embedaddon/pcre/pcre_compile.c - diff

Return to pcre_compile.c CVS log

Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / pcre

Diff for /embedaddon/pcre/pcre_compile.c between versions 1.1.1.1 and 1.1.1.3

version 1.1.1.1, 2012/02/21 23:05:51	version 1.1.1.3, 2012/10/09 09:19:17
Line 6	Line 6
and semantics are as close as possible to those of the Perl 5 language.	and semantics are as close as possible to those of the Perl 5 language.

Written by Philip Hazel	Written by Philip Hazel
Copyright (c) 1997-2011 University of Cambridge	Copyright (c) 1997-2012 University of Cambridge

-----------------------------------------------------------------------------	-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without	Redistribution and use in source and binary forms, with or without
Line 53 supporting internal functions that are not used by oth	Line 53 supporting internal functions that are not used by oth
#include "pcre_internal.h"	#include "pcre_internal.h"


/* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is	/* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which
also used by pcretest. PCRE_DEBUG is not defined when building a production	is also used by pcretest. PCRE_DEBUG is not defined when building a production
library. */	library. We do not need to select pcre16_printint.c specially, because the
	COMPILE_PCREx macro will already be appropriately set. */

#ifdef PCRE_DEBUG	#ifdef PCRE_DEBUG
#include "pcre_printint.src"	/* pcre_printint.c should not include any headers */
	#define PCRE_INCLUDED
	#include "pcre_printint.c"
	#undef PCRE_INCLUDED
#endif	#endif


Line 104 overrun before it actually does run off the end of the	Line 108 overrun before it actually does run off the end of the

#define WORK_SIZE_SAFETY_MARGIN (100)	#define WORK_SIZE_SAFETY_MARGIN (100)

	/* Private flags added to firstchar and reqchar. */

	#define REQ_CASELESS 0x10000000l /* Indicates caselessness */
	#define REQ_VARY 0x20000000l /* Reqchar followed non-literal item */

	/* Repeated character flags. */

	#define UTF_LENGTH 0x10000000l /* The char contains its length. */

/* Table for handling escaped characters in the range '0'-'z'. Positive returns	/* Table for handling escaped characters in the range '0'-'z'. Positive returns
are simple data values; negative values are for special things like \d and so	are simple data values; negative values are for special things like \d and so
on. Zero means further processing is needed (for things like \x), or the escape	on. Zero means further processing is needed (for things like \x), or the escape
Line 238 static const char posix_names[] =	Line 250 static const char posix_names[] =
STRING_graph0 STRING_print0 STRING_punct0 STRING_space0	STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
STRING_word0 STRING_xdigit;	STRING_word0 STRING_xdigit;

static const uschar posix_name_lengths[] = {	static const pcre_uint8 posix_name_lengths[] = {
5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };	5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };

/* Table of class bit maps for each POSIX class. Each class is formed from a	/* Table of class bit maps for each POSIX class. Each class is formed from a
Line 273 substitutes must be in the order of the names, defined	Line 285 substitutes must be in the order of the names, defined
both positive and negative cases. NULL means no substitute. */	both positive and negative cases. NULL means no substitute. */

#ifdef SUPPORT_UCP	#ifdef SUPPORT_UCP
static const uschar *substitutes[] = {	static const pcre_uchar string_PNd[] = {
(uschar )"\\P{Nd}", / \D */	CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
(uschar )"\\p{Nd}", / \d */	CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
(uschar )"\\P{Xsp}", / \S / / NOTE: Xsp is Perl space */	static const pcre_uchar string_pNd[] = {
(uschar )"\\p{Xsp}", / \s */	CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
(uschar )"\\P{Xwd}", / \W */	CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
(uschar )"\\p{Xwd}" / \w */	static const pcre_uchar string_PXsp[] = {
	CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
	CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
	static const pcre_uchar string_pXsp[] = {
	CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
	CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
	static const pcre_uchar string_PXwd[] = {
	CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
	CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
	static const pcre_uchar string_pXwd[] = {
	CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
	CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };

	static const pcre_uchar *substitutes[] = {
	string_PNd, /* \D */
	string_pNd, /* \d */
	string_PXsp, /* \S / / NOTE: Xsp is Perl space */
	string_pXsp, /* \s */
	string_PXwd, /* \W */
	string_pXwd /* \w */
};	};

static const uschar *posix_substitutes[] = {	static const pcre_uchar string_pL[] = {
(uschar )"\\p{L}", / alpha */	CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
(uschar )"\\p{Ll}", / lower */	CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
(uschar )"\\p{Lu}", / upper */	static const pcre_uchar string_pLl[] = {
(uschar )"\\p{Xan}", / alnum */	CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
NULL, /* ascii */	CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
(uschar )"\\h", / blank */	static const pcre_uchar string_pLu[] = {
NULL, /* cntrl */	CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
(uschar )"\\p{Nd}", / digit */	CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
NULL, /* graph */	static const pcre_uchar string_pXan[] = {
NULL, /* print */	CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
NULL, /* punct */	CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
(uschar )"\\p{Xps}", / space / / NOTE: Xps is POSIX space */	static const pcre_uchar string_h[] = {
(uschar )"\\p{Xwd}", / word */	CHAR_BACKSLASH, CHAR_h, '\0' };
NULL, /* xdigit */	static const pcre_uchar string_pXps[] = {
	CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
	CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
	static const pcre_uchar string_PL[] = {
	CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
	CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
	static const pcre_uchar string_PLl[] = {
	CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
	CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
	static const pcre_uchar string_PLu[] = {
	CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
	CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
	static const pcre_uchar string_PXan[] = {
	CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
	CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
	static const pcre_uchar string_H[] = {
	CHAR_BACKSLASH, CHAR_H, '\0' };
	static const pcre_uchar string_PXps[] = {
	CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
	CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };

	static const pcre_uchar *posix_substitutes[] = {
	string_pL, /* alpha */
	string_pLl, /* lower */
	string_pLu, /* upper */
	string_pXan, /* alnum */
	NULL, /* ascii */
	string_h, /* blank */
	NULL, /* cntrl */
	string_pNd, /* digit */
	NULL, /* graph */
	NULL, /* print */
	NULL, /* punct */
	string_pXps, /* space / / NOTE: Xps is POSIX space */
	string_pXwd, /* word */
	NULL, /* xdigit */
/* Negated cases */	/* Negated cases */
(uschar )"\\P{L}", / ^alpha */	string_PL, /* ^alpha */
(uschar )"\\P{Ll}", / ^lower */	string_PLl, /* ^lower */
(uschar )"\\P{Lu}", / ^upper */	string_PLu, /* ^upper */
(uschar )"\\P{Xan}", / ^alnum */	string_PXan, /* ^alnum */
NULL, /* ^ascii */	NULL, /* ^ascii */
(uschar )"\\H", / ^blank */	string_H, /* ^blank */
NULL, /* ^cntrl */	NULL, /* ^cntrl */
(uschar )"\\P{Nd}", / ^digit */	string_PNd, /* ^digit */
NULL, /* ^graph */	NULL, /* ^graph */
NULL, /* ^print */	NULL, /* ^print */
NULL, /* ^punct */	NULL, /* ^punct */
(uschar )"\\P{Xps}", / ^space / / NOTE: Xps is POSIX space */	string_PXps, /* ^space / / NOTE: Xps is POSIX space */
(uschar )"\\P{Xwd}", / ^word */	string_PXwd, /* ^word */
NULL /* ^xdigit */	NULL /* ^xdigit */
};	};
#define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))	#define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))
#endif	#endif

#define STRING(a) # a	#define STRING(a) # a
Line 372 static const char error_texts[] =	Line 438 static const char error_texts[] =
/* 30 */	/* 30 */
"unknown POSIX class name\0"	"unknown POSIX class name\0"
"POSIX collating elements are not supported\0"	"POSIX collating elements are not supported\0"
"this version of PCRE is not compiled with PCRE_UTF8 support\0"	"this version of PCRE is compiled without UTF support\0"
"spare error\0" / DEAD /	"spare error\0" / DEAD /
"character value in \\x{...} sequence is too large\0"	"character value in \\x{...} sequence is too large\0"
/* 35 */	/* 35 */
Line 395 static const char error_texts[] =	Line 461 static const char error_texts[] =
"too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"	"too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
/* 50 */	/* 50 */
"repeated subpattern is too long\0" / DEAD /	"repeated subpattern is too long\0" / DEAD /
"octal value is greater than \\377 (not in UTF-8 mode)\0"	"octal value is greater than \\377 in 8-bit non-UTF-8 mode\0"
"internal error: overran compiling workspace\0"	"internal error: overran compiling workspace\0"
"internal error: previously-checked referenced subpattern not found\0"	"internal error: previously-checked referenced subpattern not found\0"
"DEFINE group contains more than one branch\0"	"DEFINE group contains more than one branch\0"
Line 414 static const char error_texts[] =	Line 480 static const char error_texts[] =
/* 65 */	/* 65 */
"different names for subpatterns of the same number are not allowed\0"	"different names for subpatterns of the same number are not allowed\0"
"(*MARK) must have an argument\0"	"(*MARK) must have an argument\0"
"this version of PCRE is not compiled with PCRE_UCP support\0"	"this version of PCRE is not compiled with Unicode property support\0"
"\\c must be followed by an ASCII character\0"	"\\c must be followed by an ASCII character\0"
"\\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"
/* 70 */	/* 70 */
"internal error: unknown opcode in find_fixedlength()\0"	"internal error: unknown opcode in find_fixedlength()\0"
"\\N is not supported in a class\0"	"\\N is not supported in a class\0"
"too many forward references\0"	"too many forward references\0"
	"disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
	"invalid UTF-16 string\0"
	/* 75 */
	"name is too long in (MARK), (PRUNE), (SKIP), or (THEN)\0"
	"character value in \\u.... sequence is too large\0"
;	;

/* Table to identify digits and hex digits. This is used when compiling	/* Table to identify digits and hex digits. This is used when compiling
Line 439 For convenience, we use the same bit definitions as in	Line 510 For convenience, we use the same bit definitions as in

Then we can use ctype_digit and ctype_xdigit in the code. */	Then we can use ctype_digit and ctype_xdigit in the code. */

	/* Using a simple comparison for decimal numbers rather than a memory read
	is much faster, and the resulting code is simpler (the compiler turns it
	into a subtraction and unsigned comparison). */

	#define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9)

#ifndef EBCDIC	#ifndef EBCDIC

/* This is the "normal" case, for ASCII systems, and EBCDIC systems running in	/* This is the "normal" case, for ASCII systems, and EBCDIC systems running in
UTF-8 mode. */	UTF-8 mode. */

static const unsigned char digitab[] =	static const pcre_uint8 digitab[] =
{	{
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 */	0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */	0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */
Line 483 static const unsigned char digitab[] =	Line 560 static const unsigned char digitab[] =

/* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */	/* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */

static const unsigned char digitab[] =	static const pcre_uint8 digitab[] =
{	{
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 0 */	0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 0 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */	0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */
Line 518 static const unsigned char digitab[] =	Line 595 static const unsigned char digitab[] =
0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /* 0 - 7 F0 */	0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /* 0 - 7 F0 */
0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/* 8 -255 */	0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/* 8 -255 */

static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */	static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */
0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 0- 7 */	0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 0- 7 */
0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /* 8- 15 */	0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /* 8- 15 */
0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 16- 23 */	0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 16- 23 */
Line 557 static const unsigned char ebcdic_chartab[] = { /* cha	Line 634 static const unsigned char ebcdic_chartab[] = { /* cha
/* Definition to allow mutual recursion */	/* Definition to allow mutual recursion */

static BOOL	static BOOL
compile_regex(int, uschar , const uschar , int *, BOOL, BOOL, int, int,	compile_regex(int, pcre_uchar , const pcre_uchar , int *, BOOL, BOOL, int, int,
int , int , branch_chain , compile_data , int *);	int , int , branch_chain , compile_data , int *);


Line 604 Returns: 0 if all went well, else an error number	Line 681 Returns: 0 if all went well, else an error number
static int	static int
expand_workspace(compile_data *cd)	expand_workspace(compile_data *cd)
{	{
uschar *newspace;	pcre_uchar *newspace;
int newsize = cd->workspace_size * 2;	int newsize = cd->workspace_size * 2;

if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;	if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;
Line 612 if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX \|\|	Line 689 if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX \|\|
newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)	newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)
return ERR72;	return ERR72;

newspace = (pcre_malloc)(newsize);	newspace = (PUBL(malloc))(IN_UCHARS(newsize));
if (newspace == NULL) return ERR21;	if (newspace == NULL) return ERR21;
	memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar));
memcpy(newspace, cd->start_workspace, cd->workspace_size);	cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace);
cd->hwm = (uschar *)newspace + (cd->hwm - cd->start_workspace);
if (cd->workspace_size > COMPILE_WORK_SIZE)	if (cd->workspace_size > COMPILE_WORK_SIZE)
(pcre_free)((void *)cd->start_workspace);	(PUBL(free))((void *)cd->start_workspace);
cd->start_workspace = newspace;	cd->start_workspace = newspace;
cd->workspace_size = newsize;	cd->workspace_size = newsize;
return 0;	return 0;
Line 642 Returns: TRUE or FALSE	Line 718 Returns: TRUE or FALSE
*/	*/

static BOOL	static BOOL
is_counted_repeat(const uschar *p)	is_counted_repeat(const pcre_uchar *p)
{	{
if ((digitab[*p++] & ctype_digit) == 0) return FALSE;	if (!IS_DIGIT(*p)) return FALSE;
while ((digitab[*p] & ctype_digit) != 0) p++;	p++;
	while (IS_DIGIT(*p)) p++;
if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;	if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;

if (*p++ != CHAR_COMMA) return FALSE;	if (*p++ != CHAR_COMMA) return FALSE;
if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;	if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;

if ((digitab[*p++] & ctype_digit) == 0) return FALSE;	if (!IS_DIGIT(*p)) return FALSE;
while ((digitab[*p] & ctype_digit) != 0) p++;	p++;
	while (IS_DIGIT(*p)) p++;

return (*p == CHAR_RIGHT_CURLY_BRACKET);	return (*p == CHAR_RIGHT_CURLY_BRACKET);
}	}
Line 684 Returns: zero or positive => a data character	Line 762 Returns: zero or positive => a data character
*/	*/

static int	static int
check_escape(const uschar *ptrptr, int errorcodeptr, int bracount,	check_escape(const pcre_uchar *ptrptr, int errorcodeptr, int bracount,
int options, BOOL isclass)	int options, BOOL isclass)
{	{
BOOL utf8 = (options & PCRE_UTF8) != 0;	/* PCRE_UTF16 has the same value as PCRE_UTF8. */
const uschar ptr = ptrptr + 1;	BOOL utf = (options & PCRE_UTF8) != 0;
int c, i;	const pcre_uchar ptr = ptrptr + 1;
	pcre_int32 c;
	int i;

GETCHARINCTEST(c, ptr); /* Get character value, increment pointer */	GETCHARINCTEST(c, ptr); /* Get character value, increment pointer */
ptr--; /* Set pointer back to the last byte */	ptr--; /* Set pointer back to the last byte */
Line 703 in a table. A non-zero result is something that can be	Line 783 in a table. A non-zero result is something that can be
Otherwise further processing may be required. */	Otherwise further processing may be required. */

#ifndef EBCDIC /* ASCII/UTF-8 coding */	#ifndef EBCDIC /* ASCII/UTF-8 coding */
else if (c < CHAR_0 \|\| c > CHAR_z) {} /* Not alphanumeric */	/* Not alphanumeric */
	else if (c < CHAR_0 \|\| c > CHAR_z) {}
else if ((i = escapes[c - CHAR_0]) != 0) c = i;	else if ((i = escapes[c - CHAR_0]) != 0) c = i;

#else /* EBCDIC coding */	#else /* EBCDIC coding */
else if (c < 'a' \|\| (ebcdic_chartab[c] & 0x0E) == 0) {} /* Not alphanumeric */	/* Not alphanumeric */
	else if (c < 'a' \|\| (!MAX_255(c) \|\| (ebcdic_chartab[c] & 0x0E) == 0)) {}
else if ((i = escapes[c - 0x48]) != 0) c = i;	else if ((i = escapes[c - 0x48]) != 0) c = i;
#endif	#endif

Line 715 else if ((i = escapes[c - 0x48]) != 0) c = i;	Line 797 else if ((i = escapes[c - 0x48]) != 0) c = i;

else	else
{	{
const uschar *oldptr;	const pcre_uchar *oldptr;
BOOL braced, negated;	BOOL braced, negated;

switch (c)	switch (c)
Line 733 else	Line 815 else
{	{
/* In JavaScript, \u must be followed by four hexadecimal numbers.	/* In JavaScript, \u must be followed by four hexadecimal numbers.
Otherwise it is a lowercase u letter. */	Otherwise it is a lowercase u letter. */
if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0	if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
&& (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0)	&& MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
	&& MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
	&& MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
{	{
c = 0;	c = 0;
for (i = 0; i < 4; ++i)	for (i = 0; i < 4; ++i)
Line 748 else	Line 832 else
c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));	c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
#endif	#endif
}	}

	#ifdef COMPILE_PCRE8
	if (c > (utf ? 0x10ffff : 0xff))
	#else
	#ifdef COMPILE_PCRE16
	if (c > (utf ? 0x10ffff : 0xffff))
	#endif
	#endif
	{
	*errorcodeptr = ERR76;
	}
	else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
}	}
}	}
else	else
Line 788 else	Line 884 else

if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)	if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
{	{
const uschar *p;	const pcre_uchar *p;
for (p = ptr+2; p != 0 && p != CHAR_RIGHT_CURLY_BRACKET; p++)	for (p = ptr+2; p != 0 && p != CHAR_RIGHT_CURLY_BRACKET; p++)
if (p != CHAR_MINUS && (digitab[p] & ctype_digit) == 0) break;	if (p != CHAR_MINUS && !IS_DIGIT(p)) break;
if (p != 0 && p != CHAR_RIGHT_CURLY_BRACKET)	if (p != 0 && p != CHAR_RIGHT_CURLY_BRACKET)
{	{
c = -ESC_k;	c = -ESC_k;
Line 808 else	Line 904 else
}	}
else negated = FALSE;	else negated = FALSE;

	/* The integer range is limited by the machine's int representation. */
c = 0;	c = 0;
while ((digitab[ptr[1]] & ctype_digit) != 0)	while (IS_DIGIT(ptr[1]))
	{
	if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
	{
	c = -1;
	break;
	}
c = c * 10 + *(++ptr) - CHAR_0;	c = c * 10 + *(++ptr) - CHAR_0;
	}
if (c < 0) /* Integer overflow */	if (((unsigned int)c) > INT_MAX) /* Integer overflow */
{	{
	while (IS_DIGIT(ptr[1]))
	ptr++;
*errorcodeptr = ERR61;	*errorcodeptr = ERR61;
break;	break;
}	}
Line 861 else	Line 966 else
if (!isclass)	if (!isclass)
{	{
oldptr = ptr;	oldptr = ptr;
	/* The integer range is limited by the machine's int representation. */
c -= CHAR_0;	c -= CHAR_0;
while ((digitab[ptr[1]] & ctype_digit) != 0)	while (IS_DIGIT(ptr[1]))
	{
	if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
	{
	c = -1;
	break;
	}
c = c * 10 + *(++ptr) - CHAR_0;	c = c * 10 + *(++ptr) - CHAR_0;
if (c < 0) /* Integer overflow */	}
	if (((unsigned int)c) > INT_MAX) /* Integer overflow */
{	{
	while (IS_DIGIT(ptr[1]))
	ptr++;
*errorcodeptr = ERR61;	*errorcodeptr = ERR61;
break;	break;
}	}
Line 891 else	Line 1006 else
/* \0 always starts an octal number, but we may drop through to here with a	/* \0 always starts an octal number, but we may drop through to here with a
larger first octal digit. The original code used just to take the least	larger first octal digit. The original code used just to take the least
significant 8 bits of octal numbers (I think this is what early Perls used	significant 8 bits of octal numbers (I think this is what early Perls used
to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more	to do). Nowadays we allow for larger numbers in UTF-8 mode and 16-bit mode,
than 3 octal digits. */	but no more than 3 octal digits. */

case CHAR_0:	case CHAR_0:
c -= CHAR_0;	c -= CHAR_0;
while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)	while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
c = c * 8 + *(++ptr) - CHAR_0;	c = c * 8 + *(++ptr) - CHAR_0;
if (!utf8 && c > 255) *errorcodeptr = ERR51;	#ifdef COMPILE_PCRE8
	if (!utf && c > 0xff) *errorcodeptr = ERR51;
	#endif
break;	break;

/* \x is complicated. \x{ddd} is a character number which can be greater	/* \x is complicated. \x{ddd} is a character number which can be greater
than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is	than 0xff in utf or non-8bit mode, but only if the ddd are hex digits.
treated as a data character. */	If not, { is treated as a data character. */

case CHAR_x:	case CHAR_x:
if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)	if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
{	{
/* In JavaScript, \x must be followed by two hexadecimal numbers.	/* In JavaScript, \x must be followed by two hexadecimal numbers.
Otherwise it is a lowercase x letter. */	Otherwise it is a lowercase x letter. */
if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0)	if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
	&& MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
{	{
c = 0;	c = 0;
for (i = 0; i < 2; ++i)	for (i = 0; i < 2; ++i)
Line 930 else	Line 1048 else

if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)	if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
{	{
const uschar *pt = ptr + 2;	const pcre_uchar *pt = ptr + 2;
int count = 0;

c = 0;	c = 0;
while ((digitab[*pt] & ctype_xdigit) != 0)	while (MAX_255(pt) && (digitab[pt] & ctype_xdigit) != 0)
{	{
register int cc = *pt++;	register int cc = *pt++;
if (c == 0 && cc == CHAR_0) continue; /* Leading zeroes */	if (c == 0 && cc == CHAR_0) continue; /* Leading zeroes */
count++;

#ifndef EBCDIC /* ASCII/UTF-8 coding */	#ifndef EBCDIC /* ASCII/UTF-8 coding */
if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */	if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */
Line 947 else	Line 1063 else
if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */	if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */
c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));	c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
#endif	#endif

