| 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" |
"\\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, uschar **, const uschar **, int *, BOOL, BOOL, int, int *, |
compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int, |
| 552 |
int *, branch_chain *, compile_data *, int *); |
int *, int *, branch_chain *, compile_data *, int *); |
| 553 |
|
|
| 554 |
|
|
| 555 |
|
|
| 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 |
|
|
| 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 |
| 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 |
| 1554 |
/* We only need to continue for OP_CBRA (normal capturing bracket) and |
/* 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 |
OP_BRA (normal non-capturing bracket) because the other variants of these |
| 1556 |
opcodes are all concerned with unlimited repeated groups, which of course |
opcodes are all concerned with unlimited repeated groups, which of course |
| 1557 |
are not of fixed length. They will cause a -1 response from the default |
are not of fixed length. */ |
|
case of this switch. */ |
|
| 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), 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; |
| 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 |
Note that we must not include the OP_KETRxxx opcodes here, because they |
the same code. Note that we must not include the OP_KETRxxx opcodes here, |
| 1575 |
all imply an unlimited repeat. */ |
because they all imply an unlimited repeat. */ |
| 1576 |
|
|
| 1577 |
case OP_ALT: |
case OP_ALT: |
| 1578 |
case OP_KET: |
case OP_KET: |
| 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; |
| 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; |
| 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 |
| 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: |
| 1721 |
|
|
| 1722 |
switch (*cc) |
switch (*cc) |
| 1723 |
{ |
{ |
| 1724 |
|
case OP_CRPLUS: |
| 1725 |
|
case OP_CRMINPLUS: |
| 1726 |
case OP_CRSTAR: |
case OP_CRSTAR: |
| 1727 |
case OP_CRMINSTAR: |
case OP_CRMINSTAR: |
| 1728 |
case OP_CRQUERY: |
case OP_CRQUERY: |
| 1743 |
|
|
| 1744 |
/* Anything else is variable length */ |
/* Anything else is variable length */ |
| 1745 |
|
|
| 1746 |
default: |
case OP_ANYNL: |
| 1747 |
|
case OP_BRAMINZERO: |
| 1748 |
|
case OP_BRAPOS: |
| 1749 |
|
case OP_BRAPOSZERO: |
| 1750 |
|
case OP_BRAZERO: |
| 1751 |
|
case OP_CBRAPOS: |
| 1752 |
|
case OP_EXTUNI: |
| 1753 |
|
case OP_KETRMAX: |
| 1754 |
|
case OP_KETRMIN: |
| 1755 |
|
case OP_KETRPOS: |
| 1756 |
|
case OP_MINPLUS: |
| 1757 |
|
case OP_MINPLUSI: |
| 1758 |
|
case OP_MINQUERY: |
| 1759 |
|
case OP_MINQUERYI: |
| 1760 |
|
case OP_MINSTAR: |
| 1761 |
|
case OP_MINSTARI: |
| 1762 |
|
case OP_MINUPTO: |
| 1763 |
|
case OP_MINUPTOI: |
| 1764 |
|
case OP_NOTMINPLUS: |
| 1765 |
|
case OP_NOTMINPLUSI: |
| 1766 |
|
case OP_NOTMINQUERY: |
| 1767 |
|
case OP_NOTMINQUERYI: |
| 1768 |
|
case OP_NOTMINSTAR: |
| 1769 |
|
case OP_NOTMINSTARI: |
| 1770 |
|
case OP_NOTMINUPTO: |
| 1771 |
|
case OP_NOTMINUPTOI: |
| 1772 |
|
case OP_NOTPLUS: |
| 1773 |
|
case OP_NOTPLUSI: |
| 1774 |
|
case OP_NOTPOSPLUS: |
| 1775 |
|
case OP_NOTPOSPLUSI: |
| 1776 |
|
case OP_NOTPOSQUERY: |
| 1777 |
|
case OP_NOTPOSQUERYI: |
| 1778 |
|
case OP_NOTPOSSTAR: |
| 1779 |
|
case OP_NOTPOSSTARI: |
| 1780 |
|
case OP_NOTPOSUPTO: |
| 1781 |
|
case OP_NOTPOSUPTOI: |
| 1782 |
|
case OP_NOTQUERY: |
| 1783 |
|
case OP_NOTQUERYI: |
| 1784 |
|
case OP_NOTSTAR: |
| 1785 |
|
case OP_NOTSTARI: |
| 1786 |
|
case OP_NOTUPTO: |
| 1787 |
|
case OP_NOTUPTOI: |
| 1788 |
|
case OP_PLUS: |
| 1789 |
|
case OP_PLUSI: |
| 1790 |
|
case OP_POSPLUS: |
| 1791 |
|
case OP_POSPLUSI: |
| 1792 |
|
case OP_POSQUERY: |
| 1793 |
|
case OP_POSQUERYI: |
| 1794 |
|
case OP_POSSTAR: |
| 1795 |
|
case OP_POSSTARI: |
| 1796 |
|
case OP_POSUPTO: |
| 1797 |
|
case OP_POSUPTOI: |
| 1798 |
|
case OP_QUERY: |
| 1799 |
|
case OP_QUERYI: |
| 1800 |
|
case OP_REF: |
| 1801 |
|
case OP_REFI: |
| 1802 |
|
case OP_SBRA: |
| 1803 |
|
case OP_SBRAPOS: |
| 1804 |
|
case OP_SCBRA: |
| 1805 |
|
case OP_SCBRAPOS: |
| 1806 |
|
case OP_SCOND: |
| 1807 |
|
case OP_SKIPZERO: |
| 1808 |
|
case OP_STAR: |
| 1809 |
|
case OP_STARI: |
| 1810 |
|
case OP_TYPEMINPLUS: |
| 1811 |
|
case OP_TYPEMINQUERY: |
| 1812 |
|
case OP_TYPEMINSTAR: |
| 1813 |
|
case OP_TYPEMINUPTO: |
| 1814 |
|
case OP_TYPEPLUS: |
| 1815 |
|
case OP_TYPEPOSPLUS: |
| 1816 |
|
case OP_TYPEPOSQUERY: |
| 1817 |
|
case OP_TYPEPOSSTAR: |
| 1818 |
|
case OP_TYPEPOSUPTO: |
| 1819 |
|
case OP_TYPEQUERY: |
| 1820 |
|
case OP_TYPESTAR: |
| 1821 |
|
case OP_TYPEUPTO: |
| 1822 |
|
case OP_UPTO: |
| 1823 |
|
case OP_UPTOI: |
| 1824 |
return -1; |
return -1; |
| 1825 |
|
|
| 1826 |
|
/* Catch unrecognized opcodes so that when new ones are added they |
| 1827 |
|
are not forgotten, as has happened in the past. */ |
| 1828 |
|
|
| 1829 |
|
default: |
| 1830 |
|
return -4; |
| 1831 |
} |
} |
| 1832 |
} |
} |
| 1833 |
/* Control never gets here */ |
/* Control never gets here */ |
