| 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 |
|
|
| 92 |
|
|
| 93 |
#define COMPILE_WORK_SIZE (4096) |
#define COMPILE_WORK_SIZE (4096) |
| 94 |
|
|
| 95 |
|
/* The overrun tests check for a slightly smaller size so that they detect the |
| 96 |
|
overrun before it actually does run off the end of the data block. */ |
| 97 |
|
|
| 98 |
|
#define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100) |
| 99 |
|
|
| 100 |
|
|
| 101 |
/* Table for handling escaped characters in the range '0'-'z'. Positive returns |
/* Table for handling escaped characters in the range '0'-'z'. Positive returns |
| 102 |
are simple data values; negative values are for special things like \d and so |
are simple data values; negative values are for special things like \d and so |
| 268 |
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 |
| 269 |
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 |
| 270 |
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 |
| 271 |
because these strings are used only when there is a compilation error. */ |
because these strings are used only when there is a compilation error. |
| 272 |
|
|
| 273 |
|
Each substring ends with \0 to insert a null character. This includes the final |
| 274 |
|
substring, so that the whole string ends with \0\0, which can be detected when |
| 275 |
|
counting through. */ |
| 276 |
|
|
| 277 |
static const char error_texts[] = |
static const char error_texts[] = |
| 278 |
"no error\0" |
"no error\0" |
| 353 |
"digit expected after (?+\0" |
"digit expected after (?+\0" |
| 354 |
"] is an invalid data character in JavaScript compatibility mode\0" |
"] is an invalid data character in JavaScript compatibility mode\0" |
| 355 |
/* 65 */ |
/* 65 */ |
| 356 |
"different names for subpatterns of the same number are not allowed"; |
"different names for subpatterns of the same number are not allowed\0"; |
|
|
|
| 357 |
|
|
| 358 |
/* Table to identify digits and hex digits. This is used when compiling |
/* Table to identify digits and hex digits. This is used when compiling |
| 359 |
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 |
| 511 |
find_error_text(int n) |
find_error_text(int n) |
| 512 |
{ |
{ |
| 513 |
const char *s = error_texts; |
const char *s = error_texts; |
| 514 |
for (; n > 0; n--) while (*s++ != 0) {}; |
for (; n > 0; n--) |
| 515 |
|
{ |
| 516 |
|
while (*s++ != 0) {}; |
| 517 |
|
if (*s == 0) return "Error text not found (please report)"; |
| 518 |
|
} |
| 519 |
return s; |
return s; |
| 520 |
} |
} |
| 521 |
|
|
| 1115 |
if (name != NULL && lorn == ptr - thisname && |
if (name != NULL && lorn == ptr - thisname && |
| 1116 |
strncmp((const char *)name, (const char *)thisname, lorn) == 0) |
strncmp((const char *)name, (const char *)thisname, lorn) == 0) |
| 1117 |
return *count; |
return *count; |
| 1118 |
term++; |
term++; |
| 1119 |
} |
} |
| 1120 |
} |
} |
| 1121 |
} |
} |
| 1161 |
break; |
break; |
| 1162 |
} |
} |
| 1163 |
else if (!negate_class && ptr[1] == CHAR_CIRCUMFLEX_ACCENT) |
else if (!negate_class && ptr[1] == CHAR_CIRCUMFLEX_ACCENT) |
| 1164 |
{ |
{ |
| 1165 |
negate_class = TRUE; |
negate_class = TRUE; |
| 1166 |
ptr++; |
ptr++; |
| 1167 |
} |
} |
| 1168 |
else break; |
else break; |
| 1169 |
} |
} |
| 1170 |
|
|
| 1353 |
|
|
| 1354 |
/* Scan a branch 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, |
| 1355 |
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. |
| 1356 |
In UTF8 mode, the result is in characters rather than bytes. The branch is |
In UTF8 mode, the result is in characters rather than bytes. The branch is |
| 1357 |
temporarily terminated with OP_END when this function is called. |
temporarily terminated with OP_END when this function is called. |
| 1358 |
|
|
| 1359 |
This function is called when a backward assertion is encountered, so that if it |
This function is called when a backward assertion is encountered, so that if it |
| 1360 |
fails, the error message can point to the correct place in the pattern. |
fails, the error message can point to the correct place in the pattern. |
| 1361 |
However, we cannot do this when the assertion contains subroutine calls, |
However, we cannot do this when the assertion contains subroutine calls, |
| 1362 |
because they can be forward references. We solve this by remembering this case |
because they can be forward references. We solve this by remembering this case |
| 1363 |
and doing the check at the end; a flag specifies which mode we are running in. |
and doing the check at the end; a flag specifies which mode we are running in. |
| 1364 |
|
|
| 1365 |
Arguments: |
Arguments: |
| 1366 |
code points to the start of the pattern (the bracket) |
code points to the start of the pattern (the bracket) |
| 1367 |
options the compiling options |
options the compiling options |
| 1368 |
atend TRUE if called when the pattern is complete |
atend TRUE if called when the pattern is complete |
| 1369 |
cd the "compile data" structure |