	#ifdef COMPILE_PCRE8
	if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }
	#else
	#ifdef COMPILE_PCRE16
	if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }
	#endif
	#endif
}	}

	if (c < 0)
	{
	while (MAX_255(pt) && (digitab[pt] & ctype_xdigit) != 0) pt++;
	*errorcodeptr = ERR34;
	}

if (*pt == CHAR_RIGHT_CURLY_BRACKET)	if (*pt == CHAR_RIGHT_CURLY_BRACKET)
{	{
if (c < 0 \|\| count > (utf8? 8 : 2)) *errorcodeptr = ERR34;	if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
ptr = pt;	ptr = pt;
break;	break;
}	}
Line 963 else	Line 1093 else
/* Read just a single-byte hex-defined char */	/* Read just a single-byte hex-defined char */

c = 0;	c = 0;
while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)	while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
{	{
int cc; /* Some compilers don't like */	int cc; /* Some compilers don't like */
cc = (++ptr); / ++ in initializers */	cc = (++ptr); / ++ in initializers */
Line 1061 Returns: type value from ucp_type_table, or -1	Line 1191 Returns: type value from ucp_type_table, or -1
*/	*/

static int	static int
get_ucp(const uschar *ptrptr, BOOL negptr, int dptr, int errorcodeptr)	get_ucp(const pcre_uchar *ptrptr, BOOL negptr, int dptr, int errorcodeptr)
{	{
int c, i, bot, top;	int c, i, bot, top;
const uschar ptr = ptrptr;	const pcre_uchar ptr = ptrptr;
char name[32];	pcre_uchar name[32];

c = *(++ptr);	c = *(++ptr);
if (c == 0) goto ERROR_RETURN;	if (c == 0) goto ERROR_RETURN;
Line 1082 if (c == CHAR_LEFT_CURLY_BRACKET)	Line 1212 if (c == CHAR_LEFT_CURLY_BRACKET)
*negptr = TRUE;	*negptr = TRUE;
ptr++;	ptr++;
}	}
for (i = 0; i < (int)sizeof(name) - 1; i++)	for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
{	{
c = *(++ptr);	c = *(++ptr);
if (c == 0) goto ERROR_RETURN;	if (c == 0) goto ERROR_RETURN;
Line 1106 else	Line 1236 else
/* Search for a recognized property name using binary chop */	/* Search for a recognized property name using binary chop */

bot = 0;	bot = 0;
top = _pcre_utt_size;	top = PRIV(utt_size);

while (bot < top)	while (bot < top)
{	{
i = (bot + top) >> 1;	i = (bot + top) >> 1;
c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);	c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
if (c == 0)	if (c == 0)
{	{
*dptr = _pcre_utt[i].value;	*dptr = PRIV(utt)[i].value;
return _pcre_utt[i].type;	return PRIV(utt)[i].type;
}	}
if (c > 0) bot = i + 1; else top = i;	if (c > 0) bot = i + 1; else top = i;
}	}
Line 1153 Returns: pointer to '}' on success;	Line 1283 Returns: pointer to '}' on success;
current ptr on error, with errorcodeptr set non-zero	current ptr on error, with errorcodeptr set non-zero
*/	*/

static const uschar *	static const pcre_uchar *
read_repeat_counts(const uschar p, int minp, int maxp, int errorcodeptr)	read_repeat_counts(const pcre_uchar p, int minp, int maxp, int errorcodeptr)
{	{
int min = 0;	int min = 0;
int max = -1;	int max = -1;
Line 1162 int max = -1;	Line 1292 int max = -1;
/* Read the minimum value and do a paranoid check: a negative value indicates	/* Read the minimum value and do a paranoid check: a negative value indicates
an integer overflow. */	an integer overflow. */

while ((digitab[p] & ctype_digit) != 0) min = min 10 + *p++ - CHAR_0;	while (IS_DIGIT(p)) min = min 10 + *p++ - CHAR_0;
if (min < 0 \|\| min > 65535)	if (min < 0 \|\| min > 65535)
{	{
*errorcodeptr = ERR5;	*errorcodeptr = ERR5;
Line 1177 if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else	Line 1307 if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else
if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)	if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
{	{
max = 0;	max = 0;
while((digitab[p] & ctype_digit) != 0) max = max 10 + *p++ - CHAR_0;	while(IS_DIGIT(p)) max = max 10 + *p++ - CHAR_0;
if (max < 0 \|\| max > 65535)	if (max < 0 \|\| max > 65535)
{	{
*errorcodeptr = ERR5;	*errorcodeptr = ERR5;
Line 1232 Arguments:	Line 1362 Arguments:
name name to seek, or NULL if seeking a numbered subpattern	name name to seek, or NULL if seeking a numbered subpattern
lorn name length, or subpattern number if name is NULL	lorn name length, or subpattern number if name is NULL
xmode TRUE if we are in /x mode	xmode TRUE if we are in /x mode
utf8 TRUE if we are in UTF-8 mode	utf TRUE if we are in UTF-8 / UTF-16 mode
count pointer to the current capturing subpattern number (updated)	count pointer to the current capturing subpattern number (updated)

Returns: the number of the named subpattern, or -1 if not found	Returns: the number of the named subpattern, or -1 if not found
*/	*/

static int	static int
find_parens_sub(uschar *ptrptr, compile_data cd, const uschar *name, int lorn,	find_parens_sub(pcre_uchar *ptrptr, compile_data cd, const pcre_uchar *name, int lorn,
BOOL xmode, BOOL utf8, int *count)	BOOL xmode, BOOL utf, int *count)
{	{
uschar ptr = ptrptr;	pcre_uchar ptr = ptrptr;
int start_count = *count;	int start_count = *count;
int hwm_count = start_count;	int hwm_count = start_count;
BOOL dup_parens = FALSE;	BOOL dup_parens = FALSE;
Line 1309 if (ptr[0] == CHAR_LEFT_PARENTHESIS)	Line 1439 if (ptr[0] == CHAR_LEFT_PARENTHESIS)
ptr[1] != CHAR_EQUALS_SIGN) \|\| *ptr == CHAR_APOSTROPHE)	ptr[1] != CHAR_EQUALS_SIGN) \|\| *ptr == CHAR_APOSTROPHE)
{	{
int term;	int term;
const uschar *thisname;	const pcre_uchar *thisname;
*count += 1;	*count += 1;
if (name == NULL && count == lorn) return count;	if (name == NULL && count == lorn) return count;
term = *ptr++;	term = *ptr++;
Line 1317 if (ptr[0] == CHAR_LEFT_PARENTHESIS)	Line 1447 if (ptr[0] == CHAR_LEFT_PARENTHESIS)
thisname = ptr;	thisname = ptr;
while (*ptr != term) ptr++;	while (*ptr != term) ptr++;
if (name != NULL && lorn == ptr - thisname &&	if (name != NULL && lorn == ptr - thisname &&
strncmp((const char )name, (const char )thisname, lorn) == 0)	STRNCMP_UC_UC(name, thisname, lorn) == 0)
return *count;	return *count;
term++;	term++;
}	}
Line 1360 for (; ptr < cd->end_pattern; ptr++)	Line 1490 for (; ptr < cd->end_pattern; ptr++)
{	{
if (ptr[2] == CHAR_E)	if (ptr[2] == CHAR_E)
ptr+= 2;	ptr+= 2;
else if (strncmp((const char *)ptr+2,	else if (STRNCMP_UC_C8(ptr + 2,
STR_Q STR_BACKSLASH STR_E, 3) == 0)	STR_Q STR_BACKSLASH STR_E, 3) == 0)
ptr += 4;	ptr += 4;
else	else
Line 1408 for (; ptr < cd->end_pattern; ptr++)	Line 1538 for (; ptr < cd->end_pattern; ptr++)
{	{
if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }	if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
ptr++;	ptr++;
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;	if (utf) FORWARDCHAR(ptr);
#endif	#endif
}	}
if (*ptr == 0) goto FAIL_EXIT;	if (*ptr == 0) goto FAIL_EXIT;
Line 1420 for (; ptr < cd->end_pattern; ptr++)	Line 1550 for (; ptr < cd->end_pattern; ptr++)

if (*ptr == CHAR_LEFT_PARENTHESIS)	if (*ptr == CHAR_LEFT_PARENTHESIS)
{	{
int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);	int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
if (rc > 0) return rc;	if (rc > 0) return rc;
if (*ptr == 0) goto FAIL_EXIT;	if (*ptr == 0) goto FAIL_EXIT;
}	}
Line 1466 Arguments:	Line 1596 Arguments:
name name to seek, or NULL if seeking a numbered subpattern	name name to seek, or NULL if seeking a numbered subpattern
lorn name length, or subpattern number if name is NULL	lorn name length, or subpattern number if name is NULL
xmode TRUE if we are in /x mode	xmode TRUE if we are in /x mode
utf8 TRUE if we are in UTF-8 mode	utf TRUE if we are in UTF-8 / UTF-16 mode

Returns: the number of the found subpattern, or -1 if not found	Returns: the number of the found subpattern, or -1 if not found
*/	*/

static int	static int
find_parens(compile_data cd, const uschar name, int lorn, BOOL xmode,	find_parens(compile_data cd, const pcre_uchar name, int lorn, BOOL xmode,
BOOL utf8)	BOOL utf)
{	{
uschar ptr = (uschar )cd->start_pattern;	pcre_uchar ptr = (pcre_uchar )cd->start_pattern;
int count = 0;	int count = 0;
int rc;	int rc;

Line 1486 matching closing parens. That is why we have to have a	Line 1616 matching closing parens. That is why we have to have a

for (;;)	for (;;)
{	{
rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);	rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
if (rc > 0 \|\| *ptr++ == 0) break;	if (rc > 0 \|\| *ptr++ == 0) break;
}	}

Line 1513 Arguments:	Line 1643 Arguments:
Returns: pointer to the first significant opcode	Returns: pointer to the first significant opcode
*/	*/

static const uschar*	static const pcre_uchar*
first_significant_code(const uschar *code, BOOL skipassert)	first_significant_code(const pcre_uchar *code, BOOL skipassert)
{	{
for (;;)	for (;;)
{	{
Line 1525 for (;;)	Line 1655 for (;;)
case OP_ASSERTBACK_NOT:	case OP_ASSERTBACK_NOT:
if (!skipassert) return code;	if (!skipassert) return code;
do code += GET(code, 1); while (*code == OP_ALT);	do code += GET(code, 1); while (*code == OP_ALT);
code += _pcre_OP_lengths[*code];	code += PRIV(OP_lengths)[*code];
break;	break;

case OP_WORD_BOUNDARY:	case OP_WORD_BOUNDARY:
Line 1539 for (;;)	Line 1669 for (;;)
case OP_RREF:	case OP_RREF:
case OP_NRREF:	case OP_NRREF:
case OP_DEF:	case OP_DEF:
code += _pcre_OP_lengths[*code];	code += PRIV(OP_lengths)[*code];
break;	break;

default:	default:
Line 1569 and doing the check at the end; a flag specifies which	Line 1699 and doing the check at the end; a flag specifies which

Arguments:	Arguments:
code points to the start of the pattern (the bracket)	code points to the start of the pattern (the bracket)
utf8 TRUE in UTF-8 mode	utf TRUE in UTF-8 / UTF-16 mode
atend TRUE if called when the pattern is complete	atend TRUE if called when the pattern is complete
cd the "compile data" structure	cd the "compile data" structure

Line 1581 Returns: the fixed length,	Line 1711 Returns: the fixed length,
*/	*/

static int	static int
find_fixedlength(uschar code, BOOL utf8, BOOL atend, compile_data cd)	find_fixedlength(pcre_uchar code, BOOL utf, BOOL atend, compile_data cd)
{	{
int length = -1;	int length = -1;

register int branchlength = 0;	register int branchlength = 0;
register uschar *cc = code + 1 + LINK_SIZE;	register pcre_uchar *cc = code + 1 + LINK_SIZE;

/* Scan along the opcodes for this branch. If we get to the end of the	/* Scan along the opcodes for this branch. If we get to the end of the
branch, check the length against that of the other branches. */	branch, check the length against that of the other branches. */
Line 1594 branch, check the length against that of the other bra	Line 1724 branch, check the length against that of the other bra
for (;;)	for (;;)
{	{
int d;	int d;
uschar ce, cs;	pcre_uchar ce, cs;
register int op = *cc;	register int op = *cc;

switch (op)	switch (op)
{	{
/* We only need to continue for OP_CBRA (normal capturing bracket) and	/* We only need to continue for OP_CBRA (normal capturing bracket) and
Line 1608 for (;;)	Line 1739 for (;;)
case OP_ONCE:	case OP_ONCE:
case OP_ONCE_NC:	case OP_ONCE_NC:
case OP_COND:	case OP_COND:
d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd);	d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd);
if (d < 0) return d;	if (d < 0) return d;
branchlength += d;	branchlength += d;
do cc += GET(cc, 1); while (*cc == OP_ALT);	do cc += GET(cc, 1); while (*cc == OP_ALT);
Line 1639 for (;;)	Line 1770 for (;;)

case OP_RECURSE:	case OP_RECURSE:
if (!atend) return -3;	if (!atend) return -3;
cs = ce = (uschar )cd->start_code + GET(cc, 1); / Start subpattern */	cs = ce = (pcre_uchar )cd->start_code + GET(cc, 1); / Start subpattern */
do ce += GET(ce, 1); while (ce == OP_ALT); / End subpattern */	do ce += GET(ce, 1); while (ce == OP_ALT); / End subpattern */
if (cc > cs && cc < ce) return -1; /* Recursion */	if (cc > cs && cc < ce) return -1; /* Recursion */
d = find_fixedlength(cs + 2, utf8, atend, cd);	d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
if (d < 0) return d;	if (d < 0) return d;
branchlength += d;	branchlength += d;
cc += 1 + LINK_SIZE;	cc += 1 + LINK_SIZE;
Line 1655 for (;;)	Line 1786 for (;;)
case OP_ASSERTBACK:	case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:	case OP_ASSERTBACK_NOT:
do cc += GET(cc, 1); while (*cc == OP_ALT);	do cc += GET(cc, 1); while (*cc == OP_ALT);
/* Fall through */	cc += PRIV(OP_lengths)[*cc];
	break;

/* Skip over things that don't match chars */	/* Skip over things that don't match chars */

Line 1663 for (;;)	Line 1795 for (;;)
case OP_PRUNE_ARG:	case OP_PRUNE_ARG:
case OP_SKIP_ARG:	case OP_SKIP_ARG:
case OP_THEN_ARG:	case OP_THEN_ARG:
cc += cc[1] + _pcre_OP_lengths[*cc];	cc += cc[1] + PRIV(OP_lengths)[*cc];
break;	break;

case OP_CALLOUT:	case OP_CALLOUT:
Line 1690 for (;;)	Line 1822 for (;;)
case OP_SOM:	case OP_SOM:
case OP_THEN:	case OP_THEN:
case OP_WORD_BOUNDARY:	case OP_WORD_BOUNDARY:
cc += _pcre_OP_lengths[*cc];	cc += PRIV(OP_lengths)[*cc];
break;	break;

/* Handle literal characters */	/* Handle literal characters */
Line 1701 for (;;)	Line 1833 for (;;)
case OP_NOTI:	case OP_NOTI:
branchlength++;	branchlength++;
cc += 2;	cc += 2;
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];	if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif	#endif
break;	break;

Line 1714 for (;;)	Line 1846 for (;;)
case OP_NOTEXACT:	case OP_NOTEXACT:
case OP_NOTEXACTI:	case OP_NOTEXACTI:
branchlength += GET2(cc,1);	branchlength += GET2(cc,1);
cc += 4;	cc += 2 + IMM2_SIZE;
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];	if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif	#endif
break;	break;

case OP_TYPEEXACT:	case OP_TYPEEXACT:
branchlength += GET2(cc,1);	branchlength += GET2(cc,1);
if (cc[3] == OP_PROP \|\| cc[3] == OP_NOTPROP) cc += 2;	if (cc[1 + IMM2_SIZE] == OP_PROP \|\| cc[1 + IMM2_SIZE] == OP_NOTPROP) cc += 2;
cc += 4;	cc += 1 + IMM2_SIZE + 1;
break;	break;

/* Handle single-char matchers */	/* Handle single-char matchers */
Line 1757 for (;;)	Line 1889 for (;;)

/* Check a class for variable quantification */	/* Check a class for variable quantification */

#ifdef SUPPORT_UTF8	#if defined SUPPORT_UTF \|\| defined COMPILE_PCRE16
case OP_XCLASS:	case OP_XCLASS:
cc += GET(cc, 1) - 33;	cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
/* Fall through */	/* Fall through */
#endif	#endif

case OP_CLASS:	case OP_CLASS:
case OP_NCLASS:	case OP_NCLASS:
cc += 33;	cc += PRIV(OP_lengths)[OP_CLASS];

switch (*cc)	switch (*cc)
{	{
Line 1779 for (;;)	Line 1911 for (;;)

case OP_CRRANGE:	case OP_CRRANGE:
case OP_CRMINRANGE:	case OP_CRMINRANGE:
if (GET2(cc,1) != GET2(cc,3)) return -1;	if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
branchlength += GET2(cc,1);	branchlength += GET2(cc,1);
cc += 5;	cc += 1 + 2 * IMM2_SIZE;
break;	break;

default:	default:
Line 1896 length.	Line 2028 length.

Arguments:	Arguments:
code points to start of expression	code points to start of expression
utf8 TRUE in UTF-8 mode	utf TRUE in UTF-8 / UTF-16 mode
number the required bracket number or negative to find a lookbehind	number the required bracket number or negative to find a lookbehind

Returns: pointer to the opcode for the bracket, or NULL if not found	Returns: pointer to the opcode for the bracket, or NULL if not found
*/	*/

const uschar *	const pcre_uchar *
_pcre_find_bracket(const uschar *code, BOOL utf8, int number)	PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
{	{
for (;;)	for (;;)
{	{
Line 1921 for (;;)	Line 2053 for (;;)

else if (c == OP_REVERSE)	else if (c == OP_REVERSE)
{	{
if (number < 0) return (uschar *)code;	if (number < 0) return (pcre_uchar *)code;
code += _pcre_OP_lengths[c];	code += PRIV(OP_lengths)[c];
}	}

/* Handle capturing bracket */	/* Handle capturing bracket */
Line 1931 for (;;)	Line 2063 for (;;)
c == OP_CBRAPOS \|\| c == OP_SCBRAPOS)	c == OP_CBRAPOS \|\| c == OP_SCBRAPOS)
{	{
int n = GET2(code, 1+LINK_SIZE);	int n = GET2(code, 1+LINK_SIZE);
if (n == number) return (uschar *)code;	if (n == number) return (pcre_uchar *)code;
code += _pcre_OP_lengths[c];	code += PRIV(OP_lengths)[c];
}	}

/* Otherwise, we can get the item's length from the table, except that for	/* Otherwise, we can get the item's length from the table, except that for
Line 1960 for (;;)	Line 2092 for (;;)
case OP_TYPEMINUPTO:	case OP_TYPEMINUPTO:
case OP_TYPEEXACT:	case OP_TYPEEXACT:
case OP_TYPEPOSUPTO:	case OP_TYPEPOSUPTO:
if (code[3] == OP_PROP \|\| code[3] == OP_NOTPROP) code += 2;	if (code[1 + IMM2_SIZE] == OP_PROP
	\|\| code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;
break;	break;

case OP_MARK:	case OP_MARK:
Line 1976 for (;;)	Line 2109 for (;;)

/* Add in the fixed length from the table */	/* Add in the fixed length from the table */

code += _pcre_OP_lengths[c];	code += PRIV(OP_lengths)[c];

/* In UTF-8 mode, opcodes that are followed by a character may be followed by	/* In UTF-8 mode, opcodes that are followed by a character may be followed by
a multi-byte character. The length in the table is a minimum, so we have to	a multi-byte character. The length in the table is a minimum, so we have to
arrange to skip the extra bytes. */	arrange to skip the extra bytes. */

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8) switch(c)	if (utf) switch(c)
{	{
case OP_CHAR:	case OP_CHAR:
case OP_CHARI:	case OP_CHARI:
Line 2013 for (;;)	Line 2146 for (;;)
case OP_MINQUERYI:	case OP_MINQUERYI:
case OP_POSQUERY:	case OP_POSQUERY:
case OP_POSQUERYI:	case OP_POSQUERYI:
if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];	if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
break;	break;
}	}
#else	#else
(void)(utf8); /* Keep compiler happy by referencing function argument */	(void)(utf); /* Keep compiler happy by referencing function argument */
#endif	#endif
}	}
}	}
Line 2034 instance of OP_RECURSE.	Line 2167 instance of OP_RECURSE.

Arguments:	Arguments:
code points to start of expression	code points to start of expression
utf8 TRUE in UTF-8 mode	utf TRUE in UTF-8 / UTF-16 mode

Returns: pointer to the opcode for OP_RECURSE, or NULL if not found	Returns: pointer to the opcode for OP_RECURSE, or NULL if not found
*/	*/

static const uschar *	static const pcre_uchar *
find_recurse(const uschar *code, BOOL utf8)	find_recurse(const pcre_uchar *code, BOOL utf)
{	{
for (;;)	for (;;)
{	{
Line 2079 for (;;)	Line 2212 for (;;)
case OP_TYPEUPTO:	case OP_TYPEUPTO:
case OP_TYPEMINUPTO:	case OP_TYPEMINUPTO:
case OP_TYPEEXACT:	case OP_TYPEEXACT:
if (code[3] == OP_PROP \|\| code[3] == OP_NOTPROP) code += 2;	if (code[1 + IMM2_SIZE] == OP_PROP
	\|\| code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;
break;	break;

case OP_MARK:	case OP_MARK:
Line 2095 for (;;)	Line 2229 for (;;)

/* Add in the fixed length from the table */	/* Add in the fixed length from the table */

code += _pcre_OP_lengths[c];	code += PRIV(OP_lengths)[c];

/* In UTF-8 mode, opcodes that are followed by a character may be followed	/* In UTF-8 mode, opcodes that are followed by a character may be followed
by a multi-byte character. The length in the table is a minimum, so we have	by a multi-byte character. The length in the table is a minimum, so we have
to arrange to skip the extra bytes. */	to arrange to skip the extra bytes. */

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8) switch(c)	if (utf) switch(c)
{	{
case OP_CHAR:	case OP_CHAR:
case OP_CHARI:	case OP_CHARI:
	case OP_NOT:
	case OP_NOTI:
case OP_EXACT:	case OP_EXACT:
case OP_EXACTI:	case OP_EXACTI:
	case OP_NOTEXACT:
	case OP_NOTEXACTI:
case OP_UPTO:	case OP_UPTO:
case OP_UPTOI:	case OP_UPTOI:
	case OP_NOTUPTO:
	case OP_NOTUPTOI:
case OP_MINUPTO:	case OP_MINUPTO:
case OP_MINUPTOI:	case OP_MINUPTOI:
	case OP_NOTMINUPTO:
	case OP_NOTMINUPTOI:
case OP_POSUPTO:	case OP_POSUPTO:
case OP_POSUPTOI:	case OP_POSUPTOI:
	case OP_NOTPOSUPTO:
	case OP_NOTPOSUPTOI:
case OP_STAR:	case OP_STAR:
case OP_STARI:	case OP_STARI:
	case OP_NOTSTAR:
	case OP_NOTSTARI:
case OP_MINSTAR:	case OP_MINSTAR:
case OP_MINSTARI:	case OP_MINSTARI:
	case OP_NOTMINSTAR:
	case OP_NOTMINSTARI:
case OP_POSSTAR:	case OP_POSSTAR:
case OP_POSSTARI:	case OP_POSSTARI:
	case OP_NOTPOSSTAR:
	case OP_NOTPOSSTARI:
case OP_PLUS:	case OP_PLUS:
case OP_PLUSI:	case OP_PLUSI:
	case OP_NOTPLUS:
	case OP_NOTPLUSI:
case OP_MINPLUS:	case OP_MINPLUS:
case OP_MINPLUSI:	case OP_MINPLUSI:
	case OP_NOTMINPLUS:
	case OP_NOTMINPLUSI:
case OP_POSPLUS:	case OP_POSPLUS:
case OP_POSPLUSI:	case OP_POSPLUSI:
	case OP_NOTPOSPLUS:
	case OP_NOTPOSPLUSI:
case OP_QUERY:	case OP_QUERY:
case OP_QUERYI:	case OP_QUERYI:
	case OP_NOTQUERY:
	case OP_NOTQUERYI:
case OP_MINQUERY:	case OP_MINQUERY:
case OP_MINQUERYI:	case OP_MINQUERYI:
	case OP_NOTMINQUERY:
	case OP_NOTMINQUERYI:
case OP_POSQUERY:	case OP_POSQUERY:
case OP_POSQUERYI:	case OP_POSQUERYI:
if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];	case OP_NOTPOSQUERY:
	case OP_NOTPOSQUERYI:
	if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
break;	break;
}	}
#else	#else
(void)(utf8); /* Keep compiler happy by referencing function argument */	(void)(utf); /* Keep compiler happy by referencing function argument */
#endif	#endif
}	}
}	}
Line 2159 bracket whose current branch will already have been sc	Line 2321 bracket whose current branch will already have been sc
Arguments:	Arguments:
code points to start of search	code points to start of search
endcode points to where to stop	endcode points to where to stop
utf8 TRUE if in UTF8 mode	utf TRUE if in UTF- utf TRUE if in UTF-8 / UTF-16 mode
cd contains pointers to tables etc.	cd contains pointers to tables etc.