| 1922 |
break; |
break; |
| 1923 |
|
|
| 1924 |
case OP_THEN_ARG: |
case OP_THEN_ARG: |
| 1925 |
code += code[1+LINK_SIZE]; |
code += code[1]; |
| 1926 |
break; |
break; |
| 1927 |
} |
} |
| 1928 |
|
|
| 2041 |
break; |
break; |
| 2042 |
|
|
| 2043 |
case OP_THEN_ARG: |
case OP_THEN_ARG: |
| 2044 |
code += code[1+LINK_SIZE]; |
code += code[1]; |
| 2045 |
break; |
break; |
| 2046 |
} |
} |
| 2047 |
|
|
| 2143 |
/* For a recursion/subroutine call, if its end has been reached, which |
/* For a recursion/subroutine call, if its end has been reached, which |
| 2144 |
implies a backward reference subroutine call, we can scan it. If it's a |
implies a backward reference subroutine call, we can scan it. If it's a |
| 2145 |
forward reference subroutine call, we can't. To detect forward reference |
forward reference subroutine call, we can't. To detect forward reference |
| 2146 |
we have to scan up the list that is kept in the workspace. This function is |
we have to scan up the list that is kept in the workspace. This function is |
| 2147 |
called only when doing the real compile, not during the pre-compile that |
called only when doing the real compile, not during the pre-compile that |
| 2148 |
measures the size of the compiled pattern. */ |
measures the size of the compiled pattern. */ |
| 2149 |
|
|
| 2150 |
if (c == OP_RECURSE) |
if (c == OP_RECURSE) |
| 2151 |
{ |
{ |
| 2152 |
const uschar *scode; |
const uschar *scode; |
| 2153 |
BOOL empty_branch; |
BOOL empty_branch; |
| 2154 |
|
|
| 2155 |
/* Test for forward reference */ |
/* Test for forward reference */ |
| 2156 |
|
|
| 2157 |
for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE) |
for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE) |
| 2158 |
if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE; |
if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE; |
| 2159 |
|
|
| 2160 |
/* Not a forward reference, test for completed backward reference */ |
/* Not a forward reference, test for completed backward reference */ |
| 2161 |
|
|
| 2162 |
empty_branch = FALSE; |
empty_branch = FALSE; |
| 2163 |
scode = cd->start_code + GET(code, 1); |
scode = cd->start_code + GET(code, 1); |
| 2164 |
if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ |
if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ |
| 2165 |
|
|
| 2166 |
/* Completed backwards reference */ |
/* Completed backwards reference */ |
| 2167 |
|
|
| 2168 |
do |
do |
| 2169 |
{ |
{ |
| 2170 |
if (could_be_empty_branch(scode, endcode, utf8, cd)) |
if (could_be_empty_branch(scode, endcode, utf8, cd)) |
| 2175 |
scode += GET(scode, 1); |
scode += GET(scode, 1); |
| 2176 |
} |
} |
| 2177 |
while (*scode == OP_ALT); |
while (*scode == OP_ALT); |
| 2178 |
|
|
| 2179 |
if (!empty_branch) return FALSE; /* All branches are non-empty */ |
if (!empty_branch) return FALSE; /* All branches are non-empty */ |
| 2180 |
continue; |
continue; |
| 2181 |
} |
} |
| 2206 |
|
|
| 2207 |
if (c == OP_BRA || c == OP_BRAPOS || |
if (c == OP_BRA || c == OP_BRAPOS || |
| 2208 |
c == OP_CBRA || c == OP_CBRAPOS || |
c == OP_CBRA || c == OP_CBRAPOS || |
| 2209 |
c == OP_ONCE || c == OP_COND) |
c == OP_ONCE || c == OP_ONCE_NC || |
| 2210 |
|
c == OP_COND) |
| 2211 |
{ |
{ |
| 2212 |
BOOL empty_branch; |
BOOL empty_branch; |
| 2213 |
if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ |
if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ |
| 2379 |
break; |
break; |
| 2380 |
|
|
| 2381 |
case OP_THEN_ARG: |
case OP_THEN_ARG: |
| 2382 |
code += code[1+LINK_SIZE]; |
code += code[1]; |
| 2383 |
break; |
break; |
| 2384 |
|
|
| 2385 |
/* None of the remaining opcodes are required to match a character. */ |
/* None of the remaining opcodes are required to match a character. */ |
| 2402 |
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 |
| 2403 |
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, |
| 2404 |
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. |
| 2405 |
This function is called only during the real compile, not during the |
This function is called only during the real compile, not during the |
| 2406 |
pre-compile. |
pre-compile. |
| 2407 |
|
|
| 2408 |
Arguments: |
Arguments: |
| 2454 |
"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, |
| 2455 |
I think. |
I think. |
| 2456 |
|
|
| 2457 |
|
A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not. |
| 2458 |
|
It seems that the appearance of a nested POSIX class supersedes an apparent |
| 2459 |
|
external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or |
| 2460 |
|
a digit. |
| 2461 |
|
|
| 2462 |
|
In Perl, unescaped square brackets may also appear as part of class names. For |
| 2463 |
|
example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for |
| 2464 |
|
[:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not |
| 2465 |
|
seem right at all. PCRE does not allow closing square brackets in POSIX class |
| 2466 |
|
names. |
| 2467 |
|
|
| 2468 |
Arguments: |
Arguments: |
| 2469 |
ptr pointer to the initial [ |