cd the "compile data" structure |
| 1370 |
|
|
| 1371 |
Returns: the fixed length, |
Returns: the fixed length, |
| 1372 |
or -1 if there is no fixed length, |
or -1 if there is no fixed length, |
| 1373 |
or -2 if \C was encountered |
or -2 if \C was encountered |
| 1374 |
or -3 if an OP_RECURSE item was encountered and atend is FALSE |
or -3 if an OP_RECURSE item was encountered and atend is FALSE |
| 1418 |
cc += 1 + LINK_SIZE; |
cc += 1 + LINK_SIZE; |
| 1419 |
branchlength = 0; |
branchlength = 0; |
| 1420 |
break; |
break; |
| 1421 |
|
|
| 1422 |
/* A true recursion implies not fixed length, but a subroutine call may |
/* A true recursion implies not fixed length, but a subroutine call may |
| 1423 |
be OK. If the subroutine is a forward reference, we can't deal with |
be OK. If the subroutine is a forward reference, we can't deal with |
| 1424 |
it until the end of the pattern, so return -3. */ |
it until the end of the pattern, so return -3. */ |
| 1425 |
|
|
| 1426 |
case OP_RECURSE: |
case OP_RECURSE: |
| 1427 |
if (!atend) return -3; |
if (!atend) return -3; |
| 1428 |
cs = ce = (uschar *)cd->start_code + GET(cc, 1); /* Start subpattern */ |
cs = ce = (uschar *)cd->start_code + GET(cc, 1); /* Start subpattern */ |
| 1429 |
do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ |
do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ |
| 1430 |
if (cc > cs && cc < ce) return -1; /* Recursion */ |
if (cc > cs && cc < ce) return -1; /* Recursion */ |
| 1431 |
d = find_fixedlength(cs + 2, options, atend, cd); |
d = find_fixedlength(cs + 2, options, atend, cd); |
| 1432 |
if (d < 0) return d; |
if (d < 0) return d; |
| 1433 |
branchlength += d; |
branchlength += d; |
| 1434 |
cc += 1 + LINK_SIZE; |
cc += 1 + LINK_SIZE; |
| 1435 |
break; |
break; |
| 1436 |
|
|
| 1437 |
/* Skip over assertive subpatterns */ |
/* Skip over assertive subpatterns */ |
| 1438 |
|
|
| 1455 |
case OP_CALLOUT: |
case OP_CALLOUT: |
| 1456 |
case OP_SOD: |
case OP_SOD: |
| 1457 |
case OP_SOM: |
case OP_SOM: |
| 1458 |
|
case OP_SET_SOM: |
| 1459 |
case OP_EOD: |
case OP_EOD: |
| 1460 |
case OP_EODN: |
case OP_EODN: |
| 1461 |
case OP_CIRC: |
case OP_CIRC: |
| 1473 |
branchlength++; |
branchlength++; |
| 1474 |
cc += 2; |
cc += 2; |
| 1475 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 1476 |
if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0) |
if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0) |
| 1477 |
cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
| 1478 |
#endif |
#endif |
| 1479 |
break; |
break; |
| 1485 |
branchlength += GET2(cc,1); |
branchlength += GET2(cc,1); |
| 1486 |
cc += 4; |
cc += 4; |
| 1487 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 1488 |
if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0) |
if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0) |
| 1489 |
cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
| 1490 |
#endif |
#endif |
| 1491 |
break; |
break; |
| 1570 |
|
|
| 1571 |
/* This little function scans through a compiled pattern until it finds a |
/* This little function scans through a compiled pattern until it finds a |
| 1572 |
capturing bracket with the given number, or, if the number is negative, an |
capturing bracket with the given number, or, if the number is negative, an |
| 1573 |
instance of OP_REVERSE for a lookbehind. The function is global in the C sense |
instance of OP_REVERSE for a lookbehind. The function is global in the C sense |
| 1574 |
so that it can be called from pcre_study() when finding the minimum matching |
so that it can be called from pcre_study() when finding the minimum matching |
| 1575 |
length. |
length. |
| 1576 |
|
|
| 1577 |
Arguments: |
Arguments: |
| 1595 |
the table is zero; the actual length is stored in the compiled code. */ |
the table is zero; the actual length is stored in the compiled code. */ |
| 1596 |
|
|
| 1597 |
if (c == OP_XCLASS) code += GET(code, 1); |
if (c == OP_XCLASS) code += GET(code, 1); |
| 1598 |
|
|
| 1599 |
/* Handle recursion */ |
/* Handle recursion */ |
| 1600 |
|
|
| 1601 |
else if (c == OP_REVERSE) |
else if (c == OP_REVERSE) |
| 1602 |
{ |
{ |
| 1603 |
if (number < 0) return (uschar *)code; |
if (number < 0) return (uschar *)code; |
| 1604 |
code += _pcre_OP_lengths[c]; |
code += _pcre_OP_lengths[c]; |
| 1605 |
} |
} |
| 1606 |
|
|
| 1790 |
code points to start of search |
code points to start of search |
| 1791 |
endcode points to where to stop |
endcode points to where to stop |
| 1792 |
utf8 TRUE if in UTF8 mode |
utf8 TRUE if in UTF8 mode |
| 1793 |
|
cd contains pointers to tables etc. |
| 1794 |
|
|
| 1795 |
Returns: TRUE if what is matched could be empty |
Returns: TRUE if what is matched could be empty |
| 1796 |
*/ |
*/ |
| 1797 |
|
|
| 1798 |
static BOOL |
static BOOL |
| 1799 |
could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8) |
could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8, |
| 1800 |
|
compile_data *cd) |
| 1801 |
{ |
{ |
| 1802 |
