| 393 |
"internal error: previously-checked referenced subpattern not found\0" |
"internal error: previously-checked referenced subpattern not found\0" |
| 394 |
"DEFINE group contains more than one branch\0" |
"DEFINE group contains more than one branch\0" |
| 395 |
/* 55 */ |
/* 55 */ |
| 396 |
"repeating a DEFINE group is not allowed\0" |
"repeating a DEFINE group is not allowed\0" /** DEAD **/ |
| 397 |
"inconsistent NEWLINE options\0" |
"inconsistent NEWLINE options\0" |
| 398 |
"\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0" |
"\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0" |
| 399 |
"a numbered reference must not be zero\0" |
"a numbered reference must not be zero\0" |
| 409 |
"(*MARK) must have an argument\0" |
"(*MARK) must have an argument\0" |
| 410 |
"this version of PCRE is not compiled with PCRE_UCP support\0" |
"this version of PCRE is not compiled with PCRE_UCP support\0" |
| 411 |
"\\c must be followed by an ASCII character\0" |
"\\c must be followed by an ASCII character\0" |
| 412 |
|
"\\k is not followed by a braced, angle-bracketed, or quoted name\0" |
| 413 |
; |
; |
| 414 |
|
|
| 415 |
/* Table to identify digits and hex digits. This is used when compiling |
/* Table to identify digits and hex digits. This is used when compiling |
| 546 |
/* Definition to allow mutual recursion */ |
/* Definition to allow mutual recursion */ |
| 547 |
|
|
| 548 |
static BOOL |
static BOOL |
| 549 |
compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int *, |
compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int, |
| 550 |
int *, branch_chain *, compile_data *, int *); |
int *, int *, branch_chain *, compile_data *, int *); |
| 551 |
|
|
| 552 |
|
|
| 553 |
|
|
| 578 |
|
|
| 579 |
|
|
| 580 |
/************************************************* |
/************************************************* |
| 581 |
|
* Check for counted repeat * |
| 582 |
|
*************************************************/ |
| 583 |
|
|
| 584 |
|
/* This function is called when a '{' is encountered in a place where it might |
| 585 |
|
start a quantifier. It looks ahead to see if it really is a quantifier or not. |
| 586 |
|
It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd} |
| 587 |
|
where the ddds are digits. |
| 588 |
|
|
| 589 |
|
Arguments: |
| 590 |
|
p pointer to the first char after '{' |
| 591 |
|
|
| 592 |
|
Returns: TRUE or FALSE |
| 593 |
|
*/ |
| 594 |
|
|
| 595 |
|
static BOOL |
| 596 |
|
is_counted_repeat(const uschar *p) |
| 597 |
|
{ |
| 598 |
|
if ((digitab[*p++] & ctype_digit) == 0) return FALSE; |
| 599 |
|
while ((digitab[*p] & ctype_digit) != 0) p++; |
| 600 |
|
if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; |
| 601 |
|
|
| 602 |
|
if (*p++ != CHAR_COMMA) return FALSE; |
| 603 |
|
if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; |
| 604 |
|
|
| 605 |
|
if ((digitab[*p++] & ctype_digit) == 0) return FALSE; |
| 606 |
|
while ((digitab[*p] & ctype_digit) != 0) p++; |
| 607 |
|
|
| 608 |
|
return (*p == CHAR_RIGHT_CURLY_BRACKET); |
| 609 |
|
} |
| 610 |
|
|
| 611 |
|
|
| 612 |
|
|
| 613 |
|
/************************************************* |
| 614 |
* Handle escapes * |
* Handle escapes * |
| 615 |
*************************************************/ |
*************************************************/ |
| 616 |
|
|
| 676 |
|
|
| 677 |
case CHAR_l: |
case CHAR_l: |
| 678 |
case CHAR_L: |
case CHAR_L: |
| 679 |
|
*errorcodeptr = ERR37; |
| 680 |
|
break; |
| 681 |
|
|
| 682 |
case CHAR_u: |
case CHAR_u: |
| 683 |
|
if ((options & PCRE_JAVASCRIPT_COMPAT) != 0) |
| 684 |
|
{ |
| 685 |
|
/* In JavaScript, \u must be followed by four hexadecimal numbers. |
| 686 |
|
Otherwise it is a lowercase u letter. */ |
| 687 |
|
if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0 |
| 688 |
|
&& (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0) |
| 689 |
|
{ |
| 690 |
|
int i; |
| 691 |
|
c = 0; |
| 692 |
|
for (i = 0; i < 4; ++i) |
| 693 |
|
{ |
| 694 |
|
register int cc = *(++ptr); |
| 695 |
|
#ifndef EBCDIC /* ASCII/UTF-8 coding */ |
| 696 |
|
if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ |
| 697 |
|
c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); |
| 698 |
|
#else /* EBCDIC coding */ |
| 699 |
|
if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ |
| 700 |
|
c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); |
| 701 |
|
#endif |
| 702 |
|
} |
| 703 |
|
} |
| 704 |
|
} |
| 705 |
|
else |
| 706 |
|
*errorcodeptr = ERR37; |
| 707 |
|
break; |
| 708 |
|
|
| 709 |
case CHAR_U: |
case CHAR_U: |
| 710 |
*errorcodeptr = ERR37; |
/* In JavaScript, \U is an uppercase U letter. */ |
| 711 |
|
if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37; |
| 712 |
break; |
break; |
| 713 |
|
|
| 714 |
/* \g must be followed by one of a number of specific things: |
/* In a character class, \g is just a literal "g". Outside a character |
| 715 |
|
class, \g must be followed by one of a number of specific things: |
| 716 |
|
|
| 717 |
(1) A number, either plain or braced. If positive, it is an absolute |
(1) A number, either plain or braced. If positive, it is an absolute |
| 718 |
backreference. If negative, it is a relative backreference. This is a Perl |
backreference. If negative, it is a relative backreference. This is a Perl |
| 729 |
the -ESC_g code (cf \k). */ |
the -ESC_g code (cf \k). */ |
| 730 |
|
|
| 731 |
case CHAR_g: |
case CHAR_g: |
| 732 |
|
if (isclass) break; |
| 733 |
if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE) |
if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE) |
| 734 |
{ |
{ |
| 735 |
c = -ESC_g; |
c = -ESC_g; |
| 858 |
treated as a data character. */ |
treated as a data character. */ |
| 859 |
|
|
| 860 |
case CHAR_x: |
case CHAR_x: |
| 861 |
|
if ((options & PCRE_JAVASCRIPT_COMPAT) != 0) |
| 862 |
|
{ |
| 863 |
|
/* In JavaScript, \x must be followed by two hexadecimal numbers. |
| 864 |
|
Otherwise it is a lowercase x letter. */ |
| 865 |
|
if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0) |