ptr pointer to the initial [ |
| 2470 |
endptr where to return the end pointer |
endptr where to return the end pointer |
| 2479 |
terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ |
terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ |
| 2480 |
for (++ptr; *ptr != 0; ptr++) |
for (++ptr; *ptr != 0; ptr++) |
| 2481 |
{ |
{ |
| 2482 |
if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else |
if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) |
| 2483 |
|
ptr++; |
| 2484 |
|
else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE; |
| 2485 |
|
else |
| 2486 |
{ |
{ |
|
if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE; |
|
| 2487 |
if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) |
if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) |
| 2488 |
{ |
{ |
| 2489 |
*endptr = ptr; |
*endptr = ptr; |
| 2490 |
return TRUE; |
return TRUE; |
| 2491 |
} |
} |
| 2492 |
|
if (*ptr == CHAR_LEFT_SQUARE_BRACKET && |
| 2493 |
|
(ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || |
| 2494 |
|
ptr[1] == CHAR_EQUALS_SIGN) && |
| 2495 |
|
check_posix_syntax(ptr, endptr)) |
| 2496 |
|
return FALSE; |
| 2497 |
} |
} |
| 2498 |
} |
} |
| 2499 |
return FALSE; |
return FALSE; |
| 3203 |
firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) |
firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) |
| 3204 |
reqbyteptr set to the last literal character required, else < 0 |
reqbyteptr set to the last literal character required, else < 0 |
| 3205 |
bcptr points to current branch chain |
bcptr points to current branch chain |
| 3206 |
|
cond_depth conditional nesting depth |
| 3207 |
cd contains pointers to tables etc. |
cd contains pointers to tables etc. |
| 3208 |
lengthptr NULL during the real compile phase |
lengthptr NULL during the real compile phase |
| 3209 |
points to length accumulator during pre-compile phase |
points to length accumulator during pre-compile phase |
| 3215 |
static BOOL |
static BOOL |
| 3216 |
compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr, |
compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr, |
| 3217 |
int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, |
int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, |
| 3218 |
compile_data *cd, int *lengthptr) |
int cond_depth, compile_data *cd, int *lengthptr) |
| 3219 |
{ |
{ |
| 3220 |
int repeat_type, op_type; |
int repeat_type, op_type; |
| 3221 |
int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ |
int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ |
| 3243 |
uschar classbits[32]; |
uschar classbits[32]; |
| 3244 |
|
|
| 3245 |
/* We can fish out the UTF-8 setting once and for all into a BOOL, but we |
/* We can fish out the UTF-8 setting once and for all into a BOOL, but we |
| 3246 |
must not do this for other options (e.g. PCRE_EXTENDED) because they may change |
must not do this for other options (e.g. PCRE_EXTENDED) because they may change |
| 3247 |
dynamically as we process the pattern. */ |
dynamically as we process the pattern. */ |
| 3248 |
|
|
| 3249 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 3254 |
uschar utf8_char[6]; |
uschar utf8_char[6]; |
| 3255 |
#else |
#else |
| 3256 |
BOOL utf8 = FALSE; |
BOOL utf8 = FALSE; |
|
uschar *utf8_char = NULL; |
|
| 3257 |
#endif |
#endif |
| 3258 |
|
|
| 3259 |
#ifdef PCRE_DEBUG |
#ifdef PCRE_DEBUG |
| 3304 |
int subfirstbyte; |
int subfirstbyte; |
| 3305 |
int terminator; |
int terminator; |
| 3306 |
int mclength; |
int mclength; |
| 3307 |
|
int tempbracount; |
| 3308 |
uschar mcbuffer[8]; |
uschar mcbuffer[8]; |
| 3309 |
|
|
| 3310 |
/* Get next byte in the pattern */ |
/* Get next byte in the pattern */ |
| 4395 |
op_type = 0; /* Default single-char op codes */ |
op_type = 0; /* Default single-char op codes */ |
| 4396 |
possessive_quantifier = FALSE; /* Default not possessive quantifier */ |
possessive_quantifier = FALSE; /* Default not possessive quantifier */ |
| 4397 |
|
|
| 4398 |
/* 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 |
| 4399 |
for an inserted OP_ONCE for the additional '+' extension. */ |
insert something before it. */ |
| 4400 |
|
|
| 4401 |
tempcode = previous; |
tempcode = previous; |
| 4402 |
|
|
| 4418 |
ptr++; |
ptr++; |
| 4419 |
} |
} |
| 4420 |
else repeat_type = greedy_default; |
else repeat_type = greedy_default; |
| 4421 |
|
|
| 4422 |
/* If previous was a recursion call, wrap it in atomic brackets so that |
/* If previous was a recursion call, wrap it in atomic brackets so that |
| 4423 |
previous becomes the atomic group. All recursions were so wrapped in the |
previous becomes the atomic group. All recursions were so wrapped in the |
| 4424 |
past, but it no longer happens for non-repeated recursions. In fact, the |
past, but it no longer happens for non-repeated recursions. In fact, the |
| 4425 |
repeated ones could be re-implemented independently so as not to need this, |
repeated ones could be re-implemented independently so as not to need this, |
| 4426 |
