| 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 |
|
/* 70 */ |
| 414 |
|
"internal error: unknown opcode in find_fixedlength()\0" |
| 415 |
; |
; |
| 416 |
|
|
| 417 |
/* Table to identify digits and hex digits. This is used when compiling |
/* Table to identify digits and hex digits. This is used when compiling |
| 548 |
/* Definition to allow mutual recursion */ |
/* Definition to allow mutual recursion */ |
| 549 |
|
|
| 550 |
static BOOL |
static BOOL |
| 551 |
compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int, |
compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int, |
| 552 |
int *, int *, branch_chain *, compile_data *, int *); |
int *, int *, branch_chain *, compile_data *, int *); |
| 553 |
|
|
| 554 |
|
|
| 580 |
|
|
| 581 |
|
|
| 582 |
/************************************************* |
/************************************************* |
| 583 |
|
* Check for counted repeat * |
| 584 |
|
*************************************************/ |
| 585 |
|
|
| 586 |
|
/* This function is called when a '{' is encountered in a place where it might |
| 587 |
|
start a quantifier. It looks ahead to see if it really is a quantifier or not. |
| 588 |
|
It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd} |
| 589 |
|
where the ddds are digits. |
| 590 |
|
|
| 591 |
|
Arguments: |
| 592 |
|
p pointer to the first char after '{' |
| 593 |
|
|
| 594 |
|
Returns: TRUE or FALSE |
| 595 |
|
*/ |
| 596 |
|
|
| 597 |
|
static BOOL |
| 598 |
|
is_counted_repeat(const uschar *p) |
| 599 |
|
{ |
| 600 |
|
if ((digitab[*p++] & ctype_digit) == 0) return FALSE; |
| 601 |
|
while ((digitab[*p] & ctype_digit) != 0) p++; |
| 602 |
|
if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; |
| 603 |
|
|
| 604 |
|
if (*p++ != CHAR_COMMA) return FALSE; |
| 605 |
|
if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; |
| 606 |
|
|
| 607 |
|
if ((digitab[*p++] & ctype_digit) == 0) return FALSE; |
| 608 |
|
while ((digitab[*p] & ctype_digit) != 0) p++; |
| 609 |
|
|
| 610 |
|
return (*p == CHAR_RIGHT_CURLY_BRACKET); |
| 611 |
|
} |
| 612 |
|
|
| 613 |
|
|
| 614 |
|
|
| 615 |
|
/************************************************* |
| 616 |
* Handle escapes * |
* Handle escapes * |
| 617 |
*************************************************/ |
*************************************************/ |
| 618 |
|
|
| 678 |
|
|
| 679 |
case CHAR_l: |
case CHAR_l: |
| 680 |
case CHAR_L: |
case CHAR_L: |
| 681 |
|
*errorcodeptr = ERR37; |
| 682 |
|
break; |
| 683 |
|
|
| 684 |
case CHAR_u: |
case CHAR_u: |
| 685 |
|
if ((options & PCRE_JAVASCRIPT_COMPAT) != 0) |
| 686 |
|
{ |
| 687 |
|
/* In JavaScript, \u must be followed by four hexadecimal numbers. |
| 688 |
|
Otherwise it is a lowercase u letter. */ |
| 689 |
|
if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0 |
| 690 |
|
&& (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0) |
| 691 |
|
{ |
| 692 |
|
c = 0; |
| 693 |
|
for (i = 0; i < 4; ++i) |
| 694 |
|
{ |
| 695 |
|
register int cc = *(++ptr); |
| 696 |
|
#ifndef EBCDIC /* ASCII/UTF-8 coding */ |
| 697 |
|
if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ |
| 698 |
|
c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); |
| 699 |
|
#else /* EBCDIC coding */ |
| 700 |
|
if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ |
| 701 |
|
c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); |
| 702 |
|
#endif |
| 703 |
|
} |
| 704 |
|
} |
| 705 |
|
} |
| 706 |
|
else |
| 707 |
|
*errorcodeptr = ERR37; |
| 708 |
|
break; |
| 709 |
|
|
| 710 |
case CHAR_U: |
case CHAR_U: |
| 711 |
*errorcodeptr = ERR37; |
/* In JavaScript, \U is an uppercase U letter. */ |
| 712 |
|
if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37; |
| 713 |
break; |
break; |
| 714 |
|
|
| 715 |
/* \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 |
| 716 |
|
class, \g must be followed by one of a number of specific things: |
| 717 |
|
|
| 718 |
(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 |
| 719 |
backreference. If negative, it is a relative backreference. This is a Perl |
backreference. If negative, it is a relative backreference. This is a Perl |
| 730 |
the -ESC_g code (cf \k). */ |
the -ESC_g code (cf \k). */ |
| 731 |
|
|
| 732 |
case CHAR_g: |
case CHAR_g: |
| 733 |
|
if (isclass) break; |
| 734 |
if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE) |
if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE) |
| 735 |
{ |
{ |
| 736 |
c = -ESC_g; |
c = -ESC_g; |
| 859 |
treated as a data character. */ |
treated as a data character. */ |
| 860 |
|
|
| 861 |
case CHAR_x: |
case CHAR_x: |
| 862 |
|
if ((options & PCRE_JAVASCRIPT_COMPAT) != 0) |
| 863 |
|
{ |
| 864 |
|
/* In JavaScript, \x must be followed by two hexadecimal numbers. |
| 865 |
|
Otherwise it is a lowercase x letter. */ |
| 866 |
|
if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0) |
| 867 |
|
{ |
| 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 |
|
|
| 1461 |
|
|
| 1462 |
Arguments: |
Arguments: |
| 1463 |
code pointer to the start of the group |
code pointer to the start of the group |
|
options pointer to external options |
|
|
optbit the option bit whose changing is significant, or |
|
|
zero if none are |
|
| 1464 |
skipassert TRUE if certain assertions are to be skipped |
skipassert TRUE if certain assertions are to be skipped |
| 1465 |
|
|
| 1466 |
Returns: pointer to the first significant opcode |
Returns: pointer to the first significant opcode |
| 1467 |
*/ |
*/ |
| 1468 |
|
|
| 1469 |
static const uschar* |
static const uschar* |
| 1470 |
first_significant_code(const uschar *code, int *options, int optbit, |
first_significant_code(const uschar *code, BOOL skipassert) |
|
BOOL skipassert) |
|
| 1471 |
{ |
{ |
| 1472 |
for (;;) |
for (;;) |
| 1473 |
{ |
{ |
| 1522 |
|
|
| 1523 |
Arguments: |
Arguments: |
| 1524 |
code points to the start of the pattern (the bracket) |
code points to the start of the pattern (the bracket) |
| 1525 |
options the compiling options |
utf8 TRUE in UTF-8 mode |
| 1526 |
atend TRUE if called when the pattern is complete |
atend TRUE if called when the pattern is complete |
| 1527 |
cd the "compile data" structure |
cd the "compile data" structure |
| 1528 |
|
|
| 1529 |
Returns: the fixed length, |
Returns: the fixed length, |
| 1530 |
or -1 if there is no fixed length, |
or -1 if there is no fixed length, |
| 1531 |
or -2 if \C was encountered |
or -2 if \C was encountered (in UTF-8 mode only) |
| 1532 |
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 |
| 1533 |
|
or -4 if an unknown opcode was encountered (internal error) |
| 1534 |
*/ |
*/ |
| 1535 |
|
|
| 1536 |
static int |
static int |
| 1537 |
find_fixedlength(uschar *code, int options, BOOL atend, compile_data *cd) |
find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd) |
| 1538 |
{ |
{ |
| 1539 |
int length = -1; |
int length = -1; |
| 1540 |
|
|
| 1551 |
register int op = *cc; |
register int op = *cc; |
| 1552 |
switch (op) |
switch (op) |
| 1553 |
{ |
{ |
| 1554 |
|
/* We only need to continue for OP_CBRA (normal capturing bracket) and |
| 1555 |
|
OP_BRA (normal non-capturing bracket) because the other variants of these |
| 1556 |
|
opcodes are all concerned with unlimited repeated groups, which of course |
| 1557 |
|
are not of fixed length. */ |
| 1558 |
|
|
| 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), options, 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; |
| 1566 |
branchlength += d; |
branchlength += d; |
| 1567 |
do cc += GET(cc, 1); while (*cc == OP_ALT); |
do cc += GET(cc, 1); while (*cc == OP_ALT); |
| 1568 |
cc += 1 + LINK_SIZE; |
cc += 1 + LINK_SIZE; |
| 1569 |
break; |
break; |
| 1570 |
|
|
| 1571 |
/* Reached end of a branch; if it's a ket it is the end of a nested |
/* Reached end of a branch; if it's a ket it is the end of a nested call. |
| 1572 |
call. If it's ALT it is an alternation in a nested call. If it is |
If it's ALT it is an alternation in a nested call. An ACCEPT is effectively |
| 1573 |
END it's the end of the outer call. All can be handled by the same code. */ |
an ALT. If it is END it's the end of the outer call. All can be handled by |
| 1574 |
|
the same code. Note that we must not include the OP_KETRxxx opcodes here, |
| 1575 |
|
because they all imply an unlimited repeat. */ |
| 1576 |
|
|
| 1577 |
case OP_ALT: |
case OP_ALT: |
| 1578 |
case OP_KET: |
case OP_KET: |
|
case OP_KETRMAX: |
|
|
case OP_KETRMIN: |
|
| 1579 |
case OP_END: |
case OP_END: |
| 1580 |
|
case OP_ACCEPT: |
| 1581 |
|
case OP_ASSERT_ACCEPT: |
| 1582 |
if (length < 0) length = branchlength; |
if (length < 0) length = branchlength; |
| 1583 |
else if (length != branchlength) return -1; |
else if (length != branchlength) return -1; |
| 1584 |
if (*cc != OP_ALT) return length; |
if (*cc != OP_ALT) return length; |
| 1595 |
cs = ce = (uschar *)cd->start_code + GET(cc, 1); /* Start subpattern */ |
cs = ce = (uschar *)cd->start_code + GET(cc, 1); /* Start subpattern */ |
| 1596 |
do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ |
do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ |
| 1597 |
if (cc > cs && cc < ce) return -1; /* Recursion */ |
if (cc > cs && cc < ce) return -1; /* Recursion */ |
| 1598 |
d = find_fixedlength(cs + 2, options, atend, cd); |
d = find_fixedlength(cs + 2, utf8, atend, cd); |
| 1599 |
if (d < 0) return d; |
if (d < 0) return d; |
| 1600 |
branchlength += d; |
branchlength += d; |
| 1601 |
cc += 1 + LINK_SIZE; |
cc += 1 + LINK_SIZE; |
| 1612 |
|
|
| 1613 |
/* Skip over things that don't match chars */ |
/* Skip over things that don't match chars */ |
| 1614 |
|
|
| 1615 |
case OP_REVERSE: |
case OP_MARK: |
| 1616 |
case OP_CREF: |