| 866 |
|
{ |
| 867 |
|
int i; |
| 868 |
|
c = 0; |
| 869 |
|
for (i = 0; i < 2; ++i) |
| 870 |
|
{ |
| 871 |
|
register int cc = *(++ptr); |
| 872 |
|
#ifndef EBCDIC /* ASCII/UTF-8 coding */ |
| 873 |
|
if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ |
| 874 |
|
c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); |
| 875 |
|
#else /* EBCDIC coding */ |
| 876 |
|
if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ |
| 877 |
|
c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); |
| 878 |
|
#endif |
| 879 |
|
} |
| 880 |
|
} |
| 881 |
|
break; |
| 882 |
|
} |
| 883 |
|
|
| 884 |
if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) |
if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) |
| 885 |
{ |
{ |
| 886 |
const uschar *pt = ptr + 2; |
const uschar *pt = ptr + 2; |
| 974 |
} |
} |
| 975 |
|
|
| 976 |
/* Perl supports \N{name} for character names, as well as plain \N for "not |
/* Perl supports \N{name} for character names, as well as plain \N for "not |
| 977 |
newline". PCRE does not support \N{name}. */ |
newline". PCRE does not support \N{name}. However, it does support |
| 978 |
|
quantification such as \N{2,3}. */ |
| 979 |
|
|
| 980 |
if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET) |
if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET && |
| 981 |
|
!is_counted_repeat(ptr+2)) |
| 982 |
*errorcodeptr = ERR37; |
*errorcodeptr = ERR37; |
| 983 |
|
|
| 984 |
/* If PCRE_UCP is set, we change the values for \d etc. */ |
/* If PCRE_UCP is set, we change the values for \d etc. */ |
| 1088 |
|
|
| 1089 |
|
|
| 1090 |
/************************************************* |
/************************************************* |
|
* Check for counted repeat * |
|
|
*************************************************/ |
|
|
|
|
|
/* This function is called when a '{' is encountered in a place where it might |
|
|
start a quantifier. It looks ahead to see if it really is a quantifier or not. |
|
|
It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd} |
|
|
where the ddds are digits. |
|
|
|
|
|
Arguments: |
|
|
p pointer to the first char after '{' |
|
|
|
|
|
Returns: TRUE or FALSE |
|
|
*/ |
|
|
|
|
|
static BOOL |
|
|
is_counted_repeat(const uschar *p) |
|
|
{ |
|
|
if ((digitab[*p++] & ctype_digit) == 0) return FALSE; |
|
|
while ((digitab[*p] & ctype_digit) != 0) p++; |
|
|
if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; |
|
|
|
|
|
if (*p++ != CHAR_COMMA) return FALSE; |
|
|
if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; |
|
|
|
|
|
if ((digitab[*p++] & ctype_digit) == 0) return FALSE; |
|
|
while ((digitab[*p] & ctype_digit) != 0) p++; |
|
|
|
|
|
return (*p == CHAR_RIGHT_CURLY_BRACKET); |
|
|
} |
|
|
|
|
|
|
|
|
|
|
|
/************************************************* |
|
| 1091 |
* Read repeat counts * |
* Read repeat counts * |
| 1092 |
*************************************************/ |
*************************************************/ |
| 1093 |
|
|
| 1559 |
case OP_CBRA: |
case OP_CBRA: |
| 1560 |
case OP_BRA: |
case OP_BRA: |
| 1561 |
case OP_ONCE: |
case OP_ONCE: |
| 1562 |
|
case OP_ONCE_NC: |
| 1563 |
case OP_COND: |
case OP_COND: |
| 1564 |
d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd); |
d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd); |
| 1565 |
if (d < 0) return d; |
if (d < 0) return d; |
| 1753 |
for (;;) |
for (;;) |
| 1754 |
{ |
{ |
| 1755 |
register int c = *code; |
register int c = *code; |
| 1756 |
|
|
| 1757 |
if (c == OP_END) return NULL; |
if (c == OP_END) return NULL; |
| 1758 |
|
|
| 1759 |
/* XCLASS is used for classes that cannot be represented just by a bit |
/* XCLASS is used for classes that cannot be represented just by a bit |
| 1815 |
break; |
break; |
| 1816 |
|
|
| 1817 |
case OP_THEN_ARG: |
case OP_THEN_ARG: |
| 1818 |
code += code[1+LINK_SIZE]; |
code += code[1]; |
| 1819 |
break; |
break; |
| 1820 |
} |
} |
| 1821 |
|
|
| 1934 |
break; |
break; |
| 1935 |
|
|
| 1936 |
case OP_THEN_ARG: |
case OP_THEN_ARG: |
| 1937 |
code += code[1+LINK_SIZE]; |
code += code[1]; |
| 1938 |
break; |
break; |
| 1939 |
} |
} |
| 1940 |
|
|
| 2034 |
} |
} |
| 2035 |
|
|
| 2036 |
/* For a recursion/subroutine call, if its end has been reached, which |
/* For a recursion/subroutine call, if its end has been reached, which |
| 2037 |
implies a subroutine call, we can scan it. */ |
implies a backward reference subroutine call, we can scan it. If it's a |
| 2038 |
|
forward reference subroutine call, we can't. To detect forward reference |
| 2039 |
|
we have to scan up the list that is kept in the workspace. This function is |
| 2040 |
|
called only when doing the real compile, not during the pre-compile that |
| 2041 |
|
measures the size of the compiled pattern. */ |
| 2042 |
|
|
| 2043 |
if (c == OP_RECURSE) |
if (c == OP_RECURSE) |
| 2044 |
{ |
{ |
| 2045 |
BOOL empty_branch = FALSE; |
const uschar *scode; |
| 2046 |
const uschar *scode = cd->start_code + GET(code, 1); |
BOOL empty_branch; |
| 2047 |
|
|
| 2048 |
|
/* Test for forward reference */ |
| 2049 |
|
|
| 2050 |
|
for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE) |
| 2051 |
|
if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE; |
| 2052 |
|
|
| 2053 |
|
/* Not a forward reference, test for completed backward reference */ |
| 2054 |
|
|
| 2055 |
|
empty_branch = FALSE; |
| 2056 |
|
scode = cd->start_code + GET(code, 1); |
| 2057 |
if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ |
if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ |
| 2058 |
|
|
| 2059 |
|
/* Completed backwards reference */ |
| 2060 |
|
|
| 2061 |
do |
do |
| 2062 |
{ |
{ |
| 2063 |
if (could_be_empty_branch(scode, endcode, utf8, cd)) |
if (could_be_empty_branch(scode, endcode, utf8, cd)) |
| 2068 |
scode += GET(scode, 1); |
scode += GET(scode, 1); |
| 2069 |
} |
} |
| 2070 |
while (*scode == OP_ALT); |
while (*scode == OP_ALT); |
| 2071 |
|
|
| 2072 |
if (!empty_branch) return FALSE; /* All branches are non-empty */ |
