| 6 |
and semantics are as close as possible to those of the Perl 5 language. |
and semantics are as close as possible to those of the Perl 5 language. |
| 7 |
|
|
| 8 |
Written by Philip Hazel |
Written by Philip Hazel |
| 9 |
Copyright (c) 1997-2009 University of Cambridge |
Copyright (c) 1997-2010 University of Cambridge |
| 10 |
|
|
| 11 |
----------------------------------------------------------------------------- |
----------------------------------------------------------------------------- |
| 12 |
Redistribution and use in source and binary forms, with or without |
Redistribution and use in source and binary forms, with or without |
| 53 |
#include "pcre_internal.h" |
#include "pcre_internal.h" |
| 54 |
|
|
| 55 |
|
|
| 56 |
/* When DEBUG is defined, we need the pcre_printint() function, which is also |
/* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is |
| 57 |
used by pcretest. DEBUG is not defined when building a production library. */ |
also used by pcretest. PCRE_DEBUG is not defined when building a production |
| 58 |
|
library. */ |
| 59 |
|
|
| 60 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
| 61 |
#include "pcre_printint.src" |
#include "pcre_printint.src" |
| 62 |
#endif |
#endif |
| 63 |
|
|
| 263 |
it is now one long string. We cannot use a table of offsets, because the |
it is now one long string. We cannot use a table of offsets, because the |
| 264 |
lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we |
lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we |
| 265 |
simply count through to the one we want - this isn't a performance issue |
simply count through to the one we want - this isn't a performance issue |
| 266 |
because these strings are used only when there is a compilation error. */ |
because these strings are used only when there is a compilation error. |
| 267 |
|
|
| 268 |
|
Each substring ends with \0 to insert a null character. This includes the final |
| 269 |
|
substring, so that the whole string ends with \0\0, which can be detected when |
| 270 |
|
counting through. */ |
| 271 |
|
|
| 272 |
static const char error_texts[] = |
static const char error_texts[] = |
| 273 |
"no error\0" |
"no error\0" |
| 346 |
"number is too big\0" |
"number is too big\0" |
| 347 |
"subpattern name expected\0" |
"subpattern name expected\0" |
| 348 |
"digit expected after (?+\0" |
"digit expected after (?+\0" |
| 349 |
"] is an invalid data character in JavaScript compatibility mode"; |
"] is an invalid data character in JavaScript compatibility mode\0" |
| 350 |
|
/* 65 */ |
| 351 |
|
"different names for subpatterns of the same number are not allowed\0"; |
| 352 |
|
|
| 353 |
/* Table to identify digits and hex digits. This is used when compiling |
/* Table to identify digits and hex digits. This is used when compiling |
| 354 |
patterns. Note that the tables in chartables are dependent on the locale, and |
patterns. Note that the tables in chartables are dependent on the locale, and |
| 506 |
find_error_text(int n) |
find_error_text(int n) |
| 507 |
{ |
{ |
| 508 |
const char *s = error_texts; |
const char *s = error_texts; |
| 509 |
for (; n > 0; n--) while (*s++ != 0) {}; |
for (; n > 0; n--) |
| 510 |
|
{ |
| 511 |
|
while (*s++ != 0) {}; |
| 512 |
|
if (*s == 0) return "Error text not found (please report)"; |
| 513 |
|
} |
| 514 |
return s; |
return s; |
| 515 |
} |
} |
| 516 |
|
|
| 1110 |
if (name != NULL && lorn == ptr - thisname && |
if (name != NULL && lorn == ptr - thisname && |
| 1111 |
strncmp((const char *)name, (const char *)thisname, lorn) == 0) |
strncmp((const char *)name, (const char *)thisname, lorn) == 0) |
| 1112 |
return *count; |
return *count; |
| 1113 |
|
term++; |
| 1114 |
} |
} |
| 1115 |
} |
} |
| 1116 |
} |
} |
| 1145 |
BOOL negate_class = FALSE; |
BOOL negate_class = FALSE; |
| 1146 |
for (;;) |
for (;;) |
| 1147 |
{ |
{ |
| 1148 |
int c = *(++ptr); |
if (ptr[1] == CHAR_BACKSLASH) |
|
if (c == CHAR_BACKSLASH) |
|
| 1149 |
{ |
{ |
| 1150 |
if (ptr[1] == CHAR_E) |
if (ptr[2] == CHAR_E) |
| 1151 |
ptr++; |
ptr+= 2; |
| 1152 |
else if (strncmp((const char *)ptr+1, |
else if (strncmp((const char *)ptr+2, |
| 1153 |
STR_Q STR_BACKSLASH STR_E, 3) == 0) |
STR_Q STR_BACKSLASH STR_E, 3) == 0) |
| 1154 |
ptr += 3; |
ptr += 4; |
| 1155 |
else |
else |
| 1156 |
break; |
break; |
| 1157 |
} |
} |
| 1158 |
else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT) |
else if (!negate_class && ptr[1] == CHAR_CIRCUMFLEX_ACCENT) |
| 1159 |
|
{ |
| 1160 |
negate_class = TRUE; |
negate_class = TRUE; |
| 1161 |
|
ptr++; |
| 1162 |
|
} |
| 1163 |
else break; |
else break; |
| 1164 |
} |
} |
| 1165 |
|
|
| 1325 |
|
|
| 1326 |
case OP_CALLOUT: |
case OP_CALLOUT: |
| 1327 |
case OP_CREF: |
case OP_CREF: |
| 1328 |
|
case OP_NCREF: |
| 1329 |
case OP_RREF: |
case OP_RREF: |
| 1330 |
|
case OP_NRREF: |
| 1331 |
case OP_DEF: |
case OP_DEF: |
| 1332 |
code += _pcre_OP_lengths[*code]; |
code += _pcre_OP_lengths[*code]; |
| 1333 |
break; |
break; |
| 1343 |
|
|
| 1344 |
|
|
| 1345 |
/************************************************* |
/************************************************* |
| 1346 |
* Find the fixed length of a pattern * |
* Find the fixed length of a branch * |
| 1347 |
