Parent Directory
|
Revision Log
|
Patch
revision 1247 by zherczeg, Mon Feb 11 21:37:46 2013 UTC | revision 1624 by zherczeg, Fri Feb 5 13:47:43 2016 UTC | |
---|---|---|
# | Line 6 | Line 6 |
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. |
7 | ||
8 | Written by Philip Hazel | Written by Philip Hazel |
9 | Copyright (c) 1997-2012 University of Cambridge | Copyright (c) 1997-2013 University of Cambridge |
10 | ||
11 | The machine code generator part (this module) was written by Zoltan Herczeg | The machine code generator part (this module) was written by Zoltan Herczeg |
12 | Copyright (c) 2010-2012 | Copyright (c) 2010-2013 |
13 | ||
14 | ----------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
15 | Redistribution and use in source and binary forms, with or without | Redistribution and use in source and binary forms, with or without |
# | Line 52 POSSIBILITY OF SUCH DAMAGE. | Line 52 POSSIBILITY OF SUCH DAMAGE. |
52 | we just include it. This way we don't need to touch the build | we just include it. This way we don't need to touch the build |
53 | system files. */ | system files. */ |
54 | ||
55 | #define SLJIT_MALLOC(size) (PUBL(malloc))(size) | #define SLJIT_MALLOC(size, allocator_data) (PUBL(malloc))(size) |
56 | #define SLJIT_FREE(ptr) (PUBL(free))(ptr) | #define SLJIT_FREE(ptr, allocator_data) (PUBL(free))(ptr) |
57 | #define SLJIT_CONFIG_AUTO 1 | #define SLJIT_CONFIG_AUTO 1 |
58 | #define SLJIT_CONFIG_STATIC 1 | #define SLJIT_CONFIG_STATIC 1 |
59 | #define SLJIT_VERBOSE 0 | #define SLJIT_VERBOSE 0 |
# | Line 71 system files. */ | Line 71 system files. */ |
71 | 2 - Enable capture_last_ptr (includes option 1). */ | 2 - Enable capture_last_ptr (includes option 1). */ |
72 | /* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */ | /* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */ |
73 | ||
74 | /* 1 - Always have a control head. */ | |
75 | /* #define DEBUG_FORCE_CONTROL_HEAD 1 */ | |
76 | ||
77 | /* Allocate memory for the regex stack on the real machine stack. | /* Allocate memory for the regex stack on the real machine stack. |
78 | Fast, but limited size. */ | Fast, but limited size. */ |
79 | #define MACHINE_STACK_SIZE 32768 | #define MACHINE_STACK_SIZE 32768 |
# | Line 165 typedef struct jit_arguments { | Line 168 typedef struct jit_arguments { |
168 | pcre_uchar *mark_ptr; | pcre_uchar *mark_ptr; |
169 | void *callout_data; | void *callout_data; |
170 | /* Everything else after. */ | /* Everything else after. */ |
171 | pcre_uint32 limit_match; | |
172 | int real_offset_count; | |
173 | int offset_count; | int offset_count; |
int call_limit; | ||
174 | pcre_uint8 notbol; | pcre_uint8 notbol; |
175 | pcre_uint8 noteol; | pcre_uint8 noteol; |
176 | pcre_uint8 notempty; | pcre_uint8 notempty; |
# | Line 175 typedef struct jit_arguments { | Line 179 typedef struct jit_arguments { |
179 | ||
180 | typedef struct executable_functions { | typedef struct executable_functions { |
181 | void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; | void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; |
182 | void *read_only_data_heads[JIT_NUMBER_OF_COMPILE_MODES]; | |
183 | sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; | |
184 | PUBL(jit_callback) callback; | PUBL(jit_callback) callback; |
185 | void *userdata; | void *userdata; |
186 | pcre_uint32 top_bracket; | pcre_uint32 top_bracket; |
187 | sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; | pcre_uint32 limit_match; |
188 | } executable_functions; | } executable_functions; |
189 | ||
190 | typedef struct jump_list { | typedef struct jump_list { |
# | Line 186 typedef struct jump_list { | Line 192 typedef struct jump_list { |
192 | struct jump_list *next; | struct jump_list *next; |
193 | } jump_list; | } jump_list; |
194 | ||
enum stub_types { stack_alloc }; | ||
195 | typedef struct stub_list { | typedef struct stub_list { |
enum stub_types type; | ||
int data; | ||
196 | struct sljit_jump *start; | struct sljit_jump *start; |
197 | struct sljit_label *quit; | struct sljit_label *quit; |
198 | struct stub_list *next; | struct stub_list *next; |
199 | } stub_list; | } stub_list; |
200 | ||
201 | typedef struct label_addr_list { | |
202 | struct sljit_label *label; | |
203 | sljit_uw *update_addr; | |
204 | struct label_addr_list *next; | |
205 | } label_addr_list; | |
206 | ||
207 | enum frame_types { | |
208 | no_frame = -1, | |
209 | no_stack = -2 | |
210 | }; | |
211 | ||
212 | enum control_types { | |
213 | type_mark = 0, | |
214 | type_then_trap = 1 | |
215 | }; | |
216 | ||
217 | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); |
218 | ||
219 | /* The following structure is the key data type for the recursive | /* The following structure is the key data type for the recursive |
220 | code generator. It is allocated by compile_matchingpath, and contains | code generator. It is allocated by compile_matchingpath, and contains |
221 | the aguments for compile_backtrackingpath. Must be the first member | the arguments for compile_backtrackingpath. Must be the first member |
222 | of its descendants. */ | of its descendants. */ |
223 | typedef struct backtrack_common { | typedef struct backtrack_common { |
224 | /* Concatenation stack. */ | /* Concatenation stack. */ |
# | Line 216 typedef struct backtrack_common { | Line 234 typedef struct backtrack_common { |
234 | typedef struct assert_backtrack { | typedef struct assert_backtrack { |
235 | backtrack_common common; | backtrack_common common; |
236 | jump_list *condfailed; | jump_list *condfailed; |
237 | /* Less than 0 (-1) if a frame is not needed. */ | /* Less than 0 if a frame is not needed. */ |
238 | int framesize; | int framesize; |
239 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
240 | int private_data_ptr; | int private_data_ptr; |
# | Line 237 typedef struct bracket_backtrack { | Line 255 typedef struct bracket_backtrack { |
255 | /* Both for OP_COND, OP_SCOND. */ | /* Both for OP_COND, OP_SCOND. */ |
256 | jump_list *condfailed; | jump_list *condfailed; |
257 | assert_backtrack *assert; | assert_backtrack *assert; |
258 | /* For OP_ONCE. -1 if not needed. */ | /* For OP_ONCE. Less than 0 if not needed. */ |
259 | int framesize; | int framesize; |
260 | } u; | } u; |
261 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