if (!empty_branch) return FALSE; /* All branches are non-empty */ |
| 2073 |
continue; |
continue; |
| 2074 |
} |
} |
| 2099 |
|
|
| 2100 |
if (c == OP_BRA || c == OP_BRAPOS || |
if (c == OP_BRA || c == OP_BRAPOS || |
| 2101 |
c == OP_CBRA || c == OP_CBRAPOS || |
c == OP_CBRA || c == OP_CBRAPOS || |
| 2102 |
c == OP_ONCE || c == OP_COND) |
c == OP_ONCE || c == OP_ONCE_NC || |
| 2103 |
|
c == OP_COND) |
| 2104 |
{ |
{ |
| 2105 |
BOOL empty_branch; |
BOOL empty_branch; |
| 2106 |
if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ |
if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ |
| 2272 |
break; |
break; |
| 2273 |
|
|
| 2274 |
case OP_THEN_ARG: |
case OP_THEN_ARG: |
| 2275 |
code += code[1+LINK_SIZE]; |
code += code[1]; |
| 2276 |
break; |
break; |
| 2277 |
|
|
| 2278 |
/* None of the remaining opcodes are required to match a character. */ |
/* None of the remaining opcodes are required to match a character. */ |
| 2295 |
the current branch of the current pattern to see if it could match the empty |
the current branch of the current pattern to see if it could match the empty |
| 2296 |
string. If it could, we must look outwards for branches at other levels, |
string. If it could, we must look outwards for branches at other levels, |
| 2297 |
stopping when we pass beyond the bracket which is the subject of the recursion. |
stopping when we pass beyond the bracket which is the subject of the recursion. |
| 2298 |
|
This function is called only during the real compile, not during the |
| 2299 |
|
pre-compile. |
| 2300 |
|
|
| 2301 |
Arguments: |
Arguments: |
| 2302 |
code points to start of the recursion |
code points to start of the recursion |
| 2347 |
"l\ower". This is a lesser evil that not diagnosing bad classes when Perl does, |
"l\ower". This is a lesser evil that not diagnosing bad classes when Perl does, |
| 2348 |
I think. |
I think. |
| 2349 |
|
|
| 2350 |
|
A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not. |
| 2351 |
|
It seems that the appearance of a nested POSIX class supersedes an apparent |
| 2352 |
|
external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or |
| 2353 |
|
a digit. |
| 2354 |
|
|
| 2355 |
|
In Perl, unescaped square brackets may also appear as part of class names. For |
| 2356 |
|
example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for |
| 2357 |
|
[:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not |
| 2358 |
|
seem right at all. PCRE does not allow closing square brackets in POSIX class |
| 2359 |
|
names. |
| 2360 |
|
|
| 2361 |
Arguments: |
Arguments: |
| 2362 |
ptr pointer to the initial [ |
ptr pointer to the initial [ |
| 2363 |
endptr where to return the end pointer |
endptr where to return the end pointer |
| 2372 |
terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ |
terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ |
| 2373 |
for (++ptr; *ptr != 0; ptr++) |
for (++ptr; *ptr != 0; ptr++) |
| 2374 |
{ |
{ |
| 2375 |
if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else |
if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) |
| 2376 |
|
ptr++; |
| 2377 |
|
else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE; |
| 2378 |
|
else |
| 2379 |
{ |
{ |
|
if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE; |
|
| 2380 |
if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) |
if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) |
| 2381 |
{ |
{ |
| 2382 |
*endptr = ptr; |
*endptr = ptr; |
| 2383 |
return TRUE; |
return TRUE; |
| 2384 |
} |
} |
| 2385 |
|
if (*ptr == CHAR_LEFT_SQUARE_BRACKET && |
| 2386 |
|
(ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || |
| 2387 |
|
ptr[1] == CHAR_EQUALS_SIGN) && |
| 2388 |
|
check_posix_syntax(ptr, endptr)) |
| 2389 |
|
return FALSE; |
| 2390 |
} |
} |
| 2391 |
} |
} |
| 2392 |
return FALSE; |
return FALSE; |
| 3096 |
firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) |
firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) |
| 3097 |
reqbyteptr set to the last literal character required, else < 0 |
reqbyteptr set to the last literal character required, else < 0 |
| 3098 |
bcptr points to current branch chain |
bcptr points to current branch chain |
| 3099 |
|
cond_depth conditional nesting depth |
| 3100 |
cd contains pointers to tables etc. |
cd contains pointers to tables etc. |
| 3101 |
lengthptr NULL during the real compile phase |
lengthptr NULL during the real compile phase |
| 3102 |
points to length accumulator during pre-compile phase |
points to length accumulator during pre-compile phase |
| 3108 |
static BOOL |
static BOOL |
| 3109 |
compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr, |
compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr, |
| 3110 |
int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, |
int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, |
| 3111 |
compile_data *cd, int *lengthptr) |
int cond_depth, compile_data *cd, int *lengthptr) |
| 3112 |
{ |
{ |
| 3113 |
int repeat_type, op_type; |
int repeat_type, op_type; |
| 3114 |
int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ |
int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ |
| 3117 |
int firstbyte, reqbyte; |
int firstbyte, reqbyte; |
| 3118 |
int zeroreqbyte, zerofirstbyte; |
int zeroreqbyte, zerofirstbyte; |
| 3119 |
int req_caseopt, reqvary, tempreqvary; |
int req_caseopt, reqvary, tempreqvary; |
| 3120 |
int options = *optionsptr; |
int options = *optionsptr; /* May change dynamically */ |
| 3121 |
int after_manual_callout = 0; |
int after_manual_callout = 0; |
| 3122 |
int length_prevgroup = 0; |
int length_prevgroup = 0; |
| 3123 |
register int c; |
register int c; |
| 3135 |
uschar *save_hwm = NULL; |
uschar *save_hwm = NULL; |
| 3136 |
uschar classbits[32]; |
uschar classbits[32]; |
| 3137 |
|
|
| 3138 |
|
/* We can fish out the UTF-8 setting once and for all into a BOOL, but we |
| 3139 |
|
must not do this for other options (e.g. PCRE_EXTENDED) because they may change |
| 3140 |
|