*************************************************/ |
*************************************************/ |
| 1348 |
|
|
| 1349 |
/* Scan a pattern and compute the fixed length of subject that will match it, |
/* Scan a branch and compute the fixed length of subject that will match it, |
| 1350 |
if the length is fixed. This is needed for dealing with backward assertions. |
if the length is fixed. This is needed for dealing with backward assertions. |
| 1351 |
In UTF8 mode, the result is in characters rather than bytes. |
In UTF8 mode, the result is in characters rather than bytes. The branch is |
| 1352 |
|
temporarily terminated with OP_END when this function is called. |
| 1353 |
|
|
| 1354 |
|
This function is called when a backward assertion is encountered, so that if it |
| 1355 |
|
fails, the error message can point to the correct place in the pattern. |
| 1356 |
|
However, we cannot do this when the assertion contains subroutine calls, |
| 1357 |
|
because they can be forward references. We solve this by remembering this case |
| 1358 |
|
and doing the check at the end; a flag specifies which mode we are running in. |
| 1359 |
|
|
| 1360 |
Arguments: |
Arguments: |
| 1361 |
code points to the start of the pattern (the bracket) |
code points to the start of the pattern (the bracket) |
| 1362 |
options the compiling options |
options the compiling options |
| 1363 |
|
atend TRUE if called when the pattern is complete |
| 1364 |
|
cd the "compile data" structure |
| 1365 |
|
|
| 1366 |
Returns: the fixed length, or -1 if there is no fixed length, |
Returns: the fixed length, |
| 1367 |
|
or -1 if there is no fixed length, |
| 1368 |
or -2 if \C was encountered |
or -2 if \C was encountered |
| 1369 |
|
or -3 if an OP_RECURSE item was encountered and atend is FALSE |
| 1370 |
*/ |
*/ |
| 1371 |
|
|
| 1372 |
static int |
static int |
| 1373 |
find_fixedlength(uschar *code, int options) |
find_fixedlength(uschar *code, int options, BOOL atend, compile_data *cd) |
| 1374 |
{ |
{ |
| 1375 |
int length = -1; |
int length = -1; |
| 1376 |
|
|
| 1383 |
for (;;) |
for (;;) |
| 1384 |
{ |
{ |
| 1385 |
int d; |
int d; |
| 1386 |
|
uschar *ce, *cs; |
| 1387 |
register int op = *cc; |
register int op = *cc; |
| 1388 |
switch (op) |
switch (op) |
| 1389 |
{ |
{ |
| 1391 |
case OP_BRA: |
case OP_BRA: |
| 1392 |
case OP_ONCE: |
case OP_ONCE: |
| 1393 |
case OP_COND: |
case OP_COND: |
| 1394 |
d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), options); |
d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), options, atend, cd); |
| 1395 |
if (d < 0) return d; |
if (d < 0) return d; |
| 1396 |
branchlength += d; |
branchlength += d; |
| 1397 |
do cc += GET(cc, 1); while (*cc == OP_ALT); |
do cc += GET(cc, 1); while (*cc == OP_ALT); |
| 1414 |
branchlength = 0; |
branchlength = 0; |
| 1415 |
break; |
break; |
| 1416 |
|
|
| 1417 |
|
/* A true recursion implies not fixed length, but a subroutine call may |
| 1418 |
|
be OK. If the subroutine is a forward reference, we can't deal with |
| 1419 |
|
it until the end of the pattern, so return -3. */ |
| 1420 |
|
|
| 1421 |
|
case OP_RECURSE: |
| 1422 |
|
if (!atend) return -3; |
| 1423 |
|
cs = ce = (uschar *)cd->start_code + GET(cc, 1); /* Start subpattern */ |
| 1424 |
|
do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ |
| 1425 |
|
if (cc > cs && cc < ce) return -1; /* Recursion */ |
| 1426 |
|
d = find_fixedlength(cs + 2, options, atend, cd); |
| 1427 |
|
if (d < 0) return d; |
| 1428 |
|
branchlength += d; |
| 1429 |
|
cc += 1 + LINK_SIZE; |
| 1430 |
|
break; |
| 1431 |
|
|
| 1432 |
/* Skip over assertive subpatterns */ |
/* Skip over assertive subpatterns */ |
| 1433 |
|
|
| 1434 |
case OP_ASSERT: |
case OP_ASSERT: |
| 1442 |
|
|
| 1443 |
case OP_REVERSE: |
case OP_REVERSE: |
| 1444 |
case OP_CREF: |
case OP_CREF: |
| 1445 |
|
case OP_NCREF: |
| 1446 |
case OP_RREF: |
case OP_RREF: |
| 1447 |
|
case OP_NRREF: |
| 1448 |
case OP_DEF: |
case OP_DEF: |
| 1449 |
case OP_OPT: |
case OP_OPT: |
| 1450 |
case OP_CALLOUT: |
case OP_CALLOUT: |
| 1467 |
branchlength++; |
branchlength++; |
| 1468 |
cc += 2; |
cc += 2; |
| 1469 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 1470 |
if ((options & PCRE_UTF8) != 0) |
if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0) |
| 1471 |
{ |
cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
|
while ((*cc & 0xc0) == 0x80) cc++; |
|
|
} |
|
| 1472 |
#endif |
#endif |
| 1473 |
break; |
break; |
| 1474 |
|
|
| 1479 |
branchlength += GET2(cc,1); |
branchlength += GET2(cc,1); |
| 1480 |
cc += 4; |
cc += 4; |
| 1481 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 1482 |
if ((options & PCRE_UTF8) != 0) |
if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0) |
| 1483 |
{ |
cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
|
while((*cc & 0x80) == 0x80) cc++; |
|
|
} |
|
| 1484 |
#endif |
#endif |
| 1485 |
break; |
break; |
| 1486 |
|
|
| 1559 |
|
|
| 1560 |
|
|
| 1561 |
/************************************************* |
/************************************************* |
| 1562 |
* Scan compiled regex for numbered bracket * |
* Scan compiled regex for specific bracket * |
| 1563 |
*************************************************/ |
*************************************************/ |
| 1564 |
