blob: 76bc08525cf718eb91b4f8a2e5fcd7388f72193d [file] [log] [blame]
/*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module contains the external function pcre2_dfa_match(), which is an
alternative matching function that uses a sort of DFA algorithm (not a true
FSM). This is NOT Perl-compatible, but it has advantages in certain
applications. */
/* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved
the performance of his patterns greatly. I could not use it as it stood, as it
was not thread safe, and made assumptions about pattern sizes. Also, it caused
test 7 to loop, and test 9 to crash with a segfault.
The issue is the check for duplicate states, which is done by a simple linear
search up the state list. (Grep for "duplicate" below to find the code.) For
many patterns, there will never be many states active at one time, so a simple
linear search is fine. In patterns that have many active states, it might be a
bottleneck. The suggested code used an indexing scheme to remember which states
had previously been used for each character, and avoided the linear search when
it knew there was no chance of a duplicate. This was implemented when adding
states to the state lists.
I wrote some thread-safe, not-limited code to try something similar at the time
of checking for duplicates (instead of when adding states), using index vectors
on the stack. It did give a 13% improvement with one specially constructed
pattern for certain subject strings, but on other strings and on many of the
simpler patterns in the test suite it did worse. The major problem, I think,
was the extra time to initialize the index. This had to be done for each call
of internal_dfa_match(). (The supplied patch used a static vector, initialized
only once - I suspect this was the cause of the problems with the tests.)
Overall, I concluded that the gains in some cases did not outweigh the losses
in others, so I abandoned this code. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#define NLBLOCK mb /* Block containing newline information */
#define PSSTART start_subject /* Field containing processed string start */
#define PSEND end_subject /* Field containing processed string end */
#include "pcre2_internal.h"
#define PUBLIC_DFA_MATCH_OPTIONS \
(PCRE2_ANCHORED|PCRE2_NOTBOL|PCRE2_NOTEOL|PCRE2_NOTEMPTY| \
PCRE2_NOTEMPTY_ATSTART|PCRE2_NO_UTF_CHECK|PCRE2_PARTIAL_HARD| \
PCRE2_PARTIAL_SOFT|PCRE2_DFA_SHORTEST|PCRE2_DFA_RESTART)
/*************************************************
* Code parameters and static tables *
*************************************************/
/* These are offsets that are used to turn the OP_TYPESTAR and friends opcodes
into others, under special conditions. A gap of 20 between the blocks should be
enough. The resulting opcodes don't have to be less than 256 because they are
never stored, so we push them well clear of the normal opcodes. */
#define OP_PROP_EXTRA 300
#define OP_EXTUNI_EXTRA 320
#define OP_ANYNL_EXTRA 340
#define OP_HSPACE_EXTRA 360
#define OP_VSPACE_EXTRA 380
/* This table identifies those opcodes that are followed immediately by a
character that is to be tested in some way. This makes it possible to
centralize the loading of these characters. In the case of Type * etc, the
"character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a
small value. Non-zero values in the table are the offsets from the opcode where
the character is to be found. ***NOTE*** If the start of this table is
modified, the three tables that follow must also be modified. */
static const uint8_t coptable[] = {
0, /* End */
0, 0, 0, 0, 0, /* \A, \G, \K, \B, \b */
0, 0, 0, 0, 0, 0, /* \D, \d, \S, \s, \W, \w */
0, 0, 0, /* Any, AllAny, Anybyte */
0, 0, /* \P, \p */
0, 0, 0, 0, 0, /* \R, \H, \h, \V, \v */
0, /* \X */
0, 0, 0, 0, 0, 0, /* \Z, \z, $, $M, ^, ^M */
1, /* Char */
1, /* Chari */
1, /* not */
1, /* noti */
/* Positive single-char repeats */
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
1+IMM2_SIZE, 1+IMM2_SIZE, /* upto, minupto */
1+IMM2_SIZE, /* exact */
1, 1, 1, 1+IMM2_SIZE, /* *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* *I, *?I, +I, +?I, ?I, ??I */
1+IMM2_SIZE, 1+IMM2_SIZE, /* upto I, minupto I */
1+IMM2_SIZE, /* exact I */
1, 1, 1, 1+IMM2_SIZE, /* *+I, ++I, ?+I, upto+I */
/* Negative single-char repeats - only for chars < 256 */
1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */
1+IMM2_SIZE, 1+IMM2_SIZE, /* NOT upto, minupto */
1+IMM2_SIZE, /* NOT exact */
1, 1, 1, 1+IMM2_SIZE, /* NOT *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* NOT *I, *?I, +I, +?I, ?I, ??I */
1+IMM2_SIZE, 1+IMM2_SIZE, /* NOT upto I, minupto I */
1+IMM2_SIZE, /* NOT exact I */
1, 1, 1, 1+IMM2_SIZE, /* NOT *+I, ++I, ?+I, upto+I */
/* Positive type repeats */
1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */
1+IMM2_SIZE, 1+IMM2_SIZE, /* Type upto, minupto */
1+IMM2_SIZE, /* Type exact */
1, 1, 1, 1+IMM2_SIZE, /* Type *+, ++, ?+, upto+ */
/* Character class & ref repeats */
0, 0, 0, 0, 0, 0, /* *, *?, +, +?, ?, ?? */
0, 0, /* CRRANGE, CRMINRANGE */
0, 0, 0, 0, /* Possessive *+, ++, ?+, CRPOSRANGE */
0, /* CLASS */
0, /* NCLASS */
0, /* XCLASS - variable length */
0, /* REF */
0, /* REFI */
0, /* DNREF */
0, /* DNREFI */
0, /* RECURSE */
0, /* CALLOUT */
0, /* CALLOUT_STR */
0, /* Alt */
0, /* Ket */
0, /* KetRmax */
0, /* KetRmin */
0, /* KetRpos */
0, /* Reverse */
0, /* Assert */
0, /* Assert not */
0, /* Assert behind */
0, /* Assert behind not */
0, 0, /* ONCE, ONCE_NC */
0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */
0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */
0, 0, /* CREF, DNCREF */
0, 0, /* RREF, DNRREF */
0, 0, /* FALSE, TRUE */
0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */
0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */
0, 0, 0, 0, /* SKIP, SKIP_ARG, THEN, THEN_ARG */
0, 0, 0, 0, /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT */
0, 0, 0 /* CLOSE, SKIPZERO, DEFINE */
};
/* This table identifies those opcodes that inspect a character. It is used to
remember the fact that a character could have been inspected when the end of
the subject is reached. ***NOTE*** If the start of this table is modified, the
two tables that follow must also be modified. */
static const uint8_t poptable[] = {
0, /* End */
0, 0, 0, 1, 1, /* \A, \G, \K, \B, \b */
1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */
1, 1, 1, /* Any, AllAny, Anybyte */
1, 1, /* \P, \p */
1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */
1, /* \X */
0, 0, 0, 0, 0, 0, /* \Z, \z, $, $M, ^, ^M */
1, /* Char */
