Improve single character iterators, add special path to dotall.
This commit is contained in:
Zoltán Herczeg
parent
dea540877b
commit
cc51779d88
|
|
@ -3371,6 +3371,35 @@ else
|
|||
JUMPHERE(jump);
|
||||
}
|
||||
|
||||
static void process_partial_match(compiler_common *common)
|
||||
{
|
||||
DEFINE_COMPILER;
|
||||
struct sljit_jump *jump;
|
||||
|
||||
/* Partial matching mode. */
|
||||
if (common->mode == PCRE2_JIT_PARTIAL_SOFT)
|
||||
{
|
||||
jump = CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0);
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, 0);
|
||||
JUMPHERE(jump);
|
||||
}
|
||||
else if (common->mode == PCRE2_JIT_PARTIAL_HARD)
|
||||
{
|
||||
if (common->partialmatchlabel != NULL)
|
||||
CMPTO(SLJIT_LESS, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0, common->partialmatchlabel);
|
||||
else
|
||||
add_jump(compiler, &common->partialmatch, CMP(SLJIT_LESS, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0));
|
||||
}
|
||||
}
|
||||
|
||||
static void detect_partial_match_to(compiler_common *common, struct sljit_label *label)
|
||||
{
|
||||
DEFINE_COMPILER;
|
||||
|
||||
CMPTO(SLJIT_LESS, STR_PTR, 0, STR_END, 0, label);
|
||||
process_partial_match(common);
|
||||
}
|
||||
|
||||
static void peek_char(compiler_common *common, sljit_u32 max, sljit_s32 dst, sljit_sw dstw, jump_list **backtracks)
|
||||
{
|
||||
/* Reads the character into TMP1, keeps STR_PTR.
|
||||
|
|
@ -11466,18 +11495,74 @@ switch(opcode)
|
|||
JUMPTO(SLJIT_JUMP, label);
|
||||
if (jump != NULL)
|
||||
JUMPHERE(jump);
|
||||
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
|
||||
break;
|
||||
}
|
||||
#ifdef SUPPORT_UNICODE
|
||||
else if (type == OP_ALLANY && !common->invalid_utf)
|
||||
#else
|
||||
else if (type == OP_ALLANY)
|
||||
#endif
|
||||
{
|
||||
if (opcode == OP_STAR)
|
||||
{
|
||||
if (private_data_ptr == 0)
|
||||
allocate_stack(common, 2);
|
||||
|
||||
OP1(SLJIT_MOV, base, offset0, STR_END, 0);
|
||||
OP1(SLJIT_MOV, base, offset1, STR_PTR, 0);
|
||||
|
||||
OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0);
|
||||
process_partial_match(common);
|
||||
|
||||
if (fast_str_ptr != 0)
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), fast_str_ptr, STR_END, 0);
|
||||
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
|
||||
break;
|
||||
}
|
||||
#ifdef SUPPORT_UNICODE
|
||||
else if (!common->utf)
|
||||
#else
|
||||
else
|
||||
#endif
|
||||
{
|
||||
if (private_data_ptr == 0)
|
||||
allocate_stack(common, 2);
|
||||
|
||||
OP1(SLJIT_MOV, base, offset1, STR_PTR, 0);
|
||||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(max));
|
||||
|
||||
if (common->mode == PCRE2_JIT_COMPLETE)
|
||||
{
|
||||
OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, STR_PTR, 0, STR_END, 0);
|
||||
CMOV(SLJIT_GREATER, STR_PTR, STR_END, 0);
|
||||
}
|
||||
else
|
||||
{
|
||||
jump = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, STR_END, 0);
|
||||
process_partial_match(common);
|
||||
JUMPHERE(jump);
|
||||
}
|
||||
|
||||
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
|
||||
|
||||
if (fast_str_ptr != 0)
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), fast_str_ptr, STR_PTR, 0);
|
||||
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
charpos_enabled = FALSE;
|
||||
charpos_char = 0;
|
||||
charpos_othercasebit = 0;
|
||||
|
||||
if ((type != OP_CHAR && type != OP_CHARI) && (*end == OP_CHAR || *end == OP_CHARI))
|
||||
{
|
||||
charpos_enabled = TRUE;
|
||||
#ifdef SUPPORT_UNICODE
|
||||
charpos_enabled = !common->utf || !HAS_EXTRALEN(end[1]);
|
||||
#else
|
||||
charpos_enabled = TRUE;
|
||||
#endif
|
||||
if (charpos_enabled && *end == OP_CHARI && char_has_othercase(common, end + 1))
|
||||
{
|
||||
|
|
@ -11587,17 +11672,19 @@ switch(opcode)
|
|||
if (opcode == OP_UPTO)
|
||||
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max);
|
||||
|
||||
detect_partial_match(common, &no_match);
|
||||
label = LABEL();
|
||||
compile_char1_matchingpath(common, type, cc, &no_match, TRUE);
|
||||
compile_char1_matchingpath(common, type, cc, &no_match, FALSE);
|
||||
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
|
||||
|
||||
if (opcode == OP_UPTO)
|
||||
{
|
||||
OP2(SLJIT_SUB | SLJIT_SET_Z, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
|
||||
JUMPTO(SLJIT_NOT_ZERO, label);
|
||||
add_jump(compiler, &no_match, JUMP(SLJIT_ZERO));
|
||||
}
|
||||
else
|
||||
JUMPTO(SLJIT_JUMP, label);
|
||||
|
||||
detect_partial_match_to(common, label);
|
||||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
|
||||
|
||||
set_jumps(no_match, LABEL());
|
||||
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
|
||||
|
|
@ -11614,17 +11701,17 @@ switch(opcode)
|
|||
if (opcode == OP_UPTO)
|
||||
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max);
|
||||
|
||||
label = LABEL();
|
||||
detect_partial_match(common, &no_match);
|
||||
label = LABEL();
|
||||
compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