Returns: TRUE if what is matched could be empty	Returns: TRUE if what is matched could be empty
*/	*/

static BOOL	static BOOL
could_be_empty_branch(const uschar code, const uschar endcode, BOOL utf8,	could_be_empty_branch(const pcre_uchar code, const pcre_uchar endcode,
compile_data *cd)	BOOL utf, compile_data *cd)
{	{
register int c;	register int c;
for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE);	for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
code < endcode;	code < endcode;
code = first_significant_code(code + _pcre_OP_lengths[c], TRUE))	code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
{	{
const uschar *ccode;	const pcre_uchar *ccode;

c = *code;	c = *code;

Line 2197 for (code = first_significant_code(code + _pcre_OP_len	Line 2359 for (code = first_significant_code(code + _pcre_OP_len

if (c == OP_RECURSE)	if (c == OP_RECURSE)
{	{
const uschar *scode;	const pcre_uchar *scode;
BOOL empty_branch;	BOOL empty_branch;

/* Test for forward reference */	/* Test for forward reference */
Line 2215 for (code = first_significant_code(code + _pcre_OP_len	Line 2377 for (code = first_significant_code(code + _pcre_OP_len

do	do
{	{
if (could_be_empty_branch(scode, endcode, utf8, cd))	if (could_be_empty_branch(scode, endcode, utf, cd))
{	{
empty_branch = TRUE;	empty_branch = TRUE;
break;	break;
Line 2233 for (code = first_significant_code(code + _pcre_OP_len	Line 2395 for (code = first_significant_code(code + _pcre_OP_len
if (c == OP_BRAZERO \|\| c == OP_BRAMINZERO \|\| c == OP_SKIPZERO \|\|	if (c == OP_BRAZERO \|\| c == OP_BRAMINZERO \|\| c == OP_SKIPZERO \|\|
c == OP_BRAPOSZERO)	c == OP_BRAPOSZERO)
{	{
code += _pcre_OP_lengths[c];	code += PRIV(OP_lengths)[c];
do code += GET(code, 1); while (*code == OP_ALT);	do code += GET(code, 1); while (*code == OP_ALT);
c = *code;	c = *code;
continue;	continue;
Line 2271 for (code = first_significant_code(code + _pcre_OP_len	Line 2433 for (code = first_significant_code(code + _pcre_OP_len
empty_branch = FALSE;	empty_branch = FALSE;
do	do
{	{
if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))	if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
empty_branch = TRUE;	empty_branch = TRUE;
code += GET(code, 1);	code += GET(code, 1);
}	}
Line 2289 for (code = first_significant_code(code + _pcre_OP_len	Line 2451 for (code = first_significant_code(code + _pcre_OP_len
{	{
/* Check for quantifiers after a class. XCLASS is used for classes that	/* Check for quantifiers after a class. XCLASS is used for classes that
cannot be represented just by a bit map. This includes negated single	cannot be represented just by a bit map. This includes negated single
high-valued characters. The length in _pcre_OP_lengths[] is zero; the	high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
actual length is stored in the compiled code, so we must update "code"	actual length is stored in the compiled code, so we must update "code"
here. */	here. */

#ifdef SUPPORT_UTF8	#if defined SUPPORT_UTF \|\| !defined COMPILE_PCRE8
case OP_XCLASS:	case OP_XCLASS:
ccode = code += GET(code, 1);	ccode = code += GET(code, 1);
goto CHECK_CLASS_REPEAT;	goto CHECK_CLASS_REPEAT;
Line 2301 for (code = first_significant_code(code + _pcre_OP_len	Line 2463 for (code = first_significant_code(code + _pcre_OP_len

case OP_CLASS:	case OP_CLASS:
case OP_NCLASS:	case OP_NCLASS:
ccode = code + 33;	ccode = code + PRIV(OP_lengths)[OP_CLASS];

#ifdef SUPPORT_UTF8	#if defined SUPPORT_UTF \|\| !defined COMPILE_PCRE8
CHECK_CLASS_REPEAT:	CHECK_CLASS_REPEAT:
#endif	#endif

Line 2376 for (code = first_significant_code(code + _pcre_OP_len	Line 2538 for (code = first_significant_code(code + _pcre_OP_len
case OP_TYPEUPTO:	case OP_TYPEUPTO:
case OP_TYPEMINUPTO:	case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:	case OP_TYPEPOSUPTO:
if (code[3] == OP_PROP \|\| code[3] == OP_NOTPROP) code += 2;	if (code[1 + IMM2_SIZE] == OP_PROP
	\|\| code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;
break;	break;

/* End of branch */	/* End of branch */
Line 2391 for (code = first_significant_code(code + _pcre_OP_len	Line 2554 for (code = first_significant_code(code + _pcre_OP_len
/* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,	/* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
MINUPTO, and POSUPTO may be followed by a multibyte character */	MINUPTO, and POSUPTO may be followed by a multibyte character */

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
case OP_STAR:	case OP_STAR:
case OP_STARI:	case OP_STARI:
case OP_MINSTAR:	case OP_MINSTAR:
Line 2404 for (code = first_significant_code(code + _pcre_OP_len	Line 2567 for (code = first_significant_code(code + _pcre_OP_len
case OP_MINQUERYI:	case OP_MINQUERYI:
case OP_POSQUERY:	case OP_POSQUERY:
case OP_POSQUERYI:	case OP_POSQUERYI:
if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];	if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
break;	break;

case OP_UPTO:	case OP_UPTO:
Line 2413 for (code = first_significant_code(code + _pcre_OP_len	Line 2576 for (code = first_significant_code(code + _pcre_OP_len
case OP_MINUPTOI:	case OP_MINUPTOI:
case OP_POSUPTO:	case OP_POSUPTO:
case OP_POSUPTOI:	case OP_POSUPTOI:
if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];	if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
break;	break;
#endif	#endif

Line 2457 Arguments:	Line 2620 Arguments:
code points to start of the recursion	code points to start of the recursion
endcode points to where to stop (current RECURSE item)	endcode points to where to stop (current RECURSE item)
bcptr points to the chain of current (unclosed) branch starts	bcptr points to the chain of current (unclosed) branch starts
utf8 TRUE if in UTF-8 mode	utf TRUE if in UTF-8 / UTF-16 mode
cd pointers to tables etc	cd pointers to tables etc

Returns: TRUE if what is matched could be empty	Returns: TRUE if what is matched could be empty
*/	*/

static BOOL	static BOOL
could_be_empty(const uschar code, const uschar endcode, branch_chain *bcptr,	could_be_empty(const pcre_uchar code, const pcre_uchar endcode,
BOOL utf8, compile_data *cd)	branch_chain bcptr, BOOL utf, compile_data cd)
{	{
while (bcptr != NULL && bcptr->current_branch >= code)	while (bcptr != NULL && bcptr->current_branch >= code)
{	{
if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))	if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
return FALSE;	return FALSE;
bcptr = bcptr->outer;	bcptr = bcptr->outer;
}	}
Line 2521 Returns: TRUE or FALSE	Line 2684 Returns: TRUE or FALSE
*/	*/

static BOOL	static BOOL
check_posix_syntax(const uschar ptr, const uschar *endptr)	check_posix_syntax(const pcre_uchar ptr, const pcre_uchar *endptr)
{	{
int terminator; /* Don't combine these lines; the Solaris cc */	int terminator; /* Don't combine these lines; the Solaris cc */
terminator = (++ptr); / compiler warns about "non-constant" initializer. */	terminator = (++ptr); / compiler warns about "non-constant" initializer. */
Line 2565 Returns: a value representing the name, or -1 if u	Line 2728 Returns: a value representing the name, or -1 if u
*/	*/

static int	static int
check_posix_name(const uschar *ptr, int len)	check_posix_name(const pcre_uchar *ptr, int len)
{	{
const char *pn = posix_names;	const char *pn = posix_names;
register int yield = 0;	register int yield = 0;
while (posix_name_lengths[yield] != 0)	while (posix_name_lengths[yield] != 0)
{	{
if (len == posix_name_lengths[yield] &&	if (len == posix_name_lengths[yield] &&
strncmp((const char *)ptr, pn, len) == 0) return yield;	STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;
pn += posix_name_lengths[yield] + 1;	pn += posix_name_lengths[yield] + 1;
yield++;	yield++;
}	}
Line 2604 value in the reference (which is a group number).	Line 2767 value in the reference (which is a group number).
Arguments:	Arguments:
group points to the start of the group	group points to the start of the group
adjust the amount by which the group is to be moved	adjust the amount by which the group is to be moved
utf8 TRUE in UTF-8 mode	utf TRUE in UTF-8 / UTF-16 mode
cd contains pointers to tables etc.	cd contains pointers to tables etc.
save_hwm the hwm forward reference pointer at the start of the group	save_hwm the hwm forward reference pointer at the start of the group

Line 2612 Returns: nothing	Line 2775 Returns: nothing
*/	*/

static void	static void
adjust_recurse(uschar group, int adjust, BOOL utf8, compile_data cd,	adjust_recurse(pcre_uchar group, int adjust, BOOL utf, compile_data cd,
uschar *save_hwm)	pcre_uchar *save_hwm)
{	{
uschar *ptr = group;	pcre_uchar *ptr = group;

while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)	while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
{	{
int offset;	int offset;
uschar *hc;	pcre_uchar *hc;

/* See if this recursion is on the forward reference list. If so, adjust the	/* See if this recursion is on the forward reference list. If so, adjust the
reference. */	reference. */
Line 2665 Arguments:	Line 2828 Arguments:
Returns: new code pointer	Returns: new code pointer
*/	*/

static uschar *	static pcre_uchar *
auto_callout(uschar code, const uschar ptr, compile_data *cd)	auto_callout(pcre_uchar code, const pcre_uchar ptr, compile_data *cd)
{	{
*code++ = OP_CALLOUT;	*code++ = OP_CALLOUT;
*code++ = 255;	*code++ = 255;
PUT(code, 0, (int)(ptr - cd->start_pattern)); /* Pattern offset */	PUT(code, 0, (int)(ptr - cd->start_pattern)); /* Pattern offset */
PUT(code, LINK_SIZE, 0); /* Default length */	PUT(code, LINK_SIZE, 0); /* Default length */
return code + 2*LINK_SIZE;	return code + 2 * LINK_SIZE;
}	}


Line 2694 Returns: nothing	Line 2857 Returns: nothing
*/	*/

static void	static void
complete_callout(uschar previous_callout, const uschar ptr, compile_data *cd)	complete_callout(pcre_uchar previous_callout, const pcre_uchar ptr, compile_data *cd)
{	{
int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));	int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
PUT(previous_callout, 2 + LINK_SIZE, length);	PUT(previous_callout, 2 + LINK_SIZE, length);
Line 2777 switch(ptype)	Line 2940 switch(ptype)
prop->chartype == ucp_Lt) == negated;	prop->chartype == ucp_Lt) == negated;

case PT_GC:	case PT_GC:
return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;	return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;

case PT_PC:	case PT_PC:
return (pdata == prop->chartype) == negated;	return (pdata == prop->chartype) == negated;
Line 2788 switch(ptype)	Line 2951 switch(ptype)
/* These are specials */	/* These are specials */

case PT_ALNUM:	case PT_ALNUM:
return (_pcre_ucp_gentype[prop->chartype] == ucp_L \|\|	return (PRIV(ucp_gentype)[prop->chartype] == ucp_L \|\|
_pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;	PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;

case PT_SPACE: /* Perl space */	case PT_SPACE: /* Perl space */
return (_pcre_ucp_gentype[prop->chartype] == ucp_Z \|\|	return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z \|\|
c == CHAR_HT \|\| c == CHAR_NL \|\| c == CHAR_FF \|\| c == CHAR_CR)	c == CHAR_HT \|\| c == CHAR_NL \|\| c == CHAR_FF \|\| c == CHAR_CR)
== negated;	== negated;

case PT_PXSPACE: /* POSIX space */	case PT_PXSPACE: /* POSIX space */
return (_pcre_ucp_gentype[prop->chartype] == ucp_Z \|\|	return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z \|\|
c == CHAR_HT \|\| c == CHAR_NL \|\| c == CHAR_VT \|\|	c == CHAR_HT \|\| c == CHAR_NL \|\| c == CHAR_VT \|\|
c == CHAR_FF \|\| c == CHAR_CR)	c == CHAR_FF \|\| c == CHAR_CR)
== negated;	== negated;

case PT_WORD:	case PT_WORD:
return (_pcre_ucp_gentype[prop->chartype] == ucp_L \|\|	return (PRIV(ucp_gentype)[prop->chartype] == ucp_L \|\|
_pcre_ucp_gentype[prop->chartype] == ucp_N \|\|	PRIV(ucp_gentype)[prop->chartype] == ucp_N \|\|
c == CHAR_UNDERSCORE) == negated;	c == CHAR_UNDERSCORE) == negated;
}	}
return FALSE;	return FALSE;
Line 2823 sense to automatically possessify the repeated item.	Line 2986 sense to automatically possessify the repeated item.

Arguments:	Arguments:
previous pointer to the repeated opcode	previous pointer to the repeated opcode
utf8 TRUE in UTF-8 mode	utf TRUE in UTF-8 / UTF-16 mode
ptr next character in pattern	ptr next character in pattern
options options bits	options options bits
cd contains pointers to tables etc.	cd contains pointers to tables etc.
Line 2832 Returns: TRUE if possessifying is wanted	Line 2995 Returns: TRUE if possessifying is wanted
*/	*/

static BOOL	static BOOL
check_auto_possessive(const uschar previous, BOOL utf8, const uschar ptr,	check_auto_possessive(const pcre_uchar *previous, BOOL utf,
int options, compile_data *cd)	const pcre_uchar ptr, int options, compile_data cd)
{	{
int c, next;	pcre_int32 c, next;
int op_code = *previous++;	int op_code = *previous++;

/* Skip whitespace and comments in extended mode */	/* Skip whitespace and comments in extended mode */
Line 2844 if ((options & PCRE_EXTENDED) != 0)	Line 3007 if ((options & PCRE_EXTENDED) != 0)
{	{
for (;;)	for (;;)
{	{
while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;	while (MAX_255(ptr) && (cd->ctypes[ptr] & ctype_space) != 0) ptr++;
if (*ptr == CHAR_NUMBER_SIGN)	if (*ptr == CHAR_NUMBER_SIGN)
{	{
ptr++;	ptr++;
Line 2852 if ((options & PCRE_EXTENDED) != 0)	Line 3015 if ((options & PCRE_EXTENDED) != 0)
{	{
if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }	if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
ptr++;	ptr++;
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;	if (utf) FORWARDCHAR(ptr);
#endif	#endif
}	}
}	}
Line 2871 if (*ptr == CHAR_BACKSLASH)	Line 3034 if (*ptr == CHAR_BACKSLASH)
if (temperrorcode != 0) return FALSE;	if (temperrorcode != 0) return FALSE;
ptr++; /* Point after the escape sequence */	ptr++; /* Point after the escape sequence */
}	}
	else if (!MAX_255(ptr) \|\| (cd->ctypes[ptr] & ctype_meta) == 0)
else if ((cd->ctypes[*ptr] & ctype_meta) == 0)
{	{
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8) { GETCHARINC(next, ptr); } else	if (utf) { GETCHARINC(next, ptr); } else
#endif	#endif
next = *ptr++;	next = *ptr++;
}	}

else return FALSE;	else return FALSE;

/* Skip whitespace and comments in extended mode */	/* Skip whitespace and comments in extended mode */
Line 2888 if ((options & PCRE_EXTENDED) != 0)	Line 3049 if ((options & PCRE_EXTENDED) != 0)
{	{
for (;;)	for (;;)
{	{
while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;	while (MAX_255(ptr) && (cd->ctypes[ptr] & ctype_space) != 0) ptr++;
if (*ptr == CHAR_NUMBER_SIGN)	if (*ptr == CHAR_NUMBER_SIGN)
{	{
ptr++;	ptr++;
Line 2896 if ((options & PCRE_EXTENDED) != 0)	Line 3057 if ((options & PCRE_EXTENDED) != 0)
{	{
if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }	if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
ptr++;	ptr++;
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;	if (utf) FORWARDCHAR(ptr);
#endif	#endif
}	}
}	}
Line 2908 if ((options & PCRE_EXTENDED) != 0)	Line 3069 if ((options & PCRE_EXTENDED) != 0)
/* If the next thing is itself optional, we have to give up. */	/* If the next thing is itself optional, we have to give up. */

if (ptr == CHAR_ASTERISK \|\| ptr == CHAR_QUESTION_MARK \|\|	if (ptr == CHAR_ASTERISK \|\| ptr == CHAR_QUESTION_MARK \|\|
strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)	STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
return FALSE;	return FALSE;

/* Now compare the next item with the previous opcode. First, handle cases when	/* Now compare the next item with the previous opcode. First, handle cases when
Line 2917 the next item is a character. */	Line 3078 the next item is a character. */
if (next >= 0) switch(op_code)	if (next >= 0) switch(op_code)
{	{
case OP_CHAR:	case OP_CHAR:
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
GETCHARTEST(c, previous);	GETCHARTEST(c, previous);
#else	#else
c = *previous;	c = *previous;
Line 2929 if (next >= 0) switch(op_code)	Line 3090 if (next >= 0) switch(op_code)
high-valued characters. */	high-valued characters. */

case OP_CHARI:	case OP_CHARI:
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
GETCHARTEST(c, previous);	GETCHARTEST(c, previous);
#else	#else
c = *previous;	c = *previous;
#endif	#endif
if (c == next) return FALSE;	if (c == next) return FALSE;
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8)	if (utf)
{	{
unsigned int othercase;	unsigned int othercase;
if (next < 128) othercase = cd->fcc[next]; else	if (next < 128) othercase = cd->fcc[next]; else
Line 2948 if (next >= 0) switch(op_code)	Line 3109 if (next >= 0) switch(op_code)
return (unsigned int)c != othercase;	return (unsigned int)c != othercase;
}	}
else	else
#endif /* SUPPORT_UTF8 */	#endif /* SUPPORT_UTF */
return (c != cd->fcc[next]); /* Non-UTF-8 mode */	return (c != TABLE_GET((unsigned int)next, cd->fcc, next)); /* Non-UTF-8 mode */

/* For OP_NOT and OP_NOTI, the data is always a single-byte character. These
opcodes are not used for multi-byte characters, because they are coded using
an XCLASS instead. */

case OP_NOT:	case OP_NOT:
return (c = *previous) == next;	#ifdef SUPPORT_UTF
	GETCHARTEST(c, previous);
	#else
	c = *previous;
	#endif
	return c == next;

case OP_NOTI:	case OP_NOTI:
if ((c = *previous) == next) return TRUE;	#ifdef SUPPORT_UTF
#ifdef SUPPORT_UTF8	GETCHARTEST(c, previous);
if (utf8)	#else
	c = *previous;
	#endif
	if (c == next) return TRUE;
	#ifdef SUPPORT_UTF
	if (utf)
{	{
unsigned int othercase;	unsigned int othercase;
if (next < 128) othercase = cd->fcc[next]; else	if (next < 128) othercase = cd->fcc[next]; else
#ifdef SUPPORT_UCP	#ifdef SUPPORT_UCP
othercase = UCD_OTHERCASE(next);	othercase = UCD_OTHERCASE((unsigned int)next);
#else	#else
othercase = NOTACHAR;	othercase = NOTACHAR;
#endif	#endif
return (unsigned int)c == othercase;	return (unsigned int)c == othercase;
}	}
else	else
#endif /* SUPPORT_UTF8 */	#endif /* SUPPORT_UTF */
return (c == cd->fcc[next]); /* Non-UTF-8 mode */	return (c == TABLE_GET((unsigned int)next, cd->fcc, next)); /* Non-UTF-8 mode */

/* Note that OP_DIGIT etc. are generated only when PCRE_UCP is not set.	/* Note that OP_DIGIT etc. are generated only when PCRE_UCP is not set.
When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */	When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */

case OP_DIGIT:	case OP_DIGIT:
return next > 127 \|\| (cd->ctypes[next] & ctype_digit) == 0;	return next > 255 \|\| (cd->ctypes[next] & ctype_digit) == 0;

case OP_NOT_DIGIT:	case OP_NOT_DIGIT:
return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;	return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;

case OP_WHITESPACE:	case OP_WHITESPACE:
return next > 127 \|\| (cd->ctypes[next] & ctype_space) == 0;	return next > 255 \|\| (cd->ctypes[next] & ctype_space) == 0;

case OP_NOT_WHITESPACE:	case OP_NOT_WHITESPACE:
return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;	return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;

case OP_WORDCHAR:	case OP_WORDCHAR:
return next > 127 \|\| (cd->ctypes[next] & ctype_word) == 0;	return next > 255 \|\| (cd->ctypes[next] & ctype_word) == 0;

case OP_NOT_WORDCHAR:	case OP_NOT_WORDCHAR:
return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;	return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;

case OP_HSPACE:	case OP_HSPACE:
case OP_NOT_HSPACE:	case OP_NOT_HSPACE:
Line 3065 switch(op_code)	Line 3232 switch(op_code)
{	{
case OP_CHAR:	case OP_CHAR:
case OP_CHARI:	case OP_CHARI:
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
GETCHARTEST(c, previous);	GETCHARTEST(c, previous);
#else	#else
c = *previous;	c = *previous;
Line 3073 switch(op_code)	Line 3240 switch(op_code)
switch(-next)	switch(-next)
{	{
case ESC_d:	case ESC_d:
return c > 127 \|\| (cd->ctypes[c] & ctype_digit) == 0;	return c > 255 \|\| (cd->ctypes[c] & ctype_digit) == 0;

case ESC_D:	case ESC_D:
return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;	return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;

case ESC_s:	case ESC_s:
return c > 127 \|\| (cd->ctypes[c] & ctype_space) == 0;	return c > 255 \|\| (cd->ctypes[c] & ctype_space) == 0;

case ESC_S:	case ESC_S:
return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;	return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;

case ESC_w:	case ESC_w:
return c > 127 \|\| (cd->ctypes[c] & ctype_word) == 0;	return c > 255 \|\| (cd->ctypes[c] & ctype_word) == 0;

case ESC_W:	case ESC_W:
return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;	return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;

case ESC_h:	case ESC_h:
case ESC_H:	case ESC_H:
Line 3170 switch(op_code)	Line 3337 switch(op_code)
to the original \d etc. At this point, ptr will point to a zero byte. */	to the original \d etc. At this point, ptr will point to a zero byte. */

if (ptr == CHAR_ASTERISK \|\| ptr == CHAR_QUESTION_MARK \|\|	if (ptr == CHAR_ASTERISK \|\| ptr == CHAR_QUESTION_MARK \|\|
strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)	STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
return FALSE;	return FALSE;

/* Do the property check. */	/* Do the property check. */
Line 3197 switch(op_code)	Line 3364 switch(op_code)
return next == -ESC_d;	return next == -ESC_d;

case OP_WHITESPACE:	case OP_WHITESPACE:
return next == -ESC_S \|\| next == -ESC_d \|\| next == -ESC_w \|\| next == -ESC_R;	return next == -ESC_S \|\| next == -ESC_d \|\| next == -ESC_w;

case OP_NOT_WHITESPACE:	case OP_NOT_WHITESPACE:
return next == -ESC_s \|\| next == -ESC_h \|\| next == -ESC_v;	return next == -ESC_s \|\| next == -ESC_h \|\| next == -ESC_v \|\| next == -ESC_R;

case OP_HSPACE:	case OP_HSPACE:
return next == -ESC_S \|\| next == -ESC_H \|\| next == -ESC_d \|\|	return next == -ESC_S \|\| next == -ESC_H \|\| next == -ESC_d \|\|
Line 3248 Arguments:	Line 3415 Arguments:
codeptr points to the pointer to the current code point	codeptr points to the pointer to the current code point
ptrptr points to the current pattern pointer	ptrptr points to the current pattern pointer
errorcodeptr points to error code variable	errorcodeptr points to error code variable
firstbyteptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)	firstcharptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)
reqbyteptr set to the last literal character required, else < 0	reqcharptr set to the last literal character required, else < 0
bcptr points to current branch chain	bcptr points to current branch chain
cond_depth conditional nesting depth	cond_depth conditional nesting depth
cd contains pointers to tables etc.	cd contains pointers to tables etc.
Line 3261 Returns: TRUE on success	Line 3428 Returns: TRUE on success
*/	*/

static BOOL	static BOOL
compile_branch(int optionsptr, uschar codeptr, const uschar *ptrptr,	compile_branch(int optionsptr, pcre_uchar *codeptr,
int errorcodeptr, int firstbyteptr, int reqbyteptr, branch_chain bcptr,	const pcre_uchar *ptrptr, int errorcodeptr, pcre_int32 *firstcharptr,
int cond_depth, compile_data cd, int lengthptr)	pcre_int32 reqcharptr, branch_chain bcptr, int cond_depth,
	compile_data cd, int lengthptr)
{	{
int repeat_type, op_type;	int repeat_type, op_type;
int repeat_min = 0, repeat_max = 0; /* To please picky compilers */	int repeat_min = 0, repeat_max = 0; /* To please picky compilers */
int bravalue = 0;	int bravalue = 0;
int greedy_default, greedy_non_default;	int greedy_default, greedy_non_default;
int firstbyte, reqbyte;	pcre_int32 firstchar, reqchar;
int zeroreqbyte, zerofirstbyte;	pcre_int32 zeroreqchar, zerofirstchar;
int req_caseopt, reqvary, tempreqvary;	pcre_int32 req_caseopt, reqvary, tempreqvary;
int options = optionsptr; / May change dynamically */	int options = optionsptr; / May change dynamically */
int after_manual_callout = 0;	int after_manual_callout = 0;
int length_prevgroup = 0;	int length_prevgroup = 0;
register int c;	register int c;
register uschar code = codeptr;	register pcre_uchar code = codeptr;
uschar *last_code = code;	pcre_uchar *last_code = code;
uschar *orig_code = code;	pcre_uchar *orig_code = code;
uschar *tempcode;	pcre_uchar *tempcode;
BOOL inescq = FALSE;	BOOL inescq = FALSE;
BOOL groupsetfirstbyte = FALSE;	BOOL groupsetfirstchar = FALSE;
const uschar ptr = ptrptr;	const pcre_uchar ptr = ptrptr;
const uschar *tempptr;	const pcre_uchar *tempptr;
const uschar *nestptr = NULL;	const pcre_uchar *nestptr = NULL;
uschar *previous = NULL;	pcre_uchar *previous = NULL;
uschar *previous_callout = NULL;	pcre_uchar *previous_callout = NULL;
uschar *save_hwm = NULL;	pcre_uchar *save_hwm = NULL;
uschar classbits[32];	pcre_uint8 classbits[32];

/* 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
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
dynamically as we process the pattern. */	dynamically as we process the pattern. */

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
BOOL class_utf8;	/* PCRE_UTF16 has the same value as PCRE_UTF8. */
BOOL utf8 = (options & PCRE_UTF8) != 0;	BOOL utf = (options & PCRE_UTF8) != 0;
uschar *class_utf8data;	pcre_uchar utf_chars[6];
uschar *class_utf8data_base;
uschar utf8_char[6];
#else	#else
BOOL utf8 = FALSE;	BOOL utf = FALSE;
#endif	#endif

	/* Helper variables for OP_XCLASS opcode (for characters > 255). */

	#if defined SUPPORT_UTF \|\| !defined COMPILE_PCRE8
	BOOL xclass;
	pcre_uchar *class_uchardata;
	pcre_uchar *class_uchardata_base;
	#endif

#ifdef PCRE_DEBUG	#ifdef PCRE_DEBUG
if (lengthptr != NULL) DPRINTF((">> start branch\n"));	if (lengthptr != NULL) DPRINTF((">> start branch\n"));
#endif	#endif
Line 3315 greedy_non_default = greedy_default ^ 1;	Line 3489 greedy_non_default = greedy_default ^ 1;

/* Initialize no first byte, no required byte. REQ_UNSET means "no char	/* Initialize no first byte, no required byte. REQ_UNSET means "no char
matching encountered yet". It gets changed to REQ_NONE if we hit something that	matching encountered yet". It gets changed to REQ_NONE if we hit something that
matches a non-fixed char first char; reqbyte just remains unset if we never	matches a non-fixed char first char; reqchar just remains unset if we never
find one.	find one.