# | Line 272 typedef struct recurse_entry { | Line 290 typedef struct recurse_entry { |
290 | /* Collects the calls until the function is not created. */ | /* Collects the calls until the function is not created. */ |
291 | jump_list *calls; | jump_list *calls; |
292 | /* Points to the starting opcode. */ | /* Points to the starting opcode. */ |
293 | int start; | sljit_sw start; |
294 | } recurse_entry; | } recurse_entry; |
295 | ||
296 | typedef struct recurse_backtrack { | typedef struct recurse_backtrack { |
297 | backtrack_common common; | backtrack_common common; |
298 | BOOL inlined_pattern; | |
299 | } recurse_backtrack; | } recurse_backtrack; |
300 | ||
301 | #define MAX_RANGE_SIZE 6 | #define OP_THEN_TRAP OP_TABLE_LENGTH |
302 | ||
303 | typedef struct then_trap_backtrack { | |
304 | backtrack_common common; | |
305 | /* If then_trap is not NULL, this structure contains the real | |
306 | then_trap for the backtracking path. */ | |
307 | struct then_trap_backtrack *then_trap; | |
308 | /* Points to the starting opcode. */ | |
309 | sljit_sw start; | |
310 | /* Exit point for the then opcodes of this alternative. */ | |
311 | jump_list *quit; | |
312 | /* Frame size of the current alternative. */ | |
313 | int framesize; | |
314 | } then_trap_backtrack; | |
315 | ||
316 | #define MAX_RANGE_SIZE 4 | |
317 | ||
318 | typedef struct compiler_common { | typedef struct compiler_common { |
319 | /* The sljit ceneric compiler. */ | |
320 | struct sljit_compiler *compiler; | struct sljit_compiler *compiler; |
321 | /* First byte code. */ | |
322 | pcre_uchar *start; | pcre_uchar *start; |
323 | /* Maps private data offset to each opcode. */ | /* Maps private data offset to each opcode. */ |
324 | int *private_data_ptrs; | sljit_si *private_data_ptrs; |
325 | /* Chain list of read-only data ptrs. */ | |
326 | void *read_only_data_head; | |
327 | /* Tells whether the capturing bracket is optimized. */ | /* Tells whether the capturing bracket is optimized. */ |
328 | pcre_uint8 *optimized_cbracket; | pcre_uint8 *optimized_cbracket; |
329 | /* Tells whether the starting offset is a target of then. */ | |
330 | pcre_uint8 *then_offsets; | |
331 | /* Current position where a THEN must jump. */ | |
332 | then_trap_backtrack *then_trap; | |
333 | /* Starting offset of private data for capturing brackets. */ | /* Starting offset of private data for capturing brackets. */ |
334 | int cbraptr; | int cbra_ptr; |
335 | /* OVector starting point. Must be divisible by 2. */ | /* Output vector starting point. Must be divisible by 2. */ |
336 | int ovector_start; | int ovector_start; |
337 | /* Last known position of the requested byte. */ | /* Last known position of the requested byte. */ |
338 | int req_char_ptr; | int req_char_ptr; |
# | Line 305 typedef struct compiler_common { | Line 346 typedef struct compiler_common { |
346 | int first_line_end; | int first_line_end; |
347 | /* Points to the marked string. */ | /* Points to the marked string. */ |
348 | int mark_ptr; | int mark_ptr; |
349 | /* Recursive control verb management chain. */ | |
350 | int control_head_ptr; | |
351 | /* Points to the last matched capture block index. */ | /* Points to the last matched capture block index. */ |
352 | int capture_last_ptr; | int capture_last_ptr; |
353 | /* Points to the starting position of the current match. */ | |
354 | int start_ptr; | |
355 | ||
356 | /* Flipped and lower case tables. */ | /* Flipped and lower case tables. */ |
357 | const pcre_uint8 *fcc; | const pcre_uint8 *fcc; |
358 | sljit_sw lcc; | sljit_sw lcc; |
359 | /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */ | /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */ |
360 | int mode; | int mode; |
361 | /* TRUE, when minlength is greater than 0. */ | |
362 | BOOL might_be_empty; | |
363 | /* \K is found in the pattern. */ | |
364 | BOOL has_set_som; | |
365 | /* (*SKIP:arg) is found in the pattern. */ | |
366 | BOOL has_skip_arg; | |
367 | /* (*THEN) is found in the pattern. */ | |
368 | BOOL has_then; | |
369 | /* Needs to know the start position anytime. */ | |
370 | BOOL needs_start_ptr; | |
371 | /* Currently in recurse or negative assert. */ | |
372 | BOOL local_exit; | |
373 | /* Currently in a positive assert. */ | |
374 | BOOL positive_assert; | |
375 | /* Newline control. */ | /* Newline control. */ |
376 | int nltype; | int nltype; |
377 | pcre_uint32 nlmax; | |
378 | pcre_uint32 nlmin; | |
379 | int newline; | int newline; |
380 | int bsr_nltype; | int bsr_nltype; |
381 | pcre_uint32 bsr_nlmax; | |
382 | pcre_uint32 bsr_nlmin; | |
383 | /* Dollar endonly. */ | /* Dollar endonly. */ |
384 | int endonly; | int endonly; |
BOOL has_set_som; | ||
385 | /* Tables. */ | /* Tables. */ |
386 | sljit_sw ctypes; | sljit_sw ctypes; |
int digits[2 + MAX_RANGE_SIZE]; | ||
387 | /* Named capturing brackets. */ | /* Named capturing brackets. */ |
388 | sljit_uw name_table; | pcre_uchar *name_table; |
389 | sljit_sw name_count; | sljit_sw name_count; |
390 | sljit_sw name_entry_size; | sljit_sw name_entry_size; |
391 | ||
# | Line 333 typedef struct compiler_common { | Line 394 typedef struct compiler_common { |
394 | struct sljit_label *quit_label; | struct sljit_label *quit_label; |
395 | struct sljit_label *forced_quit_label; | struct sljit_label *forced_quit_label; |
396 | struct sljit_label *accept_label; | struct sljit_label *accept_label; |
397 | struct sljit_label *ff_newline_shortcut; | |
398 | stub_list *stubs; | stub_list *stubs; |
399 | label_addr_list *label_addrs; | |
400 | recurse_entry *entries; | recurse_entry *entries; |
401 | recurse_entry *currententry; | recurse_entry *currententry; |
402 | jump_list *partialmatch; | jump_list *partialmatch; |
403 | jump_list *quit; | jump_list *quit; |
404 | jump_list *positive_assert_quit; | |
405 | jump_list *forced_quit; | jump_list *forced_quit; |
406 | jump_list *accept; | jump_list *accept; |
407 | jump_list *calllimit; | jump_list *calllimit; |
# | Line 349 typedef struct compiler_common { | Line 413 typedef struct compiler_common { |
413 | jump_list *vspace; | jump_list *vspace; |
414 | jump_list *casefulcmp; | jump_list *casefulcmp; |
415 | jump_list *caselesscmp; | jump_list *caselesscmp; |
416 | jump_list *reset_match; | |
417 | BOOL jscript_compat; | BOOL jscript_compat; |
418 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