dynamically as we process the pattern. */ |
| 3141 |
|
|
| 3142 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 3143 |
BOOL class_utf8; |
BOOL class_utf8; |
| 3144 |
BOOL utf8 = (options & PCRE_UTF8) != 0; |
BOOL utf8 = (options & PCRE_UTF8) != 0; |
| 3147 |
uschar utf8_char[6]; |
uschar utf8_char[6]; |
| 3148 |
#else |
#else |
| 3149 |
BOOL utf8 = FALSE; |
BOOL utf8 = FALSE; |
|
uschar *utf8_char = NULL; |
|
| 3150 |
#endif |
#endif |
| 3151 |
|
|
| 3152 |
#ifdef PCRE_DEBUG |
#ifdef PCRE_DEBUG |
| 3197 |
int subfirstbyte; |
int subfirstbyte; |
| 3198 |
int terminator; |
int terminator; |
| 3199 |
int mclength; |
int mclength; |
| 3200 |
|
int tempbracount; |
| 3201 |
uschar mcbuffer[8]; |
uschar mcbuffer[8]; |
| 3202 |
|
|
| 3203 |
/* Get next byte in the pattern */ |
/* Get next byte in the pattern */ |
| 3319 |
previous_callout = NULL; |
previous_callout = NULL; |
| 3320 |
} |
} |
| 3321 |
|
|
| 3322 |
/* In extended mode, skip white space and comments */ |
/* In extended mode, skip white space and comments. */ |
| 3323 |
|
|
| 3324 |
if ((options & PCRE_EXTENDED) != 0) |
if ((options & PCRE_EXTENDED) != 0) |
| 3325 |
{ |
{ |
| 4288 |
op_type = 0; /* Default single-char op codes */ |
op_type = 0; /* Default single-char op codes */ |
| 4289 |
possessive_quantifier = FALSE; /* Default not possessive quantifier */ |
possessive_quantifier = FALSE; /* Default not possessive quantifier */ |
| 4290 |
|
|
| 4291 |
/* Save start of previous item, in case we have to move it up to make space |
/* Save start of previous item, in case we have to move it up in order to |
| 4292 |
for an inserted OP_ONCE for the additional '+' extension. */ |
insert something before it. */ |
| 4293 |
|
|
| 4294 |
tempcode = previous; |
tempcode = previous; |
| 4295 |
|
|
| 4312 |
} |
} |
| 4313 |
else repeat_type = greedy_default; |
else repeat_type = greedy_default; |
| 4314 |
|
|
| 4315 |
|
/* If previous was a recursion call, wrap it in atomic brackets so that |
| 4316 |
|
previous becomes the atomic group. All recursions were so wrapped in the |
| 4317 |
|
past, but it no longer happens for non-repeated recursions. In fact, the |
| 4318 |
|
repeated ones could be re-implemented independently so as not to need this, |
| 4319 |
|
but for the moment we rely on the code for repeating groups. */ |
| 4320 |
|
|
| 4321 |
|
if (*previous == OP_RECURSE) |
| 4322 |
|
{ |
| 4323 |
|
memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE); |
| 4324 |
|
*previous = OP_ONCE; |
| 4325 |
|
PUT(previous, 1, 2 + 2*LINK_SIZE); |
| 4326 |
|
previous[2 + 2*LINK_SIZE] = OP_KET; |
| 4327 |
|
PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE); |
| 4328 |
|
code += 2 + 2 * LINK_SIZE; |
| 4329 |
|
length_prevgroup = 3 + 3*LINK_SIZE; |
| 4330 |
|
|
| 4331 |
|
/* When actually compiling, we need to check whether this was a forward |
| 4332 |
|
reference, and if so, adjust the offset. */ |
| 4333 |
|
|
| 4334 |
|
if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE) |
| 4335 |
|
{ |
| 4336 |
|
int offset = GET(cd->hwm, -LINK_SIZE); |
| 4337 |
|
if (offset == previous + 1 - cd->start_code) |
| 4338 |
|
PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE); |
| 4339 |
|
} |
| 4340 |
|
} |
| 4341 |
|
|
| 4342 |
|
/* Now handle repetition for the different types of item. */ |
| 4343 |
|
|
| 4344 |
/* If previous was a character match, abolish the item and generate a |
/* If previous was a character match, abolish the item and generate a |
| 4345 |
repeat item instead. If a char item has a minumum of more than one, ensure |
repeat item instead. If a char item has a minumum of more than one, ensure |
| 4346 |
that it is set in reqbyte - it might not be if a sequence such as x{3} is |
that it is set in reqbyte - it might not be if a sequence such as x{3} is |
| 4632 |
} |
} |
| 4633 |
|
|
| 4634 |
/* If previous was a bracket group, we may have to replicate it in certain |
/* If previous was a bracket group, we may have to replicate it in certain |
| 4635 |
cases. Note that at this point we can encounter only the "basic" BRA and |
cases. Note that at this point we can encounter only the "basic" bracket |
| 4636 |
KET opcodes, as this is the place where they get converted into the more |
opcodes such as BRA and CBRA, as this is the place where they get converted |
| 4637 |
special varieties. */ |
into the more special varieties such as BRAPOS and SBRA. A test for >= |
| 4638 |
|
OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK, |
| 4639 |
|
ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow |
| 4640 |
|
repetition of assertions, but now it does, for Perl compatibility. */ |
| 4641 |
|
|
| 4642 |
else if (*previous == OP_BRA || *previous == OP_CBRA || |
else if (*previous >= OP_ASSERT && *previous <= OP_COND) |
|
*previous == OP_ONCE || *previous == OP_COND) |
|
| 4643 |
{ |
{ |
| 4644 |
register int i; |
register int i; |
| 4645 |
int len = (int)(code - previous); |
int len = (int)(code - previous); |
| 4646 |
uschar *bralink = NULL; |
uschar *bralink = NULL; |
| 4647 |
uschar *brazeroptr = NULL; |
uschar *brazeroptr = NULL; |
| 4648 |
|
|
| 4649 |
/* Repeating a DEFINE group is pointless */ |
/* Repeating a DEFINE group is pointless, but Perl allows the syntax, so |
| 4650 |
|
we just ignore the repeat. */ |
| 4651 |
|
|
| 4652 |
if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF) |
if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF) |
| 4653 |
|
goto END_REPEAT; |
| 4654 |
|
|
| 4655 |
|
/* There is no sense in actually repeating assertions. The only potential |
| 4656 |
|
use of repetition is in cases when the assertion is optional. Therefore, |
| 4657 |
|
if the minimum is greater than zero, just ignore the repeat. If the |
| 4658 |
|
maximum is not not zero or one, set it to 1. */ |
| 4659 |
|
|
| 4660 |
|
if (*previous < OP_ONCE) /* Assertion */ |
| 4661 |
{ |
{ |
| 4662 |
*errorcodeptr = ERR55; |
if (repeat_min > 0) goto END_REPEAT; |
| 4663 |
goto FAILED; |
if (repeat_max < 0 || repeat_max > 1) repeat_max = 1; |