|
|
| 1565 |
/* This little function scans through a compiled pattern until it finds a |
/* This little function scans through a compiled pattern until it finds a |
| 1566 |
capturing bracket with the given number. |
capturing bracket with the given number, or, if the number is negative, an |
| 1567 |
|
instance of OP_REVERSE for a lookbehind. The function is global in the C sense |
| 1568 |
|
so that it can be called from pcre_study() when finding the minimum matching |
| 1569 |
|
length. |
| 1570 |
|
|
| 1571 |
Arguments: |
Arguments: |
| 1572 |
code points to start of expression |
code points to start of expression |
| 1573 |
utf8 TRUE in UTF-8 mode |
utf8 TRUE in UTF-8 mode |
| 1574 |
number the required bracket number |
number the required bracket number or negative to find a lookbehind |
| 1575 |
|
|
| 1576 |
Returns: pointer to the opcode for the bracket, or NULL if not found |
Returns: pointer to the opcode for the bracket, or NULL if not found |
| 1577 |
*/ |
*/ |
| 1578 |
|
|
| 1579 |
static const uschar * |
const uschar * |
| 1580 |
find_bracket(const uschar *code, BOOL utf8, int number) |
_pcre_find_bracket(const uschar *code, BOOL utf8, int number) |
| 1581 |
{ |
{ |
| 1582 |
for (;;) |
for (;;) |
| 1583 |
{ |
{ |
| 1590 |
|
|
| 1591 |
if (c == OP_XCLASS) code += GET(code, 1); |
if (c == OP_XCLASS) code += GET(code, 1); |
| 1592 |
|
|
| 1593 |
|
/* Handle recursion */ |
| 1594 |
|
|
| 1595 |
|
else if (c == OP_REVERSE) |
| 1596 |
|
{ |
| 1597 |
|
if (number < 0) return (uschar *)code; |
| 1598 |
|
code += _pcre_OP_lengths[c]; |
| 1599 |
|
} |
| 1600 |
|
|
| 1601 |
/* Handle capturing bracket */ |
/* Handle capturing bracket */ |
| 1602 |
|
|
| 1603 |
else if (c == OP_CBRA) |
else if (c == OP_CBRA) |
| 1963 |
case OP_QUERY: |
case OP_QUERY: |
| 1964 |
case OP_MINQUERY: |
case OP_MINQUERY: |
| 1965 |
case OP_POSQUERY: |
case OP_POSQUERY: |
| 1966 |
|
if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f]; |
| 1967 |
|
break; |
| 1968 |
|
|
| 1969 |
case OP_UPTO: |
case OP_UPTO: |
| 1970 |
case OP_MINUPTO: |
case OP_MINUPTO: |
| 1971 |
case OP_POSUPTO: |
case OP_POSUPTO: |
| 1972 |
if (utf8) while ((code[2] & 0xc0) == 0x80) code++; |
if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f]; |
| 1973 |
break; |
break; |
| 1974 |
#endif |
#endif |
| 1975 |
} |
} |
| 2002 |
could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr, |
could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr, |
| 2003 |
BOOL utf8) |
BOOL utf8) |
| 2004 |
{ |
{ |
| 2005 |
while (bcptr != NULL && bcptr->current >= code) |
while (bcptr != NULL && bcptr->current_branch >= code) |
| 2006 |
{ |
{ |
| 2007 |
if (!could_be_empty_branch(bcptr->current, endcode, utf8)) return FALSE; |
if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8)) |
| 2008 |
|
return FALSE; |
| 2009 |
bcptr = bcptr->outer; |
bcptr = bcptr->outer; |
| 2010 |
} |
} |
| 2011 |
return TRUE; |
return TRUE; |
| 2667 |
uschar *utf8_char = NULL; |
uschar *utf8_char = NULL; |
| 2668 |
#endif |
#endif |
| 2669 |
|
|
| 2670 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
| 2671 |
if (lengthptr != NULL) DPRINTF((">> start branch\n")); |
if (lengthptr != NULL) DPRINTF((">> start branch\n")); |
| 2672 |
#endif |
#endif |
| 2673 |
|
|
| 2726 |
|
|
| 2727 |
if (lengthptr != NULL) |
if (lengthptr != NULL) |
| 2728 |
{ |
{ |
| 2729 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
| 2730 |
if (code > cd->hwm) cd->hwm = code; /* High water info */ |
if (code > cd->hwm) cd->hwm = code; /* High water info */ |
| 2731 |
#endif |
#endif |
| 2732 |
if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */ |
if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */ |
| 3924 |
|
|
| 3925 |
if (repeat_max == 0) goto END_REPEAT; |
if (repeat_max == 0) goto END_REPEAT; |
| 3926 |
|
|
| 3927 |
|
/*--------------------------------------------------------------------*/ |
| 3928 |
|
/* This code is obsolete from release 8.00; the restriction was finally |
| 3929 |
|
removed: */ |
| 3930 |
|
|
| 3931 |
/* All real repeats make it impossible to handle partial matching (maybe |
/* All real repeats make it impossible to handle partial matching (maybe |
| 3932 |
one day we will be able to remove this restriction). */ |
one day we will be able to remove this restriction). */ |
| 3933 |
|
|
| 3934 |
if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; |
/* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */ |
| 3935 |
|
/*--------------------------------------------------------------------*/ |
| 3936 |
|
|
| 3937 |
/* Combine the op_type with the repeat_type */ |
/* Combine the op_type with the repeat_type */ |
| 3938 |
|
|
| 4079 |
goto END_REPEAT; |
goto END_REPEAT; |
| 4080 |
} |
} |
| 4081 |
|
|
| 4082 |
|
/*--------------------------------------------------------------------*/ |
| 4083 |
|
/* This code is obsolete from release 8.00; the restriction was finally |
| 4084 |
|
removed: */ |
| 4085 |
|
|
| 4086 |
/* All real repeats make it impossible to handle partial matching (maybe |
/* All real repeats make it impossible to handle partial matching (maybe |
| 4087 |
one day we will be able to remove this restriction). */ |
one day we will be able to remove this restriction). */ |