When we hit a repeat whose minimum is zero, we may have to adjust these values	When we hit a repeat whose minimum is zero, we may have to adjust these values
to take the zero repeat into account. This is implemented by setting them to	to take the zero repeat into account. This is implemented by setting them to
zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual	zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
item types that can be repeated set these backoff variables appropriately. */	item types that can be repeated set these backoff variables appropriately. */

firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;	firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;

/* The variable req_caseopt contains either the REQ_CASELESS value or zero,	/* The variable req_caseopt contains either the REQ_CASELESS value
according to the current setting of the caseless flag. REQ_CASELESS is a bit	or zero, according to the current setting of the caseless flag. The
value > 255. It is added into the firstbyte or reqbyte variables to record the	REQ_CASELESS leaves the lower 28 bit empty. It is added into the
case status of the value. This is used only for ASCII characters. */	firstchar or reqchar variables to record the case status of the
	value. This is used only for ASCII characters. */

req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;	req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;

/* Switch on next character until the end of the branch */	/* Switch on next character until the end of the branch */

Line 3342 for (;; ptr++)	Line 3517 for (;; ptr++)
BOOL is_quantifier;	BOOL is_quantifier;
BOOL is_recurse;	BOOL is_recurse;
BOOL reset_bracount;	BOOL reset_bracount;
int class_charcount;	int class_has_8bitchar;
int class_lastchar;	int class_single_char;
int newoptions;	int newoptions;
int recno;	int recno;
int refsign;	int refsign;
int skipbytes;	int skipbytes;
int subreqbyte;	int subreqchar;
int subfirstbyte;	int subfirstchar;
int terminator;	int terminator;
int mclength;	int mclength;
int tempbracount;	int tempbracount;
uschar mcbuffer[8];	pcre_uchar mcbuffer[8];

/* Get next byte in the pattern */	/* Get next character in the pattern */

c = *ptr;	c = *ptr;

Line 3401 for (;; ptr++)	Line 3576 for (;; ptr++)
}	}

*lengthptr += (int)(code - last_code);	*lengthptr += (int)(code - last_code);
DPRINTF(("length=%d added %d c=%c\n", *lengthptr, (int)(code - last_code),	DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
c));	(int)(code - last_code), c, c));

/* 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
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,
Line 3412 for (;; ptr++)	Line 3587 for (;; ptr++)
{	{
if (previous > orig_code)	if (previous > orig_code)
{	{
memmove(orig_code, previous, code - previous);	memmove(orig_code, previous, IN_UCHARS(code - previous));
code -= previous - orig_code;	code -= previous - orig_code;
previous = orig_code;	previous = orig_code;
}	}
Line 3481 for (;; ptr++)	Line 3656 for (;; ptr++)

if ((options & PCRE_EXTENDED) != 0)	if ((options & PCRE_EXTENDED) != 0)
{	{
if ((cd->ctypes[c] & ctype_space) != 0) continue;	if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
if (c == CHAR_NUMBER_SIGN)	if (c == CHAR_NUMBER_SIGN)
{	{
ptr++;	ptr++;
Line 3489 for (;; ptr++)	Line 3664 for (;; ptr++)
{	{
if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }	if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
ptr++;	ptr++;
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;	if (utf) FORWARDCHAR(ptr);
#endif	#endif
}	}
if (*ptr != 0) continue;	if (*ptr != 0) continue;
Line 3514 for (;; ptr++)	Line 3689 for (;; ptr++)
case 0: /* The branch terminates at string end */	case 0: /* The branch terminates at string end */
case CHAR_VERTICAL_LINE: /* or \| or ) */	case CHAR_VERTICAL_LINE: /* or \| or ) */
case CHAR_RIGHT_PARENTHESIS:	case CHAR_RIGHT_PARENTHESIS:
*firstbyteptr = firstbyte;	*firstcharptr = firstchar;
*reqbyteptr = reqbyte;	*reqcharptr = reqchar;
*codeptr = code;	*codeptr = code;
*ptrptr = ptr;	*ptrptr = ptr;
if (lengthptr != NULL)	if (lengthptr != NULL)
Line 3539 for (;; ptr++)	Line 3714 for (;; ptr++)
previous = NULL;	previous = NULL;
if ((options & PCRE_MULTILINE) != 0)	if ((options & PCRE_MULTILINE) != 0)
{	{
if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;	if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
*code++ = OP_CIRCM;	*code++ = OP_CIRCM;
}	}
else *code++ = OP_CIRC;	else *code++ = OP_CIRC;
Line 3551 for (;; ptr++)	Line 3726 for (;; ptr++)
break;	break;

/* There can never be a first char if '.' is first, whatever happens about	/* There can never be a first char if '.' is first, whatever happens about
repeats. The value of reqbyte doesn't change either. */	repeats. The value of reqchar doesn't change either. */

case CHAR_DOT:	case CHAR_DOT:
if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;	if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
zerofirstbyte = firstbyte;	zerofirstchar = firstchar;
zeroreqbyte = reqbyte;	zeroreqchar = reqchar;
previous = code;	previous = code;
*code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;	*code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
break;	break;
Line 3611 for (;; ptr++)	Line 3786 for (;; ptr++)
{	{
if (ptr[1] == CHAR_E)	if (ptr[1] == CHAR_E)
ptr++;	ptr++;
else if (strncmp((const char *)ptr+1,	else if (STRNCMP_UC_C8(ptr + 1, STR_Q STR_BACKSLASH STR_E, 3) == 0)
STR_Q STR_BACKSLASH STR_E, 3) == 0)
ptr += 3;	ptr += 3;
else	else
break;	break;
Line 3631 for (;; ptr++)	Line 3805 for (;; ptr++)
(cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)	(cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
{	{
*code++ = negate_class? OP_ALLANY : OP_FAIL;	*code++ = negate_class? OP_ALLANY : OP_FAIL;
if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;	if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
zerofirstbyte = firstbyte;	zerofirstchar = firstchar;
break;	break;
}	}

Line 3642 for (;; ptr++)	Line 3816 for (;; ptr++)

should_flip_negation = FALSE;	should_flip_negation = FALSE;

/* Keep a count of chars with values < 256 so that we can optimize the case	/* For optimization purposes, we track some properties of the class.
of just a single character (as long as it's < 256). However, For higher	class_has_8bitchar will be non-zero, if the class contains at least one
valued UTF-8 characters, we don't yet do any optimization. */	< 256 character. class_single_char will be 1 if the class contains only
	a single character. */

class_charcount = 0;	class_has_8bitchar = 0;
class_lastchar = -1;	class_single_char = 0;

/* Initialize the 32-char bit map to all zeros. We build the map in a	/* Initialize the 32-char bit map to all zeros. We build the map in a
temporary bit of memory, in case the class contains only 1 character (less	temporary bit of memory, in case the class contains only 1 character (less
than 256), because in that case the compiled code doesn't use the bit map.	than 256), because in that case the compiled code doesn't use the bit map.
*/	*/

memset(classbits, 0, 32 * sizeof(uschar));	memset(classbits, 0, 32 * sizeof(pcre_uint8));

#ifdef SUPPORT_UTF8	#if defined SUPPORT_UTF \|\| !defined COMPILE_PCRE8
class_utf8 = FALSE; /* No chars >= 256 */	xclass = FALSE; /* No chars >= 256 */
class_utf8data = code + LINK_SIZE + 2; /* For UTF-8 items */	class_uchardata = code + LINK_SIZE + 2; /* For UTF-8 items */
class_utf8data_base = class_utf8data; /* For resetting in pass 1 */	class_uchardata_base = class_uchardata; /* For resetting in pass 1 */
#endif	#endif

/* Process characters until ] is reached. By writing this as a "do" it	/* Process characters until ] is reached. By writing this as a "do" it
Line 3668 for (;; ptr++)	Line 3843 for (;; ptr++)

if (c != 0) do	if (c != 0) do
{	{
const uschar *oldptr;	const pcre_uchar *oldptr;

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && c > 127)	if (utf && HAS_EXTRALEN(c))
{ /* Braces are required because the */	{ /* Braces are required because the */
GETCHARLEN(c, ptr, ptr); /* macro generates multiple statements */	GETCHARLEN(c, ptr, ptr); /* macro generates multiple statements */
}	}
	#endif

/* In the pre-compile phase, accumulate the length of any UTF-8 extra	#if defined SUPPORT_UTF \|\| !defined COMPILE_PCRE8
	/* In the pre-compile phase, accumulate the length of any extra
data and reset the pointer. This is so that very large classes that	data and reset the pointer. This is so that very large classes that
contain a zillion UTF-8 characters no longer overwrite the work space	contain a zillion > 255 characters no longer overwrite the work space
(which is on the stack). */	(which is on the stack). */

if (lengthptr != NULL)	if (lengthptr != NULL)
{	{
*lengthptr += (int)(class_utf8data - class_utf8data_base);	*lengthptr += class_uchardata - class_uchardata_base;
class_utf8data = class_utf8data_base;	class_uchardata = class_uchardata_base;
}	}

#endif	#endif

/* Inside \Q...\E everything is literal except \E */	/* Inside \Q...\E everything is literal except \E */
Line 3714 for (;; ptr++)	Line 3890 for (;; ptr++)
{	{
BOOL local_negate = FALSE;	BOOL local_negate = FALSE;
int posix_class, taboffset, tabopt;	int posix_class, taboffset, tabopt;
register const uschar *cbits = cd->cbits;	register const pcre_uint8 *cbits = cd->cbits;
uschar pbits[32];	pcre_uint8 pbits[32];

if (ptr[1] != CHAR_COLON)	if (ptr[1] != CHAR_COLON)
{	{
Line 3770 for (;; ptr++)	Line 3946 for (;; ptr++)
/* Copy in the first table (always present) */	/* Copy in the first table (always present) */

memcpy(pbits, cbits + posix_class_maps[posix_class],	memcpy(pbits, cbits + posix_class_maps[posix_class],
32 * sizeof(uschar));	32 * sizeof(pcre_uint8));

/* If there is a second table, add or remove it as required. */	/* If there is a second table, add or remove it as required. */

Line 3801 for (;; ptr++)	Line 3977 for (;; ptr++)
for (c = 0; c < 32; c++) classbits[c] \|= pbits[c];	for (c = 0; c < 32; c++) classbits[c] \|= pbits[c];

ptr = tempptr + 1;	ptr = tempptr + 1;
class_charcount = 10; /* Set > 1; assumes more than 1 per class */	/* Every class contains at least one < 256 characters. */
	class_has_8bitchar = 1;
	/* Every class contains at least two characters. */
	class_single_char = 2;
continue; /* End of POSIX syntax handling */	continue; /* End of POSIX syntax handling */
}	}

/* Backslash may introduce a single character, or it may introduce one	/* Backslash may introduce a single character, or it may introduce one
of the specials, which just set a flag. The sequence \b is a special	of the specials, which just set a flag. The sequence \b is a special
case. Inside a class (and only there) it is treated as backspace. We	case. Inside a class (and only there) it is treated as backspace. We
assume that other escapes have more than one character in them, so set	assume that other escapes have more than one character in them, so
class_charcount bigger than one. Unrecognized escapes fall through and	speculatively set both class_has_8bitchar and class_single_char bigger
are either treated as literal characters (by default), or are faulted if	than one. Unrecognized escapes fall through and are either treated
	as literal characters (by default), or are faulted if
PCRE_EXTRA is set. */	PCRE_EXTRA is set. */

if (c == CHAR_BACKSLASH)	if (c == CHAR_BACKSLASH)
Line 3837 for (;; ptr++)	Line 4017 for (;; ptr++)

if (c < 0)	if (c < 0)
{	{
register const uschar *cbits = cd->cbits;	register const pcre_uint8 *cbits = cd->cbits;
class_charcount += 2; /* Greater than 1 is what matters */	/* Every class contains at least two < 256 characters. */
	class_has_8bitchar++;
	/* Every class contains at least two characters. */
	class_single_char += 2;

switch (-c)	switch (-c)
{	{
Line 3851 for (;; ptr++)	Line 4034 for (;; ptr++)
case ESC_SU:	case ESC_SU:
nestptr = ptr;	nestptr = ptr;
ptr = substitutes[-c - ESC_DU] - 1; /* Just before substitute */	ptr = substitutes[-c - ESC_DU] - 1; /* Just before substitute */
class_charcount -= 2; /* Undo! */	class_has_8bitchar--; /* Undo! */
continue;	continue;
#endif	#endif
case ESC_d:	case ESC_d:
Line 3892 for (;; ptr++)	Line 4075 for (;; ptr++)
SETBIT(classbits, 0x09); /* VT */	SETBIT(classbits, 0x09); /* VT */
SETBIT(classbits, 0x20); /* SPACE */	SETBIT(classbits, 0x20); /* SPACE */
SETBIT(classbits, 0xa0); /* NSBP */	SETBIT(classbits, 0xa0); /* NSBP */
#ifdef SUPPORT_UTF8	#ifndef COMPILE_PCRE8
if (utf8)	xclass = TRUE;
	*class_uchardata++ = XCL_SINGLE;
	*class_uchardata++ = 0x1680;
	*class_uchardata++ = XCL_SINGLE;
	*class_uchardata++ = 0x180e;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x2000;
	*class_uchardata++ = 0x200a;
	*class_uchardata++ = XCL_SINGLE;
	*class_uchardata++ = 0x202f;
	*class_uchardata++ = XCL_SINGLE;
	*class_uchardata++ = 0x205f;
	*class_uchardata++ = XCL_SINGLE;
	*class_uchardata++ = 0x3000;
	#elif defined SUPPORT_UTF
	if (utf)
{	{
class_utf8 = TRUE;	xclass = TRUE;
*class_utf8data++ = XCL_SINGLE;	*class_uchardata++ = XCL_SINGLE;
class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x1680, class_uchardata);
*class_utf8data++ = XCL_SINGLE;	*class_uchardata++ = XCL_SINGLE;
class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x180e, class_uchardata);
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x2000, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x200a, class_uchardata);
*class_utf8data++ = XCL_SINGLE;	*class_uchardata++ = XCL_SINGLE;
class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x202f, class_uchardata);
*class_utf8data++ = XCL_SINGLE;	*class_uchardata++ = XCL_SINGLE;
class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x205f, class_uchardata);
*class_utf8data++ = XCL_SINGLE;	*class_uchardata++ = XCL_SINGLE;
class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x3000, class_uchardata);
}	}
#endif	#endif
continue;	continue;
Line 3926 for (;; ptr++)	Line 4124 for (;; ptr++)
}	}
classbits[c] \|= x;	classbits[c] \|= x;
}	}
	#ifndef COMPILE_PCRE8
#ifdef SUPPORT_UTF8	xclass = TRUE;
if (utf8)	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x0100;
	*class_uchardata++ = 0x167f;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x1681;
	*class_uchardata++ = 0x180d;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x180f;
	*class_uchardata++ = 0x1fff;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x200b;
	*class_uchardata++ = 0x202e;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x2030;
	*class_uchardata++ = 0x205e;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x2060;
	*class_uchardata++ = 0x2fff;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x3001;
	#ifdef SUPPORT_UTF
	if (utf)
	class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);
	else
	#endif
	*class_uchardata++ = 0xffff;
	#elif defined SUPPORT_UTF
	if (utf)
{	{
class_utf8 = TRUE;	xclass = TRUE;
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x167f, class_uchardata);
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x1681, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x180d, class_uchardata);
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x180f, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x1fff, class_uchardata);
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x200b, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x202e, class_uchardata);
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x2030, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x205e, class_uchardata);
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x2060, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x2fff, class_uchardata);
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x3001, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);
}	}
#endif	#endif
continue;	continue;
Line 3962 for (;; ptr++)	Line 4187 for (;; ptr++)
SETBIT(classbits, 0x0c); /* FF */	SETBIT(classbits, 0x0c); /* FF */
SETBIT(classbits, 0x0d); /* CR */	SETBIT(classbits, 0x0d); /* CR */
SETBIT(classbits, 0x85); /* NEL */	SETBIT(classbits, 0x85); /* NEL */
#ifdef SUPPORT_UTF8	#ifndef COMPILE_PCRE8
if (utf8)	xclass = TRUE;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x2028;
	*class_uchardata++ = 0x2029;
	#elif defined SUPPORT_UTF
	if (utf)
{	{
class_utf8 = TRUE;	xclass = TRUE;
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x2028, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x2029, class_uchardata);
}	}
#endif	#endif
continue;	continue;
Line 3990 for (;; ptr++)	Line 4220 for (;; ptr++)
classbits[c] \|= x;	classbits[c] \|= x;
}	}

#ifdef SUPPORT_UTF8	#ifndef COMPILE_PCRE8
if (utf8)	xclass = TRUE;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x0100;
	*class_uchardata++ = 0x2027;
	*class_uchardata++ = XCL_RANGE;
	*class_uchardata++ = 0x202a;
	#ifdef SUPPORT_UTF
	if (utf)
	class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);
	else
	#endif
	*class_uchardata++ = 0xffff;
	#elif defined SUPPORT_UTF
	if (utf)
{	{
class_utf8 = TRUE;	xclass = TRUE;
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x2027, class_uchardata);
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x202a, class_uchardata);
class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);	class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);
}	}
#endif	#endif
continue;	continue;
Line 4012 for (;; ptr++)	Line 4255 for (;; ptr++)
int pdata;	int pdata;
int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);	int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
if (ptype < 0) goto FAILED;	if (ptype < 0) goto FAILED;
class_utf8 = TRUE;	xclass = TRUE;
*class_utf8data++ = ((-c == ESC_p) != negated)?	*class_uchardata++ = ((-c == ESC_p) != negated)?
XCL_PROP : XCL_NOTPROP;	XCL_PROP : XCL_NOTPROP;
*class_utf8data++ = ptype;	*class_uchardata++ = ptype;
*class_utf8data++ = pdata;	*class_uchardata++ = pdata;
class_charcount -= 2; /* Not a < 256 character */	class_has_8bitchar--; /* Undo! */
continue;	continue;
}	}
#endif	#endif
Line 4031 for (;; ptr++)	Line 4274 for (;; ptr++)
*errorcodeptr = ERR7;	*errorcodeptr = ERR7;
goto FAILED;	goto FAILED;
}	}
class_charcount -= 2; /* Undo the default count from above */	class_has_8bitchar--; /* Undo the speculative increase. */
c = ptr; / Get the final character and fall through */	class_single_char -= 2; /* Undo the speculative increase. */
	c = ptr; / Get the final character and fall through */
break;	break;
}	}
}	}

/* Fall through if we have a single character (c >= 0). This may be	/* Fall through if we have a single character (c >= 0). This may be
greater than 256 in UTF-8 mode. */	greater than 256. */

} /* End of backslash handling */	} /* End of backslash handling */

Line 4086 for (;; ptr++)	Line 4330 for (;; ptr++)
goto LONE_SINGLE_CHARACTER;	goto LONE_SINGLE_CHARACTER;
}	}

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8)	if (utf)
{ /* Braces are required because the */	{ /* Braces are required because the */
GETCHARLEN(d, ptr, ptr); /* macro generates multiple statements */	GETCHARLEN(d, ptr, ptr); /* macro generates multiple statements */
}	}
Line 4131 for (;; ptr++)	Line 4375 for (;; ptr++)

if (d == CHAR_CR \|\| d == CHAR_NL) cd->external_flags \|= PCRE_HASCRORLF;	if (d == CHAR_CR \|\| d == CHAR_NL) cd->external_flags \|= PCRE_HASCRORLF;

	/* Since we found a character range, single character optimizations
	cannot be done anymore. */
	class_single_char = 2;

/* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless	/* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless
matching, we have to use an XCLASS with extra data items. Caseless	matching, we have to use an XCLASS with extra data items. Caseless
matching for characters > 127 is available only if UCP support is	matching for characters > 127 is available only if UCP support is
available. */	available. */

#ifdef SUPPORT_UTF8	#if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)
if (utf8 && (d > 255 \|\| ((options & PCRE_CASELESS) != 0 && d > 127)))	if ((d > 255) \|\| (utf && ((options & PCRE_CASELESS) != 0 && d > 127)))
	#elif defined SUPPORT_UTF
	if (utf && (d > 255 \|\| ((options & PCRE_CASELESS) != 0 && d > 127)))
	#elif !(defined COMPILE_PCRE8)
	if (d > 255)
	#endif
	#if defined SUPPORT_UTF \|\| !(defined COMPILE_PCRE8)
{	{
class_utf8 = TRUE;	xclass = TRUE;

/* With UCP support, we can find the other case equivalents of	/* With UCP support, we can find the other case equivalents of
the relevant characters. There may be several ranges. Optimize how	the relevant characters. There may be several ranges. Optimize how
they fit with the basic range. */	they fit with the basic range. */

#ifdef SUPPORT_UCP	#ifdef SUPPORT_UCP
	#ifndef COMPILE_PCRE8
	if (utf && (options & PCRE_CASELESS) != 0)
	#else
if ((options & PCRE_CASELESS) != 0)	if ((options & PCRE_CASELESS) != 0)
	#endif
{	{
unsigned int occ, ocd;	unsigned int occ, ocd;
unsigned int cc = c;	unsigned int cc = c;
Line 4172 for (;; ptr++)	Line 4430 for (;; ptr++)

if (occ == ocd)	if (occ == ocd)
{	{
*class_utf8data++ = XCL_SINGLE;	*class_uchardata++ = XCL_SINGLE;
}	}
else	else
{	{
*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(occ, class_utf8data);	class_uchardata += PRIV(ord2utf)(occ, class_uchardata);
}	}
class_utf8data += _pcre_ord2utf8(ocd, class_utf8data);	class_uchardata += PRIV(ord2utf)(ocd, class_uchardata);
}	}
}	}
#endif /* SUPPORT_UCP */	#endif /* SUPPORT_UCP */
Line 4187 for (;; ptr++)	Line 4445 for (;; ptr++)
/* Now record the original range, possibly modified for UCP caseless	/* Now record the original range, possibly modified for UCP caseless
overlapping ranges. */	overlapping ranges. */