register int c; |
register int c; |
| 1803 |
for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE); |
for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE); |
| 1807 |
const uschar *ccode; |
const uschar *ccode; |
| 1808 |
|
|
| 1809 |
c = *code; |
c = *code; |
| 1810 |
|
|
| 1811 |
/* Skip over forward assertions; the other assertions are skipped by |
/* Skip over forward assertions; the other assertions are skipped by |
| 1812 |
first_significant_code() with a TRUE final argument. */ |
first_significant_code() with a TRUE final argument. */ |
| 1813 |
|
|
| 1827 |
c = *code; |
c = *code; |
| 1828 |
continue; |
continue; |
| 1829 |
} |
} |
| 1830 |
|
|
| 1831 |
|
/* For a recursion/subroutine call, if its end has been reached, which |
| 1832 |
|
implies a subroutine call, we can scan it. */ |
| 1833 |
|
|
| 1834 |
|
if (c == OP_RECURSE) |
| 1835 |
|
{ |
| 1836 |
|
BOOL empty_branch = FALSE; |
| 1837 |
|
const uschar *scode = cd->start_code + GET(code, 1); |
| 1838 |
|
if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ |
| 1839 |
|
do |
| 1840 |
|
{ |
| 1841 |
|
if (could_be_empty_branch(scode, endcode, utf8, cd)) |
| 1842 |
|
{ |
| 1843 |
|
empty_branch = TRUE; |
| 1844 |
|
break; |
| 1845 |
|
} |
| 1846 |
|
scode += GET(scode, 1); |
| 1847 |
|
} |
| 1848 |
|
while (*scode == OP_ALT); |
| 1849 |
|
if (!empty_branch) return FALSE; /* All branches are non-empty */ |
| 1850 |
|
continue; |
| 1851 |
|
} |
| 1852 |
|
|
| 1853 |
/* For other groups, scan the branches. */ |
/* For other groups, scan the branches. */ |
| 1854 |
|
|
| 1868 |
empty_branch = FALSE; |
empty_branch = FALSE; |
| 1869 |
do |
do |
| 1870 |
{ |
{ |
| 1871 |
if (!empty_branch && could_be_empty_branch(code, endcode, utf8)) |
if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd)) |
| 1872 |
empty_branch = TRUE; |
empty_branch = TRUE; |
| 1873 |
code += GET(code, 1); |
code += GET(code, 1); |
| 1874 |
} |
} |
| 1995 |
case OP_POSQUERY: |
case OP_POSQUERY: |
| 1996 |
if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f]; |
if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f]; |
| 1997 |
break; |
break; |
| 1998 |
|
|
| 1999 |
case OP_UPTO: |
case OP_UPTO: |
| 2000 |
case OP_MINUPTO: |
case OP_MINUPTO: |
| 2001 |
case OP_POSUPTO: |
case OP_POSUPTO: |
| 2002 |
if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f]; |
if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f]; |
| 2003 |
break; |
break; |
| 2004 |
#endif |
#endif |
| 2005 |
|
|
| 2006 |
|
/* None of the remaining opcodes are required to match a character. */ |
| 2007 |
|
|
| 2008 |
|
default: |
| 2009 |
|
break; |
| 2010 |
} |
} |
| 2011 |
} |
} |
| 2012 |
|
|
| 2029 |
endcode points to where to stop (current RECURSE item) |
endcode points to where to stop (current RECURSE item) |
| 2030 |
bcptr points to the chain of current (unclosed) branch starts |
bcptr points to the chain of current (unclosed) branch starts |
| 2031 |
utf8 TRUE if in UTF-8 mode |
utf8 TRUE if in UTF-8 mode |
| 2032 |
|
cd pointers to tables etc |
| 2033 |
|
|
| 2034 |
Returns: TRUE if what is matched could be empty |
Returns: TRUE if what is matched could be empty |
| 2035 |
*/ |
*/ |
| 2036 |
|
|
| 2037 |
static BOOL |
static BOOL |
| 2038 |
could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr, |
could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr, |
| 2039 |
BOOL utf8) |
BOOL utf8, compile_data *cd) |
| 2040 |
{ |
{ |
| 2041 |
while (bcptr != NULL && bcptr->current >= code) |
while (bcptr != NULL && bcptr->current_branch >= code) |
| 2042 |
{ |
{ |
| 2043 |
if (!could_be_empty_branch(bcptr->current, endcode, utf8)) return FALSE; |
if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd)) |
| 2044 |
|
return FALSE; |
| 2045 |
bcptr = bcptr->outer; |
bcptr = bcptr->outer; |
| 2046 |
} |
} |
| 2047 |
return TRUE; |
return TRUE; |
| 2703 |
uschar *utf8_char = NULL; |
uschar *utf8_char = NULL; |
| 2704 |
#endif |
#endif |
| 2705 |
|
|
| 2706 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
| 2707 |
if (lengthptr != NULL) DPRINTF((">> start branch\n")); |
if (lengthptr != NULL) DPRINTF((">> start branch\n")); |
| 2708 |
#endif |
#endif |
| 2709 |
|
|
| 2762 |
|
|
| 2763 |
if (lengthptr != NULL) |
if (lengthptr != NULL) |
| 2764 |
{ |
{ |
| 2765 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
| 2766 |
if (code > cd->hwm) cd->hwm = code; /* High water info */ |
if (code > cd->hwm) cd->hwm = code; /* High water info */ |
| 2767 |
#endif |
#endif |
| 2768 |
if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */ |
if (code > cd->start_workspace + WORK_SIZE_CHECK) /* Check for overrun */ |
| 2769 |
{ |
{ |
| 2770 |
*errorcodeptr = ERR52; |
*errorcodeptr = ERR52; |
| 2771 |
goto FAILED; |
goto FAILED; |
| 2814 |
/* In the real compile phase, just check the workspace used by the forward |
/* In the real compile phase, just check the workspace used by the forward |
| 2815 |
reference list. */ |
reference list. */ |
| 2816 |
|
|
| 2817 |