| 4088 |
|
|
| 4089 |
if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; |
/* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */ |
| 4090 |
|
/*--------------------------------------------------------------------*/ |
| 4091 |
|
|
| 4092 |
if (repeat_min == 0 && repeat_max == -1) |
if (repeat_min == 0 && repeat_max == -1) |
| 4093 |
*code++ = OP_CRSTAR + repeat_type; |
*code++ = OP_CRSTAR + repeat_type; |
| 4222 |
{ |
{ |
| 4223 |
/* In the pre-compile phase, we don't actually do the replication. We |
/* In the pre-compile phase, we don't actually do the replication. We |
| 4224 |
just adjust the length as if we had. Do some paranoid checks for |
just adjust the length as if we had. Do some paranoid checks for |
| 4225 |
potential integer overflow. */ |
potential integer overflow. The INT64_OR_DOUBLE type is a 64-bit |
| 4226 |
|
integer type when available, otherwise double. */ |
| 4227 |
|
|
| 4228 |
if (lengthptr != NULL) |
if (lengthptr != NULL) |
| 4229 |
{ |
{ |
| 4230 |
int delta = (repeat_min - 1)*length_prevgroup; |
int delta = (repeat_min - 1)*length_prevgroup; |
| 4231 |
if ((double)(repeat_min - 1)*(double)length_prevgroup > |
if ((INT64_OR_DOUBLE)(repeat_min - 1)* |
| 4232 |
(double)INT_MAX || |
(INT64_OR_DOUBLE)length_prevgroup > |
| 4233 |
|
(INT64_OR_DOUBLE)INT_MAX || |
| 4234 |
OFLOW_MAX - *lengthptr < delta) |
OFLOW_MAX - *lengthptr < delta) |
| 4235 |
{ |
{ |
| 4236 |
*errorcodeptr = ERR20; |
*errorcodeptr = ERR20; |
| 4276 |
just adjust the length as if we had. For each repetition we must add 1 |
just adjust the length as if we had. For each repetition we must add 1 |
| 4277 |
to the length for BRAZERO and for all but the last repetition we must |
to the length for BRAZERO and for all but the last repetition we must |
| 4278 |
add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some |
add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some |
| 4279 |
paranoid checks to avoid integer overflow. */ |
paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is |
| 4280 |
|
a 64-bit integer type when available, otherwise double. */ |
| 4281 |
|
|
| 4282 |
if (lengthptr != NULL && repeat_max > 0) |
if (lengthptr != NULL && repeat_max > 0) |
| 4283 |
{ |
{ |
| 4284 |
int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) - |
int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) - |
| 4285 |
2 - 2*LINK_SIZE; /* Last one doesn't nest */ |
2 - 2*LINK_SIZE; /* Last one doesn't nest */ |
| 4286 |
if ((double)repeat_max * |
if ((INT64_OR_DOUBLE)repeat_max * |
| 4287 |
(double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE) |
(INT64_OR_DOUBLE)(length_prevgroup + 1 + 2 + 2*LINK_SIZE) |
| 4288 |
> (double)INT_MAX || |
> (INT64_OR_DOUBLE)INT_MAX || |
| 4289 |
OFLOW_MAX - *lengthptr < delta) |
OFLOW_MAX - *lengthptr < delta) |
| 4290 |
{ |
{ |
| 4291 |
*errorcodeptr = ERR20; |
*errorcodeptr = ERR20; |
| 4405 |
if (possessive_quantifier) |
if (possessive_quantifier) |
| 4406 |
{ |
{ |
| 4407 |
int len; |
int len; |
| 4408 |
if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT || |
|
| 4409 |
*tempcode == OP_NOTEXACT) |
if (*tempcode == OP_TYPEEXACT) |
| 4410 |
tempcode += _pcre_OP_lengths[*tempcode] + |
tempcode += _pcre_OP_lengths[*tempcode] + |
| 4411 |
((*tempcode == OP_TYPEEXACT && |
((tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP)? 2 : 0); |
| 4412 |
(tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0); |
|
| 4413 |
|
else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT) |
| 4414 |
|
{ |
| 4415 |
|
tempcode += _pcre_OP_lengths[*tempcode]; |
| 4416 |
|
#ifdef SUPPORT_UTF8 |
| 4417 |
|
if (utf8 && tempcode[-1] >= 0xc0) |
| 4418 |
|
tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f]; |
| 4419 |
|
#endif |
| 4420 |
|
} |
| 4421 |
|
|
| 4422 |
len = code - tempcode; |
len = code - tempcode; |
| 4423 |
if (len > 0) switch (*tempcode) |
if (len > 0) switch (*tempcode) |
| 4424 |
{ |
{ |
| 4437 |
case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break; |
case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break; |
| 4438 |
case OP_NOTUPTO: *tempcode = OP_NOTPOSUPTO; break; |
case OP_NOTUPTO: *tempcode = OP_NOTPOSUPTO; break; |
| 4439 |
|
|
| 4440 |
|
/* Because we are moving code along, we must ensure that any |
| 4441 |
|
pending recursive references are updated. */ |
| 4442 |
|
|
| 4443 |
default: |
default: |
| 4444 |
|
*code = OP_END; |
| 4445 |
|
adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm); |
| 4446 |
memmove(tempcode + 1+LINK_SIZE, tempcode, len); |
memmove(tempcode + 1+LINK_SIZE, tempcode, len); |
| 4447 |
code += 1 + LINK_SIZE; |
code += 1 + LINK_SIZE; |
| 4448 |
len += 1 + LINK_SIZE; |
len += 1 + LINK_SIZE; |
| 4501 |
if (namelen == verbs[i].len && |
if (namelen == verbs[i].len && |
| 4502 |
strncmp((char *)name, vn, namelen) == 0) |
strncmp((char *)name, vn, namelen) == 0) |
| 4503 |
{ |
{ |
| 4504 |
*code = verbs[i].op; |
/* Check for open captures before ACCEPT */ |
| 4505 |
if (*code++ == OP_ACCEPT) cd->had_accept = TRUE; |
|
| 4506 |
|
if (verbs[i].op == OP_ACCEPT) |
| 4507 |
|
{ |
| 4508 |
|
open_capitem *oc; |
| 4509 |
|
cd->had_accept = TRUE; |
| 4510 |
|
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
| 4511 |
|
{ |