1, /* Chari */
1, /* not */
1, /* noti */
/* Positive single-char repeats */
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
1, 1, 1, /* upto, minupto, exact */
1, 1, 1, 1, /* *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* *I, *?I, +I, +?I, ?I, ??I */
1, 1, 1, /* upto I, minupto I, exact I */
1, 1, 1, 1, /* *+I, ++I, ?+I, upto+I */
/* Negative single-char repeats - only for chars < 256 */
1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */
1, 1, 1, /* NOT upto, minupto, exact */
1, 1, 1, 1, /* NOT *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* NOT *I, *?I, +I, +?I, ?I, ??I */
1, 1, 1, /* NOT upto I, minupto I, exact I */
1, 1, 1, 1, /* NOT *+I, ++I, ?+I, upto+I */
/* Positive type repeats */
1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */
1, 1, 1, /* Type upto, minupto, exact */
1, 1, 1, 1, /* Type *+, ++, ?+, upto+ */
/* Character class & ref repeats */
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
1, 1, /* CRRANGE, CRMINRANGE */
1, 1, 1, 1, /* Possessive *+, ++, ?+, CRPOSRANGE */
1, /* CLASS */
1, /* NCLASS */
1, /* XCLASS - variable length */
0, /* REF */
0, /* REFI */
0, /* DNREF */
0, /* DNREFI */
0, /* RECURSE */
0, /* CALLOUT */
0, /* CALLOUT_STR */
0, /* Alt */
0, /* Ket */
0, /* KetRmax */
0, /* KetRmin */
0, /* KetRpos */
0, /* Reverse */
0, /* Assert */
0, /* Assert not */
0, /* Assert behind */
0, /* Assert behind not */
0, 0, /* ONCE, ONCE_NC */
0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */
0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */
0, 0, /* CREF, DNCREF */
0, 0, /* RREF, DNRREF */
0, 0, /* FALSE, TRUE */
0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */
0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */
0, 0, 0, 0, /* SKIP, SKIP_ARG, THEN, THEN_ARG */
0, 0, 0, 0, /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT */
0, 0, 0 /* CLOSE, SKIPZERO, DEFINE */
};
/* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W,
and \w */
static const uint8_t toptable1[] = {
0, 0, 0, 0, 0, 0,
ctype_digit, ctype_digit,
ctype_space, ctype_space,
ctype_word, ctype_word,
0, 0 /* OP_ANY, OP_ALLANY */
};
static const uint8_t toptable2[] = {
0, 0, 0, 0, 0, 0,
ctype_digit, 0,
ctype_space, 0,
ctype_word, 0,
1, 1 /* OP_ANY, OP_ALLANY */
};
/* Structure for holding data about a particular state, which is in effect the
current data for an active path through the match tree. It must consist
entirely of ints because the working vector we are passed, and which we put
these structures in, is a vector of ints. */
typedef struct stateblock {
int offset; /* Offset to opcode (-ve has meaning) */
int count; /* Count for repeats */
int data; /* Some use extra data */
} stateblock;
#define INTS_PER_STATEBLOCK (int)(sizeof(stateblock)/sizeof(int))
/*************************************************
* Match a Regular Expression - DFA engine *
*************************************************/
/* This internal function applies a compiled pattern to a subject string,
starting at a given point, using a DFA engine. This function is called from the
external one, possibly multiple times if the pattern is not anchored. The
function calls itself recursively for some kinds of subpattern.
Arguments:
mb the match_data block with fixed information
this_start_code the opening bracket of this subexpression's code
current_subject where we currently are in the subject string
start_offset start offset in the subject string
offsets vector to contain the matching string offsets
offsetcount size of same
workspace vector of workspace
wscount size of same
rlevel function call recursion level
Returns: > 0 => number of match offset pairs placed in offsets
= 0 => offsets overflowed; longest matches are present
-1 => failed to match
< -1 => some kind of unexpected problem
The following macros are used for adding states to the two state vectors (one
for the current character, one for the following character). */
#define ADD_ACTIVE(x,y) \
if (active_count++ < wscount) \
{ \
next_active_state->offset = (x); \
next_active_state->count = (y); \
next_active_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
#define ADD_ACTIVE_DATA(x,y,z) \
if (active_count++ < wscount) \
{ \
next_active_state->offset = (x); \
next_active_state->count = (y); \
next_active_state->data = (z); \
next_active_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
#define ADD_NEW(x,y) \
if (new_count++ < wscount) \
{ \
next_new_state->offset = (x); \
next_new_state->count = (y); \
next_new_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
#define ADD_NEW_DATA(x,y,z) \
if (new_count++ < wscount) \
{ \
next_new_state->offset = (x); \
next_new_state->count = (y); \
next_new_state->data = (z); \
next_new_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
/* And now, here is the code */
static int
internal_dfa_match(
dfa_match_block *mb,
PCRE2_SPTR this_start_code,
PCRE2_SPTR current_subject,
PCRE2_SIZE start_offset,
PCRE2_SIZE *offsets,
uint32_t offsetcount,
int *workspace,
int wscount,
int rlevel)
{
stateblock *active_states, *new_states, *temp_states;
stateblock *next_active_state, *next_new_state;
const uint8_t *ctypes, *lcc, *fcc;
PCRE2_SPTR ptr;
PCRE2_SPTR end_code;
PCRE2_SPTR first_op;
dfa_recursion_info new_recursive;
int active_count, new_count, match_count;
/* Some fields in the mb block are frequently referenced, so we load them into
independent variables in the hope that this will perform better. */
PCRE2_SPTR start_subject = mb->start_subject;
PCRE2_SPTR end_subject = mb->end_subject;
PCRE2_SPTR start_code = mb->start_code;
#ifdef SUPPORT_UNICODE
BOOL utf = (mb->poptions & PCRE2_UTF) != 0;
#else
BOOL utf = FALSE;
#endif
BOOL reset_could_continue = FALSE;
rlevel++;
offsetcount &= (-2);
wscount -= 2;
wscount = (wscount - (wscount % (INTS_PER_STATEBLOCK * 2))) /
(2 * INTS_PER_STATEBLOCK);
ctypes = mb->tables + ctypes_offset;
lcc = mb->tables + lcc_offset;
fcc = mb->tables + fcc_offset;
match_count = PCRE2_ERROR_NOMATCH; /* A negative number */
active_states = (stateblock *)(workspace + 2);
next_new_state = new_states = active_states + wscount;
new_count = 0;
first_op = this_start_code + 1 + LINK_SIZE +
((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA ||
*this_start_code == OP_CBRAPOS || *this_start_code == OP_SCBRAPOS)
? IMM2_SIZE:0);
/* The first thing in any (sub) pattern is a bracket of some sort. Push all
the alternative states onto the list, and find out where the end is. This
makes is possible to use this function recursively, when we want to stop at a
matching internal ket rather than at the end.