419 | BOOL utf; | BOOL utf; |
420 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
421 | BOOL use_ucp; | BOOL use_ucp; |
422 | #endif | #endif |
#ifndef COMPILE_PCRE32 | ||
jump_list *utfreadchar; | ||
#endif | ||
423 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
424 | jump_list *utfreadchar; | |
425 | jump_list *utfreadchar16; | |
426 | jump_list *utfreadtype8; | jump_list *utfreadtype8; |
427 | #endif | #endif |
428 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
# | Line 407 typedef struct compare_context { | Line 471 typedef struct compare_context { |
471 | /* Used for accessing the elements of the stack. */ | /* Used for accessing the elements of the stack. */ |
472 | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_sw)) | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_sw)) |
473 | ||
474 | #define TMP1 SLJIT_SCRATCH_REG1 | #define TMP1 SLJIT_R0 |
475 | #define TMP2 SLJIT_SCRATCH_REG3 | #define TMP2 SLJIT_R2 |
476 | #define TMP3 SLJIT_TEMPORARY_EREG2 | #define TMP3 SLJIT_R3 |
477 | #define STR_PTR SLJIT_SAVED_REG1 | #define STR_PTR SLJIT_S0 |
478 | #define STR_END SLJIT_SAVED_REG2 | #define STR_END SLJIT_S1 |
479 | #define STACK_TOP SLJIT_SCRATCH_REG2 | #define STACK_TOP SLJIT_R1 |
480 | #define STACK_LIMIT SLJIT_SAVED_REG3 | #define STACK_LIMIT SLJIT_S2 |
481 | #define ARGUMENTS SLJIT_SAVED_EREG1 | #define COUNT_MATCH SLJIT_S3 |
482 | #define CALL_COUNT SLJIT_SAVED_EREG2 | #define ARGUMENTS SLJIT_S4 |
483 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 | #define RETURN_ADDR SLJIT_R4 |
484 | ||
485 | /* Local space layout. */ | /* Local space layout. */ |
486 | /* These two locals can be used by the current opcode. */ | /* These two locals can be used by the current opcode. */ |
# | Line 426 typedef struct compare_context { | Line 490 typedef struct compare_context { |
490 | #define POSSESSIVE0 (2 * sizeof(sljit_sw)) | #define POSSESSIVE0 (2 * sizeof(sljit_sw)) |
491 | #define POSSESSIVE1 (3 * sizeof(sljit_sw)) | #define POSSESSIVE1 (3 * sizeof(sljit_sw)) |
492 | /* Max limit of recursions. */ | /* Max limit of recursions. */ |
493 | #define CALL_LIMIT (4 * sizeof(sljit_sw)) | #define LIMIT_MATCH (4 * sizeof(sljit_sw)) |
494 | /* The output vector is stored on the stack, and contains pointers | /* The output vector is stored on the stack, and contains pointers |
495 | to characters. The vector data is divided into two groups: the first | to characters. The vector data is divided into two groups: the first |
496 | group contains the start / end character pointers, and the second is | group contains the start / end character pointers, and the second is |
497 | the start pointers when the end of the capturing group has not yet reached. */ | the start pointers when the end of the capturing group has not yet reached. */ |
498 | #define OVECTOR_START (common->ovector_start) | #define OVECTOR_START (common->ovector_start) |
499 | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_sw)) | #define OVECTOR(i) (OVECTOR_START + (i) * (sljit_sw)sizeof(sljit_sw)) |
500 | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_sw)) | #define OVECTOR_PRIV(i) (common->cbra_ptr + (i) * (sljit_sw)sizeof(sljit_sw)) |
501 | #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start]) | #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start]) |
502 | ||
503 | #if defined COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
# | Line 475 the start pointers when the end of the c | Line 539 the start pointers when the end of the c |
539 | #define GET_LOCAL_BASE(dst, dstw, offset) \ | #define GET_LOCAL_BASE(dst, dstw, offset) \ |
540 | sljit_get_local_base(compiler, (dst), (dstw), (offset)) | sljit_get_local_base(compiler, (dst), (dstw), (offset)) |
541 | ||
542 | static pcre_uchar* bracketend(pcre_uchar* cc) | #define READ_CHAR_MAX 0x7fffffff |
543 | ||
544 | static pcre_uchar *bracketend(pcre_uchar *cc) | |
545 | { | { |
546 | SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)); | SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)); |
547 | do cc += GET(cc, 1); while (*cc == OP_ALT); | do cc += GET(cc, 1); while (*cc == OP_ALT); |
# | Line 484 cc += 1 + LINK_SIZE; | Line 550 cc += 1 + LINK_SIZE; |
550 | return cc; | return cc; |
551 | } | } |
552 | ||
553 | static int no_alternatives(pcre_uchar *cc) | |
554 | { | |
555 | int count = 0; | |
556 | SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)); | |
557 | do | |
558 | { | |
559 | cc += GET(cc, 1); | |
560 | count++; | |
561 | } | |
562 | while (*cc == OP_ALT); | |
563 | SLJIT_ASSERT(*cc >= OP_KET && *cc <= OP_KETRPOS); | |
564 | return count; | |
565 | } | |
566 | ||
567 | static int ones_in_half_byte[16] = { | |
568 | /* 0 */ 0, 1, 1, 2, /* 4 */ 1, 2, 2, 3, | |
569 | /* 8 */ 1, 2, 2, 3, /* 12 */ 2, 3, 3, 4 | |
570 | }; | |
571 | ||
572 | /* Functions whose might need modification for all new supported opcodes: | /* Functions whose might need modification for all new supported opcodes: |
573 | next_opcode | next_opcode |
574 | get_private_data_length | check_opcode_types |
575 | set_private_data_ptrs | set_private_data_ptrs |
576 | get_framesize | get_framesize |
577 | init_frame | init_frame |
578 | get_private_data_length_for_copy | get_private_data_copy_length |
579 | copy_private_data | copy_private_data |
580 | compile_matchingpath | compile_matchingpath |
581 | compile_backtrackingpath | compile_backtrackingpath |
# | Line 514 switch(*cc) | Line 599 switch(*cc) |
599 | case OP_WORDCHAR: | case OP_WORDCHAR: |
600 | case OP_ANY: | case OP_ANY: |
601 | case OP_ALLANY: | case OP_ALLANY: |
602 | case OP_NOTPROP: | |
603 | case OP_PROP: | |
604 | case OP_ANYNL: | case OP_ANYNL: |
605 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
606 | case OP_HSPACE: | case OP_HSPACE: |
# | Line 526 switch(*cc) | Line 613 switch(*cc) |
613 | case OP_CIRCM: | case OP_CIRCM: |
614 | case OP_DOLL: | case OP_DOLL: |
615 | case OP_DOLLM: | case OP_DOLLM: |
case OP_TYPESTAR: | ||
case OP_TYPEMINSTAR: | ||
case OP_TYPEPLUS: | ||
case OP_TYPEMINPLUS: | ||
case OP_TYPEQUERY: | ||
case OP_TYPEMINQUERY: | ||
case OP_TYPEPOSSTAR: | ||
case OP_TYPEPOSPLUS: | ||
case OP_TYPEPOSQUERY: | ||
616 | case OP_CRSTAR: | case OP_CRSTAR: |
617 | case OP_CRMINSTAR: | case OP_CRMINSTAR: |
618 | case OP_CRPLUS: | case OP_CRPLUS: |
619 | case OP_CRMINPLUS: | case OP_CRMINPLUS: |
