jump: |
4384 |
DEBUG_PRINT2 ("(to 0x%x).\n", p); |
DEBUG_PRINT2 ("(to 0x%x).\n", p); |
4385 |
break; |
break; |
4386 |
|
|
4387 |
|
|
4388 |
/* We need this opcode so we can detect where alternatives end |
/* We need this opcode so we can detect where alternatives end |
4389 |
in `group_match_null_string_p' et al. */ |
in `group_match_null_string_p' et al. */ |
4390 |
case jump_past_alt: |
case jump_past_alt: |
4419 |
|
|
4420 |
/* Have to succeed matching what follows at least n times. |
/* Have to succeed matching what follows at least n times. |
4421 |
After that, handle like `on_failure_jump'. */ |
After that, handle like `on_failure_jump'. */ |
4422 |
case succeed_n: |
case succeed_n: |
4423 |
EXTRACT_NUMBER (mcnt, p + 2); |
EXTRACT_NUMBER (mcnt, p + 2); |
4424 |
DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); |
DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); |
4425 |
|
|
4440 |
goto on_failure; |
goto on_failure; |
4441 |
} |
} |
4442 |
break; |
break; |
4443 |
|
|
4444 |
case jump_n: |
case jump_n: |
4445 |
EXTRACT_NUMBER (mcnt, p + 2); |
EXTRACT_NUMBER (mcnt, p + 2); |
4446 |
DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); |
DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); |
4447 |
|
|
4450 |
{ |
{ |
4451 |
mcnt--; |
mcnt--; |
4452 |
STORE_NUMBER (p + 2, mcnt); |
STORE_NUMBER (p + 2, mcnt); |
4453 |
goto unconditional_jump; |
goto unconditional_jump; |
4454 |
} |
} |
4455 |
/* If don't have to jump any more, skip over the rest of command. */ |
/* If don't have to jump any more, skip over the rest of command. */ |
4456 |
else |
else |
4457 |
p += 4; |
p += 4; |
4458 |
break; |
break; |
4459 |
|
|
4460 |
case set_number_at: |
case set_number_at: |
4461 |
{ |
{ |
4462 |
DEBUG_PRINT1 ("EXECUTING set_number_at.\n"); |
DEBUG_PRINT1 ("EXECUTING set_number_at.\n"); |
4500 |
if (PTR_CHAR_POS ((unsigned char *) d) >= point) |
if (PTR_CHAR_POS ((unsigned char *) d) >= point) |
4501 |
goto fail; |
goto fail; |
4502 |
break; |
break; |
4503 |
|
|
4504 |
case at_dot: |
case at_dot: |
4505 |
DEBUG_PRINT1 ("EXECUTING at_dot.\n"); |
DEBUG_PRINT1 ("EXECUTING at_dot.\n"); |
4506 |
if (PTR_CHAR_POS ((unsigned char *) d) != point) |
if (PTR_CHAR_POS ((unsigned char *) d) != point) |
4507 |
goto fail; |
goto fail; |
4508 |
break; |
break; |
4509 |
|
|
4510 |
case after_dot: |
case after_dot: |
4511 |
DEBUG_PRINT1 ("EXECUTING after_dot.\n"); |
DEBUG_PRINT1 ("EXECUTING after_dot.\n"); |
4512 |
if (PTR_CHAR_POS ((unsigned char *) d) <= point) |
if (PTR_CHAR_POS ((unsigned char *) d) <= point) |
4563 |
SET_REGS_MATCHED (); |
SET_REGS_MATCHED (); |
4564 |
d++; |
d++; |
4565 |
break; |
break; |
4566 |
|
|
4567 |
case notwordchar: |
case notwordchar: |
4568 |
DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n"); |
DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n"); |
4569 |
PREFETCH (); |
PREFETCH (); |
4573 |
d++; |
d++; |
4574 |
break; |
break; |
4575 |
#endif /* not emacs */ |
#endif /* not emacs */ |
4576 |
|
|
4577 |
default: |
default: |
4578 |
abort (); |
abort (); |
4579 |
} |
} |
4598 |
if (p < pend) |
if (p < pend) |
4599 |
{ |
{ |
4600 |
boolean is_a_jump_n = false; |
boolean is_a_jump_n = false; |
4601 |
|
|
4602 |
/* If failed to a backwards jump that's part of a repetition |
/* If failed to a backwards jump that's part of a repetition |
4603 |
loop, need to pop this failure point and use the next one. */ |
loop, need to pop this failure point and use the next one. */ |
4604 |
switch ((re_opcode_t) *p) |
switch ((re_opcode_t) *p) |
4610 |
case jump: |
case jump: |
4611 |
p1 = p + 1; |