| 4512 |
|
*code++ = OP_CLOSE; |
| 4513 |
|
PUT2INC(code, 0, oc->number); |
| 4514 |
|
} |
| 4515 |
|
} |
| 4516 |
|
*code++ = verbs[i].op; |
| 4517 |
break; |
break; |
| 4518 |
} |
} |
| 4519 |
vn += verbs[i].len + 1; |
vn += verbs[i].len + 1; |
| 4675 |
} |
} |
| 4676 |
|
|
| 4677 |
/* Otherwise (did not start with "+" or "-"), start by looking for the |
/* Otherwise (did not start with "+" or "-"), start by looking for the |
| 4678 |
name. */ |
name. If we find a name, add one to the opcode to change OP_CREF or |
| 4679 |
|
OP_RREF into OP_NCREF or OP_NRREF. These behave exactly the same, |
| 4680 |
|
except they record that the reference was originally to a name. The |
| 4681 |
|
information is used to check duplicate names. */ |
| 4682 |
|
|
| 4683 |
slot = cd->name_table; |
slot = cd->name_table; |
| 4684 |
for (i = 0; i < cd->names_found; i++) |
for (i = 0; i < cd->names_found; i++) |
| 4693 |
{ |
{ |
| 4694 |
recno = GET2(slot, 0); |
recno = GET2(slot, 0); |
| 4695 |
PUT2(code, 2+LINK_SIZE, recno); |
PUT2(code, 2+LINK_SIZE, recno); |
| 4696 |
|
code[1+LINK_SIZE]++; |
| 4697 |
} |
} |
| 4698 |
|
|
| 4699 |
/* Search the pattern for a forward reference */ |
/* Search the pattern for a forward reference */ |
| 4702 |
(options & PCRE_EXTENDED) != 0)) > 0) |
(options & PCRE_EXTENDED) != 0)) > 0) |
| 4703 |
{ |
{ |
| 4704 |
PUT2(code, 2+LINK_SIZE, i); |
PUT2(code, 2+LINK_SIZE, i); |
| 4705 |
|
code[1+LINK_SIZE]++; |
| 4706 |
} |
} |
| 4707 |
|
|
| 4708 |
/* If terminator == 0 it means that the name followed directly after |
/* If terminator == 0 it means that the name followed directly after |
| 4895 |
} |
} |
| 4896 |
} |
} |
| 4897 |
|
|
| 4898 |
/* In the real compile, create the entry in the table */ |
/* In the real compile, create the entry in the table, maintaining |
| 4899 |
|
alphabetical order. Duplicate names for different numbers are |
| 4900 |
|
permitted only if PCRE_DUPNAMES is set. Duplicate names for the same |
| 4901 |
|
number are always OK. (An existing number can be re-used if (?| |
| 4902 |
|
appears in the pattern.) In either event, a duplicate name results in |
| 4903 |
|
a duplicate entry in the table, even if the number is the same. This |
| 4904 |
|
is because the number of names, and hence the table size, is computed |
| 4905 |
|
in the pre-compile, and it affects various numbers and pointers which |
| 4906 |
|
would all have to be modified, and the compiled code moved down, if |
| 4907 |
|
duplicates with the same number were omitted from the table. This |
| 4908 |
|
doesn't seem worth the hassle. However, *different* names for the |
| 4909 |
|
same number are not permitted. */ |
| 4910 |
|
|
| 4911 |
else |
else |
| 4912 |
{ |
{ |
| 4913 |
|
BOOL dupname = FALSE; |
| 4914 |
slot = cd->name_table; |
slot = cd->name_table; |
| 4915 |
|
|
| 4916 |
for (i = 0; i < cd->names_found; i++) |
for (i = 0; i < cd->names_found; i++) |
| 4917 |
{ |
{ |
| 4918 |
int crc = memcmp(name, slot+2, namelen); |
int crc = memcmp(name, slot+2, namelen); |
| 4920 |
{ |
{ |
| 4921 |
if (slot[2+namelen] == 0) |
if (slot[2+namelen] == 0) |
| 4922 |
{ |
{ |
| 4923 |
if ((options & PCRE_DUPNAMES) == 0) |
if (GET2(slot, 0) != cd->bracount + 1 && |
| 4924 |
|
(options & PCRE_DUPNAMES) == 0) |
| 4925 |
{ |
{ |
| 4926 |
*errorcodeptr = ERR43; |
*errorcodeptr = ERR43; |
| 4927 |
goto FAILED; |
goto FAILED; |
| 4928 |
} |
} |
| 4929 |
|
else dupname = TRUE; |
| 4930 |
} |
} |
| 4931 |
else crc = -1; /* Current name is substring */ |
else crc = -1; /* Current name is a substring */ |
| 4932 |
} |
} |
| 4933 |
|
|
| 4934 |
|
/* Make space in the table and break the loop for an earlier |
| 4935 |
|
name. For a duplicate or later name, carry on. We do this for |
| 4936 |
|
duplicates so that in the simple case (when ?(| is not used) they |
| 4937 |
|
are in order of their numbers. */ |
| 4938 |
|
|
| 4939 |
if (crc < 0) |
if (crc < 0) |
| 4940 |
{ |
{ |
| 4941 |
memmove(slot + cd->name_entry_size, slot, |
memmove(slot + cd->name_entry_size, slot, |
| 4942 |
(cd->names_found - i) * cd->name_entry_size); |
(cd->names_found - i) * cd->name_entry_size); |
| 4943 |
break; |
break; |
| 4944 |
} |
} |
| 4945 |
|
|
| 4946 |
|
/* Continue the loop for a later or duplicate name */ |
| 4947 |
|
|
| 4948 |
slot += cd->name_entry_size; |
slot += cd->name_entry_size; |
| 4949 |
} |
} |
| 4950 |
|
|
| 4951 |
|
/* For non-duplicate names, check for a duplicate number before |
| 4952 |
|
adding the new name. */ |
| 4953 |
|
|
| 4954 |
|
if (!dupname) |
| 4955 |
|
{ |
| 4956 |
|
uschar *cslot = cd->name_table; |
| 4957 |
|
for (i = 0; i < cd->names_found; i++) |
| 4958 |
|
{ |
| 4959 |
|
if (cslot != slot) |
| 4960 |
|
{ |
| 4961 |
|
if (GET2(cslot, 0) == cd->bracount + 1) |
| 4962 |
|
{ |
| 4963 |
|
*errorcodeptr = ERR65; |
| 4964 |
|
goto FAILED; |
| 4965 |
|
} |
| 4966 |
|
} |
| 4967 |
|
else i--; |
| 4968 |
|
cslot += cd->name_entry_size; |
| 4969 |
|
} |
| 4970 |
|
} |
| 4971 |
|
|
| 4972 |
PUT2(slot, 0, cd->bracount + 1); |
PUT2(slot, 0, cd->bracount + 1); |
| 4973 |
memcpy(slot + 2, name, namelen); |
memcpy(slot + 2, name, namelen); |
| 4974 |
slot[2+namelen] = 0; |
slot[2+namelen] = 0; |
| 4975 |
} |
} |
| 4976 |
} |
} |
| 4977 |
|
|
| 4978 |