| 4664 |
} |
} |
| 4665 |
|
|
| 4666 |
/* The case of a zero minimum is special because of the need to stick |
/* The case of a zero minimum is special because of the need to stick |
| 4681 |
** goto END_REPEAT; |
** goto END_REPEAT; |
| 4682 |
** } |
** } |
| 4683 |
|
|
| 4684 |
However, that fails when a group is referenced as a subroutine from |
However, that fails when a group or a subgroup within it is referenced |
| 4685 |
elsewhere in the pattern, so now we stick in OP_SKIPZERO in front of it |
as a subroutine from elsewhere in the pattern, so now we stick in |
| 4686 |
so that it is skipped on execution. As we don't have a list of which |
OP_SKIPZERO in front of it so that it is skipped on execution. As we |
| 4687 |
groups are referenced, we cannot do this selectively. |
don't have a list of which groups are referenced, we cannot do this |
| 4688 |
|
selectively. |
| 4689 |
|
|
| 4690 |
If the maximum is 1 or unlimited, we just have to stick in the BRAZERO |
If the maximum is 1 or unlimited, we just have to stick in the BRAZERO |
| 4691 |
and do no more at this point. However, we do need to adjust any |
and do no more at this point. However, we do need to adjust any |
| 4871 |
} |
} |
| 4872 |
|
|
| 4873 |
/* If the maximum is unlimited, set a repeater in the final copy. For |
/* If the maximum is unlimited, set a repeater in the final copy. For |
| 4874 |
ONCE brackets, that's all we need to do. |
ONCE brackets, that's all we need to do. However, possessively repeated |
| 4875 |
|
ONCE brackets can be converted into non-capturing brackets, as the |
| 4876 |
|
behaviour of (?:xx)++ is the same as (?>xx)++ and this saves having to |
| 4877 |
|
deal with possessive ONCEs specially. |
| 4878 |
|
|
| 4879 |
Otherwise, if the quantifier was possessive, we convert the BRA code to |
Otherwise, if the quantifier was possessive, we convert the BRA code to |
| 4880 |
the POS form, and the KET code to KETRPOS. (It turns out to be convenient |
the POS form, and the KET code to KETRPOS. (It turns out to be convenient |
| 4896 |
uschar *ketcode = code - 1 - LINK_SIZE; |
uschar *ketcode = code - 1 - LINK_SIZE; |
| 4897 |
uschar *bracode = ketcode - GET(ketcode, 1); |
uschar *bracode = ketcode - GET(ketcode, 1); |
| 4898 |
|
|
| 4899 |
if (*bracode == OP_ONCE) |
if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) && |
| 4900 |
|
possessive_quantifier) *bracode = OP_BRA; |
| 4901 |
|
|
| 4902 |
|
if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC) |
| 4903 |
*ketcode = OP_KETRMAX + repeat_type; |
*ketcode = OP_KETRMAX + repeat_type; |
| 4904 |
else |
else |
| 4905 |
{ |
{ |
| 4950 |
there are special alternative opcodes for this case. For anything else, we |
there are special alternative opcodes for this case. For anything else, we |
| 4951 |
wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+' |
wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+' |
| 4952 |
notation is just syntactic sugar, taken from Sun's Java package, but the |
notation is just syntactic sugar, taken from Sun's Java package, but the |
| 4953 |
special opcodes can optimize it. |
special opcodes can optimize it. |
| 4954 |
|
|
| 4955 |
Possessively repeated subpatterns have already been handled in the code |
Possessively repeated subpatterns have already been handled in the code |
| 4956 |
just above, so possessive_quantifier is always FALSE for them at this |
just above, so possessive_quantifier is always FALSE for them at this |
| 4957 |
stage. |
stage. |
| 4958 |
|
|
| 4959 |
Note that the repeated item starts at tempcode, not at previous, which |
Note that the repeated item starts at tempcode, not at previous, which |
| 4960 |
might be the first part of a string whose (former) last char we repeated. |
might be the first part of a string whose (former) last char we repeated. |
| 4961 |
|
|
| 5061 |
while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {}; |
while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {}; |
| 5062 |
namelen = (int)(ptr - name); |
namelen = (int)(ptr - name); |
| 5063 |
|
|
| 5064 |
|
/* It appears that Perl allows any characters whatsoever, other than |
| 5065 |
|
a closing parenthesis, to appear in arguments, so we no longer insist on |
| 5066 |
|
letters, digits, and underscores. */ |
| 5067 |
|
|
| 5068 |
if (*ptr == CHAR_COLON) |
if (*ptr == CHAR_COLON) |
| 5069 |
{ |
{ |
| 5070 |
arg = ++ptr; |
arg = ++ptr; |
| 5071 |
while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0 |
while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; |
|
|| *ptr == '_') ptr++; |
|
| 5072 |
arglen = (int)(ptr - arg); |
arglen = (int)(ptr - arg); |
| 5073 |
} |
} |
| 5074 |
|
|
| 5085 |
if (namelen == verbs[i].len && |
if (namelen == verbs[i].len && |
| 5086 |
strncmp((char *)name, vn, namelen) == 0) |
strncmp((char *)name, vn, namelen) == 0) |
| 5087 |
{ |
{ |
| 5088 |
/* Check for open captures before ACCEPT and convert it to |
/* Check for open captures before ACCEPT and convert it to |
| 5089 |
ASSERT_ACCEPT if in an assertion. */ |
ASSERT_ACCEPT if in an assertion. */ |
| 5090 |
|
|
| 5091 |
if (verbs[i].op == OP_ACCEPT) |
if (verbs[i].op == OP_ACCEPT) |
| 5095 |
{ |
{ |
| 5096 |
*errorcodeptr = ERR59; |
*errorcodeptr = ERR59; |
| 5097 |
goto FAILED; |
goto FAILED; |
| 5098 |
} |
} |
| 5099 |
cd->had_accept = TRUE; |
cd->had_accept = TRUE; |
| 5100 |
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
| 5101 |
{ |
{ |
| 5103 |
PUT2INC(code, 0, oc->number); |
PUT2INC(code, 0, oc->number); |
| 5104 |
} |
} |
| 5105 |
*code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; |
*code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; |
| 5106 |
|
|
| 5107 |
|
/* Do not set firstbyte after *ACCEPT */ |
| 5108 |
|
if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; |
| 5109 |
} |
} |
| 5110 |
|
|
| 5111 |
/* Handle other cases with/without an argument */ |
/* Handle other cases with/without an argument */ |
| 5118 |
goto FAILED; |
goto FAILED; |
| 5119 |
} |
} |
| 5120 |