p1 = p + 1; |
4612 |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4613 |
p1 += mcnt; |
p1 += mcnt; |
4614 |
|
|
4615 |
if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n) |
if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n) |
4616 |
|| (!is_a_jump_n |
|| (!is_a_jump_n |
4641 |
|
|
4642 |
|
|
4643 |
/* We are passed P pointing to a register number after a start_memory. |
/* We are passed P pointing to a register number after a start_memory. |
4644 |
|
|
4645 |
Return true if the pattern up to the corresponding stop_memory can |
Return true if the pattern up to the corresponding stop_memory can |
4646 |
match the empty string, and false otherwise. |
match the empty string, and false otherwise. |
4647 |
|
|
4648 |
If we find the matching stop_memory, sets P to point to one past its number. |
If we find the matching stop_memory, sets P to point to one past its number. |
4649 |
Otherwise, sets P to an undefined byte less than or equal to END. |
Otherwise, sets P to an undefined byte less than or equal to END. |
4650 |
|
|
4658 |
int mcnt; |
int mcnt; |
4659 |
/* Point to after the args to the start_memory. */ |
/* Point to after the args to the start_memory. */ |
4660 |
unsigned char *p1 = *p + 2; |
unsigned char *p1 = *p + 2; |
4661 |
|
|
4662 |
while (p1 < end) |
while (p1 < end) |
4663 |
{ |
{ |
4664 |
/* Skip over opcodes that can match nothing, and return true or |
/* Skip over opcodes that can match nothing, and return true or |
4665 |
false, as appropriate, when we get to one that can't, or to the |
false, as appropriate, when we get to one that can't, or to the |
4666 |
matching stop_memory. */ |
matching stop_memory. */ |
4667 |
|
|
4668 |
switch ((re_opcode_t) *p1) |
switch ((re_opcode_t) *p1) |
4669 |
{ |
{ |
4670 |
/* Could be either a loop or a series of alternatives. */ |
/* Could be either a loop or a series of alternatives. */ |
4671 |
case on_failure_jump: |
case on_failure_jump: |
4672 |
p1++; |
p1++; |
4673 |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4674 |
|
|
4675 |
/* If the next operation is not a jump backwards in the |
/* If the next operation is not a jump backwards in the |
4676 |
pattern. */ |
pattern. */ |
4677 |
|
|
4685 |
|
|
4686 |
/on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6 |
/on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6 |
4687 |
/on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3 |
/on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3 |
4688 |
/exactn/1/c |
/exactn/1/c |
4689 |
|
|
4690 |
So, we have to first go through the first (n-1) |
So, we have to first go through the first (n-1) |
4691 |
alternatives and then deal with the last one separately. */ |
alternatives and then deal with the last one separately. */ |
4701 |
is, including the ending `jump_past_alt' and |
is, including the ending `jump_past_alt' and |
4702 |
its number. */ |
its number. */ |
4703 |
|
|
4704 |
if (!alt_match_null_string_p (p1, p1 + mcnt - 3, |
if (!alt_match_null_string_p (p1, p1 + mcnt - 3, |
4705 |
reg_info)) |
reg_info)) |
4706 |
return false; |
return false; |
4707 |
|
|
4708 |
/* Move to right after this alternative, including the |
/* Move to right after this alternative, including the |
4709 |
jump_past_alt. */ |
jump_past_alt. */ |
4710 |
p1 += mcnt; |
p1 += mcnt; |
4711 |
|
|
4712 |
/* Break if it's the beginning of an n-th alternative |
/* Break if it's the beginning of an n-th alternative |
4713 |
that doesn't begin with an on_failure_jump. */ |
that doesn't begin with an on_failure_jump. */ |
4714 |
if ((re_opcode_t) *p1 != on_failure_jump) |