case OP_PRUNE_ARG: |
| 1617 |
case OP_NCREF: |
case OP_SKIP_ARG: |
| 1618 |
case OP_RREF: |
case OP_THEN_ARG: |
| 1619 |
case OP_NRREF: |
cc += cc[1] + _pcre_OP_lengths[*cc]; |
| 1620 |
case OP_DEF: |
break; |
| 1621 |
|
|
| 1622 |
case OP_CALLOUT: |
case OP_CALLOUT: |
|
case OP_SOD: |
|
|
case OP_SOM: |
|
|
case OP_SET_SOM: |
|
|
case OP_EOD: |
|
|
case OP_EODN: |
|
| 1623 |
case OP_CIRC: |
case OP_CIRC: |
| 1624 |
case OP_CIRCM: |
case OP_CIRCM: |
| 1625 |
|
case OP_CLOSE: |
| 1626 |
|
case OP_COMMIT: |
| 1627 |
|
case OP_CREF: |
| 1628 |
|
case OP_DEF: |
| 1629 |
case OP_DOLL: |
case OP_DOLL: |
| 1630 |
case OP_DOLLM: |
case OP_DOLLM: |
| 1631 |
|
case OP_EOD: |
| 1632 |
|
case OP_EODN: |
| 1633 |
|
case OP_FAIL: |
| 1634 |
|
case OP_NCREF: |
| 1635 |
|
case OP_NRREF: |
| 1636 |
case OP_NOT_WORD_BOUNDARY: |
case OP_NOT_WORD_BOUNDARY: |
| 1637 |
|
case OP_PRUNE: |
| 1638 |
|
case OP_REVERSE: |
| 1639 |
|
case OP_RREF: |
| 1640 |
|
case OP_SET_SOM: |
| 1641 |
|
case OP_SKIP: |
| 1642 |
|
case OP_SOD: |
| 1643 |
|
case OP_SOM: |
| 1644 |
|
case OP_THEN: |
| 1645 |
case OP_WORD_BOUNDARY: |
case OP_WORD_BOUNDARY: |
| 1646 |
cc += _pcre_OP_lengths[*cc]; |
cc += _pcre_OP_lengths[*cc]; |
| 1647 |
break; |
break; |
| 1651 |
case OP_CHAR: |
case OP_CHAR: |
| 1652 |
case OP_CHARI: |
case OP_CHARI: |
| 1653 |
case OP_NOT: |
case OP_NOT: |
| 1654 |
case OP_NOTI: |
case OP_NOTI: |
| 1655 |
branchlength++; |
branchlength++; |
| 1656 |
cc += 2; |
cc += 2; |
| 1657 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 1658 |
if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0) |
if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
|
cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
|
| 1659 |
#endif |
#endif |
| 1660 |
break; |
break; |
| 1661 |
|
|
| 1663 |
need to skip over a multibyte character in UTF8 mode. */ |
need to skip over a multibyte character in UTF8 mode. */ |
| 1664 |
|
|
| 1665 |
case OP_EXACT: |
case OP_EXACT: |
| 1666 |
|
case OP_EXACTI: |
| 1667 |
|
case OP_NOTEXACT: |
| 1668 |
|
case OP_NOTEXACTI: |
| 1669 |
branchlength += GET2(cc,1); |
branchlength += GET2(cc,1); |
| 1670 |
cc += 4; |
cc += 4; |
| 1671 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 1672 |
if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0) |
if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
|
cc += _pcre_utf8_table4[cc[-1] & 0x3f]; |
|
| 1673 |
#endif |
#endif |
| 1674 |
break; |
break; |
| 1675 |
|
|
| 1686 |
cc += 2; |
cc += 2; |
| 1687 |
/* Fall through */ |
/* Fall through */ |
| 1688 |
|
|
| 1689 |
|
case OP_HSPACE: |
| 1690 |
|
case OP_VSPACE: |
| 1691 |
|
case OP_NOT_HSPACE: |
| 1692 |
|
case OP_NOT_VSPACE: |
| 1693 |
case OP_NOT_DIGIT: |
case OP_NOT_DIGIT: |
| 1694 |
case OP_DIGIT: |
case OP_DIGIT: |
| 1695 |
case OP_NOT_WHITESPACE: |
case OP_NOT_WHITESPACE: |
| 1702 |
cc++; |
cc++; |
| 1703 |
break; |
break; |
| 1704 |
|
|
| 1705 |
/* The single-byte matcher isn't allowed */ |
/* The single-byte matcher isn't allowed. This only happens in UTF-8 mode; |
| 1706 |
|
otherwise \C is coded as OP_ALLANY. */ |
| 1707 |
|
|
| 1708 |
case OP_ANYBYTE: |
case OP_ANYBYTE: |
| 1709 |
return -2; |
return -2; |
| 1722 |
|
|
| 1723 |
switch (*cc) |
switch (*cc) |
| 1724 |
{ |
{ |
| 1725 |
|
case OP_CRPLUS: |
| 1726 |
|
case OP_CRMINPLUS: |
| 1727 |
case OP_CRSTAR: |
case OP_CRSTAR: |
| 1728 |
case OP_CRMINSTAR: |
case OP_CRMINSTAR: |
| 1729 |
case OP_CRQUERY: |
case OP_CRQUERY: |
| 1744 |
|
|
| 1745 |
/* Anything else is variable length */ |
/* Anything else is variable length */ |
| 1746 |
|
|
| 1747 |
default: |
case OP_ANYNL: |
| 1748 |
|
case OP_BRAMINZERO: |
| 1749 |
|
case OP_BRAPOS: |
| 1750 |
|
case OP_BRAPOSZERO: |
| 1751 |
|
case OP_BRAZERO: |
| 1752 |
|
case OP_CBRAPOS: |
| 1753 |
|
case OP_EXTUNI: |
| 1754 |
|
case OP_KETRMAX: |
| 1755 |
|
case OP_KETRMIN: |
| 1756 |
|
case OP_KETRPOS: |
| 1757 |
|
case OP_MINPLUS: |
| 1758 |
|
case OP_MINPLUSI: |
| 1759 |
|
case OP_MINQUERY: |
| 1760 |
|
case OP_MINQUERYI: |
| 1761 |
|
case OP_MINSTAR: |
| 1762 |
|
case OP_MINSTARI: |
| 1763 |
|
case OP_MINUPTO: |
| 1764 |
|
case OP_MINUPTOI: |
| 1765 |
|
case OP_NOTMINPLUS: |
| 1766 |
|
case OP_NOTMINPLUSI: |
| 1767 |
|
case OP_NOTMINQUERY: |
| 1768 |
|
case OP_NOTMINQUERYI: |
| 1769 |
|
case OP_NOTMINSTAR: |
| 1770 |
|
case OP_NOTMINSTARI: |
| 1771 |
|
case OP_NOTMINUPTO: |
| 1772 |
|
case OP_NOTMINUPTOI: |
| 1773 |
|
case OP_NOTPLUS: |
| 1774 |
|
case OP_NOTPLUSI: |
| 1775 |
|
case OP_NOTPOSPLUS: |
| 1776 |
|
case OP_NOTPOSPLUSI: |
| 1777 |
|
case OP_NOTPOSQUERY: |
| 1778 |
|
case OP_NOTPOSQUERYI: |
| 1779 |
|
case OP_NOTPOSSTAR: |
| 1780 |
|
case OP_NOTPOSSTARI: |
| 1781 |
|
case OP_NOTPOSUPTO: |
| 1782 |
|
case OP_NOTPOSUPTOI: |
| 1783 |
|
case OP_NOTQUERY: |
| 1784 |
|
case OP_NOTQUERYI: |
| 1785 |
|
case OP_NOTSTAR: |
| 1786 |
|
case OP_NOTSTARI: |
| 1787 |
|
case OP_NOTUPTO: |
| 1788 |
|
case OP_NOTUPTOI: |
| 1789 |
|
case OP_PLUS: |
| 1790 |
|
case OP_PLUSI: |
| 1791 |
|
case OP_POSPLUS: |
| 1792 |
|
case OP_POSPLUSI: |
| 1793 |
|
case OP_POSQUERY: |
| 1794 |
|
case OP_POSQUERYI: |
| 1795 |
|
case OP_POSSTAR: |
| 1796 |
|
case OP_POSSTARI: |
| 1797 |
|
case OP_POSUPTO: |
| 1798 |
|
case OP_POSUPTOI: |
| 1799 |
|
case OP_QUERY: |
| 1800 |
|
case OP_QUERYI: |
| 1801 |
|
case OP_REF: |
| 1802 |
|
case OP_REFI: |
| 1803 |
|
case OP_SBRA: |
| 1804 |
|
case OP_SBRAPOS: |
| 1805 |
|
case OP_SCBRA: |
| 1806 |
|
case OP_SCBRAPOS: |
| 1807 |
|
case OP_SCOND: |
| 1808 |
|
case OP_SKIPZERO: |
| 1809 |
|
case OP_STAR: |
| 1810 |
|
case OP_STARI: |
| 1811 |
|
case OP_TYPEMINPLUS: |
| 1812 |
|
case OP_TYPEMINQUERY: |
| 1813 |
|
case OP_TYPEMINSTAR: |
| 1814 |
|
case OP_TYPEMINUPTO: |
| 1815 |
|
case OP_TYPEPLUS: |
| 1816 |
|
case OP_TYPEPOSPLUS: |
| 1817 |
|
case OP_TYPEPOSQUERY: |
| 1818 |
|
case OP_TYPEPOSSTAR: |
| 1819 |
|
case OP_TYPEPOSUPTO: |
| 1820 |
|
case OP_TYPEQUERY: |
| 1821 |
|
case OP_TYPESTAR: |
| 1822 |
|
case OP_TYPEUPTO: |
| 1823 |
|
case OP_UPTO: |
| 1824 |
|
case OP_UPTOI: |
| 1825 |
return -1; |
return -1; |
| 1826 |
|
|
| 1827 |
|
/* Catch unrecognized opcodes so that when new ones are added they |
| 1828 |
|
are not forgotten, as has happened in the past. */ |
| 1829 |
|
|
| 1830 |
|
default: |
| 1831 |
|
return -4; |
| 1832 |
} |
} |
| 1833 |
} |
} |
| 1834 |
/* Control never gets here */ |
/* Control never gets here */ |
| 1861 |
for (;;) |
for (;;) |
| 1862 |
{ |
{ |
| 1863 |
register int c = *code; |
register int c = *code; |
| 1864 |
|
|
| 1865 |
if (c == OP_END) return NULL; |
if (c == OP_END) return NULL; |
| 1866 |
|
|
| 1867 |
/* 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 |
| 1880 |
|
|
| 1881 |
/* Handle capturing bracket */ |
/* Handle capturing bracket */ |
| 1882 |
|
|
| 1883 |
else if (c == OP_CBRA) |
else if (c == OP_CBRA || c == OP_SCBRA || |
| 1884 |
|
c == OP_CBRAPOS || c == OP_SCBRAPOS) |
| 1885 |
{ |
{ |
| 1886 |
int n = GET2(code, 1+LINK_SIZE); |
int n = GET2(code, 1+LINK_SIZE); |
| 1887 |
if (n == number) return (uschar *)code; |
if (n == number) return (uschar *)code; |
| 1923 |
break; |
break; |
| 1924 |
|
|
| 1925 |
case OP_THEN_ARG: |
case OP_THEN_ARG: |
| 1926 |
code += code[1+LINK_SIZE]; |
code += code[1]; |
| 1927 |
break; |
break; |
| 1928 |
} |
} |
| 1929 |
|
|
| 2042 |
break; |
break; |
| 2043 |
|
|
| 2044 |
case OP_THEN_ARG: |
case OP_THEN_ARG: |
| 2045 |
code += code[1+LINK_SIZE]; |
code += code[1]; |
| 2046 |
break; |
break; |
| 2047 |
} |
} |
| 2048 |
|
|
| 2123 |
compile_data *cd) |
compile_data *cd) |
| 2124 |
{ |
{ |
| 2125 |
register int c; |
register int c; |
| 2126 |
for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE); |
for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE); |
| 2127 |
code < endcode; |
code < endcode; |
| 2128 |
code = first_significant_code(code + _pcre_OP_lengths[c], NULL, 0, TRUE)) |
code = first_significant_code(code + _pcre_OP_lengths[c], TRUE)) |
| 2129 |
{ |
{ |
| 2130 |
const uschar *ccode; |
const uschar *ccode; |
| 2131 |
|
|
| 2141 |
continue; |
continue; |
| 2142 |
} |
} |
| 2143 |
|
|
|
/* Groups with zero repeats can of course be empty; skip them. */ |
|
|
|
|
|
if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO) |
|
|
{ |
|
|
code += _pcre_OP_lengths[c]; |
|
|
do code += GET(code, 1); while (*code == OP_ALT); |
|
|
c = *code; |
|
|
continue; |
|
|
} |
|
|
|
|
| 2144 |
/* For a recursion/subroutine call, if its end has been reached, which |
/* For a recursion/subroutine call, if its end has been reached, which |
| 2145 |
implies a subroutine call, we can scan it. */ |
implies a backward reference subroutine call, we can scan it. If it's a |
| 2146 |
|
forward reference subroutine call, we can't. To detect forward reference |
| 2147 |
|
we have to scan up the list that is kept in the workspace. This function is |
| 2148 |
|
called only when doing the real compile, not during the pre-compile that |
| 2149 |
|
measures the size of the compiled pattern. */ |
| 2150 |
|
|
| 2151 |
if (c == OP_RECURSE) |
if (c == OP_RECURSE) |
| 2152 |
{ |
{ |
| 2153 |
BOOL empty_branch = FALSE; |
const uschar *scode; |
| 2154 |
const uschar *scode = cd->start_code + GET(code, 1); |
BOOL empty_branch; |
| 2155 |
|
|
| 2156 |
|
/* Test for forward reference */ |
| 2157 |
|
|
| 2158 |
|
for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE) |
| 2159 |
|
if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE; |
| 2160 |
|
|
| 2161 |
|
/* Not a forward reference, test for completed backward reference */ |
| 2162 |
|
|
| 2163 |
|
empty_branch = FALSE; |
| 2164 |
|
scode = cd->start_code + GET(code, 1); |
| 2165 |
if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ |
if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ |
| 2166 |
|
|
| 2167 |
|
/* Completed backwards reference */ |
| 2168 |
|
|
| 2169 |
do |
do |
| 2170 |
{ |
{ |
| 2171 |
if (could_be_empty_branch(scode, endcode, utf8, cd)) |
if (could_be_empty_branch(scode, endcode, utf8, cd)) |
| 2176 |
scode += GET(scode, 1); |
scode += GET(scode, 1); |
| 2177 |
} |
} |
| 2178 |
while (*scode == OP_ALT); |
while (*scode == OP_ALT); |
| 2179 |
|
|
| 2180 |
if (!empty_branch) return FALSE; /* All branches are non-empty */ |
if (!empty_branch) return FALSE; /* All branches are non-empty */ |
| 2181 |
continue; |
continue; |
| 2182 |
} |
} |
| 2183 |
|
|
| 2184 |
|
/* Groups with zero repeats can of course be empty; skip them. */ |
| 2185 |
|
|
| 2186 |
|
if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO || |
| 2187 |
|
c == OP_BRAPOSZERO) |
| 2188 |
|
{ |
| 2189 |
|
code += _pcre_OP_lengths[c]; |
| 2190 |
|
do code += GET(code, 1); while (*code == OP_ALT); |
| 2191 |
|
c = *code; |
| 2192 |
|
continue; |
| 2193 |
|
} |
| 2194 |
|
|
| 2195 |
|
/* A nested group that is already marked as "could be empty" can just be |
| 2196 |
|
skipped. */ |
| 2197 |
|
|
| 2198 |
|
if (c == OP_SBRA || c == OP_SBRAPOS || |
| 2199 |
|
c == OP_SCBRA || c == OP_SCBRAPOS) |
| 2200 |
|
{ |
| 2201 |
|
do code += GET(code, 1); while (*code == OP_ALT); |
| 2202 |
|
c = *code; |
| 2203 |
|
continue; |
| 2204 |
|
} |
| 2205 |
|
|
| 2206 |
/* For other groups, scan the branches. */ |
/* For other groups, scan the branches. */ |
| 2207 |
|
|
| 2208 |
if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND) |
if (c == OP_BRA || c == OP_BRAPOS || |
| 2209 |
|
c == OP_CBRA || c == OP_CBRAPOS || |
| 2210 |
|
c == OP_ONCE || c == OP_ONCE_NC || |
| 2211 |
|
c == OP_COND) |
| 2212 |
{ |
{ |
| 2213 |
BOOL empty_branch; |
BOOL empty_branch; |
| 2214 |
if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ |
if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ |
| 2337 |
case OP_KET: |
case OP_KET: |
| 2338 |
case OP_KETRMAX: |
case OP_KETRMAX: |
| 2339 |
case OP_KETRMIN: |
case OP_KETRMIN: |
| 2340 |
|
case OP_KETRPOS: |
| 2341 |
case OP_ALT: |
case OP_ALT: |
| 2342 |
return TRUE; |
return TRUE; |
| 2343 |
|
|
| 2380 |
break; |
break; |
| 2381 |
|
|
| 2382 |
case OP_THEN_ARG: |
case OP_THEN_ARG: |
| 2383 |
code += code[1+LINK_SIZE]; |
code += code[1]; |
| 2384 |
break; |
break; |
| 2385 |
|
|
| 2386 |
/* None of the remaining opcodes are required to match a character. */ |
/* None of the remaining opcodes are required to match a character. */ |
| 2403 |
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 |
| 2404 |
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, |
| 2405 |
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. |
| 2406 |
|
This function is called only during the real compile, not during the |
| 2407 |
|
pre-compile. |
| 2408 |
|
|
| 2409 |
Arguments: |
Arguments: |
| 2410 |
code points to start of the recursion |
code points to start of the recursion |
| 2455 |
"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, |
| 2456 |
I think. |
I think. |
| 2457 |
|
|
| 2458 |
|
A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not. |
| 2459 |
|
It seems that the appearance of a nested POSIX class supersedes an apparent |
| 2460 |
|
external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or |
| 2461 |
|
a digit. |
| 2462 |
|
|
| 2463 |
|
In Perl, unescaped square brackets may also appear as part of class names. For |
| 2464 |
|
example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for |
| 2465 |
|
[:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not |
| 2466 |
|
seem right at all. PCRE does not allow closing square brackets in POSIX class |
| 2467 |
|
names. |
| 2468 |
|
|
| 2469 |
Arguments: |
Arguments: |
| 2470 |
ptr pointer to the initial [ |
ptr pointer to the initial [ |
| 2471 |
endptr where to return the end pointer |
endptr where to return the end pointer |
| 2480 |
terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ |
terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ |
| 2481 |
for (++ptr; *ptr != 0; ptr++) |
for (++ptr; *ptr != 0; ptr++) |
| 2482 |
{ |
{ |
| 2483 |
if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else |
if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) |
| 2484 |
|
ptr++; |
| 2485 |
|
else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE; |
| 2486 |
|
else |
| 2487 |
{ |
{ |
|
if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE; |
|
| 2488 |
if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) |
if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) |
| 2489 |
{ |
{ |
| 2490 |
*endptr = ptr; |
*endptr = ptr; |
| 2491 |
return TRUE; |
return TRUE; |
| 2492 |
} |
} |
| 2493 |
|
if (*ptr == CHAR_LEFT_SQUARE_BRACKET && |
| 2494 |
|
(ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || |
| 2495 |
|
ptr[1] == CHAR_EQUALS_SIGN) && |
| 2496 |
|
check_posix_syntax(ptr, endptr)) |
| 2497 |
|
return FALSE; |
| 2498 |
} |
} |
| 2499 |
} |
} |
| 2500 |
return FALSE; |
return FALSE; |
| 2905 |
return (c != cd->fcc[next]); /* Non-UTF-8 mode */ |
return (c != cd->fcc[next]); /* Non-UTF-8 mode */ |
| 2906 |
|
|
| 2907 |
/* For OP_NOT and OP_NOTI, the data is always a single-byte character. These |
/* For OP_NOT and OP_NOTI, the data is always a single-byte character. These |
| 2908 |
opcodes are not used for multi-byte characters, because they are coded using |
opcodes are not used for multi-byte characters, because they are coded using |
| 2909 |
an XCLASS instead. */ |
an XCLASS instead. */ |
| 2910 |
|
|
| 2911 |
case OP_NOT: |
case OP_NOT: |
| 2912 |
return (c = *previous) == next; |
return (c = *previous) == next; |
| 2913 |
|
|
| 2914 |
case OP_NOTI: |
case OP_NOTI: |
| 2915 |
if ((c = *previous) == next) return TRUE; |
if ((c = *previous) == next) return TRUE; |
| 2916 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 2917 |
if (utf8) |
if (utf8) |
| 3204 |
firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) |
firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) |
| 3205 |
reqbyteptr set to the last literal character required, else < 0 |
reqbyteptr set to the last literal character required, else < 0 |
| 3206 |
bcptr points to current branch chain |
bcptr points to current branch chain |
| 3207 |
|
cond_depth conditional nesting depth |
| 3208 |
cd contains pointers to tables etc. |
cd contains pointers to tables etc. |
| 3209 |
lengthptr NULL during the real compile phase |
lengthptr NULL during the real compile phase |
| 3210 |
points to length accumulator during pre-compile phase |
points to length accumulator during pre-compile phase |
| 3216 |
static BOOL |
static BOOL |
| 3217 |
compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr, |
compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr, |
| 3218 |
int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, |
int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, |
| 3219 |
compile_data *cd, int *lengthptr) |
int cond_depth, compile_data *cd, int *lengthptr) |
| 3220 |
{ |
{ |
| 3221 |
int repeat_type, op_type; |
int repeat_type, op_type; |
| 3222 |
int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ |
int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ |
| 3225 |
int firstbyte, reqbyte; |
int firstbyte, reqbyte; |
| 3226 |
int zeroreqbyte, zerofirstbyte; |
int zeroreqbyte, zerofirstbyte; |
| 3227 |
int req_caseopt, reqvary, tempreqvary; |
int req_caseopt, reqvary, tempreqvary; |
| 3228 |
int options = *optionsptr; |
int options = *optionsptr; /* May change dynamically */ |
| 3229 |
int after_manual_callout = 0; |
int after_manual_callout = 0; |
| 3230 |
int length_prevgroup = 0; |
int length_prevgroup = 0; |
| 3231 |
register int c; |
register int c; |
| 3243 |
uschar *save_hwm = NULL; |
uschar *save_hwm = NULL; |
| 3244 |
uschar classbits[32]; |
uschar classbits[32]; |
| 3245 |
|
|
| 3246 |
|
/* We can fish out the UTF-8 setting once and for all into a BOOL, but we |
| 3247 |
|
must not do this for other options (e.g. PCRE_EXTENDED) because they may change |
| 3248 |
|
dynamically as we process the pattern. */ |
| 3249 |
|
|
| 3250 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 3251 |
BOOL class_utf8; |
BOOL class_utf8; |
| 3252 |
BOOL utf8 = (options & PCRE_UTF8) != 0; |
BOOL utf8 = (options & PCRE_UTF8) != 0; |
| 3255 |
uschar utf8_char[6]; |
uschar utf8_char[6]; |
| 3256 |
#else |
#else |
| 3257 |
BOOL utf8 = FALSE; |
BOOL utf8 = FALSE; |
|
uschar *utf8_char = NULL; |
|
| 3258 |
#endif |
#endif |
| 3259 |
|
|
| 3260 |
#ifdef PCRE_DEBUG |
#ifdef PCRE_DEBUG |
| 3305 |
int subfirstbyte; |
int subfirstbyte; |
| 3306 |
int terminator; |
int terminator; |
| 3307 |
int mclength; |
int mclength; |
| 3308 |
|
int tempbracount; |
| 3309 |
uschar mcbuffer[8]; |
uschar mcbuffer[8]; |
| 3310 |
|
|
| 3311 |
/* Get next byte in the pattern */ |
/* Get next byte in the pattern */ |
| 3353 |
} |
} |
| 3354 |
|
|
| 3355 |
*lengthptr += (int)(code - last_code); |
*lengthptr += (int)(code - last_code); |
| 3356 |
DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c)); |
DPRINTF(("length=%d added %d c=%c\n", *lengthptr, (int)(code - last_code), |
| 3357 |
|
c)); |
| 3358 |
|
|
| 3359 |
/* If "previous" is set and it is not at the start of the work space, move |
/* If "previous" is set and it is not at the start of the work space, move |
| 3360 |
it back to there, in order to avoid filling up the work space. Otherwise, |
it back to there, in order to avoid filling up the work space. Otherwise, |
| 3428 |
previous_callout = NULL; |
previous_callout = NULL; |
| 3429 |
} |
} |
| 3430 |
|
|
| 3431 |
/* In extended mode, skip white space and comments */ |