If the first opcode in the first alternative is OP_REVERSE, we are dealing with
a backward assertion. In that case, we have to find out the maximum amount to
move back, and set up each alternative appropriately. */
if (*first_op == OP_REVERSE)
{
size_t max_back = 0;
size_t gone_back;
end_code = this_start_code;
do
{
size_t back = GET(end_code, 2+LINK_SIZE);
if (back > max_back) max_back = back;
end_code += GET(end_code, 1);
}
while (*end_code == OP_ALT);
/* If we can't go back the amount required for the longest lookbehind
pattern, go back as far as we can; some alternatives may still be viable. */
#ifdef SUPPORT_UNICODE
/* In character mode we have to step back character by character */
if (utf)
{
for (gone_back = 0; gone_back < max_back; gone_back++)
{
if (current_subject <= start_subject) break;
current_subject--;
ACROSSCHAR(current_subject > start_subject, *current_subject, current_subject--);
}
}
else
#endif
/* In byte-mode we can do this quickly. */
{
size_t current_offset = (size_t)(current_subject - start_subject);
gone_back = (current_offset < max_back)? current_offset : max_back;
current_subject -= gone_back;
}
/* Save the earliest consulted character */
if (current_subject < mb->start_used_ptr)
mb->start_used_ptr = current_subject;
/* Now we can process the individual branches. */
end_code = this_start_code;
do
{
size_t back = GET(end_code, 2+LINK_SIZE);
if (back <= gone_back)
{
int bstate = (int)(end_code - start_code + 2 + 2*LINK_SIZE);
ADD_NEW_DATA(-bstate, 0, gone_back - back);
}
end_code += GET(end_code, 1);
}
while (*end_code == OP_ALT);
}
/* This is the code for a "normal" subpattern (not a backward assertion). The
start of a whole pattern is always one of these. If we are at the top level,
we may be asked to restart matching from the same point that we reached for a
previous partial match. We still have to scan through the top-level branches to
find the end state. */
else
{
end_code = this_start_code;
/* Restarting */
if (rlevel == 1 && (mb->moptions & PCRE2_DFA_RESTART) != 0)
{
do { end_code += GET(end_code, 1); } while (*end_code == OP_ALT);
new_count = workspace[1];
if (!workspace[0])
memcpy(new_states, active_states, new_count * sizeof(stateblock));
}
/* Not restarting */
else
{
int length = 1 + LINK_SIZE +
((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA ||
*this_start_code == OP_CBRAPOS || *this_start_code == OP_SCBRAPOS)
? IMM2_SIZE:0);
do
{
ADD_NEW((int)(end_code - start_code + length), 0);
end_code += GET(end_code, 1);
length = 1 + LINK_SIZE;
}
while (*end_code == OP_ALT);
}
}
workspace[0] = 0; /* Bit indicating which vector is current */
/* Loop for scanning the subject */
ptr = current_subject;
for (;;)
{
int i, j;
int clen, dlen;
uint32_t c, d;
int forced_fail = 0;
BOOL partial_newline = FALSE;
BOOL could_continue = reset_could_continue;
reset_could_continue = FALSE;
if (ptr > mb->last_used_ptr) mb->last_used_ptr = ptr;
/* Make the new state list into the active state list and empty the
new state list. */
temp_states = active_states;
active_states = new_states;
new_states = temp_states;
active_count = new_count;
new_count = 0;
workspace[0] ^= 1; /* Remember for the restarting feature */
workspace[1] = active_count;
/* Set the pointers for adding new states */
next_active_state = active_states + active_count;
next_new_state = new_states;
/* Load the current character from the subject outside the loop, as many
different states may want to look at it, and we assume that at least one
will. */
if (ptr < end_subject)
{
clen = 1; /* Number of data items in the character */
#ifdef SUPPORT_UNICODE
GETCHARLENTEST(c, ptr, clen);
#else
c = *ptr;
#endif /* SUPPORT_UNICODE */
}
else
{
clen = 0; /* This indicates the end of the subject */
c = NOTACHAR; /* This value should never actually be used */
}
/* Scan up the active states and act on each one. The result of an action
may be to add more states to the currently active list (e.g. on hitting a
parenthesis) or it may be to put states on the new list, for considering
when we move the character pointer on. */
for (i = 0; i < active_count; i++)
{
stateblock *current_state = active_states + i;
BOOL caseless = FALSE;
PCRE2_SPTR code;
int state_offset = current_state->offset;
int codevalue, rrc;
int count;
/* A negative offset is a special case meaning "hold off going to this
(negated) state until the number of characters in the data field have
been skipped". If the could_continue flag was passed over from a previous
state, arrange for it to passed on. */
if (state_offset < 0)
{
if (current_state->data > 0)
{
ADD_NEW_DATA(state_offset, current_state->count,
current_state->data - 1);
if (could_continue) reset_could_continue = TRUE;
continue;
}
else
{
current_state->offset = state_offset = -state_offset;
}
}
/* Check for a duplicate state with the same count, and skip if found.
See the note at the head of this module about the possibility of improving
performance here. */
for (j = 0; j < i; j++)
{
if (active_states[j].offset == state_offset &&
active_states[j].count == current_state->count)
goto NEXT_ACTIVE_STATE;
}
/* The state offset is the offset to the opcode */
code = start_code + state_offset;
codevalue = *code;
/* If this opcode inspects a character, but we are at the end of the
subject, remember the fact for use when testing for a partial match. */
if (clen == 0 && poptable[codevalue] != 0)
could_continue = TRUE;
/* If this opcode is followed by an inline character, load it. It is
tempting to test for the presence of a subject character here, but that
is wrong, because sometimes zero repetitions of the subject are
permitted.
We also use this mechanism for opcodes such as OP_TYPEPLUS that take an
argument that is not a data character - but is always one byte long because
the values are small. We have to take special action to deal with \P, \p,
\H, \h, \V, \v and \X in this case. To keep the other cases fast, convert
these ones to new opcodes. */
if (coptable[codevalue] > 0)
{
dlen = 1;
#ifdef SUPPORT_UNICODE
if (utf) { GETCHARLEN(d, (code + coptable[codevalue]), dlen); } else
#endif /* SUPPORT_UNICODE */
d = code[coptable[codevalue]];
if (codevalue >= OP_TYPESTAR)
{
switch(d)
{
case OP_ANYBYTE: return PCRE2_ERROR_DFA_UITEM;
case OP_NOTPROP:
case OP_PROP: codevalue += OP_PROP_EXTRA; break;
case OP_ANYNL: codevalue += OP_ANYNL_EXTRA; break;
case OP_EXTUNI: codevalue += OP_EXTUNI_EXTRA; break;
case OP_NOT_HSPACE:
case OP_HSPACE: codevalue += OP_HSPACE_EXTRA; break;
case OP_NOT_VSPACE:
case OP_VSPACE: codevalue += OP_VSPACE_EXTRA; break;
default: break;
}
}
}
else
{
dlen = 0; /* Not strictly necessary, but compilers moan */
d = NOTACHAR; /* if these variables are not set. */
}
/* Now process the individual opcodes */
switch (codevalue)
{
/* ========================================================================== */
/* These cases are never obeyed. This is a fudge that causes a compile-
time error if the vectors coptable or poptable, which are indexed by
opcode, are not the correct length. It seems to be the only way to do
such a check at compile time, as the sizeof() operator does not work
in the C preprocessor. */
case OP_TABLE_LENGTH:
case OP_TABLE_LENGTH +
((sizeof(coptable) == OP_TABLE_LENGTH) &&
(sizeof(poptable) == OP_TABLE_LENGTH)):
break;
/* ========================================================================== */
/* Reached a closing bracket. If not at the end of the pattern, carry
on with the next opcode. For repeating opcodes, also add the repeat
state. Note that KETRPOS will always be encountered at the end of the
subpattern, because the possessive subpattern repeats are always handled
using recursive calls. Thus, it never adds any new states.