*class_utf8data++ = XCL_RANGE;	*class_uchardata++ = XCL_RANGE;
class_utf8data += _pcre_ord2utf8(c, class_utf8data);	#ifdef SUPPORT_UTF
class_utf8data += _pcre_ord2utf8(d, class_utf8data);	#ifndef COMPILE_PCRE8
	if (utf)
	{
	class_uchardata += PRIV(ord2utf)(c, class_uchardata);
	class_uchardata += PRIV(ord2utf)(d, class_uchardata);
	}
	else
	{
	*class_uchardata++ = c;
	*class_uchardata++ = d;
	}
	#else
	class_uchardata += PRIV(ord2utf)(c, class_uchardata);
	class_uchardata += PRIV(ord2utf)(d, class_uchardata);
	#endif
	#else /* SUPPORT_UTF */
	*class_uchardata++ = c;
	*class_uchardata++ = d;
	#endif /* SUPPORT_UTF */

/* With UCP support, we are done. Without UCP support, there is no	/* With UCP support, we are done. Without UCP support, there is no
caseless matching for UTF-8 characters > 127; we can use the bit map	caseless matching for UTF characters > 127; we can use the bit map
for the smaller ones. */	for the smaller ones. As for 16 bit characters without UTF, we
	can still use */

#ifdef SUPPORT_UCP	#ifdef SUPPORT_UCP
continue; /* With next character in the class */	#ifndef COMPILE_PCRE8
#else	if (utf)
if ((options & PCRE_CASELESS) == 0 \|\| c > 127) continue;	#endif
	continue; /* With next character in the class */
	#endif /* SUPPORT_UCP */

	#if defined SUPPORT_UTF && !defined(SUPPORT_UCP) && !(defined COMPILE_PCRE8)
	if (utf)
	{
	if ((options & PCRE_CASELESS) == 0 \|\| c > 127) continue;
	/* Adjust upper limit and fall through to set up the map */
	d = 127;
	}
	else
	{
	if (c > 255) continue;
	/* Adjust upper limit and fall through to set up the map */
	d = 255;
	}
	#elif defined SUPPORT_UTF && !defined(SUPPORT_UCP)
	if ((options & PCRE_CASELESS) == 0 \|\| c > 127) continue;
/* Adjust upper limit and fall through to set up the map */	/* Adjust upper limit and fall through to set up the map */

d = 127;	d = 127;
	#else
#endif /* SUPPORT_UCP */	if (c > 255) continue;
	/* Adjust upper limit and fall through to set up the map */
	d = 255;
	#endif /* SUPPORT_UTF && !SUPPORT_UCP && !COMPILE_PCRE8 */
}	}
#endif /* SUPPORT_UTF8 */	#endif /* SUPPORT_UTF \|\| !COMPILE_PCRE8 */

/* We use the bit map for all cases when not in UTF-8 mode; else	/* We use the bit map for 8 bit mode, or when the characters fall
ranges that lie entirely within 0-127 when there is UCP support; else	partially or entirely to [0-255] ([0-127] for UCP) ranges. */
for partial ranges without UCP support. */

class_charcount += d - c + 1;	class_has_8bitchar = 1;
class_lastchar = d;

/* We can save a bit of time by skipping this in the pre-compile. */	/* We can save a bit of time by skipping this in the pre-compile. */

Line 4222 for (;; ptr++)	Line 4516 for (;; ptr++)
classbits[c/8] \|= (1 << (c&7));	classbits[c/8] \|= (1 << (c&7));
if ((options & PCRE_CASELESS) != 0)	if ((options & PCRE_CASELESS) != 0)
{	{
int uc = cd->fcc[c]; /* flip case */	int uc = cd->fcc[c]; /* flip case */
classbits[uc/8] \|= (1 << (uc&7));	classbits[uc/8] \|= (1 << (uc&7));
}	}
}	}
Line 4236 for (;; ptr++)	Line 4530 for (;; ptr++)

LONE_SINGLE_CHARACTER:	LONE_SINGLE_CHARACTER:

/* Handle a character that cannot go in the bit map */	/* Only the value of 1 matters for class_single_char. */

#ifdef SUPPORT_UTF8	if (class_single_char < 2) class_single_char++;
if (utf8 && (c > 255 \|\| ((options & PCRE_CASELESS) != 0 && c > 127)))
	/* If class_charcount is 1, we saw precisely one character. As long as
	there was no use of \p or \P, in other words, no use of any XCLASS
	features, we can optimize.

	The optimization throws away the bit map. We turn the item into a
	1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.
	In the positive case, it can cause firstchar to be set. Otherwise, there
	can be no first char if this item is first, whatever repeat count may
	follow. In the case of reqchar, save the previous value for reinstating. */

	if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
{	{
class_utf8 = TRUE;	ptr++;
*class_utf8data++ = XCL_SINGLE;	zeroreqchar = reqchar;
class_utf8data += _pcre_ord2utf8(c, class_utf8data);

	if (negate_class)
	{
	if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
	zerofirstchar = firstchar;
	*code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
	#ifdef SUPPORT_UTF
	if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
	code += PRIV(ord2utf)(c, code);
	else
	#endif
	*code++ = c;
	goto NOT_CHAR;
	}

	/* For a single, positive character, get the value into mcbuffer, and
	then we can handle this with the normal one-character code. */

	#ifdef SUPPORT_UTF
	if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
	mclength = PRIV(ord2utf)(c, mcbuffer);
	else
	#endif
	{
	mcbuffer[0] = c;
	mclength = 1;
	}
	goto ONE_CHAR;
	} /* End of 1-char optimization */

	/* Handle a character that cannot go in the bit map. */

	#if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)
	if ((c > 255) \|\| (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))
	#elif defined SUPPORT_UTF
	if (utf && (c > 255 \|\| ((options & PCRE_CASELESS) != 0 && c > 127)))
	#elif !(defined COMPILE_PCRE8)
	if (c > 255)
	#endif

	#if defined SUPPORT_UTF \|\| !(defined COMPILE_PCRE8)
	{
	xclass = TRUE;
	*class_uchardata++ = XCL_SINGLE;
	#ifdef SUPPORT_UTF
	#ifndef COMPILE_PCRE8
	/* In non 8 bit mode, we can get here even if we are not in UTF mode. */
	if (!utf)
	*class_uchardata++ = c;
	else
	#endif
	class_uchardata += PRIV(ord2utf)(c, class_uchardata);
	#else /* SUPPORT_UTF */
	*class_uchardata++ = c;
	#endif /* SUPPORT_UTF */

#ifdef SUPPORT_UCP	#ifdef SUPPORT_UCP
	#ifdef COMPILE_PCRE8
if ((options & PCRE_CASELESS) != 0)	if ((options & PCRE_CASELESS) != 0)
	#else
	/* In non 8 bit mode, we can get here even if we are not in UTF mode. */
	if (utf && (options & PCRE_CASELESS) != 0)
	#endif
{	{
unsigned int othercase;	unsigned int othercase;
if ((othercase = UCD_OTHERCASE(c)) != c)	if ((int)(othercase = UCD_OTHERCASE(c)) != c)
{	{
*class_utf8data++ = XCL_SINGLE;	*class_uchardata++ = XCL_SINGLE;
class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);	class_uchardata += PRIV(ord2utf)(othercase, class_uchardata);
}	}
}	}
#endif /* SUPPORT_UCP */	#endif /* SUPPORT_UCP */

}	}
else	else
#endif /* SUPPORT_UTF8 */	#endif /* SUPPORT_UTF \|\| COMPILE_PCRE16 */

/* Handle a single-byte character */	/* Handle a single-byte character */
{	{
	class_has_8bitchar = 1;
classbits[c/8] \|= (1 << (c&7));	classbits[c/8] \|= (1 << (c&7));
if ((options & PCRE_CASELESS) != 0)	if ((options & PCRE_CASELESS) != 0)
{	{
c = cd->fcc[c]; /* flip case */	c = cd->fcc[c]; /* flip case */
classbits[c/8] \|= (1 << (c&7));	classbits[c/8] \|= (1 << (c&7));
}	}
class_charcount++;
class_lastchar = c;
}	}
}	}

Line 4291 for (;; ptr++)	Line 4654 for (;; ptr++)
goto FAILED;	goto FAILED;
}	}

/* If class_charcount is 1, we saw precisely one character whose value is	/* If this is the first thing in the branch, there can be no first char
less than 256. As long as there were no characters >= 128 and there was no	setting, whatever the repeat count. Any reqchar setting must remain
use of \p or \P, in other words, no use of any XCLASS features, we can	unchanged after any kind of repeat. */
optimize.

In UTF-8 mode, we can optimize the negative case only if there were no	if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR	zerofirstchar = firstchar;
operate on single-bytes characters only. This is an historical hangover.	zeroreqchar = reqchar;
Maybe one day we can tidy these opcodes to handle multi-byte characters.

The optimization throws away the bit map. We turn the item into a
1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.
Note that OP_NOT[I] does not support multibyte characters. In the positive
case, it can cause firstbyte to be set. Otherwise, there can be no first
char if this item is first, whatever repeat count may follow. In the case
of reqbyte, save the previous value for reinstating. */

#ifdef SUPPORT_UTF8
if (class_charcount == 1 && !class_utf8 &&
(!utf8 \|\| !negate_class \|\| class_lastchar < 128))
#else
if (class_charcount == 1)
#endif
{
zeroreqbyte = reqbyte;

/* The OP_NOT[I] opcodes work on one-byte characters only. */

if (negate_class)
{
if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
zerofirstbyte = firstbyte;
*code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
*code++ = class_lastchar;
break;
}

/* For a single, positive character, get the value into mcbuffer, and
then we can handle this with the normal one-character code. */

#ifdef SUPPORT_UTF8
if (utf8 && class_lastchar > 127)
mclength = _pcre_ord2utf8(class_lastchar, mcbuffer);
else
#endif
{
mcbuffer[0] = class_lastchar;
mclength = 1;
}
goto ONE_CHAR;
} /* End of 1-char optimization */

/* The general case - not the one-char optimization. If this is the first
thing in the branch, there can be no first char setting, whatever the
repeat count. Any reqbyte setting must remain unchanged after any kind of
repeat. */

if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
zerofirstbyte = firstbyte;
zeroreqbyte = reqbyte;

/* If there are characters with values > 255, we have to compile an	/* If there are characters with values > 255, we have to compile an
extended class, with its own opcode, unless there was a negated special	extended class, with its own opcode, unless there was a negated special
such as \S in the class, and PCRE_UCP is not set, because in that case all	such as \S in the class, and PCRE_UCP is not set, because in that case all
Line 4360 for (;; ptr++)	Line 4670 for (;; ptr++)
be listed) there are no characters < 256, we can omit the bitmap in the	be listed) there are no characters < 256, we can omit the bitmap in the
actual compiled code. */	actual compiled code. */

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (class_utf8 && (!should_flip_negation \|\| (options & PCRE_UCP) != 0))	if (xclass && (!should_flip_negation \|\| (options & PCRE_UCP) != 0))
	#elif !defined COMPILE_PCRE8
	if (xclass && !should_flip_negation)
	#endif
	#if defined SUPPORT_UTF \|\| !defined COMPILE_PCRE8
{	{
class_utf8data++ = XCL_END; / Marks the end of extra data */	class_uchardata++ = XCL_END; / Marks the end of extra data */
*code++ = OP_XCLASS;	*code++ = OP_XCLASS;
code += LINK_SIZE;	code += LINK_SIZE;
*code = negate_class? XCL_NOT : 0;	*code = negate_class? XCL_NOT:0;

/* If the map is required, move up the extra data to make room for it;	/* If the map is required, move up the extra data to make room for it;
otherwise just move the code pointer to the end of the extra data. */	otherwise just move the code pointer to the end of the extra data. */

if (class_charcount > 0)	if (class_has_8bitchar > 0)
{	{
*code++ \|= XCL_MAP;	*code++ \|= XCL_MAP;
memmove(code + 32, code, class_utf8data - code);	memmove(code + (32 / sizeof(pcre_uchar)), code,
	IN_UCHARS(class_uchardata - code));
memcpy(code, classbits, 32);	memcpy(code, classbits, 32);
code = class_utf8data + 32;	code = class_uchardata + (32 / sizeof(pcre_uchar));
}	}
else code = class_utf8data;	else code = class_uchardata;

/* Now fill in the complete length of the item */	/* Now fill in the complete length of the item */

Line 4394 for (;; ptr++)	Line 4709 for (;; ptr++)
negating it if necessary. */	negating it if necessary. */

*code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;	*code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
if (negate_class)	if (lengthptr == NULL) /* Save time in the pre-compile phase */
{	{
if (lengthptr == NULL) /* Save time in the pre-compile phase */	if (negate_class)
for (c = 0; c < 32; c++) code[c] = ~classbits[c];	for (c = 0; c < 32; c++) classbits[c] = ~classbits[c];
}
else
{
memcpy(code, classbits, 32);	memcpy(code, classbits, 32);
}	}
code += 32;	code += 32 / sizeof(pcre_uchar);
	NOT_CHAR:
break;	break;


Line 4440 for (;; ptr++)	Line 4753 for (;; ptr++)

if (repeat_min == 0)	if (repeat_min == 0)
{	{
firstbyte = zerofirstbyte; /* Adjust for zero repeat */	firstchar = zerofirstchar; /* Adjust for zero repeat */
reqbyte = zeroreqbyte; /* Ditto */	reqchar = zeroreqchar; /* Ditto */
}	}

/* Remember whether this is a variable length repeat */	/* Remember whether this is a variable length repeat */
Line 4483 for (;; ptr++)	Line 4796 for (;; ptr++)

if (*previous == OP_RECURSE)	if (*previous == OP_RECURSE)
{	{
memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);	memmove(previous + 1 + LINK_SIZE, previous, IN_UCHARS(1 + LINK_SIZE));
*previous = OP_ONCE;	*previous = OP_ONCE;
PUT(previous, 1, 2 + 2*LINK_SIZE);	PUT(previous, 1, 2 + 2*LINK_SIZE);
previous[2 + 2*LINK_SIZE] = OP_KET;	previous[2 + 2*LINK_SIZE] = OP_KET;
Line 4504 for (;; ptr++)	Line 4817 for (;; ptr++)

/* Now handle repetition for the different types of item. */	/* Now handle repetition for the different types of item. */

/* If previous was a character match, abolish the item and generate a	/* If previous was a character or negated character match, abolish the item
repeat item instead. If a char item has a minumum of more than one, ensure	and generate a repeat item instead. If a char item has a minimum of more
that it is set in reqbyte - it might not be if a sequence such as x{3} is	than one, ensure that it is set in reqchar - it might not be if a sequence
the first thing in a branch because the x will have gone into firstbyte	such as x{3} is the first thing in a branch because the x will have gone
instead. */	into firstchar instead. */

if (previous == OP_CHAR \|\| previous == OP_CHARI)	if (previous == OP_CHAR \|\| previous == OP_CHARI
	\|\| previous == OP_NOT \|\| previous == OP_NOTI)
{	{
op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;	switch (*previous)
	{
	default: /* Make compiler happy. */
	case OP_CHAR: op_type = OP_STAR - OP_STAR; break;
	case OP_CHARI: op_type = OP_STARI - OP_STAR; break;
	case OP_NOT: op_type = OP_NOTSTAR - OP_STAR; break;
	case OP_NOTI: op_type = OP_NOTSTARI - OP_STAR; break;
	}

/* Deal with UTF-8 characters that take up more than one byte. It's	/* Deal with UTF characters that take up more than one character. It's
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
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 UTF_LENGTH to flag that
length rather than a small character. */	it's a length rather than a small character. */

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && (code[-1] & 0x80) != 0)	if (utf && NOT_FIRSTCHAR(code[-1]))
{	{
uschar *lastchar = code - 1;	pcre_uchar *lastchar = code - 1;
while((*lastchar & 0xc0) == 0x80) lastchar--;	BACKCHAR(lastchar);
c = (int)(code - lastchar); /* Length of UTF-8 character */	c = (int)(code - lastchar); /* Length of UTF-8 character */
memcpy(utf8_char, lastchar, c); /* Save the char */	memcpy(utf_chars, lastchar, IN_UCHARS(c)); /* Save the char */
c \|= 0x80; /* Flag c as a length */	c \|= UTF_LENGTH; /* Flag c as a length */
}	}
else	else
#endif	#endif /* SUPPORT_UTF */

/* Handle the case of a single byte - either with no UTF8 support, or	/* Handle the case of a single charater - either with no UTF support, or
with UTF-8 disabled, or for a UTF-8 character < 128. */	with UTF disabled, or for a single character UTF character. */

{	{
c = code[-1];	c = code[-1];
if (repeat_min > 1) reqbyte = c \| req_caseopt \| cd->req_varyopt;	if (*previous <= OP_CHARI && repeat_min > 1)
	reqchar = c \| req_caseopt \| cd->req_varyopt;
}	}

/* If the repetition is unlimited, it pays to see if the next thing on	/* If the repetition is unlimited, it pays to see if the next thing on
Line 4546 for (;; ptr++)	Line 4867 for (;; ptr++)

if (!possessive_quantifier &&	if (!possessive_quantifier &&
repeat_max < 0 &&	repeat_max < 0 &&
check_auto_possessive(previous, utf8, ptr + 1, options, cd))	check_auto_possessive(previous, utf, ptr + 1, options, cd))
{	{
repeat_type = 0; /* Force greedy */	repeat_type = 0; /* Force greedy */
possessive_quantifier = TRUE;	possessive_quantifier = TRUE;
Line 4555 for (;; ptr++)	Line 4876 for (;; ptr++)
goto OUTPUT_SINGLE_REPEAT; /* Code shared with single character types */	goto OUTPUT_SINGLE_REPEAT; /* Code shared with single character types */
}	}

/* If previous was a single negated character ([^a] or similar), we use
one of the special opcodes, replacing it. The code is shared with single-
character repeats by setting opt_type to add a suitable offset into
repeat_type. We can also test for auto-possessification. OP_NOT and OP_NOTI
are currently used only for single-byte chars. */

else if (previous == OP_NOT \|\| previous == OP_NOTI)
{
op_type = ((*previous == OP_NOT)? OP_NOTSTAR : OP_NOTSTARI) - OP_STAR;
c = previous[1];
if (!possessive_quantifier &&
repeat_max < 0 &&
check_auto_possessive(previous, utf8, ptr + 1, options, cd))
{
repeat_type = 0; /* Force greedy */
possessive_quantifier = TRUE;
}
goto OUTPUT_SINGLE_REPEAT;
}

/* If previous was a character type match (\d or similar), abolish it and	/* If previous was a character type match (\d or similar), abolish it and
create a suitable repeat item. The code is shared with single-character	create a suitable repeat item. The code is shared with single-character
repeats by setting op_type to add a suitable offset into repeat_type. Note	repeats by setting op_type to add a suitable offset into repeat_type. Note
Line 4584 for (;; ptr++)	Line 4885 for (;; ptr++)

else if (*previous < OP_EODN)	else if (*previous < OP_EODN)
{	{
uschar *oldcode;	pcre_uchar *oldcode;
int prop_type, prop_value;	int prop_type, prop_value;
op_type = OP_TYPESTAR - OP_STAR; /* Use type opcodes */	op_type = OP_TYPESTAR - OP_STAR; /* Use type opcodes */
c = *previous;	c = *previous;

if (!possessive_quantifier &&	if (!possessive_quantifier &&
repeat_max < 0 &&	repeat_max < 0 &&
check_auto_possessive(previous, utf8, ptr + 1, options, cd))	check_auto_possessive(previous, utf, ptr + 1, options, cd))
{	{
repeat_type = 0; /* Force greedy */	repeat_type = 0; /* Force greedy */
possessive_quantifier = TRUE;	possessive_quantifier = TRUE;
Line 4671 for (;; ptr++)	Line 4972 for (;; ptr++)
we have to insert the character for the previous code. For a repeated	we have to insert the character for the previous code. For a repeated
Unicode property match, there are two extra bytes that define the	Unicode property match, there are two extra bytes that define the
required property. In UTF-8 mode, long characters have their length in	required property. In UTF-8 mode, long characters have their length in
c, with the 0x80 bit as a flag. */	c, with the UTF_LENGTH bit as a flag. */

if (repeat_max < 0)	if (repeat_max < 0)
{	{
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && c >= 128)	if (utf && (c & UTF_LENGTH) != 0)
{	{
memcpy(code, utf8_char, c & 7);	memcpy(code, utf_chars, IN_UCHARS(c & 7));
code += c & 7;	code += c & 7;
}	}
else	else
Line 4700 for (;; ptr++)	Line 5001 for (;; ptr++)

else if (repeat_max != repeat_min)	else if (repeat_max != repeat_min)
{	{
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && c >= 128)	if (utf && (c & UTF_LENGTH) != 0)
{	{
memcpy(code, utf8_char, c & 7);	memcpy(code, utf_chars, IN_UCHARS(c & 7));
code += c & 7;	code += c & 7;
}	}
else	else
Line 4730 for (;; ptr++)	Line 5031 for (;; ptr++)

/* The character or character type itself comes last in all cases. */	/* The character or character type itself comes last in all cases. */

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && c >= 128)	if (utf && (c & UTF_LENGTH) != 0)
{	{
memcpy(code, utf8_char, c & 7);	memcpy(code, utf_chars, IN_UCHARS(c & 7));
code += c & 7;	code += c & 7;
}	}
else	else
Line 4757 for (;; ptr++)	Line 5058 for (;; ptr++)

else if (*previous == OP_CLASS \|\|	else if (*previous == OP_CLASS \|\|
*previous == OP_NCLASS \|\|	*previous == OP_NCLASS \|\|
#ifdef SUPPORT_UTF8	#if defined SUPPORT_UTF \|\| !defined COMPILE_PCRE8
*previous == OP_XCLASS \|\|	*previous == OP_XCLASS \|\|
#endif	#endif
*previous == OP_REF \|\|	*previous == OP_REF \|\|
Line 4806 for (;; ptr++)	Line 5107 for (;; ptr++)
{	{
register int i;	register int i;
int len = (int)(code - previous);	int len = (int)(code - previous);
uschar *bralink = NULL;	pcre_uchar *bralink = NULL;
uschar *brazeroptr = NULL;	pcre_uchar *brazeroptr = NULL;

/* Repeating a DEFINE group is pointless, but Perl allows the syntax, so	/* Repeating a DEFINE group is pointless, but Perl allows the syntax, so
we just ignore the repeat. */	we just ignore the repeat. */
Line 4860 for (;; ptr++)	Line 5161 for (;; ptr++)
if (repeat_max <= 1) /* Covers 0, 1, and unlimited */	if (repeat_max <= 1) /* Covers 0, 1, and unlimited */
{	{
*code = OP_END;	*code = OP_END;
adjust_recurse(previous, 1, utf8, cd, save_hwm);	adjust_recurse(previous, 1, utf, cd, save_hwm);
memmove(previous+1, previous, len);	memmove(previous + 1, previous, IN_UCHARS(len));
code++;	code++;
if (repeat_max == 0)	if (repeat_max == 0)
{	{
Line 4884 for (;; ptr++)	Line 5185 for (;; ptr++)
{	{
int offset;	int offset;
*code = OP_END;	*code = OP_END;
adjust_recurse(previous, 2 + LINK_SIZE, utf8, cd, save_hwm);	adjust_recurse(previous, 2 + LINK_SIZE, utf, cd, save_hwm);
memmove(previous + 2 + LINK_SIZE, previous, len);	memmove(previous + 2 + LINK_SIZE, previous, IN_UCHARS(len));
code += 2 + LINK_SIZE;	code += 2 + LINK_SIZE;
*previous++ = OP_BRAZERO + repeat_type;	*previous++ = OP_BRAZERO + repeat_type;
*previous++ = OP_BRA;	*previous++ = OP_BRA;
Line 4938 for (;; ptr++)	Line 5239 for (;; ptr++)

else	else
{	{
if (groupsetfirstbyte && reqbyte < 0) reqbyte = firstbyte;	if (groupsetfirstchar && reqchar < 0) reqchar = firstchar;

for (i = 1; i < repeat_min; i++)	for (i = 1; i < repeat_min; i++)
{	{
uschar *hc;	pcre_uchar *hc;
uschar *this_hwm = cd->hwm;	pcre_uchar *this_hwm = cd->hwm;
memcpy(code, previous, len);	memcpy(code, previous, IN_UCHARS(len));

while (cd->hwm > cd->start_workspace + cd->workspace_size -	while (cd->hwm > cd->start_workspace + cd->workspace_size -
WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))	WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))
Line 4953 for (;; ptr++)	Line 5254 for (;; ptr++)
int this_offset = this_hwm - cd->start_workspace;	int this_offset = this_hwm - cd->start_workspace;
*errorcodeptr = expand_workspace(cd);	*errorcodeptr = expand_workspace(cd);
if (*errorcodeptr != 0) goto FAILED;	if (*errorcodeptr != 0) goto FAILED;
save_hwm = (uschar *)cd->start_workspace + save_offset;	save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;
this_hwm = (uschar *)cd->start_workspace + this_offset;	this_hwm = (pcre_uchar *)cd->start_workspace + this_offset;
}	}

for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE)	for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE)
Line 5006 for (;; ptr++)	Line 5307 for (;; ptr++)

else for (i = repeat_max - 1; i >= 0; i--)	else for (i = repeat_max - 1; i >= 0; i--)
{	{
uschar *hc;	pcre_uchar *hc;
uschar *this_hwm = cd->hwm;	pcre_uchar *this_hwm = cd->hwm;