but for the moment we rely on the code for repeating groups. */ |
but for the moment we rely on the code for repeating groups. */ |
| 4427 |
|
|
| 4428 |
if (*previous == OP_RECURSE) |
if (*previous == OP_RECURSE) |
| 4429 |
{ |
{ |
| 4430 |
memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE); |
memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE); |
| 4434 |
PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE); |
PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE); |
| 4435 |
code += 2 + 2 * LINK_SIZE; |
code += 2 + 2 * LINK_SIZE; |
| 4436 |
length_prevgroup = 3 + 3*LINK_SIZE; |
length_prevgroup = 3 + 3*LINK_SIZE; |
| 4437 |
|
|
| 4438 |
/* When actually compiling, we need to check whether this was a forward |
/* When actually compiling, we need to check whether this was a forward |
| 4439 |
reference, and if so, adjust the offset. */ |
reference, and if so, adjust the offset. */ |
| 4440 |
|
|
| 4441 |
if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE) |
if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE) |
| 4442 |
{ |
{ |
| 4443 |
int offset = GET(cd->hwm, -LINK_SIZE); |
int offset = GET(cd->hwm, -LINK_SIZE); |
| 4444 |
if (offset == previous + 1 - cd->start_code) |
if (offset == previous + 1 - cd->start_code) |
| 4445 |
PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE); |
PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE); |
| 4446 |
} |
} |
| 4447 |
} |
} |
| 4448 |
|
|
| 4449 |
/* Now handle repetition for the different types of item. */ |
/* Now handle repetition for the different types of item. */ |
| 4450 |
|
|
| 4451 |
/* If previous was a character match, abolish the item and generate a |
/* If previous was a character match, abolish the item and generate a |
| 4739 |
} |
} |
| 4740 |
|
|
| 4741 |
/* 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 |
| 4742 |
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 |
| 4743 |
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 |
| 4744 |
special varieties. */ |
into the more special varieties such as BRAPOS and SBRA. A test for >= |
| 4745 |
|
OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK, |
| 4746 |
|
ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow |
| 4747 |
|
repetition of assertions, but now it does, for Perl compatibility. */ |
| 4748 |
|
|
| 4749 |
else if (*previous == OP_BRA || *previous == OP_CBRA || |
else if (*previous >= OP_ASSERT && *previous <= OP_COND) |
|
*previous == OP_ONCE || *previous == OP_COND) |
|
| 4750 |
{ |
{ |
| 4751 |
register int i; |
register int i; |
| 4752 |
int len = (int)(code - previous); |
int len = (int)(code - previous); |
| 4753 |
uschar *bralink = NULL; |
uschar *bralink = NULL; |
| 4754 |
uschar *brazeroptr = NULL; |
uschar *brazeroptr = NULL; |
| 4755 |
|
|
| 4756 |
/* Repeating a DEFINE group is pointless */ |
/* Repeating a DEFINE group is pointless, but Perl allows the syntax, so |
| 4757 |
|
we just ignore the repeat. */ |
| 4758 |
|
|
| 4759 |
if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF) |
if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF) |
| 4760 |
|
goto END_REPEAT; |
| 4761 |
|
|
| 4762 |
|
/* There is no sense in actually repeating assertions. The only potential |
| 4763 |
|
use of repetition is in cases when the assertion is optional. Therefore, |
| 4764 |
|
if the minimum is greater than zero, just ignore the repeat. If the |
| 4765 |
|
maximum is not not zero or one, set it to 1. */ |
| 4766 |
|
|
| 4767 |
|
if (*previous < OP_ONCE) /* Assertion */ |
| 4768 |
{ |
{ |
| 4769 |
*errorcodeptr = ERR55; |
if (repeat_min > 0) goto END_REPEAT; |
| 4770 |
goto FAILED; |
if (repeat_max < 0 || repeat_max > 1) repeat_max = 1; |
| 4771 |
} |
} |
| 4772 |
|
|
| 4773 |
/* 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 |
| 4788 |
** goto END_REPEAT; |
** goto END_REPEAT; |
| 4789 |
** } |
** } |
| 4790 |
|
|
| 4791 |
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 |
| 4792 |
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 |
| 4793 |
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 |
| 4794 |
groups are referenced, we cannot do this selectively. |
don't have a list of which groups are referenced, we cannot do this |
| 4795 |
|
selectively. |
| 4796 |
|
|
| 4797 |
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 |
| 4798 |
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 |
| 5002 |
{ |
{ |
| 5003 |
uschar *ketcode = code - 1 - LINK_SIZE; |
uschar *ketcode = code - 1 - LINK_SIZE; |
| 5004 |
uschar *bracode = ketcode - GET(ketcode, 1); |
uschar *bracode = ketcode - GET(ketcode, 1); |
| 5005 |
|
|
| 5006 |
if (*bracode == OP_ONCE && possessive_quantifier) *bracode = OP_BRA; |
if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) && |
| 5007 |
if (*bracode == OP_ONCE) |
possessive_quantifier) *bracode = OP_BRA; |
| 5008 |
|
|
| 5009 |
|
if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC) |
| 5010 |
*ketcode = OP_KETRMAX + repeat_type; |
*ketcode = OP_KETRMAX + repeat_type; |
| 5011 |
else |
else |