At the end of the (sub)pattern, unless we have an empty string and
PCRE2_NOTEMPTY is set, or PCRE2_NOTEMPTY_ATSTART is set and we are at the
start of the subject, save the match data, shifting up all previous
matches so we always have the longest first. */
case OP_KET:
case OP_KETRMIN:
case OP_KETRMAX:
case OP_KETRPOS:
if (code != end_code)
{
ADD_ACTIVE(state_offset + 1 + LINK_SIZE, 0);
if (codevalue != OP_KET)
{
ADD_ACTIVE(state_offset - GET(code, 1), 0);
}
}
else
{
if (ptr > current_subject ||
((mb->moptions & PCRE2_NOTEMPTY) == 0 &&
((mb->moptions & PCRE2_NOTEMPTY_ATSTART) == 0 ||
current_subject > start_subject + mb->start_offset)))
{
if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;
else if (match_count > 0 && ++match_count * 2 > (int)offsetcount)
match_count = 0;
count = ((match_count == 0)? (int)offsetcount : match_count * 2) - 2;
if (count > 0) memmove(offsets + 2, offsets, count * sizeof(PCRE2_SIZE));
if (offsetcount >= 2)
{
offsets[0] = (int)(current_subject - start_subject);
offsets[1] = (int)(ptr - start_subject);
}
if ((mb->moptions & PCRE2_DFA_SHORTEST) != 0) return match_count;
}
}
break;
/* ========================================================================== */
/* These opcodes add to the current list of states without looking
at the current character. */
/*-----------------------------------------------------------------*/
case OP_ALT:
do { code += GET(code, 1); } while (*code == OP_ALT);
ADD_ACTIVE((int)(code - start_code), 0);
break;
/*-----------------------------------------------------------------*/
case OP_BRA:
case OP_SBRA:
do
{
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
code += GET(code, 1);
}
while (*code == OP_ALT);
break;
/*-----------------------------------------------------------------*/
case OP_CBRA:
case OP_SCBRA:
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE + IMM2_SIZE), 0);
code += GET(code, 1);
while (*code == OP_ALT)
{
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
code += GET(code, 1);
}
break;
/*-----------------------------------------------------------------*/
case OP_BRAZERO:
case OP_BRAMINZERO:
ADD_ACTIVE(state_offset + 1, 0);
code += 1 + GET(code, 2);
while (*code == OP_ALT) code += GET(code, 1);
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
break;
/*-----------------------------------------------------------------*/
case OP_SKIPZERO:
code += 1 + GET(code, 2);
while (*code == OP_ALT) code += GET(code, 1);
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
break;
/*-----------------------------------------------------------------*/
case OP_CIRC:
if (ptr == start_subject && (mb->moptions & PCRE2_NOTBOL) == 0)
{ ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_CIRCM:
if ((ptr == start_subject && (mb->moptions & PCRE2_NOTBOL) == 0) ||
((ptr != end_subject || (mb->poptions & PCRE2_ALT_CIRCUMFLEX) != 0 )
&& WAS_NEWLINE(ptr)))
{ ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_EOD:
if (ptr >= end_subject)
{
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
could_continue = TRUE;
else { ADD_ACTIVE(state_offset + 1, 0); }
}
break;
/*-----------------------------------------------------------------*/
case OP_SOD:
if (ptr == start_subject) { ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_SOM:
if (ptr == start_subject + start_offset) { ADD_ACTIVE(state_offset + 1, 0); }
break;
/* ========================================================================== */
/* These opcodes inspect the next subject character, and sometimes
the previous one as well, but do not have an argument. The variable
clen contains the length of the current character and is zero if we are
at the end of the subject. */
/*-----------------------------------------------------------------*/
case OP_ANY:
if (clen > 0 && !IS_NEWLINE(ptr))
{
if (ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else
{
ADD_NEW(state_offset + 1, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_ALLANY:
if (clen > 0)
{ ADD_NEW(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_EODN:
if (clen == 0 && (mb->moptions & PCRE2_PARTIAL_HARD) != 0)
could_continue = TRUE;
else if (clen == 0 || (IS_NEWLINE(ptr) && ptr == end_subject - mb->nllen))
{ ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_DOLL:
if ((mb->moptions & PCRE2_NOTEOL) == 0)
{
if (clen == 0 && (mb->moptions & PCRE2_PARTIAL_HARD) != 0)
could_continue = TRUE;
else if (clen == 0 ||
((mb->poptions & PCRE2_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr) &&
(ptr == end_subject - mb->nllen)
))
{ ADD_ACTIVE(state_offset + 1, 0); }
else if (ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
{
reset_could_continue = TRUE;
ADD_NEW_DATA(-(state_offset + 1), 0, 1);
}
else could_continue = partial_newline = TRUE;
}
}
break;
/*-----------------------------------------------------------------*/
case OP_DOLLM:
if ((mb->moptions & PCRE2_NOTEOL) == 0)
{
if (clen == 0 && (mb->moptions & PCRE2_PARTIAL_HARD) != 0)
could_continue = TRUE;
else if (clen == 0 ||
((mb->poptions & PCRE2_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr)))
{ ADD_ACTIVE(state_offset + 1, 0); }
else if (ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
{
reset_could_continue = TRUE;
ADD_NEW_DATA(-(state_offset + 1), 0, 1);
}
else could_continue = partial_newline = TRUE;
}
}
else if (IS_NEWLINE(ptr))
{ ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_DIGIT:
case OP_WHITESPACE:
case OP_WORDCHAR:
if (clen > 0 && c < 256 &&
((ctypes[c] & toptable1[codevalue]) ^ toptable2[codevalue]) != 0)
{ ADD_NEW(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_NOT_DIGIT:
case OP_NOT_WHITESPACE:
case OP_NOT_WORDCHAR:
if (clen > 0 && (c >= 256 ||
((ctypes[c] & toptable1[codevalue]) ^ toptable2[codevalue]) != 0))
{ ADD_NEW(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_WORD_BOUNDARY:
case OP_NOT_WORD_BOUNDARY:
{
int left_word, right_word;
if (ptr > start_subject)
{
PCRE2_SPTR temp = ptr - 1;
if (temp < mb->start_used_ptr) mb->start_used_ptr = temp;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (utf) { BACKCHAR(temp); }
#endif
GETCHARTEST(d, temp);
#ifdef SUPPORT_UNICODE
if ((mb->poptions & PCRE2_UCP) != 0)
{
if (d == '_') left_word = TRUE; else
{
int cat = UCD_CATEGORY(d);
left_word = (cat == ucp_L || cat == ucp_N);
}
}
else
#endif
left_word = d < 256 && (ctypes[d] & ctype_word) != 0;
}
else left_word = FALSE;
if (clen > 0)
{
if (ptr >= mb->last_used_ptr)
{
PCRE2_SPTR temp = ptr + 1;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (utf) { FORWARDCHARTEST(temp, mb->end_subject); }
#endif
mb->last_used_ptr = temp;
}
#ifdef SUPPORT_UNICODE
if ((mb->poptions & PCRE2_UCP) != 0)
{
if (c == '_') right_word = TRUE; else
{
int cat = UCD_CATEGORY(c);
right_word = (cat == ucp_L || cat == ucp_N);
}
}
else
#endif
right_word = c < 256 && (ctypes[c] & ctype_word) != 0;
}
else right_word = FALSE;
if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))
{ ADD_ACTIVE(state_offset + 1, 0); }
}
break;
/*-----------------------------------------------------------------*/
/* Check the next character by Unicode property. We will get here only
if the support is in the binary; otherwise a compile-time error occurs.