620 | case OP_CRQUERY: | case OP_CRQUERY: |
621 | case OP_CRMINQUERY: | case OP_CRMINQUERY: |
622 | case OP_CRRANGE: | |
623 | case OP_CRMINRANGE: | |
624 | case OP_CRPOSSTAR: | |
625 | case OP_CRPOSPLUS: | |
626 | case OP_CRPOSQUERY: | |
627 | case OP_CRPOSRANGE: | |
628 | case OP_CLASS: | |
629 | case OP_NCLASS: | |
630 | case OP_REF: | |
631 | case OP_REFI: | |
632 | case OP_DNREF: | |
633 | case OP_DNREFI: | |
634 | case OP_RECURSE: | |
635 | case OP_CALLOUT: | |
636 | case OP_ALT: | |
637 | case OP_KET: | |
638 | case OP_KETRMAX: | |
639 | case OP_KETRMIN: | |
640 | case OP_KETRPOS: | |
641 | case OP_REVERSE: | |
642 | case OP_ASSERT: | |
643 | case OP_ASSERT_NOT: | |
644 | case OP_ASSERTBACK: | |
645 | case OP_ASSERTBACK_NOT: | |
646 | case OP_ONCE: | |
647 | case OP_ONCE_NC: | |
648 | case OP_BRA: | |
649 | case OP_BRAPOS: | |
650 | case OP_CBRA: | |
651 | case OP_CBRAPOS: | |
652 | case OP_COND: | |
653 | case OP_SBRA: | |
654 | case OP_SBRAPOS: | |
655 | case OP_SCBRA: | |
656 | case OP_SCBRAPOS: | |
657 | case OP_SCOND: | |
658 | case OP_CREF: | |
659 | case OP_DNCREF: | |
660 | case OP_RREF: | |
661 | case OP_DNRREF: | |
662 | case OP_DEF: | case OP_DEF: |
663 | case OP_BRAZERO: | case OP_BRAZERO: |
664 | case OP_BRAMINZERO: | case OP_BRAMINZERO: |
665 | case OP_BRAPOSZERO: | case OP_BRAPOSZERO: |
666 | case OP_PRUNE: | |
667 | case OP_SKIP: | |
668 | case OP_THEN: | |
669 | case OP_COMMIT: | case OP_COMMIT: |
670 | case OP_FAIL: | case OP_FAIL: |
671 | case OP_ACCEPT: | case OP_ACCEPT: |
672 | case OP_ASSERT_ACCEPT: | case OP_ASSERT_ACCEPT: |
673 | case OP_CLOSE: | |
674 | case OP_SKIPZERO: | case OP_SKIPZERO: |
675 | return cc + 1; | return cc + PRIV(OP_lengths)[*cc]; |
case OP_ANYBYTE: | ||
#ifdef SUPPORT_UTF | ||
if (common->utf) return NULL; | ||
#endif | ||
return cc + 1; | ||
676 | ||
677 | case OP_CHAR: | case OP_CHAR: |
678 | case OP_CHARI: | case OP_CHARI: |
# | Line 568 switch(*cc) | Line 684 switch(*cc) |
684 | case OP_MINPLUS: | case OP_MINPLUS: |
685 | case OP_QUERY: | case OP_QUERY: |
686 | case OP_MINQUERY: | case OP_MINQUERY: |
687 | case OP_UPTO: | |
688 | case OP_MINUPTO: | |
689 | case OP_EXACT: | |
690 | case OP_POSSTAR: | case OP_POSSTAR: |
691 | case OP_POSPLUS: | case OP_POSPLUS: |
692 | case OP_POSQUERY: | case OP_POSQUERY: |
693 | case OP_POSUPTO: | |
694 | case OP_STARI: | case OP_STARI: |
695 | case OP_MINSTARI: | case OP_MINSTARI: |
696 | case OP_PLUSI: | case OP_PLUSI: |
697 | case OP_MINPLUSI: | case OP_MINPLUSI: |
698 | case OP_QUERYI: | case OP_QUERYI: |
699 | case OP_MINQUERYI: | case OP_MINQUERYI: |
700 | case OP_UPTOI: | |
701 | case OP_MINUPTOI: | |
702 | case OP_EXACTI: | |
703 | case OP_POSSTARI: | case OP_POSSTARI: |
704 | case OP_POSPLUSI: | case OP_POSPLUSI: |
705 | case OP_POSQUERYI: | case OP_POSQUERYI: |
706 | case OP_POSUPTOI: | |
707 | case OP_NOTSTAR: | case OP_NOTSTAR: |
708 | case OP_NOTMINSTAR: | case OP_NOTMINSTAR: |
709 | case OP_NOTPLUS: | case OP_NOTPLUS: |
710 | case OP_NOTMINPLUS: | case OP_NOTMINPLUS: |
711 | case OP_NOTQUERY: | case OP_NOTQUERY: |
712 | case OP_NOTMINQUERY: | case OP_NOTMINQUERY: |
713 | case OP_NOTUPTO: | |
714 | case OP_NOTMINUPTO: | |
715 | case OP_NOTEXACT: | |
716 | case OP_NOTPOSSTAR: | case OP_NOTPOSSTAR: |
717 | case OP_NOTPOSPLUS: | case OP_NOTPOSPLUS: |
718 | case OP_NOTPOSQUERY: | case OP_NOTPOSQUERY: |
719 | case OP_NOTPOSUPTO: | |
720 | case OP_NOTSTARI: | case OP_NOTSTARI: |
721 | case OP_NOTMINSTARI: | case OP_NOTMINSTARI: |
722 | case OP_NOTPLUSI: | case OP_NOTPLUSI: |
723 | case OP_NOTMINPLUSI: | case OP_NOTMINPLUSI: |
724 | case OP_NOTQUERYI: | case OP_NOTQUERYI: |
725 | case OP_NOTMINQUERYI: | case OP_NOTMINQUERYI: |
case OP_NOTPOSSTARI: | ||
case OP_NOTPOSPLUSI: | ||
case OP_NOTPOSQUERYI: | ||
cc += 2; | ||
#ifdef SUPPORT_UTF | ||
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | ||
#endif | ||
return cc; | ||
case OP_UPTO: | ||
case OP_MINUPTO: | ||
case OP_EXACT: | ||
case OP_POSUPTO: | ||
case OP_UPTOI: | ||
case OP_MINUPTOI: | ||
case OP_EXACTI: | ||
case OP_POSUPTOI: | ||
case OP_NOTUPTO: | ||
case OP_NOTMINUPTO: | ||
case OP_NOTEXACT: | ||
case OP_NOTPOSUPTO: | ||
726 | case OP_NOTUPTOI: | case OP_NOTUPTOI: |
727 | case OP_NOTMINUPTOI: | case OP_NOTMINUPTOI: |
728 | case OP_NOTEXACTI: | case OP_NOTEXACTI: |
729 | case OP_NOTPOSSTARI: | |
730 | case OP_NOTPOSPLUSI: | |
731 | case OP_NOTPOSQUERYI: | |
732 | case OP_NOTPOSUPTOI: | case OP_NOTPOSUPTOI: |
733 | cc += 2 + IMM2_SIZE; | cc += PRIV(OP_lengths)[*cc]; |
734 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
735 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
736 | #endif | #endif |
737 | return cc; | return cc; |
738 | ||
739 | case OP_NOTPROP: | /* Special cases. */ |
740 | case OP_PROP: | case OP_TYPESTAR: |
741 | return cc + 1 + 2; | case OP_TYPEMINSTAR: |
742 | case OP_TYPEPLUS: | |
743 | case OP_TYPEMINPLUS: | |
744 | case OP_TYPEQUERY: | |
745 | case OP_TYPEMINQUERY: | |
746 | case OP_TYPEUPTO: | case OP_TYPEUPTO: |
747 | case OP_TYPEMINUPTO: | case OP_TYPEMINUPTO: |
748 | case OP_TYPEEXACT: | case OP_TYPEEXACT: |
749 | case OP_TYPEPOSSTAR: | |
750 | case OP_TYPEPOSPLUS: | |
751 | case OP_TYPEPOSQUERY: | |
752 | case OP_TYPEPOSUPTO: | case OP_TYPEPOSUPTO: |
753 | case OP_REF: | return cc + PRIV(OP_lengths)[*cc] - 1; |
case OP_REFI: | ||
case OP_CREF: | ||
case OP_NCREF: | ||
case OP_RREF: | ||
case OP_NRREF: | ||
case OP_CLOSE: | ||
cc += 1 + IMM2_SIZE; | ||
return cc; | ||
754 | ||
755 | case OP_CRRANGE: | case OP_ANYBYTE: |
756 | case OP_CRMINRANGE: | #ifdef SUPPORT_UTF |
757 | return cc + 1 + 2 * IMM2_SIZE; | if (common->utf) return NULL; |
758 | #endif | |
759 | case OP_CLASS: | return cc + 1; |
case OP_NCLASS: | ||
return cc + 1 + 32 / sizeof(pcre_uchar); | ||
760 | ||
761 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
762 | case OP_XCLASS: | case OP_XCLASS: |
763 | return cc + GET(cc, 1); | return cc + GET(cc, 1); |
764 | #endif | #endif |
765 | ||
case OP_RECURSE: | ||
case OP_ASSERT: | ||
case OP_ASSERT_NOT: | ||
case OP_ASSERTBACK: | ||
case OP_ASSERTBACK_NOT: | ||
case OP_REVERSE: | ||
case OP_ONCE: | ||
case OP_ONCE_NC: | ||
case OP_BRA: | ||
case OP_BRAPOS: | ||
case OP_COND: | ||
case OP_SBRA: | ||
case OP_SBRAPOS: | ||
case OP_SCOND: | ||
case OP_ALT: | ||
case OP_KET: | ||
case OP_KETRMAX: | ||
case OP_KETRMIN: | ||