| 5012 |
{ |
{ |
| 5057 |
there are special alternative opcodes for this case. For anything else, we |
there are special alternative opcodes for this case. For anything else, we |
| 5058 |
wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+' |
wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+' |
| 5059 |
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 |
| 5060 |
special opcodes can optimize it. |
special opcodes can optimize it. |
| 5061 |
|
|
| 5062 |
Possessively repeated subpatterns have already been handled in the code |
Possessively repeated subpatterns have already been handled in the code |
| 5063 |
just above, so possessive_quantifier is always FALSE for them at this |
just above, so possessive_quantifier is always FALSE for them at this |
| 5064 |
stage. |
stage. |
| 5065 |
|
|
| 5066 |
Note that the repeated item starts at tempcode, not at previous, which |
Note that the repeated item starts at tempcode, not at previous, which |
| 5067 |
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. |
| 5068 |
|
|
| 5168 |
while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {}; |
while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {}; |
| 5169 |
namelen = (int)(ptr - name); |
namelen = (int)(ptr - name); |
| 5170 |
|
|
| 5171 |
|
/* It appears that Perl allows any characters whatsoever, other than |
| 5172 |
|
a closing parenthesis, to appear in arguments, so we no longer insist on |
| 5173 |
|
letters, digits, and underscores. */ |
| 5174 |
|
|
| 5175 |
if (*ptr == CHAR_COLON) |
if (*ptr == CHAR_COLON) |
| 5176 |
{ |
{ |
| 5177 |
arg = ++ptr; |
arg = ++ptr; |
| 5178 |
while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0 |
while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; |
|
|| *ptr == '_') ptr++; |
|
| 5179 |
arglen = (int)(ptr - arg); |
arglen = (int)(ptr - arg); |
| 5180 |
} |
} |
| 5181 |
|
|
| 5192 |
if (namelen == verbs[i].len && |
if (namelen == verbs[i].len && |
| 5193 |
strncmp((char *)name, vn, namelen) == 0) |
strncmp((char *)name, vn, namelen) == 0) |
| 5194 |
{ |
{ |
| 5195 |
/* Check for open captures before ACCEPT and convert it to |
/* Check for open captures before ACCEPT and convert it to |
| 5196 |
ASSERT_ACCEPT if in an assertion. */ |
ASSERT_ACCEPT if in an assertion. */ |
| 5197 |
|
|
| 5198 |
if (verbs[i].op == OP_ACCEPT) |
if (verbs[i].op == OP_ACCEPT) |
| 5202 |
{ |
{ |
| 5203 |
*errorcodeptr = ERR59; |
*errorcodeptr = ERR59; |
| 5204 |
goto FAILED; |
goto FAILED; |
| 5205 |
} |
} |
| 5206 |
cd->had_accept = TRUE; |
cd->had_accept = TRUE; |
| 5207 |
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
for (oc = cd->open_caps; oc != NULL; oc = oc->next) |
| 5208 |
{ |
{ |
| 5210 |
PUT2INC(code, 0, oc->number); |
PUT2INC(code, 0, oc->number); |
| 5211 |
} |
} |
| 5212 |
*code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; |
*code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; |
| 5213 |
|
|
| 5214 |
|
/* Do not set firstbyte after *ACCEPT */ |
| 5215 |
|
if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE; |
| 5216 |
} |
} |
| 5217 |
|
|
| 5218 |
/* Handle other cases with/without an argument */ |
/* Handle other cases with/without an argument */ |
| 5225 |
goto FAILED; |
goto FAILED; |
| 5226 |
} |
} |
| 5227 |
*code = verbs[i].op; |
*code = verbs[i].op; |
| 5228 |
if (*code++ == OP_THEN) |
if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN; |
|
{ |
|
|
PUT(code, 0, code - bcptr->current_branch - 1); |
|
|
code += LINK_SIZE; |
|
|
} |
|
| 5229 |
} |
} |
| 5230 |
|
|
| 5231 |
else |
else |
| 5236 |
goto FAILED; |
goto FAILED; |
| 5237 |
} |
} |
| 5238 |
*code = verbs[i].op_arg; |
*code = verbs[i].op_arg; |
| 5239 |
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; |
|
|
} |
|
| 5240 |
*code++ = arglen; |
*code++ = arglen; |
| 5241 |
memcpy(code, arg, arglen); |
memcpy(code, arg, arglen); |
| 5242 |
code += arglen; |
code += arglen; |
| 5498 |
/* ------------------------------------------------------------ */ |
/* ------------------------------------------------------------ */ |
| 5499 |
case CHAR_EQUALS_SIGN: /* Positive lookahead */ |
case CHAR_EQUALS_SIGN: /* Positive lookahead */ |
| 5500 |
bravalue = OP_ASSERT; |
bravalue = OP_ASSERT; |
| 5501 |
cd->assert_depth += 1; |
cd->assert_depth += 1; |
| 5502 |
ptr++; |
ptr++; |
| 5503 |
break; |
break; |
| 5504 |
|
|
| 5513 |
continue; |
continue; |
| 5514 |
} |
} |
| 5515 |
bravalue = OP_ASSERT_NOT; |
bravalue = OP_ASSERT_NOT; |
| 5516 |
cd->assert_depth += 1; |
cd->assert_depth += 1; |
| 5517 |
break; |
break; |
| 5518 |
|
|
| 5519 |
|
|
| 5523 |
{ |
{ |
| 5524 |
case CHAR_EQUALS_SIGN: /* Positive lookbehind */ |
case CHAR_EQUALS_SIGN: /* Positive lookbehind */ |
| 5525 |
bravalue = OP_ASSERTBACK; |
bravalue = OP_ASSERTBACK; |
| 5526 |
cd->assert_depth += 1; |
cd->assert_depth += 1; |
| 5527 |
ptr += 2; |
ptr += 2; |
| 5528 |
break; |
break; |
| 5529 |
|
|
| 5530 |
case CHAR_EXCLAMATION_MARK: /* Negative lookbehind */ |
case CHAR_EXCLAMATION_MARK: /* Negative lookbehind */ |