*/
#ifdef SUPPORT_UNICODE
case OP_PROP:
case OP_NOTPROP:
if (clen > 0)
{
BOOL OK;
const uint32_t *cp;
const ucd_record * prop = GET_UCD(c);
switch(code[1])
{
case PT_ANY:
OK = TRUE;
break;
case PT_LAMP:
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
prop->chartype == ucp_Lt;
break;
case PT_GC:
OK = PRIV(ucp_gentype)[prop->chartype] == code[2];
break;
case PT_PC:
OK = prop->chartype == code[2];
break;
case PT_SC:
OK = prop->script == code[2];
break;
/* These are specials for combination cases. */
case PT_ALNUM:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N;
break;
/* Perl space used to exclude VT, but from Perl 5.18 it is included,
which means that Perl space and POSIX space are now identical. PCRE
was changed at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z;
break;
}
break;
case PT_WORD:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
c == CHAR_UNDERSCORE;
break;
case PT_CLIST:
cp = PRIV(ucd_caseless_sets) + code[2];
for (;;)
{
if (c < *cp) { OK = FALSE; break; }
if (c == *cp++) { OK = TRUE; break; }
}
break;
case PT_UCNC:
OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
c >= 0xe000;
break;
/* Should never occur, but keep compilers from grumbling. */
default:
OK = codevalue != OP_PROP;
break;
}
if (OK == (codevalue == OP_PROP)) { ADD_NEW(state_offset + 3, 0); }
}
break;
#endif
/* ========================================================================== */
/* These opcodes likewise inspect the subject character, but have an
argument that is not a data character. It is one of these opcodes:
OP_ANY, OP_ALLANY, OP_DIGIT, OP_NOT_DIGIT, OP_WHITESPACE, OP_NOT_SPACE,
OP_WORDCHAR, OP_NOT_WORDCHAR. The value is loaded into d. */
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (count > 0 && codevalue == OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW(state_offset, count);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSQUERY:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (codevalue == OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset + 2, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPOSSTAR:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (codevalue == OP_TYPEPOSSTAR)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_TYPEEXACT:
count = current_state->count; /* Number already matched */
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + 1 + IMM2_SIZE + 1, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/*-----------------------------------------------------------------*/
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0);
count = current_state->count; /* Number already matched */
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (codevalue == OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + 2 + IMM2_SIZE, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/* ========================================================================== */
/* These are virtual opcodes that are used when something like
OP_TYPEPLUS has OP_PROP, OP_NOTPROP, OP_ANYNL, or OP_EXTUNI as its
argument. It keeps the code above fast for the other cases. The argument
is in the d variable. */
#ifdef SUPPORT_UNICODE
case OP_PROP_EXTRA + OP_TYPEPLUS:
case OP_PROP_EXTRA + OP_TYPEMINPLUS:
case OP_PROP_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 4, 0); }
if (clen > 0)
{
BOOL OK;
const uint32_t *cp;
const ucd_record * prop = GET_UCD(c);
switch(code[2])
{
case PT_ANY:
OK = TRUE;
break;
case PT_LAMP:
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
prop->chartype == ucp_Lt;
break;
case PT_GC:
OK = PRIV(ucp_gentype)[prop->chartype] == code[3];
break;
case PT_PC:
OK = prop->chartype == code[3];
break;
case PT_SC:
OK = prop->script == code[3];
break;
/* These are specials for combination cases. */
case PT_ALNUM:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N;
break;
/* Perl space used to exclude VT, but from Perl 5.18 it is included,
which means that Perl space and POSIX space are now identical. PCRE
was changed at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z;
break;
}
break;
case PT_WORD:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
c == CHAR_UNDERSCORE;
break;
case PT_CLIST:
cp = PRIV(ucd_caseless_sets) + code[3];
for (;;)
{
if (c < *cp) { OK = FALSE; break; }
if (c == *cp++) { OK = TRUE; break; }
}
break;
case PT_UCNC:
OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
c >= 0xe000;
break;
/* Should never occur, but keep compilers from grumbling. */
default:
OK = codevalue != OP_PROP;
break;
}
if (OK == (d == OP_PROP))
{
if (count > 0 && codevalue == OP_PROP_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW(state_offset, count);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_EXTUNI_EXTRA + OP_TYPEPLUS:
case OP_EXTUNI_EXTRA + OP_TYPEMINPLUS:
case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
int lgb, rgb;
PCRE2_SPTR nptr = ptr + clen;
int ncount = 0;
if (count > 0 && codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
lgb = UCD_GRAPHBREAK(c);
while (nptr < end_subject)
{
dlen = 1;
if (!utf) d = *nptr; else { GETCHARLEN(d, nptr, dlen); }
rgb = UCD_GRAPHBREAK(d);
if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
ncount++;
lgb = rgb;
nptr += dlen;
}
count++;
ADD_NEW_DATA(-state_offset, count, ncount);
}
break;
#endif
/*-----------------------------------------------------------------*/
case OP_ANYNL_EXTRA + OP_TYPEPLUS:
case OP_ANYNL_EXTRA + OP_TYPEMINPLUS:
case OP_ANYNL_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
int ncount = 0;
switch (c)
{
case CHAR_VT:
case CHAR_FF:
case CHAR_NEL:
#ifndef EBCDIC
case 0x2028:
case 0x2029:
#endif /* Not EBCDIC */
if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) break;
goto ANYNL01;
case CHAR_CR:
if (ptr + 1 < end_subject && UCHAR21TEST(ptr + 1) == CHAR_LF) ncount = 1;
/* Fall through */
ANYNL01:
case CHAR_LF:
if (count > 0 && codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW_DATA(-state_offset, count, ncount);
break;
default:
break;
}
}
break;
/*-----------------------------------------------------------------*/
case OP_VSPACE_EXTRA + OP_TYPEPLUS:
case OP_VSPACE_EXTRA + OP_TYPEMINPLUS:
case OP_VSPACE_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
BOOL OK;
switch (c)
{
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_VSPACE))
{
if (count > 0 && codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW_DATA(-state_offset, count, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_HSPACE_EXTRA + OP_TYPEPLUS:
case OP_HSPACE_EXTRA + OP_TYPEMINPLUS:
case OP_HSPACE_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
BOOL OK;
switch (c)
{
HSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_HSPACE))
{
if (count > 0 && codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW_DATA(-state_offset, count, 0);
}
}
break;
/*-----------------------------------------------------------------*/
#ifdef SUPPORT_UNICODE
case OP_PROP_EXTRA + OP_TYPEQUERY:
case OP_PROP_EXTRA + OP_TYPEMINQUERY:
case OP_PROP_EXTRA + OP_TYPEPOSQUERY:
count = 4;
goto QS1;
case OP_PROP_EXTRA + OP_TYPESTAR:
case OP_PROP_EXTRA + OP_TYPEMINSTAR:
case OP_PROP_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS1:
ADD_ACTIVE(state_offset + 4, 0);
if (clen > 0)
{
BOOL OK;
const uint32_t *cp;
const ucd_record * prop = GET_UCD(c);
switch(code[2])
{
case PT_ANY:
OK = TRUE;
break;
case PT_LAMP:
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
prop->chartype == ucp_Lt;
break;
case PT_GC:
OK = PRIV(ucp_gentype)[prop->chartype] == code[3];
break;
case PT_PC:
OK = prop->chartype == code[3];
break;
case PT_SC:
OK = prop->script == code[3];
break;
/* These are specials for combination cases. */
case PT_ALNUM:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N;
break;
/* Perl space used to exclude VT, but from Perl 5.18 it is included,
which means that Perl space and POSIX space are now identical. PCRE
was changed at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z;
break;
}
break;
case PT_WORD:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
c == CHAR_UNDERSCORE;
break;
case PT_CLIST:
cp = PRIV(ucd_caseless_sets) + code[3];
for (;;)
{
if (c < *cp) { OK = FALSE; break; }
if (c == *cp++) { OK = TRUE; break; }
}
break;
case PT_UCNC:
OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
c >= 0xe000;
break;
/* Should never occur, but keep compilers from grumbling. */
default:
OK = codevalue != OP_PROP;
break;
}
if (OK == (d == OP_PROP))
{
if (codevalue == OP_PROP_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_PROP_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset + count, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_EXTUNI_EXTRA + OP_TYPEQUERY:
case OP_EXTUNI_EXTRA + OP_TYPEMINQUERY:
case OP_EXTUNI_EXTRA + OP_TYPEPOSQUERY:
count = 2;
goto QS2;
case OP_EXTUNI_EXTRA + OP_TYPESTAR:
case OP_EXTUNI_EXTRA + OP_TYPEMINSTAR:
case OP_EXTUNI_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS2:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
int lgb, rgb;
PCRE2_SPTR nptr = ptr + clen;
int ncount = 0;
if (codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
lgb = UCD_GRAPHBREAK(c);
while (nptr < end_subject)
{
dlen = 1;
if (!utf) d = *nptr; else { GETCHARLEN(d, nptr, dlen); }
rgb = UCD_GRAPHBREAK(d);
if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
ncount++;
lgb = rgb;
nptr += dlen;
}
ADD_NEW_DATA(-(state_offset + count), 0, ncount);
}
break;
#endif
/*-----------------------------------------------------------------*/
case OP_ANYNL_EXTRA + OP_TYPEQUERY:
case OP_ANYNL_EXTRA + OP_TYPEMINQUERY:
case OP_ANYNL_EXTRA + OP_TYPEPOSQUERY:
count = 2;
goto QS3;
case OP_ANYNL_EXTRA + OP_TYPESTAR:
case OP_ANYNL_EXTRA + OP_TYPEMINSTAR:
case OP_ANYNL_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS3:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
int ncount = 0;
switch (c)
{
case CHAR_VT:
case CHAR_FF:
case CHAR_NEL:
#ifndef EBCDIC
case 0x2028:
case 0x2029:
#endif /* Not EBCDIC */
if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) break;
goto ANYNL02;
case CHAR_CR:
if (ptr + 1 < end_subject && UCHAR21TEST(ptr + 1) == CHAR_LF) ncount = 1;
/* Fall through */
ANYNL02:
case CHAR_LF:
if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW_DATA(-(state_offset + (int)count), 0, ncount);
break;
default:
break;
}
}
break;
/*-----------------------------------------------------------------*/
case OP_VSPACE_EXTRA + OP_TYPEQUERY:
case OP_VSPACE_EXTRA + OP_TYPEMINQUERY:
case OP_VSPACE_EXTRA + OP_TYPEPOSQUERY:
count = 2;
goto QS4;
case OP_VSPACE_EXTRA + OP_TYPESTAR:
case OP_VSPACE_EXTRA + OP_TYPEMINSTAR:
case OP_VSPACE_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS4:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
BOOL OK;
switch (c)
{
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_VSPACE))
{
if (codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW_DATA(-(state_offset + (int)count), 0, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_HSPACE_EXTRA + OP_TYPEQUERY:
case OP_HSPACE_EXTRA + OP_TYPEMINQUERY:
case OP_HSPACE_EXTRA + OP_TYPEPOSQUERY:
count = 2;
goto QS5;
case OP_HSPACE_EXTRA + OP_TYPESTAR:
case OP_HSPACE_EXTRA + OP_TYPEMINSTAR:
case OP_HSPACE_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS5:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
BOOL OK;
switch (c)
{
HSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_HSPACE))
{
if (codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW_DATA(-(state_offset + (int)count), 0, 0);
}
}
break;
/*-----------------------------------------------------------------*/
#ifdef SUPPORT_UNICODE
case OP_PROP_EXTRA + OP_TYPEEXACT:
case OP_PROP_EXTRA + OP_TYPEUPTO:
case OP_PROP_EXTRA + OP_TYPEMINUPTO:
case OP_PROP_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_PROP_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 1 + IMM2_SIZE + 3, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
BOOL OK;
const uint32_t *cp;
const ucd_record * prop = GET_UCD(c);
switch(code[1 + IMM2_SIZE + 1])
{
case PT_ANY:
OK = TRUE;
break;
case PT_LAMP:
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
prop->chartype == ucp_Lt;
break;
case PT_GC:
OK = PRIV(ucp_gentype)[prop->chartype] == code[1 + IMM2_SIZE + 2];
break;
case PT_PC:
OK = prop->chartype == code[1 + IMM2_SIZE + 2];
break;
case PT_SC:
OK = prop->script == code[1 + IMM2_SIZE + 2];
break;
/* These are specials for combination cases. */
case PT_ALNUM:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N;
break;
/* Perl space used to exclude VT, but from Perl 5.18 it is included,
which means that Perl space and POSIX space are now identical. PCRE
was changed at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z;
break;
}
break;
case PT_WORD:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
c == CHAR_UNDERSCORE;
break;
case PT_CLIST:
cp = PRIV(ucd_caseless_sets) + code[1 + IMM2_SIZE + 2];
for (;;)
{
if (c < *cp) { OK = FALSE; break; }
if (c == *cp++) { OK = TRUE; break; }
}
break;
case PT_UCNC:
OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
c >= 0xe000;
break;
/* Should never occur, but keep compilers from grumbling. */
default:
OK = codevalue != OP_PROP;
break;
}
if (OK == (d == OP_PROP))
{
if (codevalue == OP_PROP_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + 1 + IMM2_SIZE + 3, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/*-----------------------------------------------------------------*/
case OP_EXTUNI_EXTRA + OP_TYPEEXACT:
case OP_EXTUNI_EXTRA + OP_TYPEUPTO:
case OP_EXTUNI_EXTRA + OP_TYPEMINUPTO:
case OP_EXTUNI_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
int lgb, rgb;
PCRE2_SPTR nptr = ptr + clen;
int ncount = 0;
if (codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
lgb = UCD_GRAPHBREAK(c);
while (nptr < end_subject)
{
dlen = 1;
if (!utf) d = *nptr; else { GETCHARLEN(d, nptr, dlen); }
rgb = UCD_GRAPHBREAK(d);
if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
ncount++;