*code = verbs[i].op; |
*code = verbs[i].op; |
| 5121 |
if (*code++ == OP_THEN) |
if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN; |
|
{ |
|
|
PUT(code, 0, code - bcptr->current_branch - 1); |
|
|
code += LINK_SIZE; |
|
|
} |
|
| 5122 |
} |
} |
| 5123 |
|
|
| 5124 |
else |
else |
| 5129 |
goto FAILED; |
goto FAILED; |
| 5130 |
} |
} |
| 5131 |
*code = verbs[i].op_arg; |
*code = verbs[i].op_arg; |
| 5132 |
if (*code++ == OP_THEN_ARG) |
if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN; |
|
{ |
|
|
PUT(code, 0, code - bcptr->current_branch - 1); |
|
|
code += LINK_SIZE; |
|
|
} |
|
| 5133 |
*code++ = arglen; |
*code++ = arglen; |
| 5134 |
memcpy(code, arg, arglen); |
memcpy(code, arg, arglen); |
| 5135 |
code += arglen; |
code += arglen; |
| 5391 |
/* ------------------------------------------------------------ */ |
/* ------------------------------------------------------------ */ |
| 5392 |
case CHAR_EQUALS_SIGN: /* Positive lookahead */ |
case CHAR_EQUALS_SIGN: /* Positive lookahead */ |
| 5393 |
bravalue = OP_ASSERT; |
bravalue = OP_ASSERT; |
| 5394 |
cd->assert_depth += 1; |
cd->assert_depth += 1; |
| 5395 |
ptr++; |
ptr++; |
| 5396 |
break; |
break; |
| 5397 |
|
|
| 5406 |
continue; |
continue; |
| 5407 |
} |
} |
| 5408 |
bravalue = OP_ASSERT_NOT; |
bravalue = OP_ASSERT_NOT; |
| 5409 |
cd->assert_depth += 1; |
cd->assert_depth += 1; |
| 5410 |
break; |
break; |
| 5411 |
|
|
| 5412 |
|
|
| 5416 |
{ |
{ |
| 5417 |
case CHAR_EQUALS_SIGN: /* Positive lookbehind */ |
case CHAR_EQUALS_SIGN: /* Positive lookbehind */ |
| 5418 |
bravalue = OP_ASSERTBACK; |
bravalue = OP_ASSERTBACK; |
| 5419 |
cd->assert_depth += 1; |
cd->assert_depth += 1; |
| 5420 |
ptr += 2; |
ptr += 2; |
| 5421 |
break; |
break; |
| 5422 |
|
|
| 5423 |
case CHAR_EXCLAMATION_MARK: /* Negative lookbehind */ |
case CHAR_EXCLAMATION_MARK: /* Negative lookbehind */ |
| 5424 |
bravalue = OP_ASSERTBACK_NOT; |
bravalue = OP_ASSERTBACK_NOT; |
| 5425 |
cd->assert_depth += 1; |
cd->assert_depth += 1; |
| 5426 |
ptr += 2; |
ptr += 2; |
| 5427 |
break; |
break; |
| 5428 |
|
|
| 5813 |
|
|
| 5814 |
/* Fudge the value of "called" so that when it is inserted as an |
/* Fudge the value of "called" so that when it is inserted as an |
| 5815 |
offset below, what it actually inserted is the reference number |
offset below, what it actually inserted is the reference number |
| 5816 |
of the group. */ |
of the group. Then remember the forward reference. */ |
| 5817 |
|
|
| 5818 |
called = cd->start_code + recno; |
called = cd->start_code + recno; |
| 5819 |
PUTINC(cd->hwm, 0, (int)(code + 2 + LINK_SIZE - cd->start_code)); |
PUTINC(cd->hwm, 0, (int)(code + 1 - cd->start_code)); |
| 5820 |
} |
} |
| 5821 |
|
|
| 5822 |
/* If not a forward reference, and the subpattern is still open, |
/* If not a forward reference, and the subpattern is still open, |
| 5823 |
this is a recursive call. We check to see if this is a left |
this is a recursive call. We check to see if this is a left |
| 5824 |
recursion that could loop for ever, and diagnose that case. */ |
recursion that could loop for ever, and diagnose that case. We |
| 5825 |
|
must not, however, do this check if we are in a conditional |
| 5826 |
|
subpattern because the condition might be testing for recursion in |
| 5827 |
|
a pattern such as /(?(R)a+|(?R)b)/, which is perfectly valid. |
| 5828 |
|
Forever loops are also detected at runtime, so those that occur in |
| 5829 |
|
conditional subpatterns will be picked up then. */ |
| 5830 |
|
|
| 5831 |
else if (GET(called, 1) == 0 && |
else if (GET(called, 1) == 0 && cond_depth <= 0 && |
| 5832 |
could_be_empty(called, code, bcptr, utf8, cd)) |
could_be_empty(called, code, bcptr, utf8, cd)) |
| 5833 |
{ |
{ |
| 5834 |
*errorcodeptr = ERR40; |
*errorcodeptr = ERR40; |
| 5836 |
} |
} |
| 5837 |
} |
} |
| 5838 |
|
|
| 5839 |
/* Insert the recursion/subroutine item, automatically wrapped inside |
/* Insert the recursion/subroutine item. */ |
|
"once" brackets. Set up a "previous group" length so that a |
|
|
subsequent quantifier will work. */ |
|
|
|
|
|
*code = OP_ONCE; |
|
|
PUT(code, 1, 2 + 2*LINK_SIZE); |
|
|
code += 1 + LINK_SIZE; |
|
| 5840 |
|
|
| 5841 |
*code = OP_RECURSE; |
*code = OP_RECURSE; |
| 5842 |
PUT(code, 1, (int)(called - cd->start_code)); |
PUT(code, 1, (int)(called - cd->start_code)); |
| 5843 |
code += 1 + LINK_SIZE; |
code += 1 + LINK_SIZE; |
|
|
|
|
*code = OP_KET; |
|
|
PUT(code, 1, 2 + 2*LINK_SIZE); |
|
|
code += 1 + LINK_SIZE; |
|
|
|
|
|
length_prevgroup = 3 + 3*LINK_SIZE; |
|
| 5844 |
} |
} |
| 5845 |
|
|
| 5846 |
/* Can't determine a first byte now */ |
/* Can't determine a first byte now */ |
| 5955 |
skipbytes = 2; |
skipbytes = 2; |
| 5956 |
} |
} |
| 5957 |
|
|
| 5958 |
/* Process nested bracketed regex. Assertions may not be repeated, but |
/* Process nested bracketed regex. Assertions used not to be repeatable, |
| 5959 |
other kinds can be. All their opcodes are >= OP_ONCE. We copy code into a |
but this was changed for Perl compatibility, so all kinds can now be |
| 5960 |
non-register variable (tempcode) in order to be able to pass its address |
repeated. We copy code into a non-register variable (tempcode) in order to |
| 5961 |
because some compilers complain otherwise. */ |
be able to pass its address because some compilers complain otherwise. */ |
| 5962 |
|
|
| 5963 |
previous = (bravalue >= OP_ONCE)? code : NULL; |
previous = code; /* For handling repetition */ |
| 5964 |
*code = bravalue; |
*code = bravalue; |
| 5965 |
tempcode = code; |
tempcode = code; |
| 5966 |
tempreqvary = cd->req_varyopt; /* Save value before bracket */ |
tempreqvary = cd->req_varyopt; /* Save value before bracket */ |
| 5967 |
length_prevgroup = 0; /* Initialize for pre-compile phase */ |
tempbracount = cd->bracount; /* Save value before bracket */ |
| 5968 |
|
length_prevgroup = 0; /* Initialize for pre-compile phase */ |
| 5969 |
|
|
| 5970 |
if (!compile_regex( |
if (!compile_regex( |
| 5971 |
newoptions, /* The complete new option state */ |