else if (cd->hwm > cd->start_workspace + COMPILE_WORK_SIZE) |
else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK) |
| 2818 |
{ |
{ |
| 2819 |
*errorcodeptr = ERR52; |
*errorcodeptr = ERR52; |
| 2820 |
goto FAILED; |
goto FAILED; |
| 3960 |
|
|
| 3961 |
if (repeat_max == 0) goto END_REPEAT; |
if (repeat_max == 0) goto END_REPEAT; |
| 3962 |
|
|
| 3963 |
/*--------------------------------------------------------------------*/ |
/*--------------------------------------------------------------------*/ |
| 3964 |
/* This code is obsolete from release 8.00; the restriction was finally |
/* This code is obsolete from release 8.00; the restriction was finally |
| 3965 |
removed: */ |
removed: */ |
| 3966 |
|
|
| 3967 |
/* All real repeats make it impossible to handle partial matching (maybe |
/* All real repeats make it impossible to handle partial matching (maybe |
| 3968 |
one day we will be able to remove this restriction). */ |
one day we will be able to remove this restriction). */ |
| 3969 |
|
|
| 3970 |
/* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */ |
/* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */ |
| 3971 |
/*--------------------------------------------------------------------*/ |
/*--------------------------------------------------------------------*/ |
| 3972 |
|
|
| 3973 |
/* Combine the op_type with the repeat_type */ |
/* Combine the op_type with the repeat_type */ |
| 3974 |
|
|
| 4115 |
goto END_REPEAT; |
goto END_REPEAT; |
| 4116 |
} |
} |
| 4117 |
|
|
| 4118 |
/*--------------------------------------------------------------------*/ |
/*--------------------------------------------------------------------*/ |
| 4119 |
/* This code is obsolete from release 8.00; the restriction was finally |
/* This code is obsolete from release 8.00; the restriction was finally |
| 4120 |
removed: */ |
removed: */ |
| 4121 |
|
|
| 4123 |
one day we will be able to remove this restriction). */ |
one day we will be able to remove this restriction). */ |
| 4124 |
|
|
| 4125 |
/* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */ |
/* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */ |
| 4126 |
/*--------------------------------------------------------------------*/ |
/*--------------------------------------------------------------------*/ |
| 4127 |
|
|
| 4128 |
if (repeat_min == 0 && repeat_max == -1) |
if (repeat_min == 0 && repeat_max == -1) |
| 4129 |
*code++ = OP_CRSTAR + repeat_type; |
*code++ = OP_CRSTAR + repeat_type; |
| 4258 |
{ |
{ |
| 4259 |
/* 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 |
| 4260 |
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 |
| 4261 |
potential integer overflow. */ |
potential integer overflow. The INT64_OR_DOUBLE type is a 64-bit |
| 4262 |
|
integer type when available, otherwise double. */ |
| 4263 |
|
|
| 4264 |
if (lengthptr != NULL) |
if (lengthptr != NULL) |
| 4265 |
{ |
{ |
| 4266 |
int delta = (repeat_min - 1)*length_prevgroup; |
int delta = (repeat_min - 1)*length_prevgroup; |
| 4267 |
if ((double)(repeat_min - 1)*(double)length_prevgroup > |
if ((INT64_OR_DOUBLE)(repeat_min - 1)* |
| 4268 |
(double)INT_MAX || |
(INT64_OR_DOUBLE)length_prevgroup > |
| 4269 |
|
(INT64_OR_DOUBLE)INT_MAX || |
| 4270 |
OFLOW_MAX - *lengthptr < delta) |
OFLOW_MAX - *lengthptr < delta) |
| 4271 |
{ |
{ |
| 4272 |
*errorcodeptr = ERR20; |
*errorcodeptr = ERR20; |
| 4312 |
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 |
| 4313 |
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 |
| 4314 |
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 |
| 4315 |
paranoid checks to avoid integer overflow. */ |
paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is |
| 4316 |
|
a 64-bit integer type when available, otherwise double. */ |
| 4317 |
|
|
| 4318 |
if (lengthptr != NULL && repeat_max > 0) |
if (lengthptr != NULL && repeat_max > 0) |
| 4319 |
{ |
{ |
| 4320 |
int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) - |
int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) - |
| 4321 |
2 - 2*LINK_SIZE; /* Last one doesn't nest */ |
2 - 2*LINK_SIZE; /* Last one doesn't nest */ |
| 4322 |
if ((double)repeat_max * |
if ((INT64_OR_DOUBLE)repeat_max * |
| 4323 |
(double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE) |
(INT64_OR_DOUBLE)(length_prevgroup + 1 + 2 + 2*LINK_SIZE) |
| 4324 |
> (double)INT_MAX || |
> (INT64_OR_DOUBLE)INT_MAX || |
| 4325 |
OFLOW_MAX - *lengthptr < delta) |
OFLOW_MAX - *lengthptr < delta) |
| 4326 |
{ |
{ |
| 4327 |
*errorcodeptr = ERR20; |
*errorcodeptr = ERR20; |
| 4398 |
uschar *scode = bracode; |
uschar *scode = bracode; |
| 4399 |
do |
do |
| 4400 |
{ |
{ |
| 4401 |
if (could_be_empty_branch(scode, ketcode, utf8)) |
if (could_be_empty_branch(scode, ketcode, utf8, cd)) |
| 4402 |
{ |
{ |
| 4403 |
*bracode += OP_SBRA - OP_BRA; |
*bracode += OP_SBRA - OP_BRA; |
| 4404 |
break; |
break; |
| 4441 |
if (possessive_quantifier) |
if (possessive_quantifier) |
| 4442 |
{ |
{ |
| 4443 |
int len; |
int len; |