| 5531 |
bravalue = OP_ASSERTBACK_NOT; |
bravalue = OP_ASSERTBACK_NOT; |
| 5532 |
cd->assert_depth += 1; |
cd->assert_depth += 1; |
| 5533 |
ptr += 2; |
ptr += 2; |
| 5534 |
break; |
break; |
| 5535 |
|
|
| 5773 |
|
|
| 5774 |
temp = cd->end_pattern; |
temp = cd->end_pattern; |
| 5775 |
cd->end_pattern = ptr; |
cd->end_pattern = ptr; |
| 5776 |
recno = find_parens(cd, name, namelen, |
recno = find_parens(cd, name, namelen, |
| 5777 |
(options & PCRE_EXTENDED) != 0, utf8); |
(options & PCRE_EXTENDED) != 0, utf8); |
| 5778 |
cd->end_pattern = temp; |
cd->end_pattern = temp; |
| 5779 |
if (recno < 0) recno = 0; /* Forward ref; set dummy number */ |
if (recno < 0) recno = 0; /* Forward ref; set dummy number */ |
| 5928 |
|
|
| 5929 |
/* If not a forward reference, and the subpattern is still open, |
/* If not a forward reference, and the subpattern is still open, |
| 5930 |
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 |
| 5931 |
recursion that could loop for ever, and diagnose that case. */ |
recursion that could loop for ever, and diagnose that case. We |
| 5932 |
|
must not, however, do this check if we are in a conditional |
| 5933 |
|
subpattern because the condition might be testing for recursion in |
| 5934 |
|
a pattern such as /(?(R)a+|(?R)b)/, which is perfectly valid. |
| 5935 |
|
Forever loops are also detected at runtime, so those that occur in |
| 5936 |
|
conditional subpatterns will be picked up then. */ |
| 5937 |
|
|
| 5938 |
else if (GET(called, 1) == 0 && |
else if (GET(called, 1) == 0 && cond_depth <= 0 && |
| 5939 |
could_be_empty(called, code, bcptr, utf8, cd)) |
could_be_empty(called, code, bcptr, utf8, cd)) |
| 5940 |
{ |
{ |
| 5941 |
*errorcodeptr = ERR40; |
*errorcodeptr = ERR40; |
| 5944 |
} |
} |
| 5945 |
|
|
| 5946 |
/* Insert the recursion/subroutine item. */ |
/* Insert the recursion/subroutine item. */ |
| 5947 |
|
|
| 5948 |
*code = OP_RECURSE; |
*code = OP_RECURSE; |
| 5949 |
PUT(code, 1, (int)(called - cd->start_code)); |
PUT(code, 1, (int)(called - cd->start_code)); |
| 5950 |
code += 1 + LINK_SIZE; |
code += 1 + LINK_SIZE; |
| 6062 |
skipbytes = 2; |
skipbytes = 2; |
| 6063 |
} |
} |
| 6064 |
|
|
| 6065 |
/* Process nested bracketed regex. Assertions may not be repeated, but |
/* Process nested bracketed regex. Assertions used not to be repeatable, |
| 6066 |
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 |
| 6067 |
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 |
| 6068 |
because some compilers complain otherwise. */ |
be able to pass its address because some compilers complain otherwise. */ |
| 6069 |
|
|
| 6070 |
previous = (bravalue >= OP_ONCE)? code : NULL; |
previous = code; /* For handling repetition */ |
| 6071 |
*code = bravalue; |
*code = bravalue; |
| 6072 |
tempcode = code; |
tempcode = code; |
| 6073 |
tempreqvary = cd->req_varyopt; /* Save value before bracket */ |
tempreqvary = cd->req_varyopt; /* Save value before bracket */ |
| 6074 |
length_prevgroup = 0; /* Initialize for pre-compile phase */ |
tempbracount = cd->bracount; /* Save value before bracket */ |
| 6075 |
|
length_prevgroup = 0; /* Initialize for pre-compile phase */ |
| 6076 |
|
|
| 6077 |
if (!compile_regex( |
if (!compile_regex( |
| 6078 |
newoptions, /* The complete new option state */ |
newoptions, /* The complete new option state */ |
| 6079 |
&tempcode, /* Where to put code (updated) */ |
&tempcode, /* Where to put code (updated) */ |
| 6080 |
&ptr, /* Input pointer (updated) */ |
&ptr, /* Input pointer (updated) */ |
| 6081 |
errorcodeptr, /* Where to put an error message */ |
errorcodeptr, /* Where to put an error message */ |
| 6082 |
(bravalue == OP_ASSERTBACK || |
(bravalue == OP_ASSERTBACK || |
| 6083 |
bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */ |
bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */ |
| 6084 |
reset_bracount, /* True if (?| group */ |
reset_bracount, /* True if (?| group */ |
| 6085 |
skipbytes, /* Skip over bracket number */ |
skipbytes, /* Skip over bracket number */ |
| 6086 |
&subfirstbyte, /* For possible first char */ |
cond_depth + |
| 6087 |
&subreqbyte, /* For possible last char */ |
((bravalue == OP_COND)?1:0), /* Depth of condition subpatterns */ |
| 6088 |
bcptr, /* Current branch chain */ |
&subfirstbyte, /* For possible first char */ |
| 6089 |
cd, /* Tables block */ |
&subreqbyte, /* For possible last char */ |
| 6090 |
(lengthptr == NULL)? NULL : /* Actual compile phase */ |
bcptr, /* Current branch chain */ |
| 6091 |
&length_prevgroup /* Pre-compile phase */ |
cd, /* Tables block */ |
| 6092 |
|
(lengthptr == NULL)? NULL : /* Actual compile phase */ |
| 6093 |
|
&length_prevgroup /* Pre-compile phase */ |
| 6094 |
)) |
)) |
| 6095 |
goto FAILED; |
goto FAILED; |
| 6096 |
|
|
| 6097 |
|
/* If this was an atomic group and there are no capturing groups within it, |
| 6098 |
|
generate OP_ONCE_NC instead of OP_ONCE. */ |
| 6099 |
|
|
| 6100 |
|
if (bravalue == OP_ONCE && cd->bracount <= tempbracount) |
| 6101 |
|