/* In extended mode, skip white space and comments. */ |
| 3432 |
|
|
| 3433 |
if ((options & PCRE_EXTENDED) != 0) |
if ((options & PCRE_EXTENDED) != 0) |
| 3434 |
{ |
{ |
| 4397 |
op_type = 0; /* Default single-char op codes */ |
op_type = 0; /* Default single-char op codes */ |
| 4398 |
possessive_quantifier = FALSE; /* Default not possessive quantifier */ |
possessive_quantifier = FALSE; /* Default not possessive quantifier */ |
| 4399 |
|
|
| 4400 |
/* 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 |
| 4401 |
for an inserted OP_ONCE for the additional '+' extension. */ |
insert something before it. */ |
| 4402 |
|
|
| 4403 |
tempcode = previous; |
tempcode = previous; |
| 4404 |
|
|
| 4421 |
} |
} |
| 4422 |
else repeat_type = greedy_default; |
else repeat_type = greedy_default; |
| 4423 |
|
|
| 4424 |
|
/* If previous was a recursion call, wrap it in atomic brackets so that |
| 4425 |
|
previous becomes the atomic group. All recursions were so wrapped in the |
| 4426 |
|
past, but it no longer happens for non-repeated recursions. In fact, the |
| 4427 |
|
repeated ones could be re-implemented independently so as not to need this, |
| 4428 |
|
but for the moment we rely on the code for repeating groups. */ |
| 4429 |
|
|
| 4430 |
|
if (*previous == OP_RECURSE) |
| 4431 |
|
{ |
| 4432 |
|
memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE); |
| 4433 |
|
*previous = OP_ONCE; |
| 4434 |
|
PUT(previous, 1, 2 + 2*LINK_SIZE); |
| 4435 |
|
previous[2 + 2*LINK_SIZE] = OP_KET; |
| 4436 |
|
PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE); |
| 4437 |
|
code += 2 + 2 * LINK_SIZE; |
| 4438 |
|
length_prevgroup = 3 + 3*LINK_SIZE; |
| 4439 |
|
|
| 4440 |
|
/* When actually compiling, we need to check whether this was a forward |
| 4441 |
|
reference, and if so, adjust the offset. */ |
| 4442 |
|
|
| 4443 |
|
if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE) |
| 4444 |
|
{ |
| 4445 |
|
int offset = GET(cd->hwm, -LINK_SIZE); |
| 4446 |
|
if (offset == previous + 1 - cd->start_code) |
| 4447 |
|
PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE); |
| 4448 |
|
} |
| 4449 |
|
} |
| 4450 |
|
|
| 4451 |
|
/* Now handle repetition for the different types of item. */ |
| 4452 |
|
|
| 4453 |
/* If previous was a character match, abolish the item and generate a |
/* If previous was a character match, abolish the item and generate a |
| 4454 |
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 |
| 4455 |
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 |
| 4459 |
if (*previous == OP_CHAR || *previous == OP_CHARI) |
if (*previous == OP_CHAR || *previous == OP_CHARI) |
| 4460 |
{ |
{ |
| 4461 |
op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR; |
op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR; |
| 4462 |
|
|
| 4463 |
/* Deal with UTF-8 characters that take up more than one byte. It's |
/* Deal with UTF-8 characters that take up more than one byte. It's |
| 4464 |
easier to write this out separately than try to macrify it. Use c to |
easier to write this out separately than try to macrify it. Use c to |
| 4465 |
hold the length of the character in bytes, plus 0x80 to flag that it's a |
hold the length of the character in bytes, plus 0x80 to flag that it's a |
| 4504 |
/* If previous was a single negated character ([^a] or similar), we use |
/* If previous was a single negated character ([^a] or similar), we use |
| 4505 |
one of the special opcodes, replacing it. The code is shared with single- |
one of the special opcodes, replacing it. The code is shared with single- |
| 4506 |
character repeats by setting opt_type to add a suitable offset into |
character repeats by setting opt_type to add a suitable offset into |
| 4507 |
repeat_type. We can also test for auto-possessification. OP_NOT and OP_NOTI |
repeat_type. We can also test for auto-possessification. OP_NOT and OP_NOTI |
| 4508 |
are currently used only for single-byte chars. */ |
are currently used only for single-byte chars. */ |
| 4509 |
|
|
| 4510 |
else if (*previous == OP_NOT || *previous == OP_NOTI) |
else if (*previous == OP_NOT || *previous == OP_NOTI) |
| 4741 |
} |
} |
| 4742 |
|
|
| 4743 |
/* 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 |
| 4744 |
cases. */ |
cases. Note that at this point we can encounter only the "basic" bracket |
| 4745 |
|
opcodes such as BRA and CBRA, as this is the place where they get converted |
| 4746 |
|
into the more special varieties such as BRAPOS and SBRA. A test for >= |
| 4747 |
|
OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK, |
| 4748 |
|
ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow |
| 4749 |
|
repetition of assertions, but now it does, for Perl compatibility. */ |
| 4750 |
|
|
| 4751 |
else if (*previous == OP_BRA || *previous == OP_CBRA || |
else if (*previous >= OP_ASSERT && *previous <= OP_COND) |
|
*previous == OP_ONCE || *previous == OP_COND) |
|
| 4752 |
{ |
{ |
| 4753 |
register int i; |
register int i; |
|
int ketoffset = 0; |
|
| 4754 |
int len = (int)(code - previous); |
int len = (int)(code - previous); |
| 4755 |
uschar *bralink = NULL; |
uschar *bralink = NULL; |
| 4756 |
|
uschar *brazeroptr = NULL; |
| 4757 |
|
|
| 4758 |
/* Repeating a DEFINE group is pointless */ |
/* Repeating a DEFINE group is pointless, but Perl allows the syntax, so |
| 4759 |
|
we just ignore the repeat. */ |
| 4760 |
|
|
| 4761 |
if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF) |
if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF) |
| 4762 |
{ |
goto END_REPEAT; |
|
*errorcodeptr = ERR55; |
|
|
goto FAILED; |
|
|
} |
|
| 4763 |
|
|
| 4764 |
/* If the maximum repeat count is unlimited, find the end of the bracket |
/* There is no sense in actually repeating assertions. The only potential |
| 4765 |
by scanning through from the start, and compute the offset back to it |
use of repetition is in cases when the assertion is optional. Therefore, |
| 4766 |
from the current code pointer. */ |
if the minimum is greater than zero, just ignore the repeat. If the |
| 4767 |
|
maximum is not not zero or one, set it to 1. */ |
| 4768 |
if (repeat_max == -1) |
|
| 4769 |
{ |
if (*previous < OP_ONCE) /* Assertion */ |
| 4770 |
register uschar *ket = previous; |
{ |
| 4771 |
do ket += GET(ket, 1); while (*ket != OP_KET); |
if (repeat_min > 0) goto END_REPEAT; |
| 4772 |
ketoffset = (int)(code - ket); |
if (repeat_max < 0 || repeat_max > 1) repeat_max = 1; |
| 4773 |
} |
} |
| 4774 |
|
|
| 4775 |
/* 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 |
| 4790 |
** goto END_REPEAT; |
** goto END_REPEAT; |
| 4791 |
** } |
** } |
| 4792 |
|
|
| 4793 |
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 |
| 4794 |
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 |
| 4795 |
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 |
| 4796 |
groups are referenced, we cannot do this selectively. |
don't have a list of which groups are referenced, we cannot do this |
| 4797 |
|
selectively. |
| 4798 |
|
|
| 4799 |
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 |
| 4800 |
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 |
| 4814 |
*previous++ = OP_SKIPZERO; |
*previous++ = OP_SKIPZERO; |
| 4815 |
goto END_REPEAT; |
goto END_REPEAT; |
| 4816 |
} |
} |
| 4817 |
|
brazeroptr = previous; /* Save for possessive optimizing */ |
| 4818 |
*previous++ = OP_BRAZERO + repeat_type; |
*previous++ = OP_BRAZERO + repeat_type; |
| 4819 |
} |
} |
| 4820 |
|
|
| 4979 |
} |
} |
| 4980 |
} |
} |
| 4981 |
|
|
| 4982 |
/* If the maximum is unlimited, set a repeater in the final copy. We |
/* If the maximum is unlimited, set a repeater in the final copy. For |
| 4983 |
can't just offset backwards from the current code point, because we |
ONCE brackets, that's all we need to do. However, possessively repeated |
| 4984 |
don't know if there's been an options resetting after the ket. The |
ONCE brackets can be converted into non-capturing brackets, as the |
| 4985 |
correct offset was computed above. |
behaviour of (?:xx)++ is the same as (?>xx)++ and this saves having to |
| 4986 |
|
deal with possessive ONCEs specially. |
| 4987 |
Then, when we are doing the actual compile phase, check to see whether |
|
| 4988 |
this group is a non-atomic one that could match an empty string. If so, |
Otherwise, when we are doing the actual compile phase, check to see |
| 4989 |
|
whether this group is one that could match an empty string. If so, |
| 4990 |
convert the initial operator to the S form (e.g. OP_BRA -> OP_SBRA) so |
convert the initial operator to the S form (e.g. OP_BRA -> OP_SBRA) so |
| 4991 |
that runtime checking can be done. [This check is also applied to |
that runtime checking can be done. [This check is also applied to ONCE |
| 4992 |
atomic groups at runtime, but in a different way.] */ |
groups at runtime, but in a different way.] |
| 4993 |
|
|
| 4994 |
|
Then, if the quantifier was possessive and the bracket is not a |
| 4995 |
|
conditional, we convert the BRA code to the POS form, and the KET code to |
| 4996 |
|
KETRPOS. (It turns out to be convenient at runtime to detect this kind of |
| 4997 |
|
subpattern at both the start and at the end.) The use of special opcodes |
| 4998 |
|
makes it possible to reduce greatly the stack usage in pcre_exec(). If |
| 4999 |
|
the group is preceded by OP_BRAZERO, convert this to OP_BRAPOSZERO. |
| 5000 |
|
|
| 5001 |
|
Then, if the minimum number of matches is 1 or 0, cancel the possessive |
| 5002 |
|
flag so that the default action below, of wrapping everything inside |
| 5003 |
|
atomic brackets, does not happen. When the minimum is greater than 1, |
| 5004 |
|
there will be earlier copies of the group, and so we still have to wrap |