*code++ = OP_BRAZERO + repeat_type;	*code++ = OP_BRAZERO + repeat_type;

Line 5023 for (;; ptr++)	Line 5324 for (;; ptr++)
PUTINC(code, 0, offset);	PUTINC(code, 0, offset);
}	}

memcpy(code, previous, len);	memcpy(code, previous, IN_UCHARS(len));

/* Ensure there is enough workspace for forward references before	/* Ensure there is enough workspace for forward references before
copying them. */	copying them. */
Line 5035 for (;; ptr++)	Line 5336 for (;; ptr++)
int this_offset = this_hwm - cd->start_workspace;	int this_offset = this_hwm - cd->start_workspace;
*errorcodeptr = expand_workspace(cd);	*errorcodeptr = expand_workspace(cd);
if (*errorcodeptr != 0) goto FAILED;	if (*errorcodeptr != 0) goto FAILED;
save_hwm = (uschar *)cd->start_workspace + save_offset;	save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;
this_hwm = (uschar *)cd->start_workspace + this_offset;	this_hwm = (pcre_uchar *)cd->start_workspace + this_offset;
}	}

for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE)	for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE)
Line 5055 for (;; ptr++)	Line 5356 for (;; ptr++)
{	{
int oldlinkoffset;	int oldlinkoffset;
int offset = (int)(code - bralink + 1);	int offset = (int)(code - bralink + 1);
uschar *bra = code - offset;	pcre_uchar *bra = code - offset;
oldlinkoffset = GET(bra, 1);	oldlinkoffset = GET(bra, 1);
bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;	bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;
*code++ = OP_KET;	*code++ = OP_KET;
Line 5091 for (;; ptr++)	Line 5392 for (;; ptr++)

else	else
{	{
uschar *ketcode = code - 1 - LINK_SIZE;	pcre_uchar *ketcode = code - 1 - LINK_SIZE;
uschar *bracode = ketcode - GET(ketcode, 1);	pcre_uchar *bracode = ketcode - GET(ketcode, 1);

/* Convert possessive ONCE brackets to non-capturing */	/* Convert possessive ONCE brackets to non-capturing */

Line 5114 for (;; ptr++)	Line 5415 for (;; ptr++)

if (lengthptr == NULL)	if (lengthptr == NULL)
{	{
uschar *scode = bracode;	pcre_uchar *scode = bracode;
do	do
{	{
if (could_be_empty_branch(scode, ketcode, utf8, cd))	if (could_be_empty_branch(scode, ketcode, utf, cd))
{	{
*bracode += OP_SBRA - OP_BRA;	*bracode += OP_SBRA - OP_BRA;
break;	break;
Line 5140 for (;; ptr++)	Line 5441 for (;; ptr++)
{	{
int nlen = (int)(code - bracode);	int nlen = (int)(code - bracode);
*code = OP_END;	*code = OP_END;
adjust_recurse(bracode, 1 + LINK_SIZE, utf8, cd, save_hwm);	adjust_recurse(bracode, 1 + LINK_SIZE, utf, cd, save_hwm);
memmove(bracode + 1+LINK_SIZE, bracode, nlen);	memmove(bracode + 1 + LINK_SIZE, bracode, IN_UCHARS(nlen));
code += 1 + LINK_SIZE;	code += 1 + LINK_SIZE;
nlen += 1 + LINK_SIZE;	nlen += 1 + LINK_SIZE;
*bracode = OP_BRAPOS;	*bracode = OP_BRAPOS;
Line 5210 for (;; ptr++)	Line 5511 for (;; ptr++)
int len;	int len;

if (*tempcode == OP_TYPEEXACT)	if (*tempcode == OP_TYPEEXACT)
tempcode += _pcre_OP_lengths[*tempcode] +	tempcode += PRIV(OP_lengths)[*tempcode] +
((tempcode[3] == OP_PROP \|\| tempcode[3] == OP_NOTPROP)? 2 : 0);	((tempcode[1 + IMM2_SIZE] == OP_PROP
	\|\| tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0);

else if (tempcode == OP_EXACT \|\| tempcode == OP_NOTEXACT)	else if (tempcode == OP_EXACT \|\| tempcode == OP_NOTEXACT)
{	{
tempcode += _pcre_OP_lengths[*tempcode];	tempcode += PRIV(OP_lengths)[*tempcode];
#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && tempcode[-1] >= 0xc0)	if (utf && HAS_EXTRALEN(tempcode[-1]))
tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f];	tempcode += GET_EXTRALEN(tempcode[-1]);
#endif	#endif
}	}

Line 5255 for (;; ptr++)	Line 5557 for (;; ptr++)

default:	default:
*code = OP_END;	*code = OP_END;
adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);	adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cd, save_hwm);
memmove(tempcode + 1+LINK_SIZE, tempcode, len);	memmove(tempcode + 1 + LINK_SIZE, tempcode, IN_UCHARS(len));
code += 1 + LINK_SIZE;	code += 1 + LINK_SIZE;
len += 1 + LINK_SIZE;	len += 1 + LINK_SIZE;
tempcode[0] = OP_ONCE;	tempcode[0] = OP_ONCE;
Line 5268 for (;; ptr++)	Line 5570 for (;; ptr++)
}	}

/* In all case we no longer have a previous item. We also set the	/* In all case we no longer have a previous item. We also set the
"follows varying string" flag for subsequently encountered reqbytes if	"follows varying string" flag for subsequently encountered reqchars if
it isn't already set and we have just passed a varying length item. */	it isn't already set and we have just passed a varying length item. */

END_REPEAT:	END_REPEAT:
Line 5291 for (;; ptr++)	Line 5593 for (;; ptr++)

/* First deal with various "verbs" that can be introduced by ''. /	/* First deal with various "verbs" that can be introduced by ''. /

if (*(++ptr) == CHAR_ASTERISK &&	ptr++;
((cd->ctypes[ptr[1]] & ctype_letter) != 0 \|\| ptr[1] == ':'))	if (ptr[0] == CHAR_ASTERISK && (ptr[1] == ':'
	\|\| (MAX_255(ptr[1]) && ((cd->ctypes[ptr[1]] & ctype_letter) != 0))))
{	{
int i, namelen;	int i, namelen;
int arglen = 0;	int arglen = 0;
const char *vn = verbnames;	const char *vn = verbnames;
const uschar *name = ptr + 1;	const pcre_uchar *name = ptr + 1;
const uschar *arg = NULL;	const pcre_uchar *arg = NULL;
previous = NULL;	previous = NULL;
while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};	ptr++;
	while (MAX_255(ptr) && (cd->ctypes[ptr] & ctype_letter) != 0) ptr++;
namelen = (int)(ptr - name);	namelen = (int)(ptr - name);

/* It appears that Perl allows any characters whatsoever, other than	/* It appears that Perl allows any characters whatsoever, other than
Line 5312 for (;; ptr++)	Line 5616 for (;; ptr++)
arg = ++ptr;	arg = ++ptr;
while (ptr != 0 && ptr != CHAR_RIGHT_PARENTHESIS) ptr++;	while (ptr != 0 && ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
arglen = (int)(ptr - arg);	arglen = (int)(ptr - arg);
	if (arglen > (int)MAX_MARK)
	{
	*errorcodeptr = ERR75;
	goto FAILED;
	}
}	}

if (*ptr != CHAR_RIGHT_PARENTHESIS)	if (*ptr != CHAR_RIGHT_PARENTHESIS)
Line 5325 for (;; ptr++)	Line 5634 for (;; ptr++)
for (i = 0; i < verbcount; i++)	for (i = 0; i < verbcount; i++)
{	{
if (namelen == verbs[i].len &&	if (namelen == verbs[i].len &&
strncmp((char *)name, vn, namelen) == 0)	STRNCMP_UC_C8(name, vn, namelen) == 0)
{	{
/* Check for open captures before ACCEPT and convert it to	/* Check for open captures before ACCEPT and convert it to
ASSERT_ACCEPT if in an assertion. */	ASSERT_ACCEPT if in an assertion. */
Line 5346 for (;; ptr++)	Line 5655 for (;; ptr++)
}	}
*code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;	*code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;

/* Do not set firstbyte after ACCEPT /	/* Do not set firstchar after ACCEPT /
if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;	if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
}	}

/* Handle other cases with/without an argument */	/* Handle other cases with/without an argument */
Line 5373 for (;; ptr++)	Line 5682 for (;; ptr++)
*code = verbs[i].op_arg;	*code = verbs[i].op_arg;
if (*code++ == OP_THEN_ARG) cd->external_flags \|= PCRE_HASTHEN;	if (*code++ == OP_THEN_ARG) cd->external_flags \|= PCRE_HASTHEN;
*code++ = arglen;	*code++ = arglen;
memcpy(code, arg, arglen);	memcpy(code, arg, IN_UCHARS(arglen));
code += arglen;	code += arglen;
*code++ = 0;	*code++ = 0;
}	}
Line 5396 for (;; ptr++)	Line 5705 for (;; ptr++)
{	{
int i, set, unset, namelen;	int i, set, unset, namelen;
int *optset;	int *optset;
const uschar *name;	const pcre_uchar *name;
uschar *slot;	pcre_uchar *slot;

switch (*(++ptr))	switch (*(++ptr))
{	{
Line 5450 for (;; ptr++)	Line 5759 for (;; ptr++)
break;	break;

/* Most other conditions use OP_CREF (a couple change to OP_RREF	/* Most other conditions use OP_CREF (a couple change to OP_RREF
below), and all need to skip 3 bytes at the start of the group. */	below), and all need to skip 1+IMM2_SIZE bytes at the start of the group. */

code[1+LINK_SIZE] = OP_CREF;	code[1+LINK_SIZE] = OP_CREF;
skipbytes = 3;	skipbytes = 1+IMM2_SIZE;
refsign = -1;	refsign = -1;

/* Check for a test for recursion in a named group. */	/* Check for a test for recursion in a named group. */
Line 5486 for (;; ptr++)	Line 5795 for (;; ptr++)

/* We now expect to read a name; any thing else is an error */	/* We now expect to read a name; any thing else is an error */

if ((cd->ctypes[ptr[1]] & ctype_word) == 0)	if (!MAX_255(ptr[1]) \|\| (cd->ctypes[ptr[1]] & ctype_word) == 0)
{	{
ptr += 1; /* To get the right offset */	ptr += 1; /* To get the right offset */
*errorcodeptr = ERR28;	*errorcodeptr = ERR28;
Line 5497 for (;; ptr++)	Line 5806 for (;; ptr++)

recno = 0;	recno = 0;
name = ++ptr;	name = ++ptr;
while ((cd->ctypes[*ptr] & ctype_word) != 0)	while (MAX_255(ptr) && (cd->ctypes[ptr] & ctype_word) != 0)
{	{
if (recno >= 0)	if (recno >= 0)
recno = ((digitab[*ptr] & ctype_digit) != 0)?	recno = (IS_DIGIT(ptr))? recno 10 + *ptr - CHAR_0 : -1;
recno * 10 + *ptr - CHAR_0 : -1;
ptr++;	ptr++;
}	}
namelen = (int)(ptr - name);	namelen = (int)(ptr - name);
Line 5549 for (;; ptr++)	Line 5857 for (;; ptr++)
slot = cd->name_table;	slot = cd->name_table;
for (i = 0; i < cd->names_found; i++)	for (i = 0; i < cd->names_found; i++)
{	{
if (strncmp((char )name, (char )slot+2, namelen) == 0) break;	if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0) break;
slot += cd->name_entry_size;	slot += cd->name_entry_size;
}	}

Line 5565 for (;; ptr++)	Line 5873 for (;; ptr++)
/* Search the pattern for a forward reference */	/* Search the pattern for a forward reference */

else if ((i = find_parens(cd, name, namelen,	else if ((i = find_parens(cd, name, namelen,
(options & PCRE_EXTENDED) != 0, utf8)) > 0)	(options & PCRE_EXTENDED) != 0, utf)) > 0)
{	{
PUT2(code, 2+LINK_SIZE, i);	PUT2(code, 2+LINK_SIZE, i);
code[1+LINK_SIZE]++;	code[1+LINK_SIZE]++;
Line 5591 for (;; ptr++)	Line 5899 for (;; ptr++)
recno = 0;	recno = 0;
for (i = 1; i < namelen; i++)	for (i = 1; i < namelen; i++)
{	{
if ((digitab[name[i]] & ctype_digit) == 0)	if (!IS_DIGIT(name[i]))
{	{
*errorcodeptr = ERR15;	*errorcodeptr = ERR15;
goto FAILED;	goto FAILED;
Line 5606 for (;; ptr++)	Line 5914 for (;; ptr++)
/* Similarly, check for the (?(DEFINE) "condition", which is always	/* Similarly, check for the (?(DEFINE) "condition", which is always
false. */	false. */

else if (namelen == 6 && strncmp((char *)name, STRING_DEFINE, 6) == 0)	else if (namelen == 6 && STRNCMP_UC_C8(name, STRING_DEFINE, 6) == 0)
{	{
code[1+LINK_SIZE] = OP_DEF;	code[1+LINK_SIZE] = OP_DEF;
skipbytes = 1;	skipbytes = 1;
Line 5669 for (;; ptr++)	Line 5977 for (;; ptr++)
break;	break;

default: /* Could be name define, else bad */	default: /* Could be name define, else bad */
if ((cd->ctypes[ptr[1]] & ctype_word) != 0) goto DEFINE_NAME;	if (MAX_255(ptr[1]) && (cd->ctypes[ptr[1]] & ctype_word) != 0)
	goto DEFINE_NAME;
ptr++; /* Correct offset for error */	ptr++; /* Correct offset for error */
*errorcodeptr = ERR24;	*errorcodeptr = ERR24;
goto FAILED;	goto FAILED;
Line 5691 for (;; ptr++)	Line 6000 for (;; ptr++)
*code++ = OP_CALLOUT;	*code++ = OP_CALLOUT;
{	{
int n = 0;	int n = 0;
while ((digitab[*(++ptr)] & ctype_digit) != 0)	ptr++;
n = n * 10 + *ptr - CHAR_0;	while(IS_DIGIT(*ptr))
	n = n * 10 + *ptr++ - CHAR_0;
if (*ptr != CHAR_RIGHT_PARENTHESIS)	if (*ptr != CHAR_RIGHT_PARENTHESIS)
{	{
*errorcodeptr = ERR39;	*errorcodeptr = ERR39;
Line 5737 for (;; ptr++)	Line 6047 for (;; ptr++)
CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;	CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
name = ++ptr;	name = ++ptr;

while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;	while (MAX_255(ptr) && (cd->ctypes[ptr] & ctype_word) != 0) ptr++;
namelen = (int)(ptr - name);	namelen = (int)(ptr - name);

/* In the pre-compile phase, just do a syntax check. */	/* In the pre-compile phase, just do a syntax check. */
Line 5754 for (;; ptr++)	Line 6064 for (;; ptr++)
*errorcodeptr = ERR49;	*errorcodeptr = ERR49;
goto FAILED;	goto FAILED;
}	}
if (namelen + 3 > cd->name_entry_size)	if (namelen + IMM2_SIZE + 1 > cd->name_entry_size)
{	{
cd->name_entry_size = namelen + 3;	cd->name_entry_size = namelen + IMM2_SIZE + 1;
if (namelen > MAX_NAME_SIZE)	if (namelen > MAX_NAME_SIZE)
{	{
*errorcodeptr = ERR48;	*errorcodeptr = ERR48;
Line 5785 for (;; ptr++)	Line 6095 for (;; ptr++)

for (i = 0; i < cd->names_found; i++)	for (i = 0; i < cd->names_found; i++)
{	{
int crc = memcmp(name, slot+2, namelen);	int crc = memcmp(name, slot+IMM2_SIZE, IN_UCHARS(namelen));
if (crc == 0)	if (crc == 0)
{	{
if (slot[2+namelen] == 0)	if (slot[IMM2_SIZE+namelen] == 0)
{	{
if (GET2(slot, 0) != cd->bracount + 1 &&	if (GET2(slot, 0) != cd->bracount + 1 &&
(options & PCRE_DUPNAMES) == 0)	(options & PCRE_DUPNAMES) == 0)
Line 5809 for (;; ptr++)	Line 6119 for (;; ptr++)
if (crc < 0)	if (crc < 0)
{	{
memmove(slot + cd->name_entry_size, slot,	memmove(slot + cd->name_entry_size, slot,
(cd->names_found - i) * cd->name_entry_size);	IN_UCHARS((cd->names_found - i) * cd->name_entry_size));
break;	break;
}	}

Line 5823 for (;; ptr++)	Line 6133 for (;; ptr++)

if (!dupname)	if (!dupname)
{	{
uschar *cslot = cd->name_table;	pcre_uchar *cslot = cd->name_table;
for (i = 0; i < cd->names_found; i++)	for (i = 0; i < cd->names_found; i++)
{	{
if (cslot != slot)	if (cslot != slot)
Line 5840 for (;; ptr++)	Line 6150 for (;; ptr++)
}	}

PUT2(slot, 0, cd->bracount + 1);	PUT2(slot, 0, cd->bracount + 1);
memcpy(slot + 2, name, namelen);	memcpy(slot + IMM2_SIZE, name, IN_UCHARS(namelen));
slot[2+namelen] = 0;	slot[IMM2_SIZE + namelen] = 0;
}	}
}	}

Line 5867 for (;; ptr++)	Line 6177 for (;; ptr++)

NAMED_REF_OR_RECURSE:	NAMED_REF_OR_RECURSE:
name = ++ptr;	name = ++ptr;
while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;	while (MAX_255(ptr) && (cd->ctypes[ptr] & ctype_word) != 0) ptr++;
namelen = (int)(ptr - name);	namelen = (int)(ptr - name);

/* In the pre-compile phase, do a syntax check. We used to just set	/* In the pre-compile phase, do a syntax check. We used to just set
Line 5879 for (;; ptr++)	Line 6189 for (;; ptr++)

if (lengthptr != NULL)	if (lengthptr != NULL)
{	{
const uschar *temp;	const pcre_uchar *temp;

if (namelen == 0)	if (namelen == 0)
{	{
Line 5909 for (;; ptr++)	Line 6219 for (;; ptr++)
temp = cd->end_pattern;	temp = cd->end_pattern;
cd->end_pattern = ptr;	cd->end_pattern = ptr;
recno = find_parens(cd, name, namelen,	recno = find_parens(cd, name, namelen,
(options & PCRE_EXTENDED) != 0, utf8);	(options & PCRE_EXTENDED) != 0, utf);
cd->end_pattern = temp;	cd->end_pattern = temp;
if (recno < 0) recno = 0; /* Forward ref; set dummy number */	if (recno < 0) recno = 0; /* Forward ref; set dummy number */
}	}
Line 5924 for (;; ptr++)	Line 6234 for (;; ptr++)
slot = cd->name_table;	slot = cd->name_table;
for (i = 0; i < cd->names_found; i++)	for (i = 0; i < cd->names_found; i++)
{	{
if (strncmp((char )name, (char )slot+2, namelen) == 0 &&	if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0 &&
slot[2+namelen] == 0)	slot[IMM2_SIZE+namelen] == 0)
break;	break;
slot += cd->name_entry_size;	slot += cd->name_entry_size;
}	}
Line 5936 for (;; ptr++)	Line 6246 for (;; ptr++)
}	}
else if ((recno = /* Forward back reference */	else if ((recno = /* Forward back reference */
find_parens(cd, name, namelen,	find_parens(cd, name, namelen,
(options & PCRE_EXTENDED) != 0, utf8)) <= 0)	(options & PCRE_EXTENDED) != 0, utf)) <= 0)
{	{
*errorcodeptr = ERR15;	*errorcodeptr = ERR15;
goto FAILED;	goto FAILED;
Line 5961 for (;; ptr++)	Line 6271 for (;; ptr++)
case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4:	case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4:
case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:	case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
{	{
const uschar *called;	const pcre_uchar *called;
terminator = CHAR_RIGHT_PARENTHESIS;	terminator = CHAR_RIGHT_PARENTHESIS;

/* Come here from the \g<...> and \g'...' code (Oniguruma	/* Come here from the \g<...> and \g'...' code (Oniguruma
Line 5975 for (;; ptr++)	Line 6285 for (;; ptr++)
if ((refsign = *ptr) == CHAR_PLUS)	if ((refsign = *ptr) == CHAR_PLUS)
{	{
ptr++;	ptr++;
if ((digitab[*ptr] & ctype_digit) == 0)	if (!IS_DIGIT(*ptr))
{	{
*errorcodeptr = ERR63;	*errorcodeptr = ERR63;
goto FAILED;	goto FAILED;
Line 5983 for (;; ptr++)	Line 6293 for (;; ptr++)
}	}
else if (refsign == CHAR_MINUS)	else if (refsign == CHAR_MINUS)
{	{
if ((digitab[ptr[1]] & ctype_digit) == 0)	if (!IS_DIGIT(ptr[1]))
goto OTHER_CHAR_AFTER_QUERY;	goto OTHER_CHAR_AFTER_QUERY;
ptr++;	ptr++;
}	}

recno = 0;	recno = 0;
while((digitab[*ptr] & ctype_digit) != 0)	while(IS_DIGIT(*ptr))
recno = recno * 10 + *ptr++ - CHAR_0;	recno = recno * 10 + *ptr++ - CHAR_0;

if (*ptr != terminator)	if (*ptr != terminator)
Line 6040 for (;; ptr++)	Line 6350 for (;; ptr++)
{	{
*code = OP_END;	*code = OP_END;
if (recno != 0)	if (recno != 0)
called = _pcre_find_bracket(cd->start_code, utf8, recno);	called = PRIV(find_bracket)(cd->start_code, utf, recno);

/* Forward reference */	/* Forward reference */

if (called == NULL)	if (called == NULL)
{	{
if (find_parens(cd, NULL, recno,	if (find_parens(cd, NULL, recno,
(options & PCRE_EXTENDED) != 0, utf8) < 0)	(options & PCRE_EXTENDED) != 0, utf) < 0)
{	{
*errorcodeptr = ERR15;	*errorcodeptr = ERR15;
goto FAILED;	goto FAILED;
Line 6077 for (;; ptr++)	Line 6387 for (;; ptr++)
conditional subpatterns will be picked up then. */	conditional subpatterns will be picked up then. */

else if (GET(called, 1) == 0 && cond_depth <= 0 &&	else if (GET(called, 1) == 0 && cond_depth <= 0 &&
could_be_empty(called, code, bcptr, utf8, cd))	could_be_empty(called, code, bcptr, utf, cd))
{	{
*errorcodeptr = ERR40;	*errorcodeptr = ERR40;
goto FAILED;	goto FAILED;
Line 6085 for (;; ptr++)	Line 6395 for (;; ptr++)
}	}

/* Insert the recursion/subroutine item. It does not have a set first	/* Insert the recursion/subroutine item. It does not have a set first
byte (relevant if it is repeated, because it will then be wrapped	character (relevant if it is repeated, because it will then be
with ONCE brackets). */	wrapped with ONCE brackets). */

*code = OP_RECURSE;	*code = OP_RECURSE;
PUT(code, 1, (int)(called - cd->start_code));	PUT(code, 1, (int)(called - cd->start_code));
code += 1 + LINK_SIZE;	code += 1 + LINK_SIZE;
groupsetfirstbyte = FALSE;	groupsetfirstchar = FALSE;
}	}

/* Can't determine a first byte now */	/* Can't determine a first byte now */

if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;	if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
continue;	continue;


Line 6153 for (;; ptr++)	Line 6463 for (;; ptr++)
both phases.	both phases.