*code = OP_ONCE_NC; |
| 6102 |
|
|
| 6103 |
if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT) |
if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT) |
| 6104 |
cd->assert_depth -= 1; |
cd->assert_depth -= 1; |
| 6105 |
|
|
| 6106 |
/* 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 |
| 6107 |
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. |
| 6108 |
and any option resetting that may follow it. The pattern pointer (ptr) |
The pattern pointer (ptr) is on the bracket. |
|
is on the bracket. */ |
|
| 6109 |
|
|
| 6110 |
/* 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 |
| 6111 |
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 |
| 6112 |
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 |
| 6113 |
not be available. */ |
not be available. */ |
| 6348 |
ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET)) |
ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET)) |
| 6349 |
{ |
{ |
| 6350 |
*errorcodeptr = ERR69; |
*errorcodeptr = ERR69; |
| 6351 |
break; |
break; |
| 6352 |
} |
} |
| 6353 |
is_recurse = FALSE; |
is_recurse = FALSE; |
| 6354 |
terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? |
terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? |
| 6355 |
CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)? |
CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)? |
| 6356 |
CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET; |
CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET; |
| 6357 |
goto NAMED_REF_OR_RECURSE; |
goto NAMED_REF_OR_RECURSE; |
| 6358 |
} |
} |
| 6359 |
|
|
| 6360 |
/* Back references are handled specially; must disable firstbyte if |
/* Back references are handled specially; must disable firstbyte if |
| 6361 |
not set to cope with cases like (?=(\w+))\1: which would otherwise set |
not set to cope with cases like (?=(\w+))\1: which would otherwise set |
| 6506 |
else firstbyte = reqbyte = REQ_NONE; |
else firstbyte = reqbyte = REQ_NONE; |
| 6507 |
} |
} |
| 6508 |
|
|
| 6509 |
/* 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 |
| 6510 |
1 or the matching is caseful. */ |
1 or the matching is caseful. */ |
| 6511 |
|
|
| 6512 |
else |
else |
| 6553 |
lookbehind TRUE if this is a lookbehind assertion |
lookbehind TRUE if this is a lookbehind assertion |
| 6554 |
reset_bracount TRUE to reset the count for each branch |
reset_bracount TRUE to reset the count for each branch |
| 6555 |
skipbytes skip this many bytes at start (for brackets and OP_COND) |
skipbytes skip this many bytes at start (for brackets and OP_COND) |
| 6556 |
|
cond_depth depth of nesting for conditional subpatterns |
| 6557 |
firstbyteptr place to put the first required character, or a negative number |
firstbyteptr place to put the first required character, or a negative number |
| 6558 |
reqbyteptr place to put the last required character, or a negative number |
reqbyteptr place to put the last required character, or a negative number |
| 6559 |
bcptr pointer to the chain of currently open branches |
bcptr pointer to the chain of currently open branches |
| 6567 |
static BOOL |
static BOOL |
| 6568 |
compile_regex(int options, uschar **codeptr, const uschar **ptrptr, |
compile_regex(int options, uschar **codeptr, const uschar **ptrptr, |
| 6569 |
int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes, |
int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes, |
| 6570 |
int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd, |
int cond_depth, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, |
| 6571 |
int *lengthptr) |
compile_data *cd, int *lengthptr) |
| 6572 |
{ |
{ |
| 6573 |
const uschar *ptr = *ptrptr; |
const uschar *ptr = *ptrptr; |
| 6574 |
uschar *code = *codeptr; |
uschar *code = *codeptr; |
| 6605 |
|
|
| 6606 |
/* 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 |
| 6607 |
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 |
| 6608 |
detect groups that contain recursive back references to themselves. Note that |
detect groups that contain recursive back references to themselves. Note that |
| 6609 |
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, |
| 6610 |
e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */ |
e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */ |
| 6611 |
|
|
| 6612 |
if (*code == OP_CBRA) |
if (*code == OP_CBRA) |
| 6647 |
into the length. */ |
into the length. */ |
| 6648 |
|
|
| 6649 |
if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte, |
if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte, |
| 6650 |
&branchreqbyte, &bc, cd, (lengthptr == NULL)? NULL : &length)) |
&branchreqbyte, &bc, cond_depth, cd, |
| 6651 |
|
(lengthptr == NULL)? NULL : &length)) |
| 6652 |
{ |
{ |
| 6653 |
*ptrptr = ptr; |
*ptrptr = ptr; |
| 6654 |
return FALSE; |
return FALSE; |
| 6723 |
} |
} |
| 6724 |
else if (fixed_length < 0) |
else if (fixed_length < 0) |
| 6725 |
{ |
{ |
| 6726 |
*errorcodeptr = (fixed_length == -2)? ERR36 : ERR25; |
*errorcodeptr = (fixed_length == -2)? ERR36 : |
| 6727 |
|