| 5005 |
|
the whole thing. */ |
| 5006 |
|
|
| 5007 |
else |
else |
| 5008 |
{ |
{ |
| 5009 |
uschar *ketcode = code - ketoffset; |
uschar *ketcode = code - 1 - LINK_SIZE; |
| 5010 |
uschar *bracode = ketcode - GET(ketcode, 1); |
uschar *bracode = ketcode - GET(ketcode, 1); |
| 5011 |
*ketcode = OP_KETRMAX + repeat_type; |
|
| 5012 |
if (lengthptr == NULL && *bracode != OP_ONCE) |
/* Convert possessive ONCE brackets to non-capturing */ |
| 5013 |
|
|
| 5014 |
|
if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) && |
| 5015 |
|
possessive_quantifier) *bracode = OP_BRA; |
| 5016 |
|
|
| 5017 |
|
/* For non-possessive ONCE brackets, all we need to do is to |
| 5018 |
|
set the KET. */ |
| 5019 |
|
|
| 5020 |
|
if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC) |
| 5021 |
|
*ketcode = OP_KETRMAX + repeat_type; |
| 5022 |
|
|
| 5023 |
|
/* Handle non-ONCE brackets and possessive ONCEs (which have been |
| 5024 |
|
converted to non-capturing above). */ |
| 5025 |
|
|
| 5026 |
|
else |
| 5027 |
{ |
{ |
| 5028 |
uschar *scode = bracode; |
/* In the compile phase, check for empty string matching. */ |
| 5029 |
do |
|
| 5030 |
|
if (lengthptr == NULL) |
| 5031 |
{ |
{ |
| 5032 |
if (could_be_empty_branch(scode, ketcode, utf8, cd)) |
uschar *scode = bracode; |
| 5033 |
|
do |
| 5034 |
{ |
{ |
| 5035 |
*bracode += OP_SBRA - OP_BRA; |
if (could_be_empty_branch(scode, ketcode, utf8, cd)) |
| 5036 |
break; |
{ |
| 5037 |
|
*bracode += OP_SBRA - OP_BRA; |
| 5038 |
|
break; |
| 5039 |
|
} |
| 5040 |
|
scode += GET(scode, 1); |
| 5041 |
} |
} |
| 5042 |
scode += GET(scode, 1); |
while (*scode == OP_ALT); |
| 5043 |
} |
} |
| 5044 |
while (*scode == OP_ALT); |
|
| 5045 |
|
/* Handle possessive quantifiers. */ |
| 5046 |
|
|
| 5047 |
|
if (possessive_quantifier) |
| 5048 |
|
{ |
| 5049 |
|
/* For COND brackets, we wrap the whole thing in a possessively |
| 5050 |
|
repeated non-capturing bracket, because we have not invented POS |
| 5051 |
|
versions of the COND opcodes. Because we are moving code along, we |
| 5052 |
|
must ensure that any pending recursive references are updated. */ |
| 5053 |
|
|
| 5054 |
|
if (*bracode == OP_COND || *bracode == OP_SCOND) |
| 5055 |
|
{ |
| 5056 |
|
int nlen = (int)(code - bracode); |
| 5057 |
|
*code = OP_END; |
| 5058 |
|
adjust_recurse(bracode, 1 + LINK_SIZE, utf8, cd, save_hwm); |
| 5059 |
|
memmove(bracode + 1+LINK_SIZE, bracode, nlen); |
| 5060 |
|
code += 1 + LINK_SIZE; |
| 5061 |
|
nlen += 1 + LINK_SIZE; |
| 5062 |
|
*bracode = OP_BRAPOS; |
| 5063 |
|
*code++ = OP_KETRPOS; |
| 5064 |
|
PUTINC(code, 0, nlen); |
| 5065 |
|
PUT(bracode, 1, nlen); |
| 5066 |
|
} |
| 5067 |
|
|
| 5068 |
|
/* For non-COND brackets, we modify the BRA code and use KETRPOS. */ |
| 5069 |
|
|
| 5070 |
|
else |
| 5071 |
|
{ |
| 5072 |
|
*bracode += 1; /* Switch to xxxPOS opcodes */ |
| 5073 |
|
*ketcode = OP_KETRPOS; |
| 5074 |
|
} |
| 5075 |
|
|
| 5076 |
|
/* If the minimum is zero, mark it as possessive, then unset the |
| 5077 |
|
possessive flag when the minimum is 0 or 1. */ |
| 5078 |
|
|
| 5079 |
|
if (brazeroptr != NULL) *brazeroptr = OP_BRAPOSZERO; |
| 5080 |
|
if (repeat_min < 2) possessive_quantifier = FALSE; |
| 5081 |
|
} |
| 5082 |
|
|
| 5083 |
|
/* Non-possessive quantifier */ |
| 5084 |
|
|
| 5085 |
|
else *ketcode = OP_KETRMAX + repeat_type; |
| 5086 |
} |
} |
| 5087 |
} |
} |
| 5088 |
} |
} |
| 5103 |
} |
} |
| 5104 |
|
|
| 5105 |
/* If the character following a repeat is '+', or if certain optimization |
/* If the character following a repeat is '+', or if certain optimization |
| 5106 |
tests above succeeded, possessive_quantifier is TRUE. For some of the |
tests above succeeded, possessive_quantifier is TRUE. For some opcodes, |
| 5107 |
simpler opcodes, there is an special alternative opcode for this. For |
there are special alternative opcodes for this case. For anything else, we |
| 5108 |
anything else, we wrap the entire repeated item inside OP_ONCE brackets. |
wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+' |
| 5109 |
The '+' notation is just syntactic sugar, taken from Sun's Java package, |
notation is just syntactic sugar, taken from Sun's Java package, but the |
| 5110 |
but the special opcodes can optimize it a bit. The repeated item starts at |
special opcodes can optimize it. |
| 5111 |
tempcode, not at previous, which might be the first part of a string whose |
|
| 5112 |
(former) last char we repeated. |
Some (but not all) possessively repeated subpatterns have already been |
| 5113 |
|
completely handled in the code just above. For them, possessive_quantifier |
| 5114 |
|
is always FALSE at this stage. |
| 5115 |
|
|
| 5116 |
|
Note that the repeated item starts at tempcode, not at previous, which |
| 5117 |
|
might be the first part of a string whose (former) last char we repeated. |
| 5118 |
|
|
| 5119 |
Possessifying an 'exact' quantifier has no effect, so we can ignore it. But |
Possessifying an 'exact' quantifier has no effect, so we can ignore it. But |
| 5120 |
an 'upto' may follow. We skip over an 'exact' item, and then test the |
an 'upto' may follow. We skip over an 'exact' item, and then test the |
| 5218 |
while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {}; |
while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {}; |
| 5219 |
namelen = (int)(ptr - name); |
namelen = (int)(ptr - name); |
| 5220 |
|
|
| 5221 |
|
/* It appears that Perl allows any characters whatsoever, other than |
| 5222 |
|
a closing parenthesis, to appear in arguments, so we no longer insist on |
| 5223 |
|
letters, digits, and underscores. */ |
| 5224 |
|
|
| 5225 |
if (*ptr == CHAR_COLON) |
if (*ptr == CHAR_COLON) |
| 5226 |
{ |
{ |
| 5227 |
arg = ++ptr; |
arg = ++ptr; |
| 5228 |
while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0 |
while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; |
|
|| *ptr == '_') ptr++; |
|
| 5229 |
arglen = (int)(ptr - arg); |
arglen = (int)(ptr - arg); |
| 5230 |
} |
} |
| 5231 |
|
|
| 5242 |
if (namelen == verbs[i].len && |
if (namelen == verbs[i].len && |
| 5243 |
strncmp((char *)name, vn, namelen) == 0) |
strncmp((char *)name, vn, namelen) == 0) |
| 5244 |
{ |
{ |
| 5245 |
/* Check for open captures before ACCEPT */ |
/* Check for open captures before ACCEPT and convert it to |
| 5246 |
|
ASSERT_ACCEPT if in an assertion. */ |
| 5247 |
|
|
| 5248 |
if (verbs[i].op == OP_ACCEPT) |
if (verbs[i].op == OP_ACCEPT) |
| 5249 |
{ |
{ |
| 5250 |
open_capitem *oc; |
open_capitem *oc; |
| 5251 |
|
if (arglen != 0) |
| 5252 |
|
{ |
| 5253 |
|
*errorcodeptr = ERR59; |
| 5254 |
|
goto FAILED; |
| 5255 |
|
} |
| 5256 |
cd->had_accept = TRUE; |
cd->had_accept = TRUE; |
| 5257 |
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
| 5258 |
{ |
{ |
| 5259 |
*code++ = OP_CLOSE; |
*code++ = OP_CLOSE; |
| 5260 |
PUT2INC(code, 0, oc->number); |
PUT2INC(code, 0, oc->number); |
| 5261 |
} |
} |
| 5262 |
|
*code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; |
| 5263 |
|
|
| 5264 |
|
/* Do not set firstbyte after *ACCEPT */ |
| 5265 |
|
if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; |
| 5266 |
} |
} |
| 5267 |
|
|
| 5268 |
/* Handle the cases with/without an argument */ |
/* Handle other cases with/without an argument */ |
| 5269 |
|
|
| 5270 |
if (arglen == 0) |
else if (arglen == 0) |
| 5271 |
{ |
{ |
| 5272 |
if (verbs[i].op < 0) /* Argument is mandatory */ |
if (verbs[i].op < 0) /* Argument is mandatory */ |
| 5273 |
{ |
{ |
| 5275 |
goto FAILED; |
goto FAILED; |
| 5276 |
} |
} |
| 5277 |
*code = verbs[i].op; |
*code = verbs[i].op; |
| 5278 |
if (*code++ == OP_THEN) |
if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN; |
|
{ |
|
|
PUT(code, 0, code - bcptr->current_branch - 1); |
|
|
code += LINK_SIZE; |
|
|
} |
|
| 5279 |
} |
} |
| 5280 |
|
|
| 5281 |
else |
else |
| 5286 |
goto FAILED; |
goto FAILED; |
| 5287 |
} |
} |
| 5288 |
*code = verbs[i].op_arg; |
*code = verbs[i].op_arg; |
| 5289 |
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; |
|
|
} |
|
| 5290 |
*code++ = arglen; |
*code++ = arglen; |
| 5291 |
memcpy(code, arg, arglen); |
memcpy(code, arg, arglen); |
| 5292 |
code += arglen; |
code += arglen; |
| 5548 |
/* ------------------------------------------------------------ */ |
/* ------------------------------------------------------------ */ |
| 5549 |
case CHAR_EQUALS_SIGN: /* Positive lookahead */ |
case CHAR_EQUALS_SIGN: /* Positive lookahead */ |
| 5550 |
bravalue = OP_ASSERT; |
bravalue = OP_ASSERT; |
| 5551 |
|
cd->assert_depth += 1; |
| 5552 |
ptr++; |
ptr++; |
| 5553 |
break; |
break; |
| 5554 |
|
|
| 5563 |
continue; |
continue; |
| 5564 |
} |
} |
| 5565 |
bravalue = OP_ASSERT_NOT; |
bravalue = OP_ASSERT_NOT; |
| 5566 |
|
cd->assert_depth += 1; |
| 5567 |
break; |
break; |
| 5568 |
|
|
| 5569 |
|
|
| 5573 |
{ |
{ |
| 5574 |
case CHAR_EQUALS_SIGN: /* Positive lookbehind */ |
case CHAR_EQUALS_SIGN: /* Positive lookbehind */ |
| 5575 |
bravalue = OP_ASSERTBACK; |
bravalue = OP_ASSERTBACK; |
| 5576 |
|
cd->assert_depth += 1; |
| 5577 |
ptr += 2; |
ptr += 2; |
| 5578 |
break; |
break; |
| 5579 |
|
|
| 5580 |
case CHAR_EXCLAMATION_MARK: /* Negative lookbehind */ |
case CHAR_EXCLAMATION_MARK: /* Negative lookbehind */ |
| 5581 |
bravalue = OP_ASSERTBACK_NOT; |
bravalue = OP_ASSERTBACK_NOT; |
| 5582 |
|
cd->assert_depth += 1; |
| 5583 |
ptr += 2; |
ptr += 2; |
| 5584 |
break; |
break; |
| 5585 |
|
|
| 5601 |
|
|
| 5602 |
/* ------------------------------------------------------------ */ |
/* ------------------------------------------------------------ */ |
| 5603 |
case CHAR_C: /* Callout - may be followed by digits; */ |