newoptions, /* The complete new option state */ |
| 5972 |
&tempcode, /* Where to put code (updated) */ |
&tempcode, /* Where to put code (updated) */ |
| 5973 |
&ptr, /* Input pointer (updated) */ |
&ptr, /* Input pointer (updated) */ |
| 5974 |
errorcodeptr, /* Where to put an error message */ |
errorcodeptr, /* Where to put an error message */ |
| 5975 |
(bravalue == OP_ASSERTBACK || |
(bravalue == OP_ASSERTBACK || |
| 5976 |
bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */ |
bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */ |
| 5977 |
reset_bracount, /* True if (?| group */ |
reset_bracount, /* True if (?| group */ |
| 5978 |
skipbytes, /* Skip over bracket number */ |
skipbytes, /* Skip over bracket number */ |
| 5979 |
&subfirstbyte, /* For possible first char */ |
cond_depth + |
| 5980 |
&subreqbyte, /* For possible last char */ |
((bravalue == OP_COND)?1:0), /* Depth of condition subpatterns */ |
| 5981 |
bcptr, /* Current branch chain */ |
&subfirstbyte, /* For possible first char */ |
| 5982 |
cd, /* Tables block */ |
&subreqbyte, /* For possible last char */ |
| 5983 |
(lengthptr == NULL)? NULL : /* Actual compile phase */ |
bcptr, /* Current branch chain */ |
| 5984 |
&length_prevgroup /* Pre-compile phase */ |
cd, /* Tables block */ |
| 5985 |
|
(lengthptr == NULL)? NULL : /* Actual compile phase */ |
| 5986 |
|
&length_prevgroup /* Pre-compile phase */ |
| 5987 |
)) |
)) |
| 5988 |
goto FAILED; |
goto FAILED; |
| 5989 |
|
|
| 5990 |
|
/* If this was an atomic group and there are no capturing groups within it, |
| 5991 |
|
generate OP_ONCE_NC instead of OP_ONCE. */ |
| 5992 |
|
|
| 5993 |
|
if (bravalue == OP_ONCE && cd->bracount <= tempbracount) |
| 5994 |
|
*code = OP_ONCE_NC; |
| 5995 |
|
|
| 5996 |
if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT) |
if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT) |
| 5997 |
cd->assert_depth -= 1; |
cd->assert_depth -= 1; |
| 5998 |
|
|
| 5999 |
/* At the end of compiling, code is still pointing to the start of the |
/* At the end of compiling, code is still pointing to the start of the |
| 6000 |
group, while tempcode has been updated to point past the end of the group |
group, while tempcode has been updated to point past the end of the group. |
| 6001 |
and any option resetting that may follow it. The pattern pointer (ptr) |
The pattern pointer (ptr) is on the bracket. |
|
is on the bracket. */ |
|
| 6002 |
|
|
| 6003 |
/* If this is a conditional bracket, check that there are no more than |
If this is a conditional bracket, check that there are no more than |
| 6004 |
two branches in the group, or just one if it's a DEFINE group. We do this |
two branches in the group, or just one if it's a DEFINE group. We do this |
| 6005 |
in the real compile phase, not in the pre-pass, where the whole group may |
in the real compile phase, not in the pre-pass, where the whole group may |
| 6006 |
not be available. */ |
not be available. */ |
| 6233 |
} |
} |
| 6234 |
|
|
| 6235 |
/* \k<name> or \k'name' is a back reference by name (Perl syntax). |
/* \k<name> or \k'name' is a back reference by name (Perl syntax). |
| 6236 |
We also support \k{name} (.NET syntax) */ |
We also support \k{name} (.NET syntax). */ |
| 6237 |
|
|
| 6238 |
if (-c == ESC_k && (ptr[1] == CHAR_LESS_THAN_SIGN || |
if (-c == ESC_k) |
|
ptr[1] == CHAR_APOSTROPHE || ptr[1] == CHAR_LEFT_CURLY_BRACKET)) |
|
| 6239 |
{ |
{ |
| 6240 |
|
if ((ptr[1] != CHAR_LESS_THAN_SIGN && |
| 6241 |
|
ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET)) |
| 6242 |
|
{ |
| 6243 |
|
*errorcodeptr = ERR69; |
| 6244 |
|
break; |
| 6245 |
|
} |
| 6246 |
is_recurse = FALSE; |
is_recurse = FALSE; |
| 6247 |
terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? |
terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? |
| 6248 |
CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)? |
CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)? |
| 6399 |
else firstbyte = reqbyte = REQ_NONE; |
else firstbyte = reqbyte = REQ_NONE; |
| 6400 |
} |
} |
| 6401 |
|
|
| 6402 |
/* firstbyte was previously set; we can set reqbyte only the length is |
/* firstbyte was previously set; we can set reqbyte only if the length is |
| 6403 |
1 or the matching is caseful. */ |
1 or the matching is caseful. */ |
| 6404 |
|
|
| 6405 |
else |
else |
| 6446 |
lookbehind TRUE if this is a lookbehind assertion |
lookbehind TRUE if this is a lookbehind assertion |
| 6447 |
reset_bracount TRUE to reset the count for each branch |
reset_bracount TRUE to reset the count for each branch |
| 6448 |
skipbytes skip this many bytes at start (for brackets and OP_COND) |
skipbytes skip this many bytes at start (for brackets and OP_COND) |
| 6449 |
|
cond_depth depth of nesting for conditional subpatterns |
| 6450 |
firstbyteptr place to put the first required character, or a negative number |
firstbyteptr place to put the first required character, or a negative number |
| 6451 |
reqbyteptr place to put the last required character, or a negative number |
reqbyteptr place to put the last required character, or a negative number |
| 6452 |
bcptr pointer to the chain of currently open branches |
bcptr pointer to the chain of currently open branches |
| 6460 |
static BOOL |
static BOOL |
| 6461 |
compile_regex(int options, uschar **codeptr, const uschar **ptrptr, |
compile_regex(int options, uschar **codeptr, const uschar **ptrptr, |
| 6462 |
int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes, |
int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes, |
| 6463 |
int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd, |
int cond_depth, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, |
| 6464 |
int *lengthptr) |
compile_data *cd, int *lengthptr) |
| 6465 |
{ |
{ |
| 6466 |
const uschar *ptr = *ptrptr; |
const uschar *ptr = *ptrptr; |
| 6467 |
uschar *code = *codeptr; |
uschar *code = *codeptr; |
| 6498 |
|
|
| 6499 |
/* 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 |