|
||||
if (opcode == OP_UPTO)
|
||||
{
|
||||
OP2(SLJIT_SUB | SLJIT_SET_Z, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
|
||||
JUMPTO(SLJIT_NOT_ZERO, label);
|
||||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
|
||||
add_jump(compiler, &no_match, JUMP(SLJIT_ZERO));
|
||||
}
|
||||
else
|
||||
JUMPTO(SLJIT_JUMP, label);
|
||||
|
||||
detect_partial_match_to(common, label);
|
||||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
|
||||
|
||||
set_jumps(no_char1_match, LABEL());
|
||||
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
|
||||
|
|
@ -11633,7 +11720,7 @@ switch(opcode)
|
|||
if (fast_str_ptr != 0)
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), fast_str_ptr, STR_PTR, 0);
|
||||
}
|
||||
}
|
||||
|
||||
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
|
||||
break;
|
||||
|
||||
|
|
@ -11670,25 +11757,47 @@ switch(opcode)
|
|||
break;
|
||||
|
||||
case OP_POSSTAR:
|
||||
#if defined SUPPORT_UNICODE
|
||||
if (type == OP_ALLANY && !common->invalid_utf)
|
||||
#else
|
||||
if (type == OP_ALLANY)
|
||||
#endif
|
||||
{
|
||||
OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0);
|
||||
process_partial_match(common);
|
||||
if (fast_str_ptr != 0)
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), fast_str_ptr, STR_END, 0);
|
||||
break;
|
||||
}
|
||||
|
||||
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
|
||||
if (common->utf)
|
||||
{
|
||||
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0);
|
||||
detect_partial_match(common, &no_match);
|
||||
label = LABEL();
|
||||
compile_char1_matchingpath(common, type, cc, &no_match, TRUE);
|
||||
compile_char1_matchingpath(common, type, cc, &no_match, FALSE);
|
||||
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0);
|
||||
JUMPTO(SLJIT_JUMP, label);
|
||||
detect_partial_match_to(common, label);
|
||||
|
||||
set_jumps(no_match, LABEL());
|
||||
OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, tmp_offset);
|
||||
if (fast_str_ptr != 0)
|
||||
{
|
||||
if (tmp_base == TMP3)
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), fast_str_ptr, TMP3, 0);
|
||||
else
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), fast_str_ptr, STR_PTR, 0);
|
||||
}
|
||||
break;
|
||||
}
|
||||
#endif
|
||||
label = LABEL();
|
||||
detect_partial_match(common, &no_match);
|
||||
label = LABEL();
|
||||
compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
|
||||
JUMPTO(SLJIT_JUMP, label);
|
||||
detect_partial_match_to(common, label);
|
||||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
|
||||
|
||||
set_jumps(no_char1_match, LABEL());
|
||||
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
|
||||
set_jumps(no_match, LABEL());
|
||||
|
|
@ -11703,23 +11812,52 @@ switch(opcode)
|
|||
{
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), POSSESSIVE1, STR_PTR, 0);
|
||||
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max);
|
||||
|
||||
detect_partial_match(common, &no_match);
|
||||
label = LABEL();
|
||||
compile_char1_matchingpath(common, type, cc, &no_match, TRUE);
|
||||
compile_char1_matchingpath(common, type, cc, &no_match, FALSE);
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), POSSESSIVE1, STR_PTR, 0);
|
||||
OP2(SLJIT_SUB | SLJIT_SET_Z, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
|
||||
JUMPTO(SLJIT_NOT_ZERO, label);
|
||||
add_jump(compiler, &no_match, JUMP(SLJIT_ZERO));
|
||||
detect_partial_match_to(common, label);
|
||||
|
||||
set_jumps(no_match, LABEL());
|
||||
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), POSSESSIVE1);
|
||||
break;
|
||||
}
|
||||
#endif
|
||||
|
||||
if (type == OP_ALLANY)
|
||||
{
|
||||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(max));
|
||||
|
||||
if (common->mode == PCRE2_JIT_COMPLETE)
|
||||
{
|
||||
OP2(SLJIT_SUB | SLJIT_SET_GREATER, SLJIT_UNUSED, 0, STR_PTR, 0, STR_END, 0);
|
||||
CMOV(SLJIT_GREATER, STR_PTR, STR_END, 0);
|
||||
}
|
||||
else
|
||||
{
|
||||
jump = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, STR_END, 0);
|
||||
process_partial_match(common);
|
||||
JUMPHERE(jump);
|
||||
}
|
||||
|
||||
if (fast_str_ptr != 0)
|
||||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), fast_str_ptr, STR_PTR, 0);
|
||||
break;
|
||||
}
|
||||
|
||||
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max);
|
||||
label = LABEL();
|
||||
|
||||
detect_partial_match(common, &no_match);
|
||||
label = LABEL();
|
||||
compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
|
||||
OP2(SLJIT_SUB | SLJIT_SET_Z, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
|
||||
JUMPTO(SLJIT_NOT_ZERO, label);
|
||||
add_jump(compiler, &no_match, JUMP(SLJIT_ZERO));
|
||||
detect_partial_match_to(common, label);
|
||||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
|
||||
|
||||
set_jumps(no_char1_match, LABEL());
|
||||
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
|
||||
set_jumps(no_match, LABEL());
|
||||
|
|
|
|||
Loading…
Reference in New Issue