If we are not at the pattern start, reset the greedy defaults and the	If we are not at the pattern start, reset the greedy defaults and the
case value for firstbyte and reqbyte. */	case value for firstchar and reqchar. */

if (*ptr == CHAR_RIGHT_PARENTHESIS)	if (*ptr == CHAR_RIGHT_PARENTHESIS)
{	{
Line 6166 for (;; ptr++)	Line 6476 for (;; ptr++)
{	{
greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);	greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);
greedy_non_default = greedy_default ^ 1;	greedy_non_default = greedy_default ^ 1;
req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;	req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
}	}

/* Change options at this level, and pass them back for use	/* Change options at this level, and pass them back for use
Line 6203 for (;; ptr++)	Line 6513 for (;; ptr++)
NUMBERED_GROUP:	NUMBERED_GROUP:
cd->bracount += 1;	cd->bracount += 1;
PUT2(code, 1+LINK_SIZE, cd->bracount);	PUT2(code, 1+LINK_SIZE, cd->bracount);
skipbytes = 2;	skipbytes = IMM2_SIZE;
}	}

/* Process nested bracketed regex. Assertions used not to be repeatable,	/* Process nested bracketed regex. Assertions used not to be repeatable,
Line 6229 for (;; ptr++)	Line 6539 for (;; ptr++)
skipbytes, /* Skip over bracket number */	skipbytes, /* Skip over bracket number */
cond_depth +	cond_depth +
((bravalue == OP_COND)?1:0), /* Depth of condition subpatterns */	((bravalue == OP_COND)?1:0), /* Depth of condition subpatterns */
&subfirstbyte, /* For possible first char */	&subfirstchar, /* For possible first char */
&subreqbyte, /* For possible last char */	&subreqchar, /* For possible last char */
bcptr, /* Current branch chain */	bcptr, /* Current branch chain */
cd, /* Tables block */	cd, /* Tables block */
(lengthptr == NULL)? NULL : /* Actual compile phase */	(lengthptr == NULL)? NULL : /* Actual compile phase */
Line 6258 for (;; ptr++)	Line 6568 for (;; ptr++)

if (bravalue == OP_COND && lengthptr == NULL)	if (bravalue == OP_COND && lengthptr == NULL)
{	{
uschar *tc = code;	pcre_uchar *tc = code;
int condcount = 0;	int condcount = 0;

do {	do {
Line 6281 for (;; ptr++)	Line 6591 for (;; ptr++)
}	}

/* A "normal" conditional group. If there is just one branch, we must not	/* A "normal" conditional group. If there is just one branch, we must not
make use of its firstbyte or reqbyte, because this is equivalent to an	make use of its firstchar or reqchar, because this is equivalent to an
empty second branch. */	empty second branch. */

else	else
Line 6291 for (;; ptr++)	Line 6601 for (;; ptr++)
*errorcodeptr = ERR27;	*errorcodeptr = ERR27;
goto FAILED;	goto FAILED;
}	}
if (condcount == 1) subfirstbyte = subreqbyte = REQ_NONE;	if (condcount == 1) subfirstchar = subreqchar = REQ_NONE;
}	}
}	}

Line 6335 for (;; ptr++)	Line 6645 for (;; ptr++)
/* Handle updating of the required and first characters for other types of	/* Handle updating of the required and first characters for other types of
group. Update for normal brackets of all kinds, and conditions with two	group. Update for normal brackets of all kinds, and conditions with two
branches (see code above). If the bracket is followed by a quantifier with	branches (see code above). If the bracket is followed by a quantifier with
zero repeat, we have to back off. Hence the definition of zeroreqbyte and	zero repeat, we have to back off. Hence the definition of zeroreqchar and
zerofirstbyte outside the main loop so that they can be accessed for the	zerofirstchar outside the main loop so that they can be accessed for the
back off. */	back off. */

zeroreqbyte = reqbyte;	zeroreqchar = reqchar;
zerofirstbyte = firstbyte;	zerofirstchar = firstchar;
groupsetfirstbyte = FALSE;	groupsetfirstchar = FALSE;

if (bravalue >= OP_ONCE)	if (bravalue >= OP_ONCE)
{	{
/* If we have not yet set a firstbyte in this branch, take it from the	/* If we have not yet set a firstchar in this branch, take it from the
subpattern, remembering that it was set here so that a repeat of more	subpattern, remembering that it was set here so that a repeat of more
than one can replicate it as reqbyte if necessary. If the subpattern has	than one can replicate it as reqchar if necessary. If the subpattern has
no firstbyte, set "none" for the whole branch. In both cases, a zero	no firstchar, set "none" for the whole branch. In both cases, a zero
repeat forces firstbyte to "none". */	repeat forces firstchar to "none". */

if (firstbyte == REQ_UNSET)	if (firstchar == REQ_UNSET)
{	{
if (subfirstbyte >= 0)	if (subfirstchar >= 0)
{	{
firstbyte = subfirstbyte;	firstchar = subfirstchar;
groupsetfirstbyte = TRUE;	groupsetfirstchar = TRUE;
}	}
else firstbyte = REQ_NONE;	else firstchar = REQ_NONE;
zerofirstbyte = REQ_NONE;	zerofirstchar = REQ_NONE;
}	}

/* If firstbyte was previously set, convert the subpattern's firstbyte	/* If firstchar was previously set, convert the subpattern's firstchar
into reqbyte if there wasn't one, using the vary flag that was in	into reqchar if there wasn't one, using the vary flag that was in
existence beforehand. */	existence beforehand. */

else if (subfirstbyte >= 0 && subreqbyte < 0)	else if (subfirstchar >= 0 && subreqchar < 0)
subreqbyte = subfirstbyte \| tempreqvary;	subreqchar = subfirstchar \| tempreqvary;

/* If the subpattern set a required byte (or set a first byte that isn't	/* If the subpattern set a required byte (or set a first byte that isn't
really the first byte - see above), set it. */	really the first byte - see above), set it. */

if (subreqbyte >= 0) reqbyte = subreqbyte;	if (subreqchar >= 0) reqchar = subreqchar;
}	}

/* For a forward assertion, we take the reqbyte, if set. This can be	/* For a forward assertion, we take the reqchar, if set. This can be
helpful if the pattern that follows the assertion doesn't set a different	helpful if the pattern that follows the assertion doesn't set a different
char. For example, it's useful for /(?=abcde).+/. We can't set firstbyte	char. For example, it's useful for /(?=abcde).+/. We can't set firstchar
for an assertion, however because it leads to incorrect effect for patterns	for an assertion, however because it leads to incorrect effect for patterns
such as /(?=a)a.+/ when the "real" "a" would then become a reqbyte instead	such as /(?=a)a.+/ when the "real" "a" would then become a reqchar instead
of a firstbyte. This is overcome by a scan at the end if there's no	of a firstchar. This is overcome by a scan at the end if there's no
firstbyte, looking for an asserted first char. */	firstchar, looking for an asserted first char. */

else if (bravalue == OP_ASSERT && subreqbyte >= 0) reqbyte = subreqbyte;	else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar;
break; /* End of processing '(' */	break; /* End of processing '(' */


Line 6416 for (;; ptr++)	Line 6726 for (;; ptr++)
/* For metasequences that actually match a character, we disable the	/* For metasequences that actually match a character, we disable the
setting of a first character if it hasn't already been set. */	setting of a first character if it hasn't already been set. */

if (firstbyte == REQ_UNSET && -c > ESC_b && -c < ESC_Z)	if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)
firstbyte = REQ_NONE;	firstchar = REQ_NONE;

/* Set values to reset to if this is followed by a zero repeat. */	/* Set values to reset to if this is followed by a zero repeat. */

zerofirstbyte = firstbyte;	zerofirstchar = firstchar;
zeroreqbyte = reqbyte;	zeroreqchar = reqchar;

/* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'	/* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'
is a subroutine call by number (Oniguruma syntax). In fact, the value	is a subroutine call by number (Oniguruma syntax). In fact, the value
Line 6433 for (;; ptr++)	Line 6743 for (;; ptr++)

if (-c == ESC_g)	if (-c == ESC_g)
{	{
const uschar *p;	const pcre_uchar *p;
save_hwm = cd->hwm; /* Normally this is set when '(' is read */	save_hwm = cd->hwm; /* Normally this is set when '(' is read */
terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?	terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;	CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
Line 6450 for (;; ptr++)	Line 6760 for (;; ptr++)

if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)	if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
{	{
BOOL isnumber = TRUE;	BOOL is_a_number = TRUE;
for (p = ptr + 1; p != 0 && p != terminator; p++)	for (p = ptr + 1; p != 0 && p != terminator; p++)
{	{
if ((cd->ctypes[*p] & ctype_digit) == 0) isnumber = FALSE;	if (!MAX_255(*p)) { is_a_number = FALSE; break; }
	if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;
if ((cd->ctypes[*p] & ctype_word) == 0) break;	if ((cd->ctypes[*p] & ctype_word) == 0) break;
}	}
if (*p != terminator)	if (*p != terminator)
Line 6461 for (;; ptr++)	Line 6772 for (;; ptr++)
*errorcodeptr = ERR57;	*errorcodeptr = ERR57;
break;	break;
}	}
if (isnumber)	if (is_a_number)
{	{
ptr++;	ptr++;
goto HANDLE_NUMERICAL_RECURSION;	goto HANDLE_NUMERICAL_RECURSION;
Line 6473 for (;; ptr++)	Line 6784 for (;; ptr++)
/* Test a signed number in angle brackets or quotes. */	/* Test a signed number in angle brackets or quotes. */

p = ptr + 2;	p = ptr + 2;
while ((digitab[*p] & ctype_digit) != 0) p++;	while (IS_DIGIT(*p)) p++;
if (*p != terminator)	if (*p != terminator)
{	{
*errorcodeptr = ERR57;	*errorcodeptr = ERR57;
Line 6501 for (;; ptr++)	Line 6812 for (;; ptr++)
goto NAMED_REF_OR_RECURSE;	goto NAMED_REF_OR_RECURSE;
}	}

/* Back references are handled specially; must disable firstbyte if	/* Back references are handled specially; must disable firstchar if
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
':' later. */	':' later. */

Line 6511 for (;; ptr++)	Line 6822 for (;; ptr++)
recno = -c - ESC_REF;	recno = -c - ESC_REF;

HANDLE_REFERENCE: /* Come here from named backref handling */	HANDLE_REFERENCE: /* Come here from named backref handling */
if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;	if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
previous = code;	previous = code;
*code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;	*code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
PUT2INC(code, 0, recno);	PUT2INC(code, 0, recno);
Line 6561 for (;; ptr++)	Line 6872 for (;; ptr++)
/* For the rest (including \X when Unicode properties are supported), we	/* For the rest (including \X when Unicode properties are supported), we
can obtain the OP value by negating the escape value in the default	can obtain the OP value by negating the escape value in the default
situation when PCRE_UCP is not set. When it is set, we substitute	situation when PCRE_UCP is not set. When it is set, we substitute
Unicode property tests. */	Unicode property tests. Note that \b and \B do a one-character
	lookbehind. */

else	else
{	{
	if ((-c == ESC_b \|\| -c == ESC_B) && cd->max_lookbehind == 0)
	cd->max_lookbehind = 1;
#ifdef SUPPORT_UCP	#ifdef SUPPORT_UCP
if (-c >= ESC_DU && -c <= ESC_wu)	if (-c >= ESC_DU && -c <= ESC_wu)
{	{
Line 6578 for (;; ptr++)	Line 6892 for (;; ptr++)

{	{
previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;	previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
*code++ = (!utf8 && c == -ESC_C)? OP_ALLANY : -c;	*code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;
}	}
}	}
continue;	continue;
Line 6588 for (;; ptr++)	Line 6902 for (;; ptr++)
a value > 127. We set its representation in the length/buffer, and then	a value > 127. We set its representation in the length/buffer, and then
handle it as a data character. */	handle it as a data character. */

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && c > 127)	if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
mclength = _pcre_ord2utf8(c, mcbuffer);	mclength = PRIV(ord2utf)(c, mcbuffer);
else	else
#endif	#endif

Line 6611 for (;; ptr++)	Line 6925 for (;; ptr++)
mclength = 1;	mclength = 1;
mcbuffer[0] = c;	mcbuffer[0] = c;

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && c >= 0xc0)	if (utf && HAS_EXTRALEN(c))
{	ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr));
while ((ptr[1] & 0xc0) == 0x80)
mcbuffer[mclength++] = *(++ptr);
}
#endif	#endif

/* At this point we have the character's bytes in mcbuffer, and the length	/* At this point we have the character's bytes in mcbuffer, and the length
Line 6634 for (;; ptr++)	Line 6945 for (;; ptr++)

/* Set the first and required bytes appropriately. If no previous first	/* Set the first and required bytes appropriately. If no previous first
byte, set it from this character, but revert to none on a zero repeat.	byte, set it from this character, but revert to none on a zero repeat.
Otherwise, leave the firstbyte value alone, and don't change it on a zero	Otherwise, leave the firstchar value alone, and don't change it on a zero
repeat. */	repeat. */

if (firstbyte == REQ_UNSET)	if (firstchar == REQ_UNSET)
{	{
zerofirstbyte = REQ_NONE;	zerofirstchar = REQ_NONE;
zeroreqbyte = reqbyte;	zeroreqchar = reqchar;

/* If the character is more than one byte long, we can set firstbyte	/* If the character is more than one byte long, we can set firstchar
only if it is not to be matched caselessly. */	only if it is not to be matched caselessly. */

if (mclength == 1 \|\| req_caseopt == 0)	if (mclength == 1 \|\| req_caseopt == 0)
{	{
firstbyte = mcbuffer[0] \| req_caseopt;	firstchar = mcbuffer[0] \| req_caseopt;
if (mclength != 1) reqbyte = code[-1] \| cd->req_varyopt;	if (mclength != 1) reqchar = code[-1] \| cd->req_varyopt;
}	}
else firstbyte = reqbyte = REQ_NONE;	else firstchar = reqchar = REQ_NONE;
}	}

/* firstbyte was previously set; we can set reqbyte only if the length is	/* firstchar was previously set; we can set reqchar only if the length is
1 or the matching is caseful. */	1 or the matching is caseful. */

else	else
{	{
zerofirstbyte = firstbyte;	zerofirstchar = firstchar;
zeroreqbyte = reqbyte;	zeroreqchar = reqchar;
if (mclength == 1 \|\| req_caseopt == 0)	if (mclength == 1 \|\| req_caseopt == 0)
reqbyte = code[-1] \| req_caseopt \| cd->req_varyopt;	reqchar = code[-1] \| req_caseopt \| cd->req_varyopt;
}	}

break; /* End of literal character handling */	break; /* End of literal character handling */
Line 6701 Arguments:	Line 7012 Arguments:
reset_bracount TRUE to reset the count for each branch	reset_bracount TRUE to reset the count for each branch
skipbytes skip this many bytes at start (for brackets and OP_COND)	skipbytes skip this many bytes at start (for brackets and OP_COND)
cond_depth depth of nesting for conditional subpatterns	cond_depth depth of nesting for conditional subpatterns
firstbyteptr place to put the first required character, or a negative number	firstcharptr place to put the first required character, or a negative number
reqbyteptr place to put the last required character, or a negative number	reqcharptr place to put the last required character, or a negative number
bcptr pointer to the chain of currently open branches	bcptr pointer to the chain of currently open branches
cd points to the data block with tables pointers etc.	cd points to the data block with tables pointers etc.
lengthptr NULL during the real compile phase	lengthptr NULL during the real compile phase
Line 6712 Returns: TRUE on success	Line 7023 Returns: TRUE on success
*/	*/

static BOOL	static BOOL
compile_regex(int options, uschar codeptr, const uschar ptrptr,	compile_regex(int options, pcre_uchar codeptr, const pcre_uchar ptrptr,
int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,	int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
int cond_depth, int firstbyteptr, int reqbyteptr, branch_chain *bcptr,	int cond_depth, pcre_int32 firstcharptr, pcre_int32 reqcharptr,
compile_data cd, int lengthptr)	branch_chain bcptr, compile_data cd, int *lengthptr)
{	{
const uschar ptr = ptrptr;	const pcre_uchar ptr = ptrptr;
uschar code = codeptr;	pcre_uchar code = codeptr;
uschar *last_branch = code;	pcre_uchar *last_branch = code;
uschar *start_bracket = code;	pcre_uchar *start_bracket = code;
uschar *reverse_count = NULL;	pcre_uchar *reverse_count = NULL;
open_capitem capitem;	open_capitem capitem;
int capnumber = 0;	int capnumber = 0;
int firstbyte, reqbyte;	pcre_int32 firstchar, reqchar;
int branchfirstbyte, branchreqbyte;	pcre_int32 branchfirstchar, branchreqchar;
int length;	int length;
int orig_bracount;	int orig_bracount;
int max_bracount;	int max_bracount;
Line 6734 branch_chain bc;	Line 7045 branch_chain bc;
bc.outer = bcptr;	bc.outer = bcptr;
bc.current_branch = code;	bc.current_branch = code;

firstbyte = reqbyte = REQ_UNSET;	firstchar = reqchar = REQ_UNSET;

/* Accumulate the length for use in the pre-compile phase. Start with the	/* Accumulate the length for use in the pre-compile phase. Start with the
length of the BRA and KET and any extra bytes that are required at the	length of the BRA and KET and any extra bytes that are required at the
Line 6793 for (;;)	Line 7104 for (;;)
/* Now compile the branch; in the pre-compile phase its length gets added	/* Now compile the branch; in the pre-compile phase its length gets added
into the length. */	into the length. */

if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte,	if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar,
&branchreqbyte, &bc, cond_depth, cd,	&branchreqchar, &bc, cond_depth, cd,
(lengthptr == NULL)? NULL : &length))	(lengthptr == NULL)? NULL : &length))
{	{
*ptrptr = ptr;	*ptrptr = ptr;
Line 6810 for (;;)	Line 7121 for (;;)

if (lengthptr == NULL)	if (lengthptr == NULL)
{	{
/* If this is the first branch, the firstbyte and reqbyte values for the	/* If this is the first branch, the firstchar and reqchar values for the
branch become the values for the regex. */	branch become the values for the regex. */

if (*last_branch != OP_ALT)	if (*last_branch != OP_ALT)
{	{
firstbyte = branchfirstbyte;	firstchar = branchfirstchar;
reqbyte = branchreqbyte;	reqchar = branchreqchar;
}	}

/* If this is not the first branch, the first char and reqbyte have to	/* If this is not the first branch, the first char and reqchar have to
match the values from all the previous branches, except that if the	match the values from all the previous branches, except that if the
previous value for reqbyte didn't have REQ_VARY set, it can still match,	previous value for reqchar didn't have REQ_VARY set, it can still match,
and we set REQ_VARY for the regex. */	and we set REQ_VARY for the regex. */

else	else
{	{
/* If we previously had a firstbyte, but it doesn't match the new branch,	/* If we previously had a firstchar, but it doesn't match the new branch,
we have to abandon the firstbyte for the regex, but if there was	we have to abandon the firstchar for the regex, but if there was
previously no reqbyte, it takes on the value of the old firstbyte. */	previously no reqchar, it takes on the value of the old firstchar. */

if (firstbyte >= 0 && firstbyte != branchfirstbyte)	if (firstchar >= 0 && firstchar != branchfirstchar)
{	{
if (reqbyte < 0) reqbyte = firstbyte;	if (reqchar < 0) reqchar = firstchar;
firstbyte = REQ_NONE;	firstchar = REQ_NONE;
}	}

/* If we (now or from before) have no firstbyte, a firstbyte from the	/* If we (now or from before) have no firstchar, a firstchar from the
branch becomes a reqbyte if there isn't a branch reqbyte. */	branch becomes a reqchar if there isn't a branch reqchar. */

if (firstbyte < 0 && branchfirstbyte >= 0 && branchreqbyte < 0)	if (firstchar < 0 && branchfirstchar >= 0 && branchreqchar < 0)
branchreqbyte = branchfirstbyte;	branchreqchar = branchfirstchar;

/* Now ensure that the reqbytes match */	/* Now ensure that the reqchars match */

if ((reqbyte & ~REQ_VARY) != (branchreqbyte & ~REQ_VARY))	if ((reqchar & ~REQ_VARY) != (branchreqchar & ~REQ_VARY))
reqbyte = REQ_NONE;	reqchar = REQ_NONE;
else reqbyte \|= branchreqbyte; /* To "or" REQ_VARY */	else reqchar \|= branchreqchar; /* To "or" REQ_VARY */
}	}

/* If lookbehind, check that this branch matches a fixed-length string, and	/* If lookbehind, check that this branch matches a fixed-length string, and
Line 6875 for (;;)	Line 7186 for (;;)
*ptrptr = ptr;	*ptrptr = ptr;
return FALSE;	return FALSE;
}	}
else { PUT(reverse_count, 0, fixed_length); }	else
	{
	if (fixed_length > cd->max_lookbehind)
	cd->max_lookbehind = fixed_length;
	PUT(reverse_count, 0, fixed_length);
	}
}	}
}	}

Line 6916 for (;;)	Line 7232 for (;;)
if (cd->open_caps->flag)	if (cd->open_caps->flag)
{	{
memmove(start_bracket + 1 + LINK_SIZE, start_bracket,	memmove(start_bracket + 1 + LINK_SIZE, start_bracket,
code - start_bracket);	IN_UCHARS(code - start_bracket));
*start_bracket = OP_ONCE;	*start_bracket = OP_ONCE;
code += 1 + LINK_SIZE;	code += 1 + LINK_SIZE;
PUT(start_bracket, 1, (int)(code - start_bracket));	PUT(start_bracket, 1, (int)(code - start_bracket));
Line 6936 for (;;)	Line 7252 for (;;)

*codeptr = code;	*codeptr = code;
*ptrptr = ptr;	*ptrptr = ptr;
*firstbyteptr = firstbyte;	*firstcharptr = firstchar;
*reqbyteptr = reqbyte;	*reqcharptr = reqchar;
if (lengthptr != NULL)	if (lengthptr != NULL)
{	{
if (OFLOW_MAX - *lengthptr < length)	if (OFLOW_MAX - *lengthptr < length)
Line 7018 Returns: TRUE or FALSE	Line 7334 Returns: TRUE or FALSE
*/	*/

static BOOL	static BOOL
is_anchored(register const uschar *code, unsigned int bracket_map,	is_anchored(register const pcre_uchar *code, unsigned int bracket_map,
unsigned int backref_map)	unsigned int backref_map)
{	{
do {	do {
const uschar scode = first_significant_code(code + _pcre_OP_lengths[code],	const pcre_uchar *scode = first_significant_code(
FALSE);	code + PRIV(OP_lengths)[*code], FALSE);
register int op = *scode;	register int op = *scode;

/* Non-capturing brackets */	/* Non-capturing brackets */
Line 7095 Returns: TRUE or FALSE	Line 7411 Returns: TRUE or FALSE
*/	*/

static BOOL	static BOOL
is_startline(const uschar *code, unsigned int bracket_map,	is_startline(const pcre_uchar *code, unsigned int bracket_map,
unsigned int backref_map)	unsigned int backref_map)
{	{
do {	do {
const uschar scode = first_significant_code(code + _pcre_OP_lengths[code],	const pcre_uchar *scode = first_significant_code(
FALSE);	code + PRIV(OP_lengths)[*code], FALSE);
register int op = *scode;	register int op = *scode;

/* 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
Line 7111 do {	Line 7427 do {
if (op == OP_COND)	if (op == OP_COND)
{	{
scode += 1 + LINK_SIZE;	scode += 1 + LINK_SIZE;
if (*scode == OP_CALLOUT) scode += _pcre_OP_lengths[OP_CALLOUT];	if (*scode == OP_CALLOUT) scode += PRIV(OP_lengths)[OP_CALLOUT];
switch (*scode)	switch (*scode)
{	{
case OP_CREF:	case OP_CREF:
Line 7198 Returns: -1 or the fixed first char	Line 7514 Returns: -1 or the fixed first char
*/	*/

static int	static int
find_firstassertedchar(const uschar *code, BOOL inassert)	find_firstassertedchar(const pcre_uchar *code, BOOL inassert)
{	{
register int c = -1;	register int c = -1;
do {	do {
int d;	int d;
int xl = (code == OP_CBRA \|\| code == OP_SCBRA \|\|	int xl = (code == OP_CBRA \|\| code == OP_SCBRA \|\|
code == OP_CBRAPOS \|\| code == OP_SCBRAPOS)? 2:0;	code == OP_CBRAPOS \|\| code == OP_SCBRAPOS)? IMM2_SIZE:0;
const uschar *scode = first_significant_code(code + 1+LINK_SIZE + xl, TRUE);	const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl,
	TRUE);
register int op = *scode;	register int op = *scode;

switch(op)	switch(op)
Line 7229 do {	Line 7546 do {
break;	break;

case OP_EXACT:	case OP_EXACT:
scode += 2;	scode += IMM2_SIZE;
/* Fall through */	/* Fall through */

case OP_CHAR:	case OP_CHAR:
Line 7242 do {	Line 7559 do {
break;	break;

case OP_EXACTI:	case OP_EXACTI:
scode += 2;	scode += IMM2_SIZE;
/* Fall through */	/* Fall through */

case OP_CHARI:	case OP_CHARI:
Line 7285 Returns: pointer to compiled data block, or NUL	Line 7602 Returns: pointer to compiled data block, or NUL
with errorptr and erroroffset set	with errorptr and erroroffset set
*/	*/