case OP_KETRPOS: | ||
return cc + 1 + LINK_SIZE; | ||
case OP_CBRA: | ||
case OP_CBRAPOS: | ||
case OP_SCBRA: | ||
case OP_SCBRAPOS: | ||
return cc + 1 + LINK_SIZE + IMM2_SIZE; | ||
766 | case OP_MARK: | case OP_MARK: |
767 | case OP_PRUNE_ARG: | |
768 | case OP_SKIP_ARG: | |
769 | case OP_THEN_ARG: | |
770 | return cc + 1 + 2 + cc[1]; | return cc + 1 + 2 + cc[1]; |
771 | ||
case OP_CALLOUT: | ||
return cc + 2 + 2 * LINK_SIZE; | ||
772 | default: | default: |
773 | /* All opcodes are supported now! */ | |
774 | SLJIT_ASSERT_STOP(); | |
775 | return NULL; | return NULL; |
776 | } | } |
777 | } | } |
778 | ||
779 | #define CASE_ITERATOR_PRIVATE_DATA_1 \ | static BOOL check_opcode_types(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) |
case OP_MINSTAR: \ | ||
case OP_MINPLUS: \ | ||
case OP_QUERY: \ | ||
case OP_MINQUERY: \ | ||
case OP_MINSTARI: \ | ||
case OP_MINPLUSI: \ | ||
case OP_QUERYI: \ | ||
case OP_MINQUERYI: \ | ||
case OP_NOTMINSTAR: \ | ||
case OP_NOTMINPLUS: \ | ||
case OP_NOTQUERY: \ | ||
case OP_NOTMINQUERY: \ | ||
case OP_NOTMINSTARI: \ | ||
case OP_NOTMINPLUSI: \ | ||
case OP_NOTQUERYI: \ | ||
case OP_NOTMINQUERYI: | ||
#define CASE_ITERATOR_PRIVATE_DATA_2A \ | ||
case OP_STAR: \ | ||
case OP_PLUS: \ | ||
case OP_STARI: \ | ||
case OP_PLUSI: \ | ||
case OP_NOTSTAR: \ | ||
case OP_NOTPLUS: \ | ||
case OP_NOTSTARI: \ | ||
case OP_NOTPLUSI: | ||
#define CASE_ITERATOR_PRIVATE_DATA_2B \ | ||
case OP_UPTO: \ | ||
case OP_MINUPTO: \ | ||
case OP_UPTOI: \ | ||
case OP_MINUPTOI: \ | ||
case OP_NOTUPTO: \ | ||
case OP_NOTMINUPTO: \ | ||
case OP_NOTUPTOI: \ | ||
case OP_NOTMINUPTOI: | ||
#define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \ | ||
case OP_TYPEMINSTAR: \ | ||
case OP_TYPEMINPLUS: \ | ||
case OP_TYPEQUERY: \ | ||
case OP_TYPEMINQUERY: | ||
#define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \ | ||
case OP_TYPESTAR: \ | ||
case OP_TYPEPLUS: | ||
#define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \ | ||
case OP_TYPEUPTO: \ | ||
case OP_TYPEMINUPTO: | ||
static int get_class_iterator_size(pcre_uchar *cc) | ||
{ | ||
switch(*cc) | ||
{ | ||
case OP_CRSTAR: | ||
case OP_CRPLUS: | ||
return 2; | ||
case OP_CRMINSTAR: | ||
case OP_CRMINPLUS: | ||
case OP_CRQUERY: | ||
case OP_CRMINQUERY: | ||
return 1; | ||
case OP_CRRANGE: | ||
case OP_CRMINRANGE: | ||
if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE)) | ||
return 0; | ||
return 2; | ||
default: | ||
return 0; | ||
} | ||
} | ||
static int get_private_data_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | ||
780 | { | { |
781 | int private_data_length = 0; | int count; |
782 | pcre_uchar *alternative; | pcre_uchar *slot; |
pcre_uchar *name; | ||
pcre_uchar *end = NULL; | ||
int space, size, i; | ||
pcre_uint32 bracketlen; | ||
783 | ||
784 | /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ | /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ |
785 | while (cc < ccend) | while (cc < ccend) |
786 | { | { |
space = 0; | ||
size = 0; | ||
bracketlen = 0; | ||
787 | switch(*cc) | switch(*cc) |
788 | { | { |
789 | case OP_SET_SOM: | case OP_SET_SOM: |
790 | common->has_set_som = TRUE; | common->has_set_som = TRUE; |
791 | common->might_be_empty = TRUE; | |
792 | cc += 1; | cc += 1; |
793 | break; | break; |
794 | ||
# | Line 800 while (cc < ccend) | Line 798 while (cc < ccend) |
798 | cc += 1 + IMM2_SIZE; | cc += 1 + IMM2_SIZE; |
799 | break; | break; |
800 | ||
case OP_ASSERT: | ||
case OP_ASSERT_NOT: | ||
case OP_ASSERTBACK: | ||
case OP_ASSERTBACK_NOT: | ||
case OP_ONCE: | ||
case OP_ONCE_NC: | ||
case OP_BRAPOS: | ||
case OP_SBRA: | ||
case OP_SBRAPOS: | ||
private_data_length += sizeof(sljit_sw); | ||
bracketlen = 1 + LINK_SIZE; | ||
break; | ||
801 | case OP_CBRAPOS: | case OP_CBRAPOS: |
802 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
private_data_length += sizeof(sljit_sw); | ||
803 | common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0; | common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0; |
804 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
805 | break; | break; |
806 | ||
807 | case OP_COND: | case OP_COND: |
# | Line 825 while (cc < ccend) | Line 809 while (cc < ccend) |
809 | /* Only AUTO_CALLOUT can insert this opcode. We do | /* Only AUTO_CALLOUT can insert this opcode. We do |
810 | not intend to support this case. */ | not intend to support this case. */ |
811 | if (cc[1 + LINK_SIZE] == OP_CALLOUT) | if (cc[1 + LINK_SIZE] == OP_CALLOUT) |
812 | return -1; | return FALSE; |
813 | cc += 1 + LINK_SIZE; | |
if (*cc == OP_COND) | ||
{ | ||
/* Might be a hidden SCOND. */ | ||
alternative = cc + GET(cc, 1); | ||
if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | ||
private_data_length += sizeof(sljit_sw); | ||
} | ||
else | ||
private_data_length += sizeof(sljit_sw); | ||
bracketlen = 1 + LINK_SIZE; | ||
814 | break; | break; |
815 | ||
816 | case OP_CREF: | case OP_CREF: |
817 | i = GET2(cc, 1); | common->optimized_cbracket[GET2(cc, 1)] = 0; |
common->optimized_cbracket[i] = 0; | ||
818 | cc += 1 + IMM2_SIZE; | cc += 1 + IMM2_SIZE; |
819 | break; | break; |
820 | ||
821 | case OP_NCREF: | case OP_DNREF: |
822 | bracketlen = GET2(cc, 1); | case OP_DNREFI: |
823 | name = (pcre_uchar *)common->name_table; | case OP_DNCREF: |
824 | alternative = name; | count = GET2(cc, 1 + IMM2_SIZE); |
825 | for (i = 0; i < common->name_count; i++) | slot = common->name_table + GET2(cc, 1) * common->name_entry_size; |
826 | { | while (count-- > 0) |
if (GET2(name, 0) == bracketlen) break; | ||
name += common->name_entry_size; | ||
} | ||
SLJIT_ASSERT(i != common->name_count); | ||
for (i = 0; i < common->name_count; i++) | ||
827 | { | { |
828 | if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0) | common->optimized_cbracket[GET2(slot, 0)] = 0; |
829 | common->optimized_cbracket[GET2(alternative, 0)] = 0; | slot += common->name_entry_size; |
alternative += common->name_entry_size; | ||
830 | } | } |
831 | bracketlen = 0; | cc += 1 + 2 * IMM2_SIZE; |
cc += 1 + IMM2_SIZE; | ||
break; | ||
case OP_BRA: | ||
bracketlen = 1 + LINK_SIZE; | ||
break; | ||
case OP_CBRA: | ||
case OP_SCBRA: | ||
bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | ||
break; | ||
CASE_ITERATOR_PRIVATE_DATA_1 | ||
space = 1; | ||
size = -2; | ||
break; | ||
CASE_ITERATOR_PRIVATE_DATA_2A | ||
space = 2; | ||
size = -2; | ||