case CHAR_C: /* Callout - may be followed by digits; */ |
| 5604 |
previous_callout = code; /* Save for later completion */ |
previous_callout = code; /* Save for later completion */ |
| 5605 |
after_manual_callout = 1; /* Skip one item before completing */ |
after_manual_callout = 1; /* Skip one item before completing */ |
| 5606 |
*code++ = OP_CALLOUT; |
*code++ = OP_CALLOUT; |
| 5607 |
{ |
{ |
| 5608 |
int n = 0; |
int n = 0; |
| 5970 |
|
|
| 5971 |
/* 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 |
| 5972 |
offset below, what it actually inserted is the reference number |
offset below, what it actually inserted is the reference number |
| 5973 |
of the group. */ |
of the group. Then remember the forward reference. */ |
| 5974 |
|
|
| 5975 |
called = cd->start_code + recno; |
called = cd->start_code + recno; |
| 5976 |
PUTINC(cd->hwm, 0, (int)(code + 2 + LINK_SIZE - cd->start_code)); |
PUTINC(cd->hwm, 0, (int)(code + 1 - cd->start_code)); |
| 5977 |
} |
} |
| 5978 |
|
|
| 5979 |
/* If not a forward reference, and the subpattern is still open, |
/* If not a forward reference, and the subpattern is still open, |
| 5980 |
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 |
| 5981 |
recursion that could loop for ever, and diagnose that case. */ |
recursion that could loop for ever, and diagnose that case. We |
| 5982 |
|
must not, however, do this check if we are in a conditional |
| 5983 |
|
subpattern because the condition might be testing for recursion in |
| 5984 |
|
a pattern such as /(?(R)a+|(?R)b)/, which is perfectly valid. |
| 5985 |
|
Forever loops are also detected at runtime, so those that occur in |
| 5986 |
|
conditional subpatterns will be picked up then. */ |
| 5987 |
|
|
| 5988 |
else if (GET(called, 1) == 0 && |
else if (GET(called, 1) == 0 && cond_depth <= 0 && |
| 5989 |
could_be_empty(called, code, bcptr, utf8, cd)) |
could_be_empty(called, code, bcptr, utf8, cd)) |
| 5990 |
{ |
{ |
| 5991 |
*errorcodeptr = ERR40; |
*errorcodeptr = ERR40; |
| 5993 |
} |
} |
| 5994 |
} |
} |
| 5995 |
|
|
| 5996 |
/* 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; |
|
| 5997 |
|
|
| 5998 |
*code = OP_RECURSE; |
*code = OP_RECURSE; |
| 5999 |
PUT(code, 1, (int)(called - cd->start_code)); |
PUT(code, 1, (int)(called - cd->start_code)); |
| 6000 |
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; |
|
| 6001 |
} |
} |
| 6002 |
|
|
| 6003 |
/* Can't determine a first byte now */ |
/* Can't determine a first byte now */ |
| 6058 |
is necessary to ensure we correctly detect the start of the pattern in |
is necessary to ensure we correctly detect the start of the pattern in |
| 6059 |
both phases. |
both phases. |
| 6060 |
|
|
| 6061 |
If we are not at the pattern start, compile code to change the ims |
If we are not at the pattern start, reset the greedy defaults and the |
| 6062 |
options if this setting actually changes any of them, and reset the |
case value for firstbyte and reqbyte. */ |
|
greedy defaults and the case value for firstbyte and reqbyte. */ |
|
| 6063 |
|
|
| 6064 |
if (*ptr == CHAR_RIGHT_PARENTHESIS) |
if (*ptr == CHAR_RIGHT_PARENTHESIS) |
| 6065 |
{ |
{ |
| 6076 |
} |
} |
| 6077 |
|
|
| 6078 |
/* Change options at this level, and pass them back for use |
/* Change options at this level, and pass them back for use |
| 6079 |
in subsequent branches. When not at the start of the pattern, this |
in subsequent branches. */ |
|
information is also necessary so that a resetting item can be |
|
|
compiled at the end of a group (if we are in a group). */ |
|
| 6080 |
|
|
| 6081 |
*optionsptr = options = newoptions; |
*optionsptr = options = newoptions; |
| 6082 |
previous = NULL; /* This item can't be repeated */ |
previous = NULL; /* This item can't be repeated */ |
| 6112 |
skipbytes = 2; |
skipbytes = 2; |
| 6113 |
} |
} |
| 6114 |
|
|
| 6115 |
/* Process nested bracketed regex. Assertions may not be repeated, but |
/* Process nested bracketed regex. Assertions used not to be repeatable, |
| 6116 |
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 |
| 6117 |
non-register variable in order to be able to pass its address because some |
repeated. We copy code into a non-register variable (tempcode) in order to |
| 6118 |
compilers complain otherwise. Pass in a new setting for the ims options if |
be able to pass its address because some compilers complain otherwise. */ |
|
they have changed. */ |
|
| 6119 |
|
|
| 6120 |
previous = (bravalue >= OP_ONCE)? code : NULL; |
previous = code; /* For handling repetition */ |
| 6121 |
*code = bravalue; |
*code = bravalue; |
| 6122 |
tempcode = code; |
tempcode = code; |
| 6123 |
tempreqvary = cd->req_varyopt; /* Save value before bracket */ |
tempreqvary = cd->req_varyopt; /* Save value before bracket */ |
| 6124 |
length_prevgroup = 0; /* Initialize for pre-compile phase */ |
tempbracount = cd->bracount; /* Save value before bracket */ |
| 6125 |
|
length_prevgroup = 0; /* Initialize for pre-compile phase */ |
| 6126 |
|
|
| 6127 |
if (!compile_regex( |
if (!compile_regex( |
| 6128 |
newoptions, /* The complete new option state */ |
newoptions, /* The complete new option state */ |
| 6129 |
options & PCRE_IMS, /* The previous ims option state */ |
&tempcode, /* Where to put code (updated) */ |
| 6130 |
&tempcode, /* Where to put code (updated) */ |
&ptr, /* Input pointer (updated) */ |
| 6131 |
&ptr, /* Input pointer (updated) */ |
errorcodeptr, /* Where to put an error message */ |
|
errorcodeptr, /* Where to put an error message */ |
|
| 6132 |
(bravalue == OP_ASSERTBACK || |
(bravalue == OP_ASSERTBACK || |
| 6133 |
bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */ |
bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */ |
| 6134 |
reset_bracount, /* True if (?| group */ |
reset_bracount, /* True if (?| group */ |
| 6135 |
skipbytes, /* Skip over bracket number */ |
skipbytes, /* Skip over bracket number */ |
| 6136 |
&subfirstbyte, /* For possible first char */ |
cond_depth + |
| 6137 |
&subreqbyte, /* For possible last char */ |
((bravalue == OP_COND)?1:0), /* Depth of condition subpatterns */ |
| 6138 |
bcptr, /* Current branch chain */ |
&subfirstbyte, /* For possible first char */ |
| 6139 |
cd, /* Tables block */ |
&subreqbyte, /* For possible last char */ |
| 6140 |
(lengthptr == NULL)? NULL : /* Actual compile phase */ |
bcptr, /* Current branch chain */ |
| 6141 |
&length_prevgroup /* Pre-compile phase */ |
cd, /* Tables block */ |
| 6142 |
|
(lengthptr == NULL)? NULL : /* Actual compile phase */ |
| 6143 |
|
&length_prevgroup /* Pre-compile phase */ |
| 6144 |
)) |
)) |
| 6145 |
goto FAILED; |
goto FAILED; |
| 6146 |
|
|
| 6147 |
|
/* If this was an atomic group and there are no capturing groups within it, |
| 6148 |
|
generate OP_ONCE_NC instead of OP_ONCE. */ |
| 6149 |
|
|
| 6150 |
|
if (bravalue == OP_ONCE && cd->bracount <= tempbracount) |
| 6151 |
|
*code = OP_ONCE_NC; |
| 6152 |
|
|
| 6153 |
|
if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT) |
| 6154 |
|
cd->assert_depth -= 1; |
| 6155 |
|
|
| 6156 |
/* 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 |
| 6157 |
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. |
| 6158 |
and any option resetting that may follow it. The pattern pointer (ptr) |
The pattern pointer (ptr) is on the bracket. |
|
is on the bracket. */ |
|
| 6159 |
|
|
| 6160 |
/* 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 |
| 6161 |
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 |
| 6162 |
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 |
| 6163 |
not be available. */ |
not be available. */ |
| 6222 |
goto FAILED; |
goto FAILED; |
| 6223 |
} |
} |
| 6224 |
*lengthptr += length_prevgroup - 2 - 2*LINK_SIZE; |
*lengthptr += length_prevgroup - 2 - 2*LINK_SIZE; |
| 6225 |
*code++ = OP_BRA; |
code++; /* This already contains bravalue */ |
| 6226 |
PUTINC(code, 0, 1 + LINK_SIZE); |
PUTINC(code, 0, 1 + LINK_SIZE); |
| 6227 |
*code++ = OP_KET; |
*code++ = OP_KET; |
| 6228 |
PUTINC(code, 0, 1 + LINK_SIZE); |
PUTINC(code, 0, 1 + LINK_SIZE); |
| 6390 |
} |
} |
| 6391 |
|
|
| 6392 |
/* \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). |
| 6393 |
We also support \k{name} (.NET syntax) */ |
We also support \k{name} (.NET syntax). */ |
| 6394 |
|
|
| 6395 |
if (-c == ESC_k && (ptr[1] == CHAR_LESS_THAN_SIGN || |
if (-c == ESC_k) |
|
ptr[1] == CHAR_APOSTROPHE || ptr[1] == CHAR_LEFT_CURLY_BRACKET)) |
|
| 6396 |
{ |
{ |
| 6397 |
|
if ((ptr[1] != CHAR_LESS_THAN_SIGN && |
| 6398 |
|
ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET)) |
| 6399 |
|
{ |
| 6400 |
|
*errorcodeptr = ERR69; |
| 6401 |
|
break; |
| 6402 |
|
} |
| 6403 |
is_recurse = FALSE; |
is_recurse = FALSE; |
| 6404 |
terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? |
terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? |
| 6405 |
CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)? |
CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)? |
| 6479 |
} |
} |
| 6480 |
else |
else |
| 6481 |
#endif |
#endif |
| 6482 |
{ |
/* In non-UTF-8 mode, we turn \C into OP_ALLANY instead of OP_ANYBYTE |
| 6483 |
|
so that it works in DFA mode and in lookbehinds. */ |
| 6484 |
|
|
| 6485 |
|
{ |
| 6486 |
previous = (-c > ESC_b && -c < ESC_Z)? code : NULL; |
previous = (-c > ESC_b && -c < ESC_Z)? code : NULL; |
| 6487 |
*code++ = -c; |
*code++ = (!utf8 && c == -ESC_C)? OP_ALLANY : -c; |
| 6488 |
} |
} |
| 6489 |
} |
} |
| 6490 |
continue; |
continue; |
| 6559 |
else firstbyte = reqbyte = REQ_NONE; |
else firstbyte = reqbyte = REQ_NONE; |
| 6560 |
} |
} |
| 6561 |
|
|
| 6562 |
/* 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 |
| 6563 |
1 or the matching is caseful. */ |
1 or the matching is caseful. */ |