	#ifdef COMPILE_PCRE8
PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION	PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
pcre_compile(const char pattern, int options, const char *errorptr,	pcre_compile(const char pattern, int options, const char *errorptr,
int erroroffset, const unsigned char tables)	int erroroffset, const unsigned char tables)
	#else
	PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
	pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr,
	int erroroffset, const unsigned char tables)
	#endif
{	{
	#ifdef COMPILE_PCRE8
return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);	return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
	#else
	return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables);
	#endif
}	}


	#ifdef COMPILE_PCRE8
PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION	PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
pcre_compile2(const char pattern, int options, int errorcodeptr,	pcre_compile2(const char pattern, int options, int errorcodeptr,
const char *errorptr, int erroroffset, const unsigned char *tables)	const char *errorptr, int erroroffset, const unsigned char *tables)
	#else
	PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION
	pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr,
	const char *errorptr, int erroroffset, const unsigned char *tables)
	#endif
{	{
real_pcre *re;	REAL_PCRE *re;
int length = 1; /* For final END opcode */	int length = 1; /* For final END opcode */
int firstbyte, reqbyte, newline;	pcre_int32 firstchar, reqchar;
	int newline;
int errorcode = 0;	int errorcode = 0;
int skipatstart = 0;	int skipatstart = 0;
BOOL utf8;	BOOL utf;
size_t size;	size_t size;
uschar *code;	pcre_uchar *code;
const uschar *codestart;	const pcre_uchar *codestart;
const uschar *ptr;	const pcre_uchar *ptr;
compile_data compile_block;	compile_data compile_block;
compile_data *cd = &compile_block;	compile_data *cd = &compile_block;

Line 7317 this purpose. The same space is used in the second pha	Line 7651 this purpose. The same space is used in the second pha
to fill in forward references to subpatterns. That may overflow, in which case	to fill in forward references to subpatterns. That may overflow, in which case
new memory is obtained from malloc(). */	new memory is obtained from malloc(). */

uschar cworkspace[COMPILE_WORK_SIZE];	pcre_uchar cworkspace[COMPILE_WORK_SIZE];

/* Set this early so that early errors get offset 0. */	/* Set this early so that early errors get offset 0. */

ptr = (const uschar *)pattern;	ptr = (const pcre_uchar *)pattern;

/* We can't pass back an error message if errorptr is NULL; I guess the best we	/* We can't pass back an error message if errorptr is NULL; I guess the best we
can do is just return NULL, but we can set a code value if there is a code	can do is just return NULL, but we can set a code value if there is a code
Line 7348 if (erroroffset == NULL)	Line 7682 if (erroroffset == NULL)

/* Set up pointers to the individual character tables */	/* Set up pointers to the individual character tables */

if (tables == NULL) tables = _pcre_default_tables;	if (tables == NULL) tables = PRIV(default_tables);
cd->lcc = tables + lcc_offset;	cd->lcc = tables + lcc_offset;
cd->fcc = tables + fcc_offset;	cd->fcc = tables + fcc_offset;
cd->cbits = tables + cbits_offset;	cd->cbits = tables + cbits_offset;
Line 7371 while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS &&	Line 7705 while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS &&
int newnl = 0;	int newnl = 0;
int newbsr = 0;	int newbsr = 0;

if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)	#ifdef COMPILE_PCRE8
	if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 5) == 0)
{ skipatstart += 7; options \|= PCRE_UTF8; continue; }	{ skipatstart += 7; options \|= PCRE_UTF8; continue; }
else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)	#endif
	#ifdef COMPILE_PCRE16
	if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0)
	{ skipatstart += 8; options \|= PCRE_UTF16; continue; }
	#endif
	else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UCP_RIGHTPAR, 4) == 0)
{ skipatstart += 6; options \|= PCRE_UCP; continue; }	{ skipatstart += 6; options \|= PCRE_UCP; continue; }
else if (strncmp((char *)(ptr+skipatstart+2), STRING_NO_START_OPT_RIGHTPAR, 13) == 0)	else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_START_OPT_RIGHTPAR, 13) == 0)
{ skipatstart += 15; options \|= PCRE_NO_START_OPTIMIZE; continue; }	{ skipatstart += 15; options \|= PCRE_NO_START_OPTIMIZE; continue; }

if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)	if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_CR_RIGHTPAR, 3) == 0)
{ skipatstart += 5; newnl = PCRE_NEWLINE_CR; }	{ skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
else if (strncmp((char *)(ptr+skipatstart+2), STRING_LF_RIGHTPAR, 3) == 0)	else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_LF_RIGHTPAR, 3) == 0)
{ skipatstart += 5; newnl = PCRE_NEWLINE_LF; }	{ skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
else if (strncmp((char *)(ptr+skipatstart+2), STRING_CRLF_RIGHTPAR, 5) == 0)	else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_CRLF_RIGHTPAR, 5) == 0)
{ skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }	{ skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
else if (strncmp((char *)(ptr+skipatstart+2), STRING_ANY_RIGHTPAR, 4) == 0)	else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_ANY_RIGHTPAR, 4) == 0)
{ skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }	{ skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
else if (strncmp((char *)(ptr+skipatstart+2), STRING_ANYCRLF_RIGHTPAR, 8) == 0)	else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_ANYCRLF_RIGHTPAR, 8) == 0)
{ skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }	{ skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }

else if (strncmp((char *)(ptr+skipatstart+2), STRING_BSR_ANYCRLF_RIGHTPAR, 12) == 0)	else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_BSR_ANYCRLF_RIGHTPAR, 12) == 0)
{ skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }	{ skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
else if (strncmp((char *)(ptr+skipatstart+2), STRING_BSR_UNICODE_RIGHTPAR, 12) == 0)	else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_BSR_UNICODE_RIGHTPAR, 12) == 0)
{ skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }	{ skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }

if (newnl != 0)	if (newnl != 0)
Line 7401 while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS &&	Line 7741 while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS &&
else break;	else break;
}	}

utf8 = (options & PCRE_UTF8) != 0;	/* PCRE_UTF16 has the same value as PCRE_UTF8. */
	utf = (options & PCRE_UTF8) != 0;

/* Can't support UTF8 unless PCRE has been compiled to include the code. The	/* Can't support UTF unless PCRE has been compiled to include the code. The
return of an error code from _pcre_valid_utf8() is a new feature, introduced in	return of an error code from PRIV(valid_utf)() is a new feature, introduced in
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
not used here. */	not used here. */

#ifdef SUPPORT_UTF8	#ifdef SUPPORT_UTF
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 &&	if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 &&
(errorcode = _pcre_valid_utf8((USPTR)pattern, -1, erroroffset)) != 0)	(errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0)
{	{
	#ifdef COMPILE_PCRE8
errorcode = ERR44;	errorcode = ERR44;
	#else
	errorcode = ERR74;
	#endif
goto PCRE_EARLY_ERROR_RETURN2;	goto PCRE_EARLY_ERROR_RETURN2;
}	}
#else	#else
if (utf8)	if (utf)
{	{
errorcode = ERR32;	errorcode = ERR32;
goto PCRE_EARLY_ERROR_RETURN;	goto PCRE_EARLY_ERROR_RETURN;
Line 7492 cd->backref_map = 0;	Line 7837 cd->backref_map = 0;
/* Reflect pattern for debugging output */	/* Reflect pattern for debugging output */

DPRINTF(("------------------------------------------------------------------\n"));	DPRINTF(("------------------------------------------------------------------\n"));
DPRINTF(("%s\n", pattern));	#ifdef PCRE_DEBUG
	print_puchar(stdout, (PCRE_PUCHAR)pattern);
	#endif
	DPRINTF(("\n"));

/* Pretend to compile the pattern while actually just accumulating the length	/* Pretend to compile the pattern while actually just accumulating the length
of memory required. This behaviour is triggered by passing a non-NULL final	of memory required. This behaviour is triggered by passing a non-NULL final
Line 7509 cd->start_code = cworkspace;	Line 7857 cd->start_code = cworkspace;
cd->hwm = cworkspace;	cd->hwm = cworkspace;
cd->start_workspace = cworkspace;	cd->start_workspace = cworkspace;
cd->workspace_size = COMPILE_WORK_SIZE;	cd->workspace_size = COMPILE_WORK_SIZE;
cd->start_pattern = (const uschar *)pattern;	cd->start_pattern = (const pcre_uchar *)pattern;
cd->end_pattern = (const uschar *)(pattern + strlen(pattern));	cd->end_pattern = (const pcre_uchar )(pattern + STRLEN_UC((const pcre_uchar )pattern));
cd->req_varyopt = 0;	cd->req_varyopt = 0;
	cd->assert_depth = 0;
	cd->max_lookbehind = 0;
cd->external_options = options;	cd->external_options = options;
cd->external_flags = 0;	cd->external_flags = 0;
cd->open_caps = NULL;	cd->open_caps = NULL;
Line 7526 ptr += skipatstart;	Line 7876 ptr += skipatstart;
code = cworkspace;	code = cworkspace;
*code = OP_BRA;	*code = OP_BRA;
(void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,	(void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,
FALSE, 0, 0, &firstbyte, &reqbyte, NULL, cd, &length);	FALSE, 0, 0, &firstchar, &reqchar, NULL, cd, &length);
if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;	if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;

DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,	DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
cd->hwm - cworkspace));	(int)(cd->hwm - cworkspace)));

if (length > MAX_PATTERN_SIZE)	if (length > MAX_PATTERN_SIZE)
{	{
Line 7543 externally provided function. Integer overflow should	Line 7893 externally provided function. Integer overflow should
because nowadays we limit the maximum value of cd->names_found and	because nowadays we limit the maximum value of cd->names_found and
cd->name_entry_size. */	cd->name_entry_size. */

size = length + sizeof(real_pcre) + cd->names_found * cd->name_entry_size;	size = sizeof(REAL_PCRE) + (length + cd->names_found * cd->name_entry_size) * sizeof(pcre_uchar);
re = (real_pcre *)(pcre_malloc)(size);	re = (REAL_PCRE *)(PUBL(malloc))(size);

if (re == NULL)	if (re == NULL)
{	{
Line 7562 re->magic_number = MAGIC_NUMBER;	Line 7912 re->magic_number = MAGIC_NUMBER;
re->size = (int)size;	re->size = (int)size;
re->options = cd->external_options;	re->options = cd->external_options;
re->flags = cd->external_flags;	re->flags = cd->external_flags;
re->dummy1 = 0;	re->first_char = 0;
re->first_byte = 0;	re->req_char = 0;
re->req_byte = 0;	re->name_table_offset = sizeof(REAL_PCRE) / sizeof(pcre_uchar);
re->name_table_offset = sizeof(real_pcre);
re->name_entry_size = cd->name_entry_size;	re->name_entry_size = cd->name_entry_size;
re->name_count = cd->names_found;	re->name_count = cd->names_found;
re->ref_count = 0;	re->ref_count = 0;
re->tables = (tables == _pcre_default_tables)? NULL : tables;	re->tables = (tables == PRIV(default_tables))? NULL : tables;
re->nullpad = NULL;	re->nullpad = NULL;

/* The starting points of the name/number translation table and of the code are	/* The starting points of the name/number translation table and of the code are
Line 7582 field; this time it's used for remembering forward ref	Line 7931 field; this time it's used for remembering forward ref
cd->final_bracount = cd->bracount; /* Save for checking forward references */	cd->final_bracount = cd->bracount; /* Save for checking forward references */
cd->assert_depth = 0;	cd->assert_depth = 0;
cd->bracount = 0;	cd->bracount = 0;
	cd->max_lookbehind = 0;
cd->names_found = 0;	cd->names_found = 0;
cd->name_table = (uschar *)re + re->name_table_offset;	cd->name_table = (pcre_uchar *)re + re->name_table_offset;
codestart = cd->name_table + re->name_entry_size * re->name_count;	codestart = cd->name_table + re->name_entry_size * re->name_count;
cd->start_code = codestart;	cd->start_code = codestart;
cd->hwm = (uschar *)(cd->start_workspace);	cd->hwm = (pcre_uchar *)(cd->start_workspace);
cd->req_varyopt = 0;	cd->req_varyopt = 0;
cd->had_accept = FALSE;	cd->had_accept = FALSE;
cd->check_lookbehind = FALSE;	cd->check_lookbehind = FALSE;
Line 7596 cd->open_caps = NULL;	Line 7946 cd->open_caps = NULL;
error, errorcode will be set non-zero, so we don't need to look at the result	error, errorcode will be set non-zero, so we don't need to look at the result
of the function here. */	of the function here. */

ptr = (const uschar *)pattern + skipatstart;	ptr = (const pcre_uchar *)pattern + skipatstart;
code = (uschar *)codestart;	code = (pcre_uchar *)codestart;
*code = OP_BRA;	*code = OP_BRA;
(void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0,	(void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0,
&firstbyte, &reqbyte, NULL, cd, NULL);	&firstchar, &reqchar, NULL, cd, NULL);
re->top_bracket = cd->bracount;	re->top_bracket = cd->bracount;
re->top_backref = cd->top_backref;	re->top_backref = cd->top_backref;
re->flags = cd->external_flags;	re->max_lookbehind = cd->max_lookbehind;
	re->flags = cd->external_flags \| PCRE_MODE;

if (cd->had_accept) reqbyte = REQ_NONE; /* Must disable after (ACCEPT) /	if (cd->had_accept) reqchar = REQ_NONE; /* Must disable after (ACCEPT) /

/* 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. */

Line 7626 references; optimize for them, as searching a large re	Line 7977 references; optimize for them, as searching a large re
if (cd->hwm > cd->start_workspace)	if (cd->hwm > cd->start_workspace)
{	{
int prev_recno = -1;	int prev_recno = -1;
const uschar *groupptr = NULL;	const pcre_uchar *groupptr = NULL;
while (errorcode == 0 && cd->hwm > cd->start_workspace)	while (errorcode == 0 && cd->hwm > cd->start_workspace)
{	{
int offset, recno;	int offset, recno;
Line 7635 if (cd->hwm > cd->start_workspace)	Line 7986 if (cd->hwm > cd->start_workspace)
recno = GET(codestart, offset);	recno = GET(codestart, offset);
if (recno != prev_recno)	if (recno != prev_recno)
{	{
groupptr = _pcre_find_bracket(codestart, utf8, recno);	groupptr = PRIV(find_bracket)(codestart, utf, recno);
prev_recno = recno;	prev_recno = recno;
}	}
if (groupptr == NULL) errorcode = ERR53;	if (groupptr == NULL) errorcode = ERR53;
else PUT(((uschar *)codestart), offset, (int)(groupptr - codestart));	else PUT(((pcre_uchar *)codestart), offset, (int)(groupptr - codestart));
}	}
}	}

/* If the workspace had to be expanded, free the new memory. */	/* If the workspace had to be expanded, free the new memory. */

if (cd->workspace_size > COMPILE_WORK_SIZE)	if (cd->workspace_size > COMPILE_WORK_SIZE)
(pcre_free)((void *)cd->start_workspace);	(PUBL(free))((void *)cd->start_workspace);

/* Give an error if there's back reference to a non-existent capturing	/* Give an error if there's back reference to a non-existent capturing
subpattern. */	subpattern. */
Line 7663 length, and set their lengths. */	Line 8014 length, and set their lengths. */

if (cd->check_lookbehind)	if (cd->check_lookbehind)
{	{
uschar cc = (uschar )codestart;	pcre_uchar cc = (pcre_uchar )codestart;

/* Loop, searching for OP_REVERSE items, and process those that do not have	/* Loop, searching for OP_REVERSE items, and process those that do not have
their length set. (Actually, it will also re-process any that have a length	their length set. (Actually, it will also re-process any that have a length
of zero, but that is a pathological case, and it does no harm.) When we find	of zero, but that is a pathological case, and it does no harm.) When we find
one, we temporarily terminate the branch it is in while we scan it. */	one, we temporarily terminate the branch it is in while we scan it. */

for (cc = (uschar *)_pcre_find_bracket(codestart, utf8, -1);	for (cc = (pcre_uchar *)PRIV(find_bracket)(codestart, utf, -1);
cc != NULL;	cc != NULL;
cc = (uschar *)_pcre_find_bracket(cc, utf8, -1))	cc = (pcre_uchar *)PRIV(find_bracket)(cc, utf, -1))
{	{
if (GET(cc, 1) == 0)	if (GET(cc, 1) == 0)
{	{
int fixed_length;	int fixed_length;
uschar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE);	pcre_uchar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE);
int end_op = *be;	int end_op = *be;
*be = OP_END;	*be = OP_END;
fixed_length = find_fixedlength(cc, (re->options & PCRE_UTF8) != 0, TRUE,	fixed_length = find_fixedlength(cc, (re->options & PCRE_UTF8) != 0, TRUE,
Line 7690 if (cd->check_lookbehind)	Line 8041 if (cd->check_lookbehind)
(fixed_length == -4)? ERR70 : ERR25;	(fixed_length == -4)? ERR70 : ERR25;
break;	break;
}	}
	if (fixed_length > cd->max_lookbehind) cd->max_lookbehind = fixed_length;
PUT(cc, 1, fixed_length);	PUT(cc, 1, fixed_length);
}	}
cc += 1 + LINK_SIZE;	cc += 1 + LINK_SIZE;
Line 7700 if (cd->check_lookbehind)	Line 8052 if (cd->check_lookbehind)

if (errorcode != 0)	if (errorcode != 0)
{	{
(pcre_free)(re);	(PUBL(free))(re);
PCRE_EARLY_ERROR_RETURN:	PCRE_EARLY_ERROR_RETURN:
erroroffset = (int)(ptr - (const uschar )pattern);	erroroffset = (int)(ptr - (const pcre_uchar )pattern);
PCRE_EARLY_ERROR_RETURN2:	PCRE_EARLY_ERROR_RETURN2:
*errorptr = find_error_text(errorcode);	*errorptr = find_error_text(errorcode);
if (errorcodeptr != NULL) *errorcodeptr = errorcode;	if (errorcodeptr != NULL) *errorcodeptr = errorcode;
Line 7725 if ((re->options & PCRE_ANCHORED) == 0)	Line 8077 if ((re->options & PCRE_ANCHORED) == 0)
re->options \|= PCRE_ANCHORED;	re->options \|= PCRE_ANCHORED;
else	else
{	{
if (firstbyte < 0)	if (firstchar < 0)
firstbyte = find_firstassertedchar(codestart, FALSE);	firstchar = find_firstassertedchar(codestart, FALSE);
if (firstbyte >= 0) /* Remove caseless flag for non-caseable chars */	if (firstchar >= 0) /* Remove caseless flag for non-caseable chars */
{	{
int ch = firstbyte & 255;	#ifdef COMPILE_PCRE8
re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&	re->first_char = firstchar & 0xff;
cd->fcc[ch] == ch)? ch : firstbyte;	#else
	#ifdef COMPILE_PCRE16
	re->first_char = firstchar & 0xffff;
	#endif
	#endif
	if ((firstchar & REQ_CASELESS) != 0)
	{
	#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)
	/* We ignore non-ASCII first chars in 8 bit mode. */
	if (utf)
	{
	if (re->first_char < 128)
	{
	if (cd->fcc[re->first_char] != re->first_char)
	re->flags \|= PCRE_FCH_CASELESS;
	}
	else if (UCD_OTHERCASE(re->first_char) != re->first_char)
	re->flags \|= PCRE_FCH_CASELESS;
	}
	else
	#endif
	if (MAX_255(re->first_char)
	&& cd->fcc[re->first_char] != re->first_char)
	re->flags \|= PCRE_FCH_CASELESS;
	}

re->flags \|= PCRE_FIRSTSET;	re->flags \|= PCRE_FIRSTSET;
}	}
else if (is_startline(codestart, 0, cd->backref_map))	else if (is_startline(codestart, 0, cd->backref_map))
Line 7743 if ((re->options & PCRE_ANCHORED) == 0)	Line 8120 if ((re->options & PCRE_ANCHORED) == 0)
variable length item in the regex. Remove the caseless flag for non-caseable	variable length item in the regex. Remove the caseless flag for non-caseable
bytes. */	bytes. */

if (reqbyte >= 0 &&	if (reqchar >= 0 &&
((re->options & PCRE_ANCHORED) == 0 \|\| (reqbyte & REQ_VARY) != 0))	((re->options & PCRE_ANCHORED) == 0 \|\| (reqchar & REQ_VARY) != 0))
{	{
int ch = reqbyte & 255;	#ifdef COMPILE_PCRE8
re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&	re->req_char = reqchar & 0xff;
cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;	#else
	#ifdef COMPILE_PCRE16
	re->req_char = reqchar & 0xffff;
	#endif
	#endif
	if ((reqchar & REQ_CASELESS) != 0)
	{
	#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8)
	/* We ignore non-ASCII first chars in 8 bit mode. */
	if (utf)
	{
	if (re->req_char < 128)
	{
	if (cd->fcc[re->req_char] != re->req_char)
	re->flags \|= PCRE_RCH_CASELESS;
	}
	else if (UCD_OTHERCASE(re->req_char) != re->req_char)
	re->flags \|= PCRE_RCH_CASELESS;
	}
	else
	#endif
	if (MAX_255(re->req_char) && cd->fcc[re->req_char] != re->req_char)
	re->flags \|= PCRE_RCH_CASELESS;
	}

re->flags \|= PCRE_REQCHSET;	re->flags \|= PCRE_REQCHSET;
}	}

Line 7763 printf("Options=%08x\n", re->options);	Line 8164 printf("Options=%08x\n", re->options);

if ((re->flags & PCRE_FIRSTSET) != 0)	if ((re->flags & PCRE_FIRSTSET) != 0)
{	{
int ch = re->first_byte & 255;	pcre_uchar ch = re->first_char;
const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?	const char *caseless =
"" : " (caseless)";	((re->flags & PCRE_FCH_CASELESS) == 0)? "" : " (caseless)";
if (isprint(ch)) printf("First char = %c%s\n", ch, caseless);	if (PRINTABLE(ch)) printf("First char = %c%s\n", ch, caseless);
else printf("First char = \\x%02x%s\n", ch, caseless);	else printf("First char = \\x%02x%s\n", ch, caseless);
}	}

if ((re->flags & PCRE_REQCHSET) != 0)	if ((re->flags & PCRE_REQCHSET) != 0)
{	{
int ch = re->req_byte & 255;	pcre_uchar ch = re->req_char;
const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?	const char *caseless =
"" : " (caseless)";	((re->flags & PCRE_RCH_CASELESS) == 0)? "" : " (caseless)";
if (isprint(ch)) printf("Req char = %c%s\n", ch, caseless);	if (PRINTABLE(ch)) printf("Req char = %c%s\n", ch, caseless);
else printf("Req char = \\x%02x%s\n", ch, caseless);	else printf("Req char = \\x%02x%s\n", ch, caseless);
}	}

pcre_printint(re, stdout, TRUE);	#ifdef COMPILE_PCRE8
	pcre_printint((pcre *)re, stdout, TRUE);
	#else
	pcre16_printint((pcre *)re, stdout, TRUE);
	#endif

/* This check is done here in the debugging case so that the code that	/* This check is done here in the debugging case so that the code that
was compiled can be seen. */	was compiled can be seen. */

if (code - codestart > length)	if (code - codestart > length)
{	{
(pcre_free)(re);	(PUBL(free))(re);
*errorptr = find_error_text(ERR23);	*errorptr = find_error_text(ERR23);
erroroffset = ptr - (uschar )pattern;	erroroffset = ptr - (pcre_uchar )pattern;
if (errorcodeptr != NULL) *errorcodeptr = ERR23;	if (errorcodeptr != NULL) *errorcodeptr = ERR23;
return NULL;	return NULL;
}	}
#endif /* PCRE_DEBUG */	#endif /* PCRE_DEBUG */

	#ifdef COMPILE_PCRE8
return (pcre *)re;	return (pcre *)re;
	#else
	return (pcre16 *)re;
	#endif
}	}

/* End of pcre_compile.c */	/* End of pcre_compile.c */

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>

Removed from v.1.1.1.1
changed lines
	Added in v.1.1.1.3