lgb = rgb;
nptr += dlen;
}
if (nptr >= end_subject && (mb->moptions & PCRE2_PARTIAL_HARD) != 0)
reset_could_continue = TRUE;
if (++count >= (int)GET2(code, 1))
{ ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, ncount); }
else
{ ADD_NEW_DATA(-state_offset, count, ncount); }
}
break;
#endif
/*-----------------------------------------------------------------*/
case OP_ANYNL_EXTRA + OP_TYPEEXACT:
case OP_ANYNL_EXTRA + OP_TYPEUPTO:
case OP_ANYNL_EXTRA + OP_TYPEMINUPTO:
case OP_ANYNL_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_ANYNL_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
int ncount = 0;
switch (c)
{
case CHAR_VT:
case CHAR_FF:
case CHAR_NEL:
#ifndef EBCDIC
case 0x2028:
case 0x2029:
#endif /* Not EBCDIC */
if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) break;
goto ANYNL03;
case CHAR_CR:
if (ptr + 1 < end_subject && UCHAR21TEST(ptr + 1) == CHAR_LF) ncount = 1;
/* Fall through */
ANYNL03:
case CHAR_LF:
if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, ncount); }
else
{ ADD_NEW_DATA(-state_offset, count, ncount); }
break;
default:
break;
}
}
break;
/*-----------------------------------------------------------------*/
case OP_VSPACE_EXTRA + OP_TYPEEXACT:
case OP_VSPACE_EXTRA + OP_TYPEUPTO:
case OP_VSPACE_EXTRA + OP_TYPEMINUPTO:
case OP_VSPACE_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_VSPACE_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
BOOL OK;
switch (c)
{
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
}
if (OK == (d == OP_VSPACE))
{
if (codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, 0); }
else
{ ADD_NEW_DATA(-state_offset, count, 0); }
}
}
break;
/*-----------------------------------------------------------------*/
case OP_HSPACE_EXTRA + OP_TYPEEXACT:
case OP_HSPACE_EXTRA + OP_TYPEUPTO:
case OP_HSPACE_EXTRA + OP_TYPEMINUPTO:
case OP_HSPACE_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_HSPACE_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
BOOL OK;
switch (c)
{
HSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_HSPACE))
{
if (codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, 0); }
else
{ ADD_NEW_DATA(-state_offset, count, 0); }
}
}
break;
/* ========================================================================== */
/* These opcodes are followed by a character that is usually compared
to the current subject character; it is loaded into d. We still get
here even if there is no subject character, because in some cases zero
repetitions are permitted. */
/*-----------------------------------------------------------------*/
case OP_CHAR:
if (clen > 0 && c == d) { ADD_NEW(state_offset + dlen + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_CHARI:
if (clen == 0) break;
#ifdef SUPPORT_UNICODE
if (utf)
{
if (c == d) { ADD_NEW(state_offset + dlen + 1, 0); } else
{
unsigned int othercase;
if (c < 128)
othercase = fcc[c];
else
othercase = UCD_OTHERCASE(c);
if (d == othercase) { ADD_NEW(state_offset + dlen + 1, 0); }
}
}
else
#endif /* SUPPORT_UNICODE */
/* Not UTF mode */
{
if (TABLE_GET(c, lcc, c) == TABLE_GET(d, lcc, d))
{ ADD_NEW(state_offset + 2, 0); }
}
break;
#ifdef SUPPORT_UNICODE
/*-----------------------------------------------------------------*/
/* This is a tricky one because it can match more than one character.
Find out how many characters to skip, and then set up a negative state
to wait for them to pass before continuing. */
case OP_EXTUNI:
if (clen > 0)
{
int lgb, rgb;
PCRE2_SPTR nptr = ptr + clen;
int ncount = 0;
lgb = UCD_GRAPHBREAK(c);
while (nptr < end_subject)
{
dlen = 1;
if (!utf) d = *nptr; else { GETCHARLEN(d, nptr, dlen); }
rgb = UCD_GRAPHBREAK(d);
if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
ncount++;
lgb = rgb;
nptr += dlen;
}
if (nptr >= end_subject && (mb->moptions & PCRE2_PARTIAL_HARD) != 0)
reset_could_continue = TRUE;
ADD_NEW_DATA(-(state_offset + 1), 0, ncount);
}
break;
#endif
/*-----------------------------------------------------------------*/
/* This is a tricky like EXTUNI because it too can match more than one
character (when CR is followed by LF). In this case, set up a negative
state to wait for one character to pass before continuing. */
case OP_ANYNL:
if (clen > 0) switch(c)
{
case CHAR_VT:
case CHAR_FF:
case CHAR_NEL:
#ifndef EBCDIC
case 0x2028:
case 0x2029:
#endif /* Not EBCDIC */
if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) break;
case CHAR_LF:
ADD_NEW(state_offset + 1, 0);
break;
case CHAR_CR:
if (ptr + 1 >= end_subject)
{
ADD_NEW(state_offset + 1, 0);
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
reset_could_continue = TRUE;
}
else if (UCHAR21TEST(ptr + 1) == CHAR_LF)
{
ADD_NEW_DATA(-(state_offset + 1), 0, 1);
}
else
{
ADD_NEW(state_offset + 1, 0);
}
break;
}
break;
/*-----------------------------------------------------------------*/
case OP_NOT_VSPACE:
if (clen > 0) switch(c)
{
VSPACE_CASES:
break;
default:
ADD_NEW(state_offset + 1, 0);
break;
}
break;
/*-----------------------------------------------------------------*/
case OP_VSPACE:
if (clen > 0) switch(c)
{
VSPACE_CASES:
ADD_NEW(state_offset + 1, 0);
break;
default:
break;
}
break;
/*-----------------------------------------------------------------*/
case OP_NOT_HSPACE:
if (clen > 0) switch(c)
{
HSPACE_CASES:
break;
default:
ADD_NEW(state_offset + 1, 0);
break;
}
break;
/*-----------------------------------------------------------------*/
case OP_HSPACE:
if (clen > 0) switch(c)
{
HSPACE_CASES:
ADD_NEW(state_offset + 1, 0);
break;
default:
break;
}
break;
/*-----------------------------------------------------------------*/
/* Match a negated single character casefully. */
case OP_NOT:
if (clen > 0 && c != d) { ADD_NEW(state_offset + dlen + 1, 0); }
break;
/*-----------------------------------------------------------------*/
/* Match a negated single character caselessly. */
case OP_NOTI:
if (clen > 0)
{
unsigned int otherd;
#ifdef SUPPORT_UNICODE
if (utf && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
if (c != d && c != otherd)
{ ADD_NEW(state_offset + dlen + 1, 0); }
}
break;
/*-----------------------------------------------------------------*/
case OP_PLUSI:
case OP_MINPLUSI:
case OP_POSPLUSI:
case OP_NOTPLUSI:
case OP_NOTMINPLUSI:
case OP_NOTPOSPLUSI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + dlen + 1, 0); }
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (count > 0 &&
(codevalue == OP_POSPLUS || codevalue == OP_NOTPOSPLUS))
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW(state_offset, count);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_QUERYI:
case OP_MINQUERYI:
case OP_POSQUERYI:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTPOSQUERYI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTPOSQUERY:
ADD_ACTIVE(state_offset + dlen + 1, 0);
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (codevalue == OP_POSQUERY || codevalue == OP_NOTPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset + dlen + 1, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_STARI:
case OP_MINSTARI:
case OP_POSSTARI:
case OP_NOTSTARI:
case OP_NOTMINSTARI:
case OP_NOTPOSSTARI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_STAR:
case OP_MINSTAR:
case OP_POSSTAR:
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPOSSTAR:
ADD_ACTIVE(state_offset + dlen + 1, 0);
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (codevalue == OP_POSSTAR || codevalue == OP_NOTPOSSTAR)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_EXACTI:
case OP_NOTEXACTI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_EXACT:
case OP_NOTEXACT:
count = current_state->count; /* Number already matched */
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + dlen + 1 + IMM2_SIZE, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/*-----------------------------------------------------------------*/
case OP_UPTOI:
case OP_MINUPTOI:
case OP_POSUPTOI:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTPOSUPTOI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTPOSUPTO:
ADD_ACTIVE(state_offset + dlen + 1 + IMM2_SIZE, 0);
count = current_state->count; /* Number already matched */
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (codevalue == OP_POSUPTO || codevalue == OP_NOTPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + dlen + 1 + IMM2_SIZE, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/* ========================================================================== */
/* These are the class-handling opcodes */
case OP_CLASS:
case OP_NCLASS:
case OP_XCLASS:
{
BOOL isinclass = FALSE;
int next_state_offset;
PCRE2_SPTR ecode;
/* For a simple class, there is always just a 32-byte table, and we
can set isinclass from it. */
if (codevalue != OP_XCLASS)
{
ecode = code + 1 + (32 / sizeof(PCRE2_UCHAR));
if (clen > 0)
{
isinclass = (c > 255)? (codevalue == OP_NCLASS) :
((((uint8_t *)(code + 1))[c/8] & (1 << (c&7))) != 0);
}
}
/* An extended class may have a table or a list of single characters,
ranges, or both, and it may be positive or negative. There's a
function that sorts all this out. */
else
{
ecode = code + GET(code, 1);
if (clen > 0) isinclass = PRIV(xclass)(c, code + 1 + LINK_SIZE, utf);
}
/* At this point, isinclass is set for all kinds of class, and ecode
points to the byte after the end of the class. If there is a
quantifier, this is where it will be. */
next_state_offset = (int)(ecode - start_code);
switch (*ecode)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRPOSSTAR:
ADD_ACTIVE(next_state_offset + 1, 0);
if (isinclass)
{
if (*ecode == OP_CRPOSSTAR)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset, 0);
}
break;
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(next_state_offset + 1, 0); }
if (isinclass)
{
if (count > 0 && *ecode == OP_CRPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW(state_offset, count);
}
break;
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSQUERY:
ADD_ACTIVE(next_state_offset + 1, 0);
if (isinclass)
{
if (*ecode == OP_CRPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(next_state_offset + 1, 0);
}
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
count = current_state->count; /* Already matched */
if (count >= (int)GET2(ecode, 1))
{ ADD_ACTIVE(next_state_offset + 1 + 2 * IMM2_SIZE, 0); }
if (isinclass)
{
int max = (int)GET2(ecode, 1 + IMM2_SIZE);
if (*ecode == OP_CRPOSRANGE)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= max && max != 0) /* Max 0 => no limit */
{ ADD_NEW(next_state_offset + 1 + 2 * IMM2_SIZE, 0); }
else
{ ADD_NEW(state_offset, count); }
}
break;
default:
if (isinclass) { ADD_NEW(next_state_offset, 0); }
break;
}
}
break;
/* ========================================================================== */
/* These are the opcodes for fancy brackets of various kinds. We have
to use recursion in order to handle them. The "always failing" assertion
(?!) is optimised to OP_FAIL when compiling, so we have to support that,
though the other "backtracking verbs" are not supported. */
case OP_FAIL:
forced_fail++; /* Count FAILs for multiple states */
break;
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
{
PCRE2_SPTR endasscode = code + GET(code, 1);
PCRE2_SIZE local_offsets[2];
int rc;
int local_workspace[1000];
while (*endasscode == OP_ALT) endasscode += GET(endasscode, 1);
rc = internal_dfa_match(
mb, /* static match data */
code, /* this subexpression's code */
ptr, /* where we currently are */
(int)(ptr - start_subject), /* start offset */
local_offsets, /* offset vector */
sizeof(local_offsets)/sizeof(PCRE2_SIZE), /* size of same */
local_workspace, /* workspace vector */
sizeof(local_workspace)/sizeof(int), /* size of same */
rlevel); /* function recursion level */
if (rc == PCRE2_ERROR_DFA_UITEM) return rc;
if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))
{ ADD_ACTIVE((int)(endasscode + LINK_SIZE + 1 - start_code), 0); }
}
break;
/*-----------------------------------------------------------------*/
case OP_COND:
case OP_SCOND:
{
PCRE2_SIZE local_offsets[1000];
int local_workspace[1000];
int codelink = GET(code, 1);
int condcode;
/* Because of the way auto-callout works during compile, a callout item
is inserted between OP_COND and an assertion condition. This does not
happen for the other conditions. */
if (code[LINK_SIZE + 1] == OP_CALLOUT
|| code[LINK_SIZE + 1] == OP_CALLOUT_STR)
{
unsigned int callout_length = (code[LINK_SIZE + 1] == OP_CALLOUT)
? PRIV(OP_lengths)[OP_CALLOUT] : GET(code, 2 + 3*LINK_SIZE);
rrc = 0;
if (mb->callout != NULL)
{
pcre2_callout_block cb;
cb.version = 1;
cb.capture_top = 1;
cb.capture_last = 0;
cb.offset_vector = offsets;
cb.mark = NULL; /* No (*MARK) support */
cb.subject = start_subject;
cb.subject_length = (PCRE2_SIZE)(end_subject - start_subject);
cb.start_match = (PCRE2_SIZE)(current_subject - start_subject);
cb.current_position = (PCRE2_SIZE)(ptr - start_subject);
cb.pattern_position = GET(code, LINK_SIZE + 2);
cb.next_item_length = GET(code, LINK_SIZE + 2 + LINK_SIZE);
if (code[LINK_SIZE + 1] == OP_CALLOUT)
{
cb.callout_number = code[2 + 3*LINK_SIZE];
cb.callout_string_offset = 0;
cb.callout_string = NULL;
cb.callout_string_length = 0;
}
else
{
cb.callout_number = 0;
cb.callout_string_offset = GET(code, 2 + 4*LINK_SIZE);
cb.callout_string = code + (2 + 5*LINK_SIZE) + 1;
cb.callout_string_length =
callout_length - (1 + 4*LINK_SIZE) - 2;
}
if ((rrc = (mb->callout)(&cb, mb->callout_data)) < 0)
return rrc; /* Abandon */
}
if (rrc > 0) break; /* Fail this thread */
code += callout_length; /* Skip callout data */
}
condcode = code[LINK_SIZE+1];
/* Back reference conditions and duplicate named recursion conditions
are not supported */
if (condcode == OP_CREF || condcode == OP_DNCREF ||
condcode == OP_DNRREF)
return PCRE2_ERROR_DFA_UCOND;
/* The DEFINE condition is always false, and the assertion (?!) is
converted to OP_FAIL. */