| 6564 |
|
|
| 6565 |
else |
else |
| 6600 |
|
|
| 6601 |
Arguments: |
Arguments: |
| 6602 |
options option bits, including any changes for this subpattern |
options option bits, including any changes for this subpattern |
|
oldims previous settings of ims option bits |
|
| 6603 |
codeptr -> the address of the current code pointer |
codeptr -> the address of the current code pointer |
| 6604 |
ptrptr -> the address of the current pattern pointer |
ptrptr -> the address of the current pattern pointer |
| 6605 |
errorcodeptr -> pointer to error code variable |
errorcodeptr -> pointer to error code variable |
| 6606 |
lookbehind TRUE if this is a lookbehind assertion |
lookbehind TRUE if this is a lookbehind assertion |
| 6607 |
reset_bracount TRUE to reset the count for each branch |
reset_bracount TRUE to reset the count for each branch |
| 6608 |
skipbytes skip this many bytes at start (for brackets and OP_COND) |
skipbytes skip this many bytes at start (for brackets and OP_COND) |
| 6609 |
|
cond_depth depth of nesting for conditional subpatterns |
| 6610 |
firstbyteptr place to put the first required character, or a negative number |
firstbyteptr place to put the first required character, or a negative number |
| 6611 |
reqbyteptr place to put the last required character, or a negative number |
reqbyteptr place to put the last required character, or a negative number |
| 6612 |
bcptr pointer to the chain of currently open branches |
bcptr pointer to the chain of currently open branches |
| 6618 |
*/ |
*/ |
| 6619 |
|
|
| 6620 |
static BOOL |
static BOOL |
| 6621 |
compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr, |
compile_regex(int options, uschar **codeptr, const uschar **ptrptr, |
| 6622 |
int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes, |
int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes, |
| 6623 |
int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd, |
int cond_depth, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, |
| 6624 |
int *lengthptr) |
compile_data *cd, int *lengthptr) |
| 6625 |
{ |
{ |
| 6626 |
const uschar *ptr = *ptrptr; |
const uschar *ptr = *ptrptr; |
| 6627 |
uschar *code = *codeptr; |
uschar *code = *codeptr; |
| 6635 |
int length; |
int length; |
| 6636 |
int orig_bracount; |
int orig_bracount; |
| 6637 |
int max_bracount; |
int max_bracount; |
|
int old_external_options = cd->external_options; |
|
| 6638 |
branch_chain bc; |
branch_chain bc; |
| 6639 |
|
|
| 6640 |
bc.outer = bcptr; |
bc.outer = bcptr; |
| 6658 |
|
|
| 6659 |
/* 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 |
| 6660 |
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 |
| 6661 |
detect groups that contain recursive back references to themselves. */ |
detect groups that contain recursive back references to themselves. Note that |
| 6662 |
|
only OP_CBRA need be tested here; changing this opcode to one of its variants, |
| 6663 |
|
e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */ |
| 6664 |
|
|
| 6665 |
if (*code == OP_CBRA) |
if (*code == OP_CBRA) |
| 6666 |
{ |
{ |
| 6700 |
into the length. */ |
into the length. */ |
| 6701 |
|
|
| 6702 |
if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte, |
if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte, |
| 6703 |
&branchreqbyte, &bc, cd, (lengthptr == NULL)? NULL : &length)) |
&branchreqbyte, &bc, cond_depth, cd, |
| 6704 |
|
(lengthptr == NULL)? NULL : &length)) |
| 6705 |
{ |
{ |
| 6706 |
*ptrptr = ptr; |
*ptrptr = ptr; |
| 6707 |
return FALSE; |
return FALSE; |
| 6708 |
} |
} |
| 6709 |
|
|
|
/* If the external options have changed during this branch, it means that we |
|
|
are at the top level, and a leading option setting has been encountered. We |
|
|
need to re-set the original option values to take account of this so that, |
|
|
during the pre-compile phase, we know to allow for a re-set at the start of |
|
|
subsequent branches. */ |
|
|
|
|
|
if (old_external_options != cd->external_options) |
|
|
oldims = cd->external_options & PCRE_IMS; |
|
|
|
|
| 6710 |
/* Keep the highest bracket count in case (?| was used and some branch |
/* Keep the highest bracket count in case (?| was used and some branch |
| 6711 |
has fewer than the rest. */ |
has fewer than the rest. */ |
| 6712 |
|
|
| 6767 |
{ |
{ |
| 6768 |
int fixed_length; |
int fixed_length; |
| 6769 |
*code = OP_END; |
*code = OP_END; |
| 6770 |
fixed_length = find_fixedlength(last_branch, options, FALSE, cd); |
fixed_length = find_fixedlength(last_branch, (options & PCRE_UTF8) != 0, |
| 6771 |
|
FALSE, cd); |
| 6772 |
DPRINTF(("fixed length = %d\n", fixed_length)); |
DPRINTF(("fixed length = %d\n", fixed_length)); |
| 6773 |
if (fixed_length == -3) |
if (fixed_length == -3) |
| 6774 |
{ |
{ |
| 6776 |
} |
} |
| 6777 |
else if (fixed_length < 0) |
else if (fixed_length < 0) |
| 6778 |
{ |
{ |
| 6779 |
*errorcodeptr = (fixed_length == -2)? ERR36 : ERR25; |
*errorcodeptr = (fixed_length == -2)? ERR36 : |
| 6780 |
|
(fixed_length == -4)? ERR70: ERR25; |
| 6781 |
*ptrptr = ptr; |
*ptrptr = ptr; |
| 6782 |
return FALSE; |
return FALSE; |
| 6783 |
} |
} |
| 6790 |
of offsets, with the field in the BRA item now becoming an offset to the |
of offsets, with the field in the BRA item now becoming an offset to the |
| 6791 |
first alternative. If there are no alternatives, it points to the end of the |
first alternative. If there are no alternatives, it points to the end of the |
| 6792 |
group. The length in the terminating ket is always the length of the whole |
group. The length in the terminating ket is always the length of the whole |
| 6793 |
bracketed item. If any of the ims options were changed inside the group, |
bracketed item. Return leaving the pointer at the terminating char. */ |
|
compile a resetting op-code following, except at the very end of the pattern. |
|
|
Return leaving the pointer at the terminating char. */ |
|
| 6794 |
|
|
| 6795 |
if (*ptr != CHAR_VERTICAL_LINE) |
if (*ptr != CHAR_VERTICAL_LINE) |
| 6796 |
{ |
{ |
| 6915 |
|
|
| 6916 |
Arguments: |
Arguments: |
| 6917 |
code points to start of expression (the bracket) |
code points to start of expression (the bracket) |
|
options points to the options setting |
|
| 6918 |
bracket_map a bitmap of which brackets we are inside while testing; this |
bracket_map a bitmap of which brackets we are inside while testing; this |
| 6919 |
handles up to substring 31; after that we just have to take |
handles up to substring 31; after that we just have to take |
| 6920 |
the less precise approach |
the less precise approach |
| 6924 |
*/ |
*/ |
| 6925 |
|
|
| 6926 |
static BOOL |
static BOOL |
| 6927 |
is_anchored(register const uschar *code, int *options, unsigned int bracket_map, |
is_anchored(register const uschar *code, unsigned int bracket_map, |
| 6928 |
unsigned int backref_map) |
unsigned int backref_map) |
| 6929 |
{ |
{ |
| 6930 |
do { |
do { |
| 6931 |
const uschar *scode = first_significant_code(code + _pcre_OP_lengths[*code], |
const uschar *scode = first_significant_code(code + _pcre_OP_lengths[*code], |
| 6932 |
options, PCRE_MULTILINE, FALSE); |
FALSE); |
| 6933 |
register int op = *scode; |
register int op = *scode; |
| 6934 |
|
|
| 6935 |
/* Non-capturing brackets */ |
/* Non-capturing brackets */ |
| 6936 |
|
|
| 6937 |
if (op == OP_BRA) |
if (op == OP_BRA || op == OP_BRAPOS || |
| 6938 |
|
op == OP_SBRA || op == OP_SBRAPOS) |
| 6939 |
{ |
{ |
| 6940 |
if (!is_anchored(scode, options, bracket_map, backref_map)) return FALSE; |
if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; |
| 6941 |
} |
} |
| 6942 |
|
|
| 6943 |
/* Capturing brackets */ |
/* Capturing brackets */ |
| 6944 |
|
|
| 6945 |
else if (op == OP_CBRA) |
else if (op == OP_CBRA || op == OP_CBRAPOS || |
| 6946 |
|
op == OP_SCBRA || op == OP_SCBRAPOS) |
| 6947 |
{ |
{ |
| 6948 |
int n = GET2(scode, 1+LINK_SIZE); |
int n = GET2(scode, 1+LINK_SIZE); |
| 6949 |
int new_map = bracket_map | ((n < 32)? (1 << n) : 1); |
int new_map = bracket_map | ((n < 32)? (1 << n) : 1); |
| 6950 |
if (!is_anchored(scode, options, new_map, backref_map)) return FALSE; |
if (!is_anchored(scode, new_map, backref_map)) return FALSE; |
| 6951 |
} |
} |
| 6952 |
|
|
| 6953 |
/* Other brackets */ |
/* Other brackets */ |
| 6954 |
|
|
| 6955 |
else if (op == OP_ASSERT || op == OP_ONCE || op == OP_COND) |
else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC || |
| 6956 |
|
op == OP_COND) |
| 6957 |
{ |
{ |
| 6958 |
if (!is_anchored(scode, options, bracket_map, backref_map)) return FALSE; |
if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; |
| 6959 |
} |
} |
| 6960 |
|
|
| 6961 |
/* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and |
/* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and |
| 7006 |
{ |
{ |
| 7007 |
do { |
do { |
| 7008 |
const uschar *scode = first_significant_code(code + _pcre_OP_lengths[*code], |
const uschar *scode = first_significant_code(code + _pcre_OP_lengths[*code], |
| 7009 |
NULL, 0, FALSE); |
FALSE); |
| 7010 |
register int op = *scode; |
register int op = *scode; |
| 7011 |
|
|
| 7012 |
/* If we are at the start of a conditional assertion group, *both* the |
/* If we are at the start of a conditional assertion group, *both* the |
| 7033 |
scode += 1 + LINK_SIZE; |
scode += 1 + LINK_SIZE; |
| 7034 |
break; |
break; |
| 7035 |
} |
} |
| 7036 |
scode = first_significant_code(scode, NULL, 0, FALSE); |
scode = first_significant_code(scode, FALSE); |
| 7037 |
op = *scode; |
op = *scode; |
| 7038 |
} |
} |
| 7039 |
|
|
| 7040 |
/* Non-capturing brackets */ |
/* Non-capturing brackets */ |
| 7041 |
|
|
| 7042 |
if (op == OP_BRA) |
if (op == OP_BRA || op == OP_BRAPOS || |