| 4444 |
|
|
| 4445 |
if (*tempcode == OP_TYPEEXACT) |
if (*tempcode == OP_TYPEEXACT) |
| 4446 |
tempcode += _pcre_OP_lengths[*tempcode] + |
tempcode += _pcre_OP_lengths[*tempcode] + |
| 4447 |
((tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP)? 2 : 0); |
((tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP)? 2 : 0); |
| 4448 |
|
|
| 4449 |
else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT) |
else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT) |
| 4450 |
{ |
{ |
| 4451 |
tempcode += _pcre_OP_lengths[*tempcode]; |
tempcode += _pcre_OP_lengths[*tempcode]; |
| 4453 |
if (utf8 && tempcode[-1] >= 0xc0) |
if (utf8 && tempcode[-1] >= 0xc0) |
| 4454 |
tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f]; |
tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f]; |
| 4455 |
#endif |
#endif |
| 4456 |
} |
} |
| 4457 |
|
|
| 4458 |
len = code - tempcode; |
len = code - tempcode; |
| 4459 |
if (len > 0) switch (*tempcode) |
if (len > 0) switch (*tempcode) |
| 4460 |
{ |
{ |
| 4473 |
case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break; |
case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break; |
| 4474 |
case OP_NOTUPTO: *tempcode = OP_NOTPOSUPTO; break; |
case OP_NOTUPTO: *tempcode = OP_NOTPOSUPTO; break; |
| 4475 |
|
|
| 4476 |
|
/* Because we are moving code along, we must ensure that any |
| 4477 |
|
pending recursive references are updated. */ |
| 4478 |
|
|
| 4479 |
default: |
default: |
| 4480 |
|
*code = OP_END; |
| 4481 |
|
adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm); |
| 4482 |
memmove(tempcode + 1+LINK_SIZE, tempcode, len); |
memmove(tempcode + 1+LINK_SIZE, tempcode, len); |
| 4483 |
code += 1 + LINK_SIZE; |
code += 1 + LINK_SIZE; |
| 4484 |
len += 1 + LINK_SIZE; |
len += 1 + LINK_SIZE; |
| 4538 |
strncmp((char *)name, vn, namelen) == 0) |
strncmp((char *)name, vn, namelen) == 0) |
| 4539 |
{ |
{ |
| 4540 |
/* Check for open captures before ACCEPT */ |
/* Check for open captures before ACCEPT */ |
| 4541 |
|
|
| 4542 |
if (verbs[i].op == OP_ACCEPT) |
if (verbs[i].op == OP_ACCEPT) |
| 4543 |
{ |
{ |
| 4544 |
open_capitem *oc; |
open_capitem *oc; |
| 4545 |
cd->had_accept = TRUE; |
cd->had_accept = TRUE; |
| 4546 |
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
| 4547 |
{ |
{ |
| 4548 |
*code++ = OP_CLOSE; |
*code++ = OP_CLOSE; |
| 4549 |
PUT2INC(code, 0, oc->number); |
PUT2INC(code, 0, oc->number); |
| 4550 |
} |
} |
| 4551 |
} |
} |
| 4552 |
*code++ = verbs[i].op; |
*code++ = verbs[i].op; |
| 4553 |
break; |
break; |
| 4554 |
} |
} |
| 4711 |
} |
} |
| 4712 |
|
|
| 4713 |
/* Otherwise (did not start with "+" or "-"), start by looking for the |
/* Otherwise (did not start with "+" or "-"), start by looking for the |
| 4714 |
name. If we find a name, add one to the opcode to change OP_CREF or |
name. If we find a name, add one to the opcode to change OP_CREF or |
| 4715 |
OP_RREF into OP_NCREF or OP_NRREF. These behave exactly the same, |
OP_RREF into OP_NCREF or OP_NRREF. These behave exactly the same, |
| 4716 |
except they record that the reference was originally to a name. The |
except they record that the reference was originally to a name. The |
| 4717 |
information is used to check duplicate names. */ |
information is used to check duplicate names. */ |
| 4718 |
|
|
| 4719 |
slot = cd->name_table; |
slot = cd->name_table; |
| 4940 |
is because the number of names, and hence the table size, is computed |
is because the number of names, and hence the table size, is computed |
| 4941 |
in the pre-compile, and it affects various numbers and pointers which |
in the pre-compile, and it affects various numbers and pointers which |
| 4942 |
would all have to be modified, and the compiled code moved down, if |
would all have to be modified, and the compiled code moved down, if |
| 4943 |
duplicates with the same number were omitted from the table. This |
duplicates with the same number were omitted from the table. This |
| 4944 |
doesn't seem worth the hassle. However, *different* names for the |
doesn't seem worth the hassle. However, *different* names for the |
| 4945 |
same number are not permitted. */ |
same number are not permitted. */ |
| 4946 |
|
|
| 4948 |
{ |
{ |
| 4949 |
BOOL dupname = FALSE; |
BOOL dupname = FALSE; |
| 4950 |
slot = cd->name_table; |
slot = cd->name_table; |
| 4951 |
|
|
| 4952 |
for (i = 0; i < cd->names_found; i++) |
for (i = 0; i < cd->names_found; i++) |
| 4953 |
{ |
{ |
| 4954 |
int crc = memcmp(name, slot+2, namelen); |
int crc = memcmp(name, slot+2, namelen); |
| 4962 |
*errorcodeptr = ERR43; |
*errorcodeptr = ERR43; |
| 4963 |
goto FAILED; |
goto FAILED; |
| 4964 |
} |
} |
| 4965 |
else dupname = TRUE; |
else dupname = TRUE; |
| 4966 |
} |
} |
| 4967 |
else crc = -1; /* Current name is a substring */ |
else crc = -1; /* Current name is a substring */ |
| 4968 |
} |
} |
| 4969 |
|
|
| 4970 |
/* Make space in the table and break the loop for an earlier |
/* Make space in the table and break the loop for an earlier |