break; | ||
CASE_ITERATOR_PRIVATE_DATA_2B | ||
space = 2; | ||
size = -(2 + IMM2_SIZE); | ||
break; | ||
CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | ||
space = 1; | ||
size = 1; | ||
break; | ||
CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | ||
if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI) | ||
space = 2; | ||
size = 1; | ||
break; | ||
CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | ||
if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) | ||
space = 2; | ||
size = 1 + IMM2_SIZE; | ||
break; | ||
case OP_CLASS: | ||
case OP_NCLASS: | ||
size += 1 + 32 / sizeof(pcre_uchar); | ||
space = get_class_iterator_size(cc + size); | ||
break; | ||
#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | ||
case OP_XCLASS: | ||
size = GET(cc, 1); | ||
space = get_class_iterator_size(cc + size); | ||
832 | break; | break; |
#endif | ||
833 | ||
834 | case OP_RECURSE: | case OP_RECURSE: |
835 | /* Set its value only once. */ | /* Set its value only once. */ |
# | Line 939 while (cc < ccend) | Line 850 while (cc < ccend) |
850 | cc += 2 + 2 * LINK_SIZE; | cc += 2 + 2 * LINK_SIZE; |
851 | break; | break; |
852 | ||
853 | case OP_THEN_ARG: | |
854 | common->has_then = TRUE; | |
855 | common->control_head_ptr = 1; | |
856 | /* Fall through. */ | |
857 | ||
858 | case OP_PRUNE_ARG: | |
859 | common->needs_start_ptr = TRUE; | |
860 | /* Fall through. */ | |
861 | ||
862 | case OP_MARK: | case OP_MARK: |
863 | if (common->mark_ptr == 0) | if (common->mark_ptr == 0) |
864 | { | { |
# | Line 948 while (cc < ccend) | Line 868 while (cc < ccend) |
868 | cc += 1 + 2 + cc[1]; | cc += 1 + 2 + cc[1]; |
869 | break; | break; |
870 | ||
871 | case OP_THEN: | |
872 | common->has_then = TRUE; | |
873 | common->control_head_ptr = 1; | |
874 | /* Fall through. */ | |
875 | ||
876 | case OP_PRUNE: | |
877 | case OP_SKIP: | |
878 | common->needs_start_ptr = TRUE; | |
879 | cc += 1; | |
880 | break; | |
881 | ||
882 | case OP_SKIP_ARG: | |
883 | common->control_head_ptr = 1; | |
884 | common->has_skip_arg = TRUE; | |
885 | cc += 1 + 2 + cc[1]; | |
886 | break; | |
887 | ||
888 | default: | default: |
889 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
890 | if (cc == NULL) | if (cc == NULL) |
891 | return -1; | return FALSE; |
892 | break; | break; |
893 | } | } |
894 | } | |
895 | return TRUE; | |
896 | } | |
897 | ||
898 | if (space > 0 && cc >= end) | static int get_class_iterator_size(pcre_uchar *cc) |
899 | private_data_length += sizeof(sljit_sw) * space; | { |
900 | switch(*cc) | |
901 | { | |
902 | case OP_CRSTAR: | |
903 | case OP_CRPLUS: | |
904 | return 2; | |
905 | ||
906 | if (size != 0) | case OP_CRMINSTAR: |
907 | case OP_CRMINPLUS: | |
908 | case OP_CRQUERY: | |
909 | case OP_CRMINQUERY: | |
910 | return 1; | |
911 | ||
912 | case OP_CRRANGE: | |
913 | case OP_CRMINRANGE: | |
914 | if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE)) | |
915 | return 0; | |
916 | return 2; | |
917 | ||
918 | default: | |
919 | return 0; | |
920 | } | |
921 | } | |
922 | ||
923 | static BOOL detect_repeat(compiler_common *common, pcre_uchar *begin) | |
924 | { | |
925 | pcre_uchar *end = bracketend(begin); | |
926 | pcre_uchar *next; | |
927 | pcre_uchar *next_end; | |
928 | pcre_uchar *max_end; | |
929 | pcre_uchar type; | |
930 | sljit_sw length = end - begin; | |
931 | int min, max, i; | |
932 | ||
933 | /* Detect fixed iterations first. */ | |
934 | if (end[-(1 + LINK_SIZE)] != OP_KET) | |
935 | return FALSE; | |
936 | ||
937 | /* Already detected repeat. */ | |
938 | if (common->private_data_ptrs[end - common->start - LINK_SIZE] != 0) | |
939 | return TRUE; | |
940 | ||
941 | next = end; | |
942 | min = 1; | |
943 | while (1) | |
944 | { | |
945 | if (*next != *begin) | |
946 | break; | |
947 | next_end = bracketend(next); | |
948 | if (next_end - next != length || memcmp(begin, next, IN_UCHARS(length)) != 0) | |
949 | break; | |
950 | next = next_end; | |
951 | min++; | |
952 | } | |
953 | ||
954 | if (min == 2) | |
955 | return FALSE; | |
956 | ||
957 | max = 0; | |
958 | max_end = next; | |
959 | if (*next == OP_BRAZERO || *next == OP_BRAMINZERO) | |
960 | { | |
961 | type = *next; | |
962 | while (1) | |
963 | { | { |
964 | if (size < 0) | if (next[0] != type || next[1] != OP_BRA || next[2 + LINK_SIZE] != *begin) |
965 | { | break; |
966 | cc += -size; | next_end = bracketend(next + 2 + LINK_SIZE); |
967 | #ifdef SUPPORT_UTF | if (next_end - next != (length + 2 + LINK_SIZE) || memcmp(begin, next + 2 + LINK_SIZE, IN_UCHARS(length)) != 0) |
968 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | break; |
969 | #endif | next = next_end; |
970 | } | max++; |
else | ||
cc += size; | ||
971 | } | } |
972 | ||
973 | if (bracketlen != 0) | if (next[0] == type && next[1] == *begin && max >= 1) |
974 | { | { |
975 | if (cc >= end) | next_end = bracketend(next + 1); |
976 | if (next_end - next == (length + 1) && memcmp(begin, next + 1, IN_UCHARS(length)) == 0) | |
977 | { | { |
978 | end = bracketend(cc); | for (i = 0; i < max; i++, next_end += 1 + LINK_SIZE) |
979 | if (end[-1 - LINK_SIZE] == OP_KET) | if (*next_end != OP_KET) |
980 | end = NULL; | break; |
981 | ||
982 | if (i == max) | |
983 | { | |
984 | common->private_data_ptrs[max_end - common->start - LINK_SIZE] = next_end - max_end; | |
985 | common->private_data_ptrs[max_end - common->start - LINK_SIZE + 1] = (type == OP_BRAZERO) ? OP_UPTO : OP_MINUPTO; | |
986 | /* +2 the original and the last. */ | |
987 | common->private_data_ptrs[max_end - common->start - LINK_SIZE + 2] = max + 2; | |
988 | if (min == 1) | |
989 | return TRUE; | |
990 | min--; | |
991 | max_end -= (1 + LINK_SIZE) + GET(max_end, -LINK_SIZE); | |
992 | } | |
993 | } | } |
cc += bracketlen; | ||
994 | } | } |
995 | } | } |
996 | return private_data_length; | |
997 | if (min >= 3) | |
998 | { | |
999 | common->private_data_ptrs[end - common->start - LINK_SIZE] = max_end - end; | |
1000 | common->private_data_ptrs[end - common->start - LINK_SIZE + 1] = OP_EXACT; | |
1001 | common->private_data_ptrs[end - common->start - LINK_SIZE + 2] = min; | |
1002 | return TRUE; | |
1003 | } | |
1004 | ||
1005 | return FALSE; | |
1006 | } | } |
1007 | ||
1008 | static void set_private_data_ptrs(compiler_common *common, int private_data_ptr, pcre_uchar *ccend) | #define CASE_ITERATOR_PRIVATE_DATA_1 \ |