| 7043 |
|
op == OP_SBRA || op == OP_SBRAPOS) |
| 7044 |
{ |
{ |
| 7045 |
if (!is_startline(scode, bracket_map, backref_map)) return FALSE; |
if (!is_startline(scode, bracket_map, backref_map)) return FALSE; |
| 7046 |
} |
} |
| 7047 |
|
|
| 7048 |
/* Capturing brackets */ |
/* Capturing brackets */ |
| 7049 |
|
|
| 7050 |
else if (op == OP_CBRA) |
else if (op == OP_CBRA || op == OP_CBRAPOS || |
| 7051 |
|
op == OP_SCBRA || op == OP_SCBRAPOS) |
| 7052 |
{ |
{ |
| 7053 |
int n = GET2(scode, 1+LINK_SIZE); |
int n = GET2(scode, 1+LINK_SIZE); |
| 7054 |
int new_map = bracket_map | ((n < 32)? (1 << n) : 1); |
int new_map = bracket_map | ((n < 32)? (1 << n) : 1); |
| 7057 |
|
|
| 7058 |
/* Other brackets */ |
/* Other brackets */ |
| 7059 |
|
|
| 7060 |
else if (op == OP_ASSERT || op == OP_ONCE) |
else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC) |
| 7061 |
{ |
{ |
| 7062 |
if (!is_startline(scode, bracket_map, backref_map)) return FALSE; |
if (!is_startline(scode, bracket_map, backref_map)) return FALSE; |
| 7063 |
} |
} |
| 7098 |
|
|
| 7099 |
Arguments: |
Arguments: |
| 7100 |
code points to start of expression (the bracket) |
code points to start of expression (the bracket) |
|
options pointer to the options (used to check casing changes) |
|
| 7101 |
inassert TRUE if in an assertion |
inassert TRUE if in an assertion |
| 7102 |
|
|
| 7103 |
Returns: -1 or the fixed first char |
Returns: -1 or the fixed first char |
| 7104 |
*/ |
*/ |
| 7105 |
|
|
| 7106 |
static int |
static int |
| 7107 |
find_firstassertedchar(const uschar *code, int *options, BOOL inassert) |
find_firstassertedchar(const uschar *code, BOOL inassert) |
| 7108 |
{ |
{ |
| 7109 |
register int c = -1; |
register int c = -1; |
| 7110 |
do { |
do { |
| 7111 |
int d; |
int d; |
| 7112 |
const uschar *scode = |
int xl = (*code == OP_CBRA || *code == OP_SCBRA || |
| 7113 |
first_significant_code(code + 1+LINK_SIZE, options, PCRE_CASELESS, TRUE); |
*code == OP_CBRAPOS || *code == OP_SCBRAPOS)? 2:0; |
| 7114 |
|
const uschar *scode = first_significant_code(code + 1+LINK_SIZE + xl, TRUE); |
| 7115 |
register int op = *scode; |
register int op = *scode; |
| 7116 |
|
|
| 7117 |
switch(op) |
switch(op) |
| 7120 |
return -1; |
return -1; |
| 7121 |
|
|
| 7122 |
case OP_BRA: |
case OP_BRA: |
| 7123 |
|
case OP_BRAPOS: |
| 7124 |
case OP_CBRA: |
case OP_CBRA: |
| 7125 |
|
case OP_SCBRA: |
| 7126 |
|
case OP_CBRAPOS: |
| 7127 |
|
case OP_SCBRAPOS: |
| 7128 |
case OP_ASSERT: |
case OP_ASSERT: |
| 7129 |
case OP_ONCE: |
case OP_ONCE: |
| 7130 |
|
case OP_ONCE_NC: |
| 7131 |
case OP_COND: |
case OP_COND: |
| 7132 |
if ((d = find_firstassertedchar(scode, options, op == OP_ASSERT)) < 0) |
if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0) |
| 7133 |
return -1; |
return -1; |
| 7134 |
if (c < 0) c = d; else if (c != d) return -1; |
if (c < 0) c = d; else if (c != d) return -1; |
| 7135 |
break; |
break; |
| 7136 |
|
|
| 7137 |
case OP_EXACT: /* Fall through */ |
case OP_EXACT: |
| 7138 |
scode += 2; |
scode += 2; |
| 7139 |
|
/* Fall through */ |
| 7140 |
|
|
| 7141 |
case OP_CHAR: |
case OP_CHAR: |
|
case OP_CHARI: |
|
| 7142 |
case OP_PLUS: |
case OP_PLUS: |
| 7143 |
case OP_MINPLUS: |
case OP_MINPLUS: |
| 7144 |
case OP_POSPLUS: |
case OP_POSPLUS: |
| 7145 |
if (!inassert) return -1; |
if (!inassert) return -1; |
| 7146 |
if (c < 0) |
if (c < 0) c = scode[1]; |
| 7147 |
{ |
else if (c != scode[1]) return -1; |
| 7148 |
c = scode[1]; |
break; |
| 7149 |
if ((*options & PCRE_CASELESS) != 0) c |= REQ_CASELESS; |
|
| 7150 |
} |
case OP_EXACTI: |
| 7151 |
else if (c != scode[1]) return -1; |
scode += 2; |
| 7152 |
|
/* Fall through */ |
| 7153 |
|
|
| 7154 |
|
case OP_CHARI: |
| 7155 |
|
case OP_PLUSI: |
| 7156 |
|
case OP_MINPLUSI: |
| 7157 |
|
case OP_POSPLUSI: |
| 7158 |
|
if (!inassert) return -1; |
| 7159 |
|
if (c < 0) c = scode[1] | REQ_CASELESS; |
| 7160 |
|
else if (c != scode[1]) return -1; |
| 7161 |
break; |
break; |
| 7162 |
} |
} |
| 7163 |
|
|
| 7308 |
|
|
| 7309 |
utf8 = (options & PCRE_UTF8) != 0; |
utf8 = (options & PCRE_UTF8) != 0; |
| 7310 |
|
|
| 7311 |
/* 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 |
| 7312 |
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 |
| 7313 |
release 8.13. The only use we make of it here is to adjust the offset value to |
release 8.13. It is passed back from pcre_[dfa_]exec(), but at the moment is |
| 7314 |
the end of the string for a short string error, for compatibility with previous |
not used here. */ |
|
versions. */ |
|
| 7315 |
|
|
| 7316 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 7317 |
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 && |
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 && |
| 7318 |
(*erroroffset = _pcre_valid_utf8((USPTR)pattern, -1, &errorcode)) >= 0) |
(errorcode = _pcre_valid_utf8((USPTR)pattern, -1, erroroffset)) != 0) |
| 7319 |
{ |
{ |
| 7320 |
errorcode = ERR44; |
errorcode = ERR44; |
| 7321 |
goto PCRE_EARLY_ERROR_RETURN2; |
goto PCRE_EARLY_ERROR_RETURN2; |
| 7340 |
|
|
| 7341 |
/* Check validity of \R options. */ |
/* Check validity of \R options. */ |
| 7342 |
|
|
| 7343 |
switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) |
if ((options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) == |
| 7344 |
|
(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) |
| 7345 |
{ |
{ |
| 7346 |
case 0: |
errorcode = ERR56; |
| 7347 |
case PCRE_BSR_ANYCRLF: |
goto PCRE_EARLY_ERROR_RETURN; |
|
case PCRE_BSR_UNICODE: |
|
|
break; |
|
|
default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN; |
|
| 7348 |
} |
} |
| 7349 |
|
|
| 7350 |
/* Handle different types of newline. The three bits give seven cases. The |
/* Handle different types of newline. The three bits give seven cases. The |
| 7429 |
ptr += skipatstart; |
ptr += skipatstart; |
| 7430 |
code = cworkspace; |
code = cworkspace; |
| 7431 |
*code = OP_BRA; |
*code = OP_BRA; |
| 7432 |
(void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS, |
(void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE, |
| 7433 |
&code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, |
FALSE, 0, 0, &firstbyte, &reqbyte, NULL, cd, &length); |
|
&length); |
|
| 7434 |
if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN; |
if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN; |
| 7435 |
|
|
| 7436 |
DPRINTF(("end pre-compile: length=%d workspace=%d\n", length, |
DPRINTF(("end pre-compile: length=%d workspace=%d\n", length, |
| 7484 |
*/ |
*/ |
| 7485 |
|
|
| 7486 |
cd->final_bracount = cd->bracount; /* Save for checking forward references */ |
cd->final_bracount = cd->bracount; /* Save for checking forward references */ |
| 7487 |
|
cd->assert_depth = 0; |
| 7488 |
cd->bracount = 0; |
cd->bracount = 0; |
| 7489 |
cd->names_found = 0; |
cd->names_found = 0; |
| 7490 |
cd->name_table = (uschar *)re + re->name_table_offset; |
cd->name_table = (uschar *)re + re->name_table_offset; |
| 7503 |
ptr = (const uschar *)pattern + skipatstart; |
ptr = (const uschar *)pattern + skipatstart; |
| 7504 |
code = (uschar *)codestart; |
code = (uschar *)codestart; |
| 7505 |
*code = OP_BRA; |
*code = OP_BRA; |
| 7506 |
(void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr, |
(void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0, |
| 7507 |
&errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL); |
&firstbyte, &reqbyte, NULL, cd, NULL); |
| 7508 |
re->top_bracket = cd->bracount; |
re->top_bracket = cd->bracount; |
| 7509 |
re->top_backref = cd->top_backref; |
re->top_backref = cd->top_backref; |
| 7510 |
re->flags = cd->external_flags; |
re->flags = cd->external_flags; |
| 7511 |
|
|
| 7512 |
if (cd->had_accept) reqbyte = -1; /* Must disable after (*ACCEPT) */ |
if (cd->had_accept) reqbyte = REQ_NONE; /* Must disable after (*ACCEPT) */ |
| 7513 |
|
|
| 7514 |
/* 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. */ |
| 7515 |
|
|
| 7570 |
uschar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); |
uschar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); |
| 7571 |
int end_op = *be; |
int end_op = *be; |
| 7572 |
*be = OP_END; |
*be = OP_END; |
| 7573 |
fixed_length = find_fixedlength(cc, re->options, TRUE, cd); |
fixed_length = find_fixedlength(cc, (re->options & PCRE_UTF8) != 0, TRUE, |
| 7574 |
|
cd); |
| 7575 |
*be = end_op; |
*be = end_op; |
| 7576 |
DPRINTF(("fixed length = %d\n", fixed_length)); |
DPRINTF(("fixed length = %d\n", fixed_length)); |
| 7577 |
if (fixed_length < 0) |
if (fixed_length < 0) |
| 7578 |
{ |
{ |
| 7579 |
errorcode = (fixed_length == -2)? ERR36 : ERR25; |
errorcode = (fixed_length == -2)? ERR36 : |
| 7580 |
|
(fixed_length == -4)? ERR70 : ERR25; |
| 7581 |
break; |
break; |
| 7582 |
} |
} |
| 7583 |
PUT(cc, 1, fixed_length); |
PUT(cc, 1, fixed_length); |
| 7611 |
|
|
| 7612 |
if ((re->options & PCRE_ANCHORED) == 0) |
if ((re->options & PCRE_ANCHORED) == 0) |
| 7613 |
{ |
{ |
| 7614 |
int temp_options = re->options; /* May get changed during these scans */ |
if (is_anchored(codestart, 0, cd->backref_map)) |
|
if (is_anchored(codestart, &temp_options, 0, cd->backref_map)) |
|
| 7615 |
re->options |= PCRE_ANCHORED; |
re->options |= PCRE_ANCHORED; |
| 7616 |
else |
else |
| 7617 |
{ |
{ |
| 7618 |
if (firstbyte < 0) |
if (firstbyte < 0) |
| 7619 |
firstbyte = find_firstassertedchar(codestart, &temp_options, FALSE); |
firstbyte = find_firstassertedchar(codestart, FALSE); |
| 7620 |
if (firstbyte >= 0) /* Remove caseless flag for non-caseable chars */ |
if (firstbyte >= 0) /* Remove caseless flag for non-caseable chars */ |
| 7621 |
{ |
{ |
| 7622 |
int ch = firstbyte & 255; |
int ch = firstbyte & 255; |