if ((re_opcode_t) *p1 != on_failure_jump) |
4715 |
break; |
break; |
4716 |
|
|
4717 |
/* Still have to check that it's not an n-th |
/* Still have to check that it's not an n-th |
4718 |
alternative that starts with an on_failure_jump. */ |
alternative that starts with an on_failure_jump. */ |
4719 |
p1++; |
p1++; |
4738 |
} /* if mcnt > 0 */ |
} /* if mcnt > 0 */ |
4739 |
break; |
break; |
4740 |
|
|
4741 |
|
|
4742 |
case stop_memory: |
case stop_memory: |
4743 |
assert (p1[1] == **p); |
assert (p1[1] == **p); |
4744 |
*p = p1 + 2; |
*p = p1 + 2; |
4745 |
return true; |
return true; |
4746 |
|
|
4747 |
|
|
4748 |
default: |
default: |
4749 |
if (!common_op_match_null_string_p (&p1, end, reg_info)) |
if (!common_op_match_null_string_p (&p1, end, reg_info)) |
4750 |
return false; |
return false; |
4751 |
} |
} |
4758 |
/* Similar to group_match_null_string_p, but doesn't deal with alternatives: |
/* Similar to group_match_null_string_p, but doesn't deal with alternatives: |
4759 |
It expects P to be the first byte of a single alternative and END one |
It expects P to be the first byte of a single alternative and END one |
4760 |
byte past the last. The alternative can contain groups. */ |
byte past the last. The alternative can contain groups. */ |
4761 |
|
|
4762 |
static boolean |
static boolean |
4763 |
alt_match_null_string_p (p, end, reg_info) |
alt_match_null_string_p (p, end, reg_info) |
4764 |
unsigned char *p, *end; |
unsigned char *p, *end; |
4766 |
{ |
{ |
4767 |
int mcnt; |
int mcnt; |
4768 |
unsigned char *p1 = p; |
unsigned char *p1 = p; |
4769 |
|
|
4770 |
while (p1 < end) |
while (p1 < end) |
4771 |
{ |
{ |
4772 |
/* Skip over opcodes that can match nothing, and break when we get |
/* Skip over opcodes that can match nothing, and break when we get |
4773 |
to one that can't. */ |
to one that can't. */ |
4774 |
|
|
4775 |
switch ((re_opcode_t) *p1) |
switch ((re_opcode_t) *p1) |
4776 |
{ |
{ |
4777 |
/* It's a loop. */ |
/* It's a loop. */ |
4780 |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4781 |
p1 += mcnt; |
p1 += mcnt; |
4782 |
break; |
break; |
4783 |
|
|
4784 |
default: |
default: |
4785 |
if (!common_op_match_null_string_p (&p1, end, reg_info)) |
if (!common_op_match_null_string_p (&p1, end, reg_info)) |
4786 |
return false; |
return false; |
4787 |
} |
} |
4792 |
|
|
4793 |
|
|
4794 |
/* Deals with the ops common to group_match_null_string_p and |
/* Deals with the ops common to group_match_null_string_p and |
4795 |
alt_match_null_string_p. |
alt_match_null_string_p. |
4796 |
|
|
4797 |
Sets P to one after the op and its arguments, if any. */ |
Sets P to one after the op and its arguments, if any. */ |
4798 |
|
|
4799 |
static boolean |
static boolean |
4828 |
reg_no = *p1; |
reg_no = *p1; |
4829 |
assert (reg_no > 0 && reg_no <= MAX_REGNUM); |
assert (reg_no > 0 && reg_no <= MAX_REGNUM); |
4830 |
ret = group_match_null_string_p (&p1, end, reg_info); |
ret = group_match_null_string_p (&p1, end, reg_info); |
4831 |
|
|
4832 |
/* Have to set this here in case we're checking a group which |
/* Have to set this here in case we're checking a group which |
4833 |
contains a group and a back reference to it. */ |
contains a group and a back reference to it. */ |
4834 |
|
|
4838 |
if (!ret) |
if (!ret) |
4839 |
return false; |
return false; |
4840 |
break; |
break; |
4841 |
|
|
4842 |
/* If this is an optimized succeed_n for zero times, make the jump. */ |
/* If this is an optimized succeed_n for zero times, make the jump. */ |
4843 |
case jump: |
case jump: |