/* In both cases, count the number of names we've encountered. */ |
/* In both pre-compile and compile, count the number of names we've |
| 4979 |
|
encountered. */ |
| 4980 |
|
|
|
ptr++; /* Move past > or ' */ |
|
| 4981 |
cd->names_found++; |
cd->names_found++; |
| 4982 |
|
ptr++; /* Move past > or ' */ |
| 4983 |
goto NUMBERED_GROUP; |
goto NUMBERED_GROUP; |
| 4984 |
|
|
| 4985 |
|
|
| 5148 |
if (lengthptr == NULL) |
if (lengthptr == NULL) |
| 5149 |
{ |
{ |
| 5150 |
*code = OP_END; |
*code = OP_END; |
| 5151 |
if (recno != 0) called = find_bracket(cd->start_code, utf8, recno); |
if (recno != 0) |
| 5152 |
|
called = _pcre_find_bracket(cd->start_code, utf8, recno); |
| 5153 |
|
|
| 5154 |
/* Forward reference */ |
/* Forward reference */ |
| 5155 |
|
|
| 5161 |
*errorcodeptr = ERR15; |
*errorcodeptr = ERR15; |
| 5162 |
goto FAILED; |
goto FAILED; |
| 5163 |
} |
} |
| 5164 |
|
|
| 5165 |
|
/* Fudge the value of "called" so that when it is inserted as an |
| 5166 |
|
offset below, what it actually inserted is the reference number |
| 5167 |
|
of the group. */ |
| 5168 |
|
|
| 5169 |
called = cd->start_code + recno; |
called = cd->start_code + recno; |
| 5170 |
PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code); |
PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code); |
| 5171 |
} |
} |
| 5270 |
{ |
{ |
| 5271 |
cd->external_options = newoptions; |
cd->external_options = newoptions; |
| 5272 |
} |
} |
| 5273 |
else |
else |
| 5274 |
{ |
{ |
| 5275 |
if ((options & PCRE_IMS) != (newoptions & PCRE_IMS)) |
if ((options & PCRE_IMS) != (newoptions & PCRE_IMS)) |
| 5276 |
{ |
{ |
| 5607 |
|
|
| 5608 |
if (-c >= ESC_REF) |
if (-c >= ESC_REF) |
| 5609 |
{ |
{ |
| 5610 |
|
open_capitem *oc; |
| 5611 |
recno = -c - ESC_REF; |
recno = -c - ESC_REF; |
| 5612 |
|
|
| 5613 |
HANDLE_REFERENCE: /* Come here from named backref handling */ |
HANDLE_REFERENCE: /* Come here from named backref handling */ |
| 5617 |
PUT2INC(code, 0, recno); |
PUT2INC(code, 0, recno); |
| 5618 |
cd->backref_map |= (recno < 32)? (1 << recno) : 1; |
cd->backref_map |= (recno < 32)? (1 << recno) : 1; |
| 5619 |
if (recno > cd->top_backref) cd->top_backref = recno; |
if (recno > cd->top_backref) cd->top_backref = recno; |
| 5620 |
|
|
| 5621 |
|
/* Check to see if this back reference is recursive, that it, it |
| 5622 |
|
is inside the group that it references. A flag is set so that the |
| 5623 |
|
group can be made atomic. */ |
| 5624 |
|
|
| 5625 |
|
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
| 5626 |
|
{ |
| 5627 |
|
if (oc->number == recno) |
| 5628 |
|
{ |
| 5629 |
|
oc->flag = TRUE; |
| 5630 |
|
break; |
| 5631 |
|
} |
| 5632 |
|
} |
| 5633 |
} |
} |
| 5634 |
|
|
| 5635 |
/* So are Unicode property matches, if supported. */ |
/* So are Unicode property matches, if supported. */ |
| 5812 |
uschar *last_branch = code; |
uschar *last_branch = code; |
| 5813 |
uschar *start_bracket = code; |
uschar *start_bracket = code; |
| 5814 |
uschar *reverse_count = NULL; |
uschar *reverse_count = NULL; |
| 5815 |
|
open_capitem capitem; |
| 5816 |
|
int capnumber = 0; |
| 5817 |
int firstbyte, reqbyte; |
int firstbyte, reqbyte; |
| 5818 |
int branchfirstbyte, branchreqbyte; |
int branchfirstbyte, branchreqbyte; |
| 5819 |
int length; |
int length; |
| 5820 |
int orig_bracount; |
int orig_bracount; |
| 5821 |
int max_bracount; |
int max_bracount; |
| 5822 |
|
int old_external_options = cd->external_options; |
| 5823 |
branch_chain bc; |
branch_chain bc; |
| 5824 |
|
|
| 5825 |
bc.outer = bcptr; |
bc.outer = bcptr; |
| 5826 |
bc.current = code; |
bc.current_branch = code; |
| 5827 |
|
|
| 5828 |
firstbyte = reqbyte = REQ_UNSET; |
firstbyte = reqbyte = REQ_UNSET; |
| 5829 |
|
|
| 5841 |
them global. It tests the value of length for (2 + 2*LINK_SIZE) in the |
them global. It tests the value of length for (2 + 2*LINK_SIZE) in the |
| 5842 |
pre-compile phase to find out whether anything has yet been compiled or not. */ |
pre-compile phase to find out whether anything has yet been compiled or not. */ |
| 5843 |
|
|
| 5844 |
|
/* If this is a capturing subpattern, add to the chain of open capturing items |
| 5845 |
|
so that we can detect them if (*ACCEPT) is encountered. This is also used to |
| 5846 |
|
detect groups that contain recursive back references to themselves. */ |
| 5847 |
|
|
| 5848 |
|
if (*code == OP_CBRA) |
| 5849 |
|
{ |
| 5850 |
|
capnumber = GET2(code, 1 + LINK_SIZE); |
| 5851 |
|
capitem.number = capnumber; |
| 5852 |
|
capitem.next = cd->open_caps; |
| 5853 |
|
capitem.flag = FALSE; |
| 5854 |
|
cd->open_caps = &capitem; |
| 5855 |
|
} |
| 5856 |
|
|
| 5857 |
/* Offset is set zero to mark that this bracket is still open */ |
/* Offset is set zero to mark that this bracket is still open */ |
| 5858 |
|
|
| 5859 |
PUT(code, 1, 0); |
PUT(code, 1, 0); |
| 5898 |
return FALSE; |
return FALSE; |
| 5899 |
} |
} |
| 5900 |
|
|
| 5901 |
|
/* If the external options have changed during this branch, it means that we |
| 5902 |
|
are at the top level, and a leading option setting has been encountered. We |
| 5903 |
|
need to re-set the original option values to take account of this so that, |
| 5904 |
|
during the pre-compile phase, we know to allow for a re-set at the start of |
| 5905 |