| 4971 |
name. For a duplicate or later name, carry on. We do this for |
name. For a duplicate or later name, carry on. We do this for |
| 4972 |
duplicates so that in the simple case (when ?(| is not used) they |
duplicates so that in the simple case (when ?(| is not used) they |
| 4973 |
are in order of their numbers. */ |
are in order of their numbers. */ |
| 4974 |
|
|
| 4975 |
if (crc < 0) |
if (crc < 0) |
| 4976 |
{ |
{ |
| 4977 |
memmove(slot + cd->name_entry_size, slot, |
memmove(slot + cd->name_entry_size, slot, |
| 4978 |
(cd->names_found - i) * cd->name_entry_size); |
(cd->names_found - i) * cd->name_entry_size); |
| 4979 |
break; |
break; |
| 4980 |
} |
} |
| 4981 |
|
|
| 4982 |
/* Continue the loop for a later or duplicate name */ |
/* Continue the loop for a later or duplicate name */ |
| 4983 |
|
|
| 4984 |
slot += cd->name_entry_size; |
slot += cd->name_entry_size; |
| 4985 |
} |
} |
| 4986 |
|
|
| 4987 |
/* For non-duplicate names, check for a duplicate number before |
/* For non-duplicate names, check for a duplicate number before |
| 4988 |
adding the new name. */ |
adding the new name. */ |
| 4989 |
|
|
| 4990 |
if (!dupname) |
if (!dupname) |
| 4991 |
{ |
{ |
| 4992 |
uschar *cslot = cd->name_table; |
uschar *cslot = cd->name_table; |
| 4998 |
{ |
{ |
| 4999 |
*errorcodeptr = ERR65; |
*errorcodeptr = ERR65; |
| 5000 |
goto FAILED; |
goto FAILED; |
| 5001 |
} |
} |
| 5002 |
} |
} |
| 5003 |
else i--; |
else i--; |
| 5004 |
cslot += cd->name_entry_size; |
cslot += cd->name_entry_size; |
| 5005 |
} |
} |
| 5006 |
} |
} |
| 5007 |
|
|
| 5008 |
PUT2(slot, 0, cd->bracount + 1); |
PUT2(slot, 0, cd->bracount + 1); |
| 5009 |
memcpy(slot + 2, name, namelen); |
memcpy(slot + 2, name, namelen); |
| 5184 |
if (lengthptr == NULL) |
if (lengthptr == NULL) |
| 5185 |
{ |
{ |
| 5186 |
*code = OP_END; |
*code = OP_END; |
| 5187 |
if (recno != 0) |
if (recno != 0) |
| 5188 |
called = _pcre_find_bracket(cd->start_code, utf8, recno); |
called = _pcre_find_bracket(cd->start_code, utf8, recno); |
| 5189 |
|
|
| 5190 |
/* Forward reference */ |
/* Forward reference */ |
| 5197 |
*errorcodeptr = ERR15; |
*errorcodeptr = ERR15; |
| 5198 |
goto FAILED; |
goto FAILED; |
| 5199 |
} |
} |
| 5200 |
|
|
| 5201 |
|
/* Fudge the value of "called" so that when it is inserted as an |
| 5202 |
|
offset below, what it actually inserted is the reference number |
| 5203 |
|
of the group. */ |
| 5204 |
|
|
| 5205 |
called = cd->start_code + recno; |
called = cd->start_code + recno; |
| 5206 |
PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code); |
PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code); |
| 5207 |
} |
} |
| 5211 |
recursion that could loop for ever, and diagnose that case. */ |
recursion that could loop for ever, and diagnose that case. */ |
| 5212 |
|
|
| 5213 |
else if (GET(called, 1) == 0 && |
else if (GET(called, 1) == 0 && |
| 5214 |
could_be_empty(called, code, bcptr, utf8)) |
could_be_empty(called, code, bcptr, utf8, cd)) |
| 5215 |
{ |
{ |
| 5216 |
*errorcodeptr = ERR40; |
*errorcodeptr = ERR40; |
| 5217 |
goto FAILED; |
goto FAILED; |
| 5306 |
{ |
{ |
| 5307 |
cd->external_options = newoptions; |
cd->external_options = newoptions; |
| 5308 |
} |
} |
| 5309 |
else |
else |
| 5310 |
{ |
{ |
| 5311 |
if ((options & PCRE_IMS) != (newoptions & PCRE_IMS)) |
if ((options & PCRE_IMS) != (newoptions & PCRE_IMS)) |
| 5312 |
{ |
{ |
| 5643 |
|
|
| 5644 |
if (-c >= ESC_REF) |
if (-c >= ESC_REF) |
| 5645 |
{ |
{ |
| 5646 |
|
open_capitem *oc; |
| 5647 |
recno = -c - ESC_REF; |
recno = -c - ESC_REF; |
| 5648 |
|
|
| 5649 |
HANDLE_REFERENCE: /* Come here from named backref handling */ |
HANDLE_REFERENCE: /* Come here from named backref handling */ |
| 5653 |
PUT2INC(code, 0, recno); |
PUT2INC(code, 0, recno); |
| 5654 |
cd->backref_map |= (recno < 32)? (1 << recno) : 1; |
cd->backref_map |= (recno < 32)? (1 << recno) : 1; |
| 5655 |
if (recno > cd->top_backref) cd->top_backref = recno; |
if (recno > cd->top_backref) cd->top_backref = recno; |
| 5656 |
|
|
| 5657 |
|
/* Check to see if this back reference is recursive, that it, it |
| 5658 |
|
is inside the group that it references. A flag is set so that the |
| 5659 |
|
group can be made atomic. */ |
| 5660 |
|
|
| 5661 |
|
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
| 5662 |
|
{ |
| 5663 |
|
if (oc->number == recno) |
| 5664 |
|
{ |
| 5665 |
|
oc->flag = TRUE; |
| 5666 |
|
break; |
| 5667 |
|
} |
| 5668 |
|
} |
| 5669 |
} |
} |
| 5670 |
|
|
| 5671 |
/* So are Unicode property matches, if supported. */ |
/* So are Unicode property matches, if supported. */ |
| 5855 |
int length; |
int length; |
| 5856 |
int orig_bracount; |
int orig_bracount; |
| 5857 |
int max_bracount; |
int max_bracount; |
| 5858 |
|
int old_external_options = cd->external_options; |
| 5859 |
branch_chain bc; |
branch_chain bc; |
| 5860 |
|
|
| 5861 |
bc.outer = bcptr; |
bc.outer = bcptr; |
| 5862 |
bc.current = code; |
bc.current_branch = code; |