4844 |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4850 |
|
|
4851 |
case succeed_n: |
case succeed_n: |
4852 |
/* Get to the number of times to succeed. */ |
/* Get to the number of times to succeed. */ |
4853 |
p1 += 2; |
p1 += 2; |
4854 |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4855 |
|
|
4856 |
if (mcnt == 0) |
if (mcnt == 0) |
4863 |
return false; |
return false; |
4864 |
break; |
break; |
4865 |
|
|
4866 |
case duplicate: |
case duplicate: |
4867 |
if (!REG_MATCH_NULL_STRING_P (reg_info[*p1])) |
if (!REG_MATCH_NULL_STRING_P (reg_info[*p1])) |
4868 |
return false; |
return false; |
4869 |
break; |
break; |
4883 |
|
|
4884 |
/* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN |
/* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN |
4885 |
bytes; nonzero otherwise. */ |
bytes; nonzero otherwise. */ |
4886 |
|
|
4887 |
static int |
static int |
4888 |
bcmp_translate (s1, s2, len, translate) |
bcmp_translate (s1, s2, len, translate) |
4889 |
unsigned char *s1, *s2; |
unsigned char *s1, *s2; |
4904 |
/* re_compile_pattern is the GNU regular expression compiler: it |
/* re_compile_pattern is the GNU regular expression compiler: it |
4905 |
compiles PATTERN (of length SIZE) and puts the result in BUFP. |
compiles PATTERN (of length SIZE) and puts the result in BUFP. |
4906 |
Returns 0 if the pattern was valid, otherwise an error string. |
Returns 0 if the pattern was valid, otherwise an error string. |
4907 |
|
|
4908 |
Assumes the `allocated' (and perhaps `buffer') and `translate' fields |
Assumes the `allocated' (and perhaps `buffer') and `translate' fields |
4909 |
are set in BUFP on entry. |
are set in BUFP on entry. |
4910 |
|
|
4911 |
We call regex_compile to do the actual compilation. */ |
We call regex_compile to do the actual compilation. */ |
4912 |
|
|
4913 |
const char * |
const char * |
4917 |
struct re_pattern_buffer *bufp; |
struct re_pattern_buffer *bufp; |
4918 |
{ |
{ |
4919 |
reg_errcode_t ret; |
reg_errcode_t ret; |
4920 |
|
|
4921 |
/* GNU code is written to assume at least RE_NREGS registers will be set |
/* GNU code is written to assume at least RE_NREGS registers will be set |
4922 |
(and at least one extra will be -1). */ |
(and at least one extra will be -1). */ |
4923 |
bufp->regs_allocated = REGS_UNALLOCATED; |
bufp->regs_allocated = REGS_UNALLOCATED; |
4924 |
|
|
4925 |
/* And GNU code determines whether or not to get register information |
/* And GNU code determines whether or not to get register information |
4926 |
by passing null for the REGS argument to re_match, etc., not by |
by passing null for the REGS argument to re_match, etc., not by |
4927 |
setting no_sub. */ |
setting no_sub. */ |
4928 |
bufp->no_sub = 0; |
bufp->no_sub = 0; |
4929 |
|
|
4930 |
/* Match anchors at newline. */ |
/* Match anchors at newline. */ |
4931 |
bufp->newline_anchor = 1; |
bufp->newline_anchor = 1; |
4932 |
|
|
4933 |
ret = regex_compile (pattern, length, re_syntax_options, bufp); |
ret = regex_compile (pattern, length, re_syntax_options, bufp); |
4934 |
|
|
4935 |
return re_error_msg[(int) ret]; |
return re_error_msg[(int) ret]; |
4936 |
} |
} |
4937 |
|
|
4938 |
/* Entry points compatible with 4.2 BSD regex library. We don't define |
/* Entry points compatible with 4.2 BSD regex library. We don't define |
4939 |
them unless specifically requested. */ |
them unless specifically requested. */ |
4948 |
const char *s; |
const char *s; |
4949 |
{ |
{ |
4950 |
reg_errcode_t ret; |
reg_errcode_t ret; |
4951 |
|
|
4952 |
if (!s) |
if (!s) |
4953 |
{ |
{ |
4954 |
if (!re_comp_buf.buffer) |
if (!re_comp_buf.buffer) |