(fixed_length == -4)? ERR70: ERR25; |
| 6728 |
*ptrptr = ptr; |
*ptrptr = ptr; |
| 6729 |
return FALSE; |
return FALSE; |
| 6730 |
} |
} |
| 6899 |
|
|
| 6900 |
/* Other brackets */ |
/* Other brackets */ |
| 6901 |
|
|
| 6902 |
else if (op == OP_ASSERT || op == OP_ONCE || op == OP_COND) |
else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC || |
| 6903 |
|
op == OP_COND) |
| 6904 |
{ |
{ |
| 6905 |
if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; |
if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; |
| 6906 |
} |
} |
| 7004 |
|
|
| 7005 |
/* Other brackets */ |
/* Other brackets */ |
| 7006 |
|
|
| 7007 |
else if (op == OP_ASSERT || op == OP_ONCE) |
else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC) |
| 7008 |
{ |
{ |
| 7009 |
if (!is_startline(scode, bracket_map, backref_map)) return FALSE; |
if (!is_startline(scode, bracket_map, backref_map)) return FALSE; |
| 7010 |
} |
} |
| 7074 |
case OP_SCBRAPOS: |
case OP_SCBRAPOS: |
| 7075 |
case OP_ASSERT: |
case OP_ASSERT: |
| 7076 |
case OP_ONCE: |
case OP_ONCE: |
| 7077 |
|
case OP_ONCE_NC: |
| 7078 |
case OP_COND: |
case OP_COND: |
| 7079 |
if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0) |
if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0) |
| 7080 |
return -1; |
return -1; |
| 7084 |
case OP_EXACT: |
case OP_EXACT: |
| 7085 |
scode += 2; |
scode += 2; |
| 7086 |
/* Fall through */ |
/* Fall through */ |
| 7087 |
|
|
| 7088 |
case OP_CHAR: |
case OP_CHAR: |
| 7089 |
case OP_PLUS: |
case OP_PLUS: |
| 7090 |
case OP_MINPLUS: |
case OP_MINPLUS: |
| 7097 |
case OP_EXACTI: |
case OP_EXACTI: |
| 7098 |
scode += 2; |
scode += 2; |
| 7099 |
/* Fall through */ |
/* Fall through */ |
| 7100 |
|
|
| 7101 |
case OP_CHARI: |
case OP_CHARI: |
| 7102 |
case OP_PLUSI: |
case OP_PLUSI: |
| 7103 |
case OP_MINPLUSI: |
case OP_MINPLUSI: |
| 7257 |
|
|
| 7258 |
/* 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 |
| 7259 |
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 |
| 7260 |
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 |
| 7261 |
not used here. */ |
not used here. */ |
| 7262 |
|
|
| 7263 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
| 7287 |
|
|
| 7288 |
/* Check validity of \R options. */ |
/* Check validity of \R options. */ |
| 7289 |
|
|
| 7290 |
switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) |
if ((options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) == |
| 7291 |
|
(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) |
| 7292 |
{ |
{ |
| 7293 |
case 0: |
errorcode = ERR56; |
| 7294 |
case PCRE_BSR_ANYCRLF: |
goto PCRE_EARLY_ERROR_RETURN; |
|
case PCRE_BSR_UNICODE: |
|
|
break; |
|
|
default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN; |
|
| 7295 |
} |
} |
| 7296 |
|
|
| 7297 |
/* Handle different types of newline. The three bits give seven cases. The |
/* Handle different types of newline. The three bits give seven cases. The |
| 7376 |
ptr += skipatstart; |
ptr += skipatstart; |
| 7377 |
code = cworkspace; |
code = cworkspace; |
| 7378 |
*code = OP_BRA; |
*code = OP_BRA; |
| 7379 |
(void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE, |
(void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE, |
| 7380 |
FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length); |
FALSE, 0, 0, &firstbyte, &reqbyte, NULL, cd, &length); |
| 7381 |
if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN; |
if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN; |
| 7382 |
|
|
| 7383 |
DPRINTF(("end pre-compile: length=%d workspace=%d\n", length, |
DPRINTF(("end pre-compile: length=%d workspace=%d\n", length, |
| 7450 |
ptr = (const uschar *)pattern + skipatstart; |
ptr = (const uschar *)pattern + skipatstart; |
| 7451 |
code = (uschar *)codestart; |
code = (uschar *)codestart; |
| 7452 |
*code = OP_BRA; |
*code = OP_BRA; |
| 7453 |
(void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, |
(void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0, |
| 7454 |
&firstbyte, &reqbyte, NULL, cd, NULL); |
&firstbyte, &reqbyte, NULL, cd, NULL); |
| 7455 |
re->top_bracket = cd->bracount; |
re->top_bracket = cd->bracount; |
| 7456 |
re->top_backref = cd->top_backref; |
re->top_backref = cd->top_backref; |
| 7457 |
re->flags = cd->external_flags; |
re->flags = cd->external_flags; |
| 7458 |
|
|
| 7459 |
if (cd->had_accept) reqbyte = -1; /* Must disable after (*ACCEPT) */ |
if (cd->had_accept) reqbyte = REQ_NONE; /* Must disable after (*ACCEPT) */ |
| 7460 |
|
|
| 7461 |
/* 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. */ |
| 7462 |
|
|
| 7523 |
DPRINTF(("fixed length = %d\n", fixed_length)); |
DPRINTF(("fixed length = %d\n", fixed_length)); |
| 7524 |
if (fixed_length < 0) |
if (fixed_length < 0) |
| 7525 |
{ |
{ |
| 7526 |
errorcode = (fixed_length == -2)? ERR36 : ERR25; |
errorcode = (fixed_length == -2)? ERR36 : |
| 7527 |
|
(fixed_length == -4)? ERR70 : ERR25; |
| 7528 |
break; |
break; |
| 7529 |
} |
} |
| 7530 |
PUT(cc, 1, fixed_length); |
PUT(cc, 1, fixed_length); |