| 5863 |
|
|
| 5864 |
firstbyte = reqbyte = REQ_UNSET; |
firstbyte = reqbyte = REQ_UNSET; |
| 5865 |
|
|
| 5878 |
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. */ |
| 5879 |
|
|
| 5880 |
/* If this is a capturing subpattern, add to the chain of open capturing items |
/* If this is a capturing subpattern, add to the chain of open capturing items |
| 5881 |
so that we can detect them if (*ACCEPT) is encountered. */ |
so that we can detect them if (*ACCEPT) is encountered. This is also used to |
| 5882 |
|
detect groups that contain recursive back references to themselves. */ |
| 5883 |
|
|
| 5884 |
if (*code == OP_CBRA) |
if (*code == OP_CBRA) |
| 5885 |
{ |
{ |
| 5886 |
capnumber = GET2(code, 1 + LINK_SIZE); |
capnumber = GET2(code, 1 + LINK_SIZE); |
| 5887 |
capitem.number = capnumber; |
capitem.number = capnumber; |
| 5888 |
capitem.next = cd->open_caps; |
capitem.next = cd->open_caps; |
| 5889 |
cd->open_caps = &capitem; |
capitem.flag = FALSE; |
| 5890 |
} |
cd->open_caps = &capitem; |
| 5891 |
|
} |
| 5892 |
|
|
| 5893 |
/* Offset is set zero to mark that this bracket is still open */ |
/* Offset is set zero to mark that this bracket is still open */ |
| 5894 |
|
|
| 5934 |
return FALSE; |
return FALSE; |
| 5935 |
} |
} |
| 5936 |
|
|
| 5937 |
|
/* If the external options have changed during this branch, it means that we |
| 5938 |
|
are at the top level, and a leading option setting has been encountered. We |
| 5939 |
|
need to re-set the original option values to take account of this so that, |
| 5940 |
|
during the pre-compile phase, we know to allow for a re-set at the start of |
| 5941 |
|
subsequent branches. */ |
| 5942 |
|
|
| 5943 |
|
if (old_external_options != cd->external_options) |
| 5944 |
|
oldims = cd->external_options & PCRE_IMS; |
| 5945 |
|
|
| 5946 |
/* Keep the highest bracket count in case (?| was used and some branch |
/* Keep the highest bracket count in case (?| was used and some branch |
| 5947 |
has fewer than the rest. */ |
has fewer than the rest. */ |
| 5948 |
|
|
| 5993 |
|
|
| 5994 |
/* If lookbehind, check that this branch matches a fixed-length string, and |
/* If lookbehind, check that this branch matches a fixed-length string, and |
| 5995 |
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 |
| 5996 |
branch with OP_END. If the branch contains OP_RECURSE, the result is -3 |
branch with OP_END. If the branch contains OP_RECURSE, the result is -3 |
| 5997 |
because there may be forward references that we can't check here. Set a |
because there may be forward references that we can't check here. Set a |
| 5998 |
flag to cause another lookbehind check at the end. Why not do it all at the |
flag to cause another lookbehind check at the end. Why not do it all at the |
| 5999 |
end? Because common, erroneous checks are picked up here and the offset of |
end? Because common, erroneous checks are picked up here and the offset of |
| 6000 |
the problem can be shown. */ |
the problem can be shown. */ |
| 6001 |
|
|
| 6002 |
if (lookbehind) |
if (lookbehind) |
| 6007 |
DPRINTF(("fixed length = %d\n", fixed_length)); |
DPRINTF(("fixed length = %d\n", fixed_length)); |
| 6008 |
if (fixed_length == -3) |
if (fixed_length == -3) |
| 6009 |
{ |
{ |
| 6010 |
cd->check_lookbehind = TRUE; |
cd->check_lookbehind = TRUE; |
| 6011 |
} |
} |
| 6012 |
else if (fixed_length < 0) |
else if (fixed_length < 0) |
| 6013 |
{ |
{ |
| 6014 |
*errorcodeptr = (fixed_length == -2)? ERR36 : ERR25; |
*errorcodeptr = (fixed_length == -2)? ERR36 : ERR25; |
| 6042 |
} |
} |
| 6043 |
while (branch_length > 0); |
while (branch_length > 0); |
| 6044 |
} |
} |
|
|
|
|
/* If it was a capturing subpattern, remove it from the chain. */ |
|
|
|
|
|
if (capnumber > 0) cd->open_caps = cd->open_caps->next; |
|
| 6045 |
|
|
| 6046 |
/* Fill in the ket */ |
/* Fill in the ket */ |
| 6047 |
|
|
| 6049 |
PUT(code, 1, code - start_bracket); |
PUT(code, 1, code - start_bracket); |
| 6050 |
code += 1 + LINK_SIZE; |
code += 1 + LINK_SIZE; |
| 6051 |
|
|
| 6052 |
/* Resetting option if needed */ |
/* If it was a capturing subpattern, check to see if it contained any |
| 6053 |
|
recursive back references. If so, we must wrap it in atomic brackets. |
| 6054 |
|
In any event, remove the block from the chain. */ |
| 6055 |
|
|
| 6056 |
|
if (capnumber > 0) |
| 6057 |
|
{ |
| 6058 |
|
if (cd->open_caps->flag) |
| 6059 |
|
{ |
| 6060 |
|
memmove(start_bracket + 1 + LINK_SIZE, start_bracket, |
| 6061 |
|
code - start_bracket); |
| 6062 |
|
*start_bracket = OP_ONCE; |
| 6063 |
|
code += 1 + LINK_SIZE; |
| 6064 |
|
PUT(start_bracket, 1, code - start_bracket); |
| 6065 |
|
*code = OP_KET; |
| 6066 |
|
PUT(code, 1, code - start_bracket); |
| 6067 |
|
code += 1 + LINK_SIZE; |
| 6068 |
|
length += 2 + 2*LINK_SIZE; |
| 6069 |
|
} |
| 6070 |
|
cd->open_caps = cd->open_caps->next; |
| 6071 |
|
} |
| 6072 |
|
|
| 6073 |
|
/* Reset options if needed. */ |
| 6074 |
|
|
| 6075 |
if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS) |