4975 |
re_comp_buf.newline_anchor = 1; |
re_comp_buf.newline_anchor = 1; |
4976 |
|
|
4977 |
ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf); |
ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf); |
4978 |
|
|
4979 |
/* Yes, we're discarding `const' here. */ |
/* Yes, we're discarding `const' here. */ |
4980 |
return (char *) re_error_msg[(int) ret]; |
return (char *) re_error_msg[(int) ret]; |
4981 |
} |
} |
5032 |
int |
int |
5033 |
regcomp (preg, pattern, cflags) |
regcomp (preg, pattern, cflags) |
5034 |
regex_t *preg; |
regex_t *preg; |
5035 |
const char *pattern; |
const char *pattern; |
5036 |
int cflags; |
int cflags; |
5037 |
{ |
{ |
5038 |
reg_errcode_t ret; |
reg_errcode_t ret; |
5044 |
preg->buffer = 0; |
preg->buffer = 0; |
5045 |
preg->allocated = 0; |
preg->allocated = 0; |
5046 |
preg->used = 0; |
preg->used = 0; |
5047 |
|
|
5048 |
/* Don't bother to use a fastmap when searching. This simplifies the |
/* Don't bother to use a fastmap when searching. This simplifies the |
5049 |
REG_NEWLINE case: if we used a fastmap, we'd have to put all the |
REG_NEWLINE case: if we used a fastmap, we'd have to put all the |
5050 |
characters after newlines into the fastmap. This way, we just try |
characters after newlines into the fastmap. This way, we just try |
5051 |
every character. */ |
every character. */ |
5052 |
preg->fastmap = 0; |
preg->fastmap = 0; |
5053 |
|
|
5054 |
if (cflags & REG_ICASE) |
if (cflags & REG_ICASE) |
5055 |
{ |
{ |
5056 |
unsigned i; |
unsigned i; |
5057 |
|
|
5058 |
preg->translate = (char *) malloc (CHAR_SET_SIZE); |
preg->translate = (char *) malloc (CHAR_SET_SIZE); |
5059 |
if (preg->translate == NULL) |
if (preg->translate == NULL) |
5060 |
return (int) REG_ESPACE; |
return (int) REG_ESPACE; |
5079 |
|
|
5080 |
preg->no_sub = !!(cflags & REG_NOSUB); |
preg->no_sub = !!(cflags & REG_NOSUB); |
5081 |
|
|
5082 |
/* POSIX says a null character in the pattern terminates it, so we |
/* POSIX says a null character in the pattern terminates it, so we |
5083 |
can use strlen here in compiling the pattern. */ |
can use strlen here in compiling the pattern. */ |
5084 |
ret = regex_compile (pattern, strlen (pattern), syntax, preg); |
ret = regex_compile (pattern, strlen (pattern), syntax, preg); |
5085 |
|
|
5086 |
/* POSIX doesn't distinguish between an unmatched open-group and an |
/* POSIX doesn't distinguish between an unmatched open-group and an |
5087 |
unmatched close-group: both are REG_EPAREN. */ |
unmatched close-group: both are REG_EPAREN. */ |
5088 |
if (ret == REG_ERPAREN) ret = REG_EPAREN; |
if (ret == REG_ERPAREN) ret = REG_EPAREN; |
5089 |
|
|
5090 |
return (int) ret; |
return (int) ret; |
5091 |
} |
} |
5092 |
|
|
5093 |
|
|
5094 |
/* regexec searches for a given pattern, specified by PREG, in the |
/* regexec searches for a given pattern, specified by PREG, in the |
5095 |
string STRING. |
string STRING. |
5096 |
|
|
5097 |
If NMATCH is zero or REG_NOSUB was set in the cflags argument to |
If NMATCH is zero or REG_NOSUB was set in the cflags argument to |
5098 |
`regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at |
`regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at |
5099 |
least NMATCH elements, and we set them to the offsets of the |
least NMATCH elements, and we set them to the offsets of the |
5100 |
corresponding matched substrings. |
corresponding matched substrings. |
5101 |
|
|
5102 |
EFLAGS specifies `execution flags' which affect matching: if |
EFLAGS specifies `execution flags' which affect matching: if |
5103 |