1009 | case OP_MINSTAR: \ | |
1010 | case OP_MINPLUS: \ | |
1011 | case OP_QUERY: \ | |
1012 | case OP_MINQUERY: \ | |
1013 | case OP_MINSTARI: \ | |
1014 | case OP_MINPLUSI: \ | |
1015 | case OP_QUERYI: \ | |
1016 | case OP_MINQUERYI: \ | |
1017 | case OP_NOTMINSTAR: \ | |
1018 | case OP_NOTMINPLUS: \ | |
1019 | case OP_NOTQUERY: \ | |
1020 | case OP_NOTMINQUERY: \ | |
1021 | case OP_NOTMINSTARI: \ | |
1022 | case OP_NOTMINPLUSI: \ | |
1023 | case OP_NOTQUERYI: \ | |
1024 | case OP_NOTMINQUERYI: | |
1025 | ||
1026 | #define CASE_ITERATOR_PRIVATE_DATA_2A \ | |
1027 | case OP_STAR: \ | |
1028 | case OP_PLUS: \ | |
1029 | case OP_STARI: \ | |
1030 | case OP_PLUSI: \ | |
1031 | case OP_NOTSTAR: \ | |
1032 | case OP_NOTPLUS: \ | |
1033 | case OP_NOTSTARI: \ | |
1034 | case OP_NOTPLUSI: | |
1035 | ||
1036 | #define CASE_ITERATOR_PRIVATE_DATA_2B \ | |
1037 | case OP_UPTO: \ | |
1038 | case OP_MINUPTO: \ | |
1039 | case OP_UPTOI: \ | |
1040 | case OP_MINUPTOI: \ | |
1041 | case OP_NOTUPTO: \ | |
1042 | case OP_NOTMINUPTO: \ | |
1043 | case OP_NOTUPTOI: \ | |
1044 | case OP_NOTMINUPTOI: | |
1045 | ||
1046 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \ | |
1047 | case OP_TYPEMINSTAR: \ | |
1048 | case OP_TYPEMINPLUS: \ | |
1049 | case OP_TYPEQUERY: \ | |
1050 | case OP_TYPEMINQUERY: | |
1051 | ||
1052 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \ | |
1053 | case OP_TYPESTAR: \ | |
1054 | case OP_TYPEPLUS: | |
1055 | ||
1056 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \ | |
1057 | case OP_TYPEUPTO: \ | |
1058 | case OP_TYPEMINUPTO: | |
1059 | ||
1060 | static void set_private_data_ptrs(compiler_common *common, int *private_data_start, pcre_uchar *ccend) | |
1061 | { | { |
1062 | pcre_uchar *cc = common->start; | pcre_uchar *cc = common->start; |
1063 | pcre_uchar *alternative; | pcre_uchar *alternative; |
1064 | pcre_uchar *end = NULL; | pcre_uchar *end = NULL; |
1065 | int private_data_ptr = *private_data_start; | |
1066 | int space, size, bracketlen; | int space, size, bracketlen; |
1067 | BOOL repeat_check = TRUE; | |
1068 | ||
1069 | while (cc < ccend) | while (cc < ccend) |
1070 | { | { |
1071 | space = 0; | space = 0; |
1072 | size = 0; | size = 0; |
1073 | bracketlen = 0; | bracketlen = 0; |
1074 | if (private_data_ptr > SLJIT_MAX_LOCAL_SIZE) | |
1075 | break; | |
1076 | ||
1077 | if (repeat_check && (*cc == OP_ONCE || *cc == OP_ONCE_NC || *cc == OP_BRA || *cc == OP_CBRA || *cc == OP_COND)) | |
1078 | { | |
1079 | if (detect_repeat(common, cc)) | |
1080 | { | |
1081 | /* These brackets are converted to repeats, so no global | |
1082 | based single character repeat is allowed. */ | |
1083 | if (cc >= end) | |
1084 | end = bracketend(cc); | |
1085 | } | |
1086 | } | |
1087 | repeat_check = TRUE; | |
1088 | ||
1089 | switch(*cc) | switch(*cc) |
1090 | { | { |
1091 | case OP_KET: | |
1092 | if (common->private_data_ptrs[cc + 1 - common->start] != 0) | |
1093 | { | |
1094 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | |
1095 | private_data_ptr += sizeof(sljit_sw); | |
1096 | cc += common->private_data_ptrs[cc + 1 - common->start]; | |
1097 | } | |
1098 | cc += 1 + LINK_SIZE; | |
1099 | break; | |
1100 | ||
1101 | case OP_ASSERT: | case OP_ASSERT: |
1102 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
1103 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
# | Line 1041 while (cc < ccend) | Line 1140 while (cc < ccend) |
1140 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; |
1141 | break; | break; |
1142 | ||
1143 | case OP_BRAZERO: | |
1144 | case OP_BRAMINZERO: | |
1145 | case OP_BRAPOSZERO: | |
1146 | repeat_check = FALSE; | |
1147 | size = 1; | |
1148 | break; | |
1149 | ||
1150 | CASE_ITERATOR_PRIVATE_DATA_1 | CASE_ITERATOR_PRIVATE_DATA_1 |
1151 | space = 1; | space = 1; |
1152 | size = -2; | size = -2; |
# | Line 1067 while (cc < ccend) | Line 1173 while (cc < ccend) |
1173 | size = 1; | size = 1; |
1174 | break; | break; |
1175 | ||
1176 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | case OP_TYPEUPTO: |
1177 | if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) | if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) |
1178 | space = 2; | space = 2; |
1179 | size = 1 + IMM2_SIZE; | size = 1 + IMM2_SIZE; |
1180 | break; | break; |
1181 | ||
1182 | case OP_TYPEMINUPTO: | |
1183 | space = 2; | |
1184 | size = 1 + IMM2_SIZE; | |
1185 | break; | |
1186 | ||
1187 | case OP_CLASS: | case OP_CLASS: |
1188 | case OP_NCLASS: | case OP_NCLASS: |
1189 | size += 1 + 32 / sizeof(pcre_uchar); | size += 1 + 32 / sizeof(pcre_uchar); |
# | Line 1092 while (cc < ccend) | Line 1203 while (cc < ccend) |
1203 | break; | break; |
1204 | } | } |
1205 | ||
1206 | /* Character iterators, which are not inside a repeated bracket, | |
1207 | gets a private slot instead of allocating it on the stack. */ | |
1208 | if (space > 0 && cc >= end) | if (space > 0 && cc >= end) |
1209 | { | { |
1210 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
# | Line 1122 while (cc < ccend) | Line 1235 while (cc < ccend) |
1235 | cc += bracketlen; | cc += bracketlen; |
1236 | } | } |
1237 | } | } |
1238 | *private_data_start = private_data_ptr; | |
1239 | } | } |
1240 | ||
1241 | /* Returns with -1 if no need for frame. */ | /* Returns with a frame_types (always < 0) if no need for frame. */ |
1242 | static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive) | static int get_framesize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL recursive, BOOL *needs_control_head) |
1243 | { | { |
pcre_uchar *ccend = bracketend(cc); | ||
1244 | int length = 0; | int length = 0; |
1245 | int possessive = 0; | int possessive = 0; |
1246 | BOOL stack_restore = FALSE; | |
1247 | BOOL setsom_found = recursive; | BOOL setsom_found = recursive; |
1248 | BOOL setmark_found = recursive; | BOOL setmark_found = recursive; |
1249 | /* The last capture is a local variable even for recursions. */ | /* The last capture is a local variable even for recursions. */ |
1250 | BOOL capture_last_found = FALSE; | BOOL capture_last_found = FALSE; |
1251 | ||
1252 | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) | #if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD |
1253 | SLJIT_ASSERT(common->control_head_ptr != 0); | |
1254 | *needs_control_head = TRUE; | |
1255 | #else | |
1256 | *needs_control_head = FALSE; | |
1257 | #endif | |
1258 | ||
1259 | if (ccend == NULL) | |
1260 | { | { |
1261 | possessive = length = (common->capture_last_ptr != 0) ? 5 : 3; | ccend = bracketend(cc) - (1 + LINK_SIZE); |
1262 | /* This is correct regardless of common->capture_last_ptr. */ | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) |
1263 | capture_last_found = TRUE; | { |
1264 | possessive = length = (common->capture_last_ptr != 0) ? 5 : 3; | |
1265 | /* This is correct regardless of common->capture_last_ptr. */ | |
1266 | capture_last_found = TRUE; | |
1267 | } | |
1268 | cc = next_opcode(common, cc); | |
1269 | } | } |
1270 | ||
cc = next_opcode(common, cc); | ||
1271 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1272 | while (cc < ccend) | while (cc < ccend) |
1273 | switch(*cc) | switch(*cc) |
1274 | { | { |
1275 | case OP_SET_SOM: | case OP_SET_SOM: |
1276 | SLJIT_ASSERT(common->has_set_som); | SLJIT_ASSERT(common->has_set_som); |
1277 | stack_restore = TRUE; | |
1278 | if (!setsom_found) | if (!setsom_found) |
1279 | { | { |
1280 | length += 2; | length += 2; |
# | Line 1158 while (cc < ccend) | Line 1284 while (cc < ccend) |
1284 | break; | break; |
1285 | ||
1286 | case OP_MARK: | case OP_MARK: |
1287 | case OP_PRUNE_ARG: | |
1288 | case OP_THEN_ARG: | |
1289 | SLJIT_ASSERT(common->mark_ptr != 0); | SLJIT_ASSERT(common->mark_ptr != 0); |
1290 | stack_restore = TRUE; | |
1291 | if (!setmark_found) | if (!setmark_found) |
1292 | { | { |
1293 | length += 2; | length += 2; |
1294 | setmark_found = TRUE; | setmark_found = TRUE; |
1295 | } | } |
1296 | if (common->control_head_ptr != 0) | |
1297 | *needs_control_head = TRUE; | |
1298 | cc += 1 + 2 + cc[1]; | cc += 1 + 2 + cc[1]; |
1299 | break; | break; |
1300 | ||
1301 | case OP_RECURSE: | case OP_RECURSE: |
1302 | stack_restore = TRUE; | |
1303 | if (common->has_set_som && !setsom_found) | if (common->has_set_som && !setsom_found) |
1304 | { | { |
1305 | length += 2; | length += 2; |
# | Line 1190 while (cc < ccend) | Line 1322 while (cc < ccend) |
1322 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1323 | case OP_SCBRA: | case OP_SCBRA: |
1324 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1325 | stack_restore = TRUE; | |
1326 | if (common->capture_last_ptr != 0 && !capture_last_found) | if (common->capture_last_ptr != 0 && !capture_last_found) |
1327 | { | { |
1328 | length += 2; | length += 2; |
# | Line 1199 while (cc < ccend) | Line 1332 while (cc < ccend) |
1332 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1333 | break; | break; |
1334 | ||
1335 | default: | case OP_THEN: |
1336 | cc = next_opcode(common, cc); | stack_restore = TRUE; |
1337 | SLJIT_ASSERT(cc != NULL); | if (common->control_head_ptr != 0) |
1338 | *needs_control_head = TRUE; | |
1339 | cc ++; | |
1340 | break; | break; |
} | ||
1341 | ||
1342 | /* Possessive quantifiers can use a special case. */ | default: |
1343 | if (SLJIT_UNLIKELY(possessive == length)) | stack_restore = TRUE; |
1344 | return -1; | /* Fall through. */ |
1345 | ||
1346 | if (length > 0) | case OP_NOT_WORD_BOUNDARY: |
1347 | return length + 1; | case OP_WORD_BOUNDARY: |
1348 | return -1; | case OP_NOT_DIGIT: |
1349 | case OP_DIGIT: | |
1350 | case OP_NOT_WHITESPACE: | |
1351 | case OP_WHITESPACE: | |
1352 | case OP_NOT_WORDCHAR: | |
1353 | case OP_WORDCHAR: | |
1354 | case OP_ANY: | |
1355 | case OP_ALLANY: | |
1356 | case OP_ANYBYTE: | |
1357 | case OP_NOTPROP: | |
1358 | case OP_PROP: | |
1359 | case OP_ANYNL: | |
1360 | case OP_NOT_HSPACE: | |
1361 | case OP_HSPACE: | |
1362 | case OP_NOT_VSPACE: | |
1363 | case OP_VSPACE: | |
1364 | case OP_EXTUNI: | |
1365 | case OP_EODN: | |
1366 | case OP_EOD: | |
1367 | case OP_CIRC: | |
1368 | case OP_CIRCM: | |
1369 | case OP_DOLL: | |
1370 | case OP_DOLLM: | |
1371 | case OP_CHAR: | |
1372 | case OP_CHARI: | |
1373 | case OP_NOT: | |
1374 | case OP_NOTI: | |
1375 | ||
1376 | case OP_EXACT: | |
1377 | case OP_POSSTAR: | |
1378 | case OP_POSPLUS: | |
1379 | case OP_POSQUERY: | |
1380 | case OP_POSUPTO: | |
1381 | ||
1382 | case OP_EXACTI: | |
1383 | case OP_POSSTARI: | |
1384 | case OP_POSPLUSI: | |
1385 | case OP_POSQUERYI: | |
1386 | case OP_POSUPTOI: | |
1387 | ||
1388 | case OP_NOTEXACT: | |
1389 | case OP_NOTPOSSTAR: | |
1390 | case OP_NOTPOSPLUS: | |
1391 | case OP_NOTPOSQUERY: | |
1392 | case OP_NOTPOSUPTO: | |
1393 | ||
1394 | case OP_NOTEXACTI: | |
1395 | case OP_NOTPOSSTARI: | |
1396 | case OP_NOTPOSPLUSI: | |
1397 | case OP_NOTPOSQUERYI: | |
1398 | case OP_NOTPOSUPTOI: | |
1399 | ||
1400 | case OP_TYPEEXACT: | |
1401 | case OP_TYPEPOSSTAR: | |
1402 | case OP_TYPEPOSPLUS: | |
1403 | case OP_TYPEPOSQUERY: | |
1404 | case OP_TYPEPOSUPTO: | |
1405 | ||
1406 | case OP_CLASS: | |
1407 | case OP_NCLASS: | |
1408 | case OP_XCLASS: | |
1409 | case OP_CALLOUT: | |
1410 | ||
1411 | cc = next_opcode(common, cc); | |
1412 | SLJIT_ASSERT(cc != NULL); | |
1413 | break; | |
1414 | } | |
1415 | ||
1416 | /* Possessive quantifiers can use a special case. */ | |
1417 | if (SLJIT_UNLIKELY(possessive == length)) | |
1418 | return stack_restore ? no_frame : no_stack; | |
1419 | ||
1420 | if (length > 0) | |
1421 | return length + 1; | |
1422 | return stack_restore ? no_frame : no_stack; | |
1423 | } | } |
1424 | ||
1425 | static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive) | static void init_frame(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, int stackpos, int stacktop, BOOL recursive) |
1426 | { | { |
1427 | DEFINE_COMPILER; | DEFINE_COMPILER; |
pcre_uchar *ccend = bracketend(cc); | ||
1428 | BOOL setsom_found = recursive; | BOOL setsom_found = recursive; |
1429 | BOOL setmark_found = recursive; | BOOL setmark_found = recursive; |
1430 | /* The last capture is a local variable even for recursions. */ | /* The last capture is a local variable even for recursions. */ |
# | Line 1229 SLJIT_UNUSED_ARG(stacktop); | Line 1436 SLJIT_UNUSED_ARG(stacktop); |
1436 | SLJIT_ASSERT(stackpos >= stacktop + 2); | SLJIT_ASSERT(stackpos >= stacktop + 2); |
1437 | ||
1438 | stackpos = STACK(stackpos); | stackpos = STACK(stackpos); |
1439 | if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS)) | if (ccend == NULL) |
1440 | cc = next_opcode(common, cc); | { |
1441 | ccend = bracketend(cc) - (1 + LINK_SIZE); | |
1442 | if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS)) | |
1443 | cc