if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS) |
| 6076 |
{ |
{ |
| 6119 |
{ |
{ |
| 6120 |
*code = OP_ALT; |
*code = OP_ALT; |
| 6121 |
PUT(code, 1, code - last_branch); |
PUT(code, 1, code - last_branch); |
| 6122 |
bc.current = last_branch = code; |
bc.current_branch = last_branch = code; |
| 6123 |
code += 1 + LINK_SIZE; |
code += 1 + LINK_SIZE; |
| 6124 |
} |
} |
| 6125 |
|
|
| 6535 |
|
|
| 6536 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 6537 |
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 && |
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 && |
| 6538 |
(*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0) |
(*erroroffset = _pcre_valid_utf8((USPTR)pattern, -1)) >= 0) |
| 6539 |
{ |
{ |
| 6540 |
errorcode = ERR44; |
errorcode = ERR44; |
| 6541 |
goto PCRE_EARLY_ERROR_RETURN2; |
goto PCRE_EARLY_ERROR_RETURN2; |
| 6732 |
|
|
| 6733 |
*code++ = OP_END; |
*code++ = OP_END; |
| 6734 |
|
|
| 6735 |
#ifndef DEBUG |
#ifndef PCRE_DEBUG |
| 6736 |
if (code - codestart > length) errorcode = ERR23; |
if (code - codestart > length) errorcode = ERR23; |
| 6737 |
#endif |
#endif |
| 6738 |
|
|
| 6755 |
|
|
| 6756 |
if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15; |
if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15; |
| 6757 |
|
|
| 6758 |
/* If there were any lookbehind assertions that contained OP_RECURSE |
/* If there were any lookbehind assertions that contained OP_RECURSE |
| 6759 |
(recursions or subroutine calls), a flag is set for them to be checked here, |
(recursions or subroutine calls), a flag is set for them to be checked here, |
| 6760 |
because they may contain forward references. Actual recursions can't be fixed |
because they may contain forward references. Actual recursions can't be fixed |
| 6761 |
length, but subroutine calls can. It is done like this so that those without |
length, but subroutine calls can. It is done like this so that those without |
| 6766 |
if (cd->check_lookbehind) |
if (cd->check_lookbehind) |
| 6767 |
{ |
{ |
| 6768 |
uschar *cc = (uschar *)codestart; |
uschar *cc = (uschar *)codestart; |
| 6769 |
|
|
| 6770 |
/* Loop, searching for OP_REVERSE items, and process those that do not have |
/* Loop, searching for OP_REVERSE items, and process those that do not have |
| 6771 |
their length set. (Actually, it will also re-process any that have a length |
their length set. (Actually, it will also re-process any that have a length |
| 6772 |
of zero, but that is a pathological case, and it does no harm.) When we find |
of zero, but that is a pathological case, and it does no harm.) When we find |
| 6773 |
one, we temporarily terminate the branch it is in while we scan it. */ |
one, we temporarily terminate the branch it is in while we scan it. */ |
| 6774 |
|
|
| 6775 |
for (cc = (uschar *)_pcre_find_bracket(codestart, utf8, -1); |
for (cc = (uschar *)_pcre_find_bracket(codestart, utf8, -1); |
| 6776 |
cc != NULL; |
cc != NULL; |
| 6777 |
cc = (uschar *)_pcre_find_bracket(cc, utf8, -1)) |
cc = (uschar *)_pcre_find_bracket(cc, utf8, -1)) |
| 6778 |
{ |
{ |
| 6779 |
if (GET(cc, 1) == 0) |
if (GET(cc, 1) == 0) |
| 6780 |
{ |
{ |
| 6781 |
int fixed_length; |
int fixed_length; |
| 6782 |
uschar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); |
uschar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); |
| 6783 |
int end_op = *be; |
int end_op = *be; |
| 6784 |
*be = OP_END; |
*be = OP_END; |
| 6785 |
fixed_length = find_fixedlength(cc, re->options, TRUE, cd); |
fixed_length = find_fixedlength(cc, re->options, TRUE, cd); |
| 6786 |
*be = end_op; |
*be = end_op; |
| 6788 |
if (fixed_length < 0) |
if (fixed_length < 0) |
| 6789 |
{ |
{ |
| 6790 |
errorcode = (fixed_length == -2)? ERR36 : ERR25; |
errorcode = (fixed_length == -2)? ERR36 : ERR25; |
| 6791 |
break; |
break; |
| 6792 |
} |
} |
| 6793 |
PUT(cc, 1, fixed_length); |
PUT(cc, 1, fixed_length); |
| 6794 |
} |
} |
| 6795 |
cc += 1 + LINK_SIZE; |
cc += 1 + LINK_SIZE; |
| 6796 |
} |
} |
| 6797 |
} |
} |
| 6798 |
|
|
| 6799 |
/* Failed to compile, or error while post-processing */ |
/* Failed to compile, or error while post-processing */ |
| 6800 |
|
|
| 6856 |
/* 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 |
| 6857 |
case when building a production library. */ |
case when building a production library. */ |
| 6858 |
|
|
| 6859 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
|
|
|
| 6860 |
printf("Length = %d top_bracket = %d top_backref = %d\n", |
printf("Length = %d top_bracket = %d top_backref = %d\n", |
| 6861 |
length, re->top_bracket, re->top_backref); |
length, re->top_bracket, re->top_backref); |
| 6862 |
|
|
| 6893 |
if (errorcodeptr != NULL) *errorcodeptr = ERR23; |
if (errorcodeptr != NULL) *errorcodeptr = ERR23; |
| 6894 |
return NULL; |
return NULL; |
| 6895 |
} |
} |
| 6896 |
#endif /* DEBUG */ |
#endif /* PCRE_DEBUG */ |
| 6897 |
|
|
| 6898 |
return (pcre *)re; |
return (pcre *)re; |
| 6899 |
} |
} |