REG_NOTBOL is set, then ^ does not match at the beginning of the |
REG_NOTBOL is set, then ^ does not match at the beginning of the |
5104 |
string; if REG_NOTEOL is set, then $ does not match at the end. |
string; if REG_NOTEOL is set, then $ does not match at the end. |
5105 |
|
|
5106 |
We return 0 if we find a match and REG_NOMATCH if not. */ |
We return 0 if we find a match and REG_NOMATCH if not. */ |
5107 |
|
|
5108 |
int |
int |
5109 |
regexec (preg, string, nmatch, pmatch, eflags) |
regexec (preg, string, nmatch, pmatch, eflags) |
5110 |
const regex_t *preg; |
const regex_t *preg; |
5111 |
const char *string; |
const char *string; |
5112 |
size_t nmatch; |
size_t nmatch; |
5113 |
regmatch_t pmatch[]; |
regmatch_t pmatch[]; |
5114 |
int eflags; |
int eflags; |
5115 |
{ |
{ |
5116 |
int ret; |
int ret; |
5120 |
boolean want_reg_info = !preg->no_sub && nmatch > 0; |
boolean want_reg_info = !preg->no_sub && nmatch > 0; |
5121 |
|
|
5122 |
private_preg = *preg; |
private_preg = *preg; |
5123 |
|
|
5124 |
private_preg.not_bol = !!(eflags & REG_NOTBOL); |
private_preg.not_bol = !!(eflags & REG_NOTBOL); |
5125 |
private_preg.not_eol = !!(eflags & REG_NOTEOL); |
private_preg.not_eol = !!(eflags & REG_NOTEOL); |
5126 |
|
|
5127 |
/* The user has told us exactly how many registers to return |
/* The user has told us exactly how many registers to return |
5128 |
information about, via `nmatch'. We have to pass that on to the |
information about, via `nmatch'. We have to pass that on to the |
5129 |
matching routines. */ |
matching routines. */ |
5130 |
private_preg.regs_allocated = REGS_FIXED; |
private_preg.regs_allocated = REGS_FIXED; |
5131 |
|
|
5132 |
if (want_reg_info) |
if (want_reg_info) |
5133 |
{ |
{ |
5134 |
regs.num_regs = nmatch; |
regs.num_regs = nmatch; |
5142 |
ret = re_search (&private_preg, string, len, |
ret = re_search (&private_preg, string, len, |
5143 |
/* start: */ 0, /* range: */ len, |
/* start: */ 0, /* range: */ len, |
5144 |
want_reg_info ? ®s : (struct re_registers *) 0); |
want_reg_info ? ®s : (struct re_registers *) 0); |
5145 |
|
|
5146 |
/* Copy the register information to the POSIX structure. */ |
/* Copy the register information to the POSIX structure. */ |
5147 |
if (want_reg_info) |
if (want_reg_info) |
5148 |
{ |
{ |
5182 |
|
|
5183 |
if (errcode < 0 |
if (errcode < 0 |
5184 |
|| errcode >= (sizeof (re_error_msg) / sizeof (re_error_msg[0]))) |
|| errcode >= (sizeof (re_error_msg) / sizeof (re_error_msg[0]))) |
5185 |
/* Only error codes returned by the rest of the code should be passed |
/* Only error codes returned by the rest of the code should be passed |
5186 |
to this routine. If we are given anything else, or if other regex |
to this routine. If we are given anything else, or if other regex |
5187 |
code generates an invalid error code, then the program has a bug. |
code generates an invalid error code, then the program has a bug. |
5188 |
Dump core so we can fix it. */ |
Dump core so we can fix it. */ |
5196 |
msg = "Success"; |
msg = "Success"; |
5197 |
|
|
5198 |
msg_size = strlen (msg) + 1; /* Includes the null. */ |
msg_size = strlen (msg) + 1; /* Includes the null. */ |
5199 |
|
|
5200 |
if (errbuf_size != 0) |
if (errbuf_size != 0) |
5201 |
{ |
{ |
5202 |
if (msg_size > errbuf_size) |
if (msg_size > errbuf_size) |
5221 |
if (preg->buffer != NULL) |
if (preg->buffer != NULL) |
5222 |
free (preg->buffer); |
free (preg->buffer); |
5223 |
preg->buffer = NULL; |
preg->buffer = NULL; |
5224 |
|
|
5225 |
preg->allocated = 0; |
preg->allocated = 0; |
5226 |
preg->used = 0; |
preg->used = 0; |
5227 |
|
|