|
subsequent branches. */ |
| 5906 |
|
|
| 5907 |
|
if (old_external_options != cd->external_options) |
| 5908 |
|
oldims = cd->external_options & PCRE_IMS; |
| 5909 |
|
|
| 5910 |
/* Keep the highest bracket count in case (?| was used and some branch |
/* Keep the highest bracket count in case (?| was used and some branch |
| 5911 |
has fewer than the rest. */ |
has fewer than the rest. */ |
| 5912 |
|
|
| 5957 |
|
|
| 5958 |
/* If lookbehind, check that this branch matches a fixed-length string, and |
/* If lookbehind, check that this branch matches a fixed-length string, and |
| 5959 |
put the length into the OP_REVERSE item. Temporarily mark the end of the |
put the length into the OP_REVERSE item. Temporarily mark the end of the |
| 5960 |
branch with OP_END. */ |
branch with OP_END. If the branch contains OP_RECURSE, the result is -3 |
| 5961 |
|
because there may be forward references that we can't check here. Set a |
| 5962 |
|
flag to cause another lookbehind check at the end. Why not do it all at the |
| 5963 |
|
end? Because common, erroneous checks are picked up here and the offset of |
| 5964 |
|
the problem can be shown. */ |
| 5965 |
|
|
| 5966 |
if (lookbehind) |
if (lookbehind) |
| 5967 |
{ |
{ |
| 5968 |
int fixed_length; |
int fixed_length; |
| 5969 |
*code = OP_END; |
*code = OP_END; |
| 5970 |
fixed_length = find_fixedlength(last_branch, options); |
fixed_length = find_fixedlength(last_branch, options, FALSE, cd); |
| 5971 |
DPRINTF(("fixed length = %d\n", fixed_length)); |
DPRINTF(("fixed length = %d\n", fixed_length)); |
| 5972 |
if (fixed_length < 0) |
if (fixed_length == -3) |
| 5973 |
|
{ |
| 5974 |
|
cd->check_lookbehind = TRUE; |
| 5975 |
|
} |
| 5976 |
|
else if (fixed_length < 0) |
| 5977 |
{ |
{ |
| 5978 |
*errorcodeptr = (fixed_length == -2)? ERR36 : ERR25; |
*errorcodeptr = (fixed_length == -2)? ERR36 : ERR25; |
| 5979 |
*ptrptr = ptr; |
*ptrptr = ptr; |
| 5980 |
return FALSE; |
return FALSE; |
| 5981 |
} |
} |
| 5982 |
PUT(reverse_count, 0, fixed_length); |
else { PUT(reverse_count, 0, fixed_length); } |
| 5983 |
} |
} |
| 5984 |
} |
} |
| 5985 |
|
|
| 6013 |
PUT(code, 1, code - start_bracket); |
PUT(code, 1, code - start_bracket); |
| 6014 |
code += 1 + LINK_SIZE; |
code += 1 + LINK_SIZE; |
| 6015 |
|
|
| 6016 |
/* Resetting option if needed */ |
/* If it was a capturing subpattern, check to see if it contained any |
| 6017 |
|
recursive back references. If so, we must wrap it in atomic brackets. |
| 6018 |
|
In any event, remove the block from the chain. */ |
| 6019 |
|
|
| 6020 |
|
if (capnumber > 0) |
| 6021 |
|
{ |
| 6022 |
|
if (cd->open_caps->flag) |
| 6023 |
|
{ |
| 6024 |
|
memmove(start_bracket + 1 + LINK_SIZE, start_bracket, |
| 6025 |
|
code - start_bracket); |
| 6026 |
|
*start_bracket = OP_ONCE; |
| 6027 |
|
code += 1 + LINK_SIZE; |
| 6028 |
|
PUT(start_bracket, 1, code - start_bracket); |
| 6029 |
|
*code = OP_KET; |
| 6030 |
|
PUT(code, 1, code - start_bracket); |
| 6031 |
|
code += 1 + LINK_SIZE; |
| 6032 |
|
length += 2 + 2*LINK_SIZE; |
| 6033 |
|
} |
| 6034 |
|
cd->open_caps = cd->open_caps->next; |
| 6035 |
|
} |
| 6036 |
|
|
| 6037 |
|
/* Reset options if needed. */ |
| 6038 |
|
|
| 6039 |
if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS) |
if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS) |
| 6040 |
{ |
{ |
| 6083 |
{ |
{ |
| 6084 |
*code = OP_ALT; |
*code = OP_ALT; |
| 6085 |
PUT(code, 1, code - last_branch); |
PUT(code, 1, code - last_branch); |
| 6086 |
bc.current = last_branch = code; |
bc.current_branch = last_branch = code; |
| 6087 |
code += 1 + LINK_SIZE; |
code += 1 + LINK_SIZE; |
| 6088 |
} |
} |
| 6089 |
|
|
| 6230 |
switch (*scode) |
switch (*scode) |
| 6231 |
{ |
{ |
| 6232 |
case OP_CREF: |
case OP_CREF: |
| 6233 |
|
case OP_NCREF: |
| 6234 |
case OP_RREF: |
case OP_RREF: |
| 6235 |
|
case OP_NRREF: |
| 6236 |
case OP_DEF: |
case OP_DEF: |
| 6237 |
return FALSE; |
return FALSE; |
| 6238 |
|
|
| 6401 |
int firstbyte, reqbyte, newline; |
int firstbyte, reqbyte, newline; |
| 6402 |
int errorcode = 0; |
int errorcode = 0; |
| 6403 |
int skipatstart = 0; |
int skipatstart = 0; |
| 6404 |
#ifdef SUPPORT_UTF8 |
BOOL utf8 = (options & PCRE_UTF8) != 0; |
|
BOOL utf8; |
|
|
#endif |
|
| 6405 |
size_t size; |
size_t size; |
| 6406 |
uschar *code; |
uschar *code; |
| 6407 |
const uschar *codestart; |
const uschar *codestart; |
| 6498 |
/* Can't support UTF8 unless PCRE has been compiled to include the code. */ |
/* Can't support UTF8 unless PCRE has been compiled to include the code. */ |
| 6499 |
|
|
| 6500 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
|
utf8 = (options & PCRE_UTF8) != 0; |
|
| 6501 |
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 && |
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 && |
| 6502 |
(*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0) |
(*erroroffset = _pcre_valid_utf8((USPTR)pattern, -1)) >= 0) |
| 6503 |
{ |
{ |
| 6504 |
errorcode = ERR44; |
errorcode = ERR44; |
| 6505 |
goto PCRE_EARLY_ERROR_RETURN2; |
goto PCRE_EARLY_ERROR_RETURN2; |
| 6506 |
} |
} |
| 6507 |
#else |
#else |
| 6508 |
if ((options & PCRE_UTF8) != 0) |
if (utf8) |
| 6509 |
{ |
{ |
| 6510 |
errorcode = ERR32; |
errorcode = ERR32; |
| 6511 |