| 6500 |
so that we can detect them if (*ACCEPT) is encountered. This is also used to |
so that we can detect them if (*ACCEPT) is encountered. This is also used to |
| 6501 |
detect groups that contain recursive back references to themselves. Note that |
detect groups that contain recursive back references to themselves. Note that |
| 6502 |
only OP_CBRA need be tested here; changing this opcode to one of its variants, |
only OP_CBRA need be tested here; changing this opcode to one of its variants, |
| 6503 |
e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */ |
e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */ |
| 6504 |
|
|
| 6505 |
if (*code == OP_CBRA) |
if (*code == OP_CBRA) |
| 6540 |
into the length. */ |
into the length. */ |
| 6541 |
|
|
| 6542 |
if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte, |
if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte, |
| 6543 |
&branchreqbyte, &bc, cd, (lengthptr == NULL)? NULL : &length)) |
&branchreqbyte, &bc, cond_depth, cd, |
| 6544 |
|
(lengthptr == NULL)? NULL : &length)) |
| 6545 |
{ |
{ |
| 6546 |
*ptrptr = ptr; |
*ptrptr = ptr; |
| 6547 |
return FALSE; |
return FALSE; |
| 6791 |
|
|
| 6792 |
/* Other brackets */ |
/* Other brackets */ |
| 6793 |
|
|
| 6794 |
else if (op == OP_ASSERT || op == OP_ONCE || op == OP_COND) |
else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC || |
| 6795 |
|
op == OP_COND) |
| 6796 |
{ |
{ |
| 6797 |
if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; |
if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; |
| 6798 |
} |
} |
| 6896 |
|
|
| 6897 |
/* Other brackets */ |
/* Other brackets */ |
| 6898 |
|
|
| 6899 |
else if (op == OP_ASSERT || op == OP_ONCE) |
else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC) |
| 6900 |
{ |
{ |
| 6901 |
if (!is_startline(scode, bracket_map, backref_map)) return FALSE; |
if (!is_startline(scode, bracket_map, backref_map)) return FALSE; |
| 6902 |
} |
} |
| 6966 |
case OP_SCBRAPOS: |
case OP_SCBRAPOS: |
| 6967 |
case OP_ASSERT: |
case OP_ASSERT: |
| 6968 |
case OP_ONCE: |
case OP_ONCE: |
| 6969 |
|
case OP_ONCE_NC: |
| 6970 |
case OP_COND: |
case OP_COND: |
| 6971 |
if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0) |
if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0) |
| 6972 |
return -1; |
return -1; |
| 6976 |
case OP_EXACT: |
case OP_EXACT: |
| 6977 |
scode += 2; |
scode += 2; |
| 6978 |
/* Fall through */ |
/* Fall through */ |
| 6979 |
|
|
| 6980 |
case OP_CHAR: |
case OP_CHAR: |
| 6981 |
case OP_PLUS: |
case OP_PLUS: |
| 6982 |
case OP_MINPLUS: |
case OP_MINPLUS: |
| 6989 |
case OP_EXACTI: |
case OP_EXACTI: |
| 6990 |
scode += 2; |
scode += 2; |
| 6991 |
/* Fall through */ |
/* Fall through */ |
| 6992 |
|
|
| 6993 |
case OP_CHARI: |
case OP_CHARI: |
| 6994 |
case OP_PLUSI: |
case OP_PLUSI: |
| 6995 |
case OP_MINPLUSI: |
case OP_MINPLUSI: |
| 7149 |
|
|
| 7150 |
/* Can't support UTF8 unless PCRE has been compiled to include the code. The |
/* Can't support UTF8 unless PCRE has been compiled to include the code. The |
| 7151 |
return of an error code from _pcre_valid_utf8() is a new feature, introduced in |
return of an error code from _pcre_valid_utf8() is a new feature, introduced in |
| 7152 |
release 8.13. It is passed back from pcre_[dfa_]exec(), but at the moment is |
release 8.13. It is passed back from pcre_[dfa_]exec(), but at the moment is |
| 7153 |
not used here. */ |
not used here. */ |
| 7154 |
|
|
| 7155 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 7179 |
|
|
| 7180 |
/* Check validity of \R options. */ |
/* Check validity of \R options. */ |
| 7181 |
|
|
| 7182 |
switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) |
if ((options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) == |
| 7183 |
|
(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) |
| 7184 |
{ |
{ |
| 7185 |
case 0: |
errorcode = ERR56; |
| 7186 |
case PCRE_BSR_ANYCRLF: |
goto PCRE_EARLY_ERROR_RETURN; |
|
case PCRE_BSR_UNICODE: |
|
|
break; |
|
|
default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN; |
|
| 7187 |
} |
} |
| 7188 |
|
|
| 7189 |
/* Handle different types of newline. The three bits give seven cases. The |
/* Handle different types of newline. The three bits give seven cases. The |
| 7268 |
ptr += skipatstart; |
ptr += skipatstart; |
| 7269 |
code = cworkspace; |
code = cworkspace; |
| 7270 |
*code = OP_BRA; |
*code = OP_BRA; |
| 7271 |
(void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE, |
(void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE, |
| 7272 |
FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length); |
FALSE, 0, 0, &firstbyte, &reqbyte, NULL, cd, &length); |
| 7273 |
if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN; |
if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN; |
| 7274 |
|
|
| 7275 |
DPRINTF(("end pre-compile: length=%d workspace=%d\n", length, |
DPRINTF(("end pre-compile: length=%d workspace=%d\n", length, |
| 7342 |
ptr = (const uschar *)pattern + skipatstart; |
ptr = (const uschar *)pattern + skipatstart; |
| 7343 |
code = (uschar *)codestart; |
code = (uschar *)codestart; |
| 7344 |
*code = OP_BRA; |
*code = OP_BRA; |
| 7345 |
(void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, |
(void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0, |
| 7346 |
&firstbyte, &reqbyte, NULL, cd, NULL); |
&firstbyte, &reqbyte, NULL, cd, NULL); |
| 7347 |
re->top_bracket = cd->bracount; |
re->top_bracket = cd->bracount; |
| 7348 |
re->top_backref = cd->top_backref; |
re->top_backref = cd->top_backref; |
| 7349 |
re->flags = cd->external_flags; |
re->flags = cd->external_flags; |
| 7350 |
|
|
| 7351 |
if (cd->had_accept) reqbyte = -1; /* Must disable after (*ACCEPT) */ |
if (cd->had_accept) reqbyte = REQ_NONE; /* Must disable after (*ACCEPT) */ |
| 7352 |
|
|
| 7353 |
/* If not reached end of pattern on success, there's an excess bracket. */ |
/* If not reached end of pattern on success, there's an excess bracket. */ |
| 7354 |
|
|