goto PCRE_EARLY_ERROR_RETURN; |
goto PCRE_EARLY_ERROR_RETURN; |
| 6594 |
cd->req_varyopt = 0; |
cd->req_varyopt = 0; |
| 6595 |
cd->external_options = options; |
cd->external_options = options; |
| 6596 |
cd->external_flags = 0; |
cd->external_flags = 0; |
| 6597 |
|
cd->open_caps = NULL; |
| 6598 |
|
|
| 6599 |
/* Now do the pre-compile. On error, errorcode will be set non-zero, so we |
/* Now do the pre-compile. On error, errorcode will be set non-zero, so we |
| 6600 |
don't need to look at the result of the function here. The initial options have |
don't need to look at the result of the function here. The initial options have |
| 6669 |
cd->hwm = cworkspace; |
cd->hwm = cworkspace; |
| 6670 |
cd->req_varyopt = 0; |
cd->req_varyopt = 0; |
| 6671 |
cd->had_accept = FALSE; |
cd->had_accept = FALSE; |
| 6672 |
|
cd->check_lookbehind = FALSE; |
| 6673 |
|
cd->open_caps = NULL; |
| 6674 |
|
|
| 6675 |
/* Set up a starting, non-extracting bracket, then compile the expression. On |
/* Set up a starting, non-extracting bracket, then compile the expression. On |
| 6676 |
error, errorcode will be set non-zero, so we don't need to look at the result |
error, errorcode will be set non-zero, so we don't need to look at the result |
| 6696 |
|
|
| 6697 |
*code++ = OP_END; |
*code++ = OP_END; |
| 6698 |
|
|
| 6699 |
#ifndef DEBUG |
#ifndef PCRE_DEBUG |
| 6700 |
if (code - codestart > length) errorcode = ERR23; |
if (code - codestart > length) errorcode = ERR23; |
| 6701 |
#endif |
#endif |
| 6702 |
|
|
| 6709 |
cd->hwm -= LINK_SIZE; |
cd->hwm -= LINK_SIZE; |
| 6710 |
offset = GET(cd->hwm, 0); |
offset = GET(cd->hwm, 0); |
| 6711 |
recno = GET(codestart, offset); |
recno = GET(codestart, offset); |
| 6712 |
groupptr = find_bracket(codestart, (re->options & PCRE_UTF8) != 0, recno); |
groupptr = _pcre_find_bracket(codestart, utf8, recno); |
| 6713 |
if (groupptr == NULL) errorcode = ERR53; |
if (groupptr == NULL) errorcode = ERR53; |
| 6714 |
else PUT(((uschar *)codestart), offset, groupptr - codestart); |
else PUT(((uschar *)codestart), offset, groupptr - codestart); |
| 6715 |
} |
} |
| 6719 |
|
|
| 6720 |
if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15; |
if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15; |
| 6721 |
|
|
| 6722 |
|
/* If there were any lookbehind assertions that contained OP_RECURSE |
| 6723 |
|
(recursions or subroutine calls), a flag is set for them to be checked here, |
| 6724 |
|
because they may contain forward references. Actual recursions can't be fixed |
| 6725 |
|
length, but subroutine calls can. It is done like this so that those without |
| 6726 |
|
OP_RECURSE that are not fixed length get a diagnosic with a useful offset. The |
| 6727 |
|
exceptional ones forgo this. We scan the pattern to check that they are fixed |
| 6728 |
|
length, and set their lengths. */ |
| 6729 |
|
|
| 6730 |
|
if (cd->check_lookbehind) |
| 6731 |
|
{ |
| 6732 |
|
uschar *cc = (uschar *)codestart; |
| 6733 |
|
|
| 6734 |
|
/* Loop, searching for OP_REVERSE items, and process those that do not have |
| 6735 |
|
their length set. (Actually, it will also re-process any that have a length |
| 6736 |
|
of zero, but that is a pathological case, and it does no harm.) When we find |
| 6737 |
|
one, we temporarily terminate the branch it is in while we scan it. */ |
| 6738 |
|
|
| 6739 |
|
for (cc = (uschar *)_pcre_find_bracket(codestart, utf8, -1); |
| 6740 |
|
cc != NULL; |
| 6741 |
|
cc = (uschar *)_pcre_find_bracket(cc, utf8, -1)) |
| 6742 |
|
{ |
| 6743 |
|
if (GET(cc, 1) == 0) |
| 6744 |
|
{ |
| 6745 |
|
int fixed_length; |
| 6746 |
|
uschar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); |
| 6747 |
|
int end_op = *be; |
| 6748 |
|
*be = OP_END; |
| 6749 |
|
fixed_length = find_fixedlength(cc, re->options, TRUE, cd); |
| 6750 |
|
*be = end_op; |
| 6751 |
|
DPRINTF(("fixed length = %d\n", fixed_length)); |
| 6752 |
|
if (fixed_length < 0) |
| 6753 |
|
{ |
| 6754 |
|
errorcode = (fixed_length == -2)? ERR36 : ERR25; |
| 6755 |
|
break; |
| 6756 |
|
} |
| 6757 |
|
PUT(cc, 1, fixed_length); |
| 6758 |
|
} |
| 6759 |
|
cc += 1 + LINK_SIZE; |
| 6760 |
|
} |
| 6761 |
|
} |
| 6762 |
|
|
| 6763 |
/* Failed to compile, or error while post-processing */ |
/* Failed to compile, or error while post-processing */ |
| 6764 |
|
|
| 6765 |
if (errorcode != 0) |
if (errorcode != 0) |
| 6820 |
/* Print out the compiled data if debugging is enabled. This is never the |
/* Print out the compiled data if debugging is enabled. This is never the |
| 6821 |
case when building a production library. */ |
case when building a production library. */ |
| 6822 |
|
|
| 6823 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
|
|
|
| 6824 |
printf("Length = %d top_bracket = %d top_backref = %d\n", |
printf("Length = %d top_bracket = %d top_backref = %d\n", |
| 6825 |
length, re->top_bracket, re->top_backref); |
length, re->top_bracket, re->top_backref); |
| 6826 |
|
|
| 6857 |
if (errorcodeptr != NULL) *errorcodeptr = ERR23; |
if (errorcodeptr != NULL) *errorcodeptr = ERR23; |
| 6858 |
return NULL; |
return NULL; |
| 6859 |
} |
} |
| 6860 |
#endif /* DEBUG */ |
#endif /* PCRE_DEBUG */ |
| 6861 |
|
|
| 6862 |
return (pcre *)re; |
return (pcre *)re; |
| 6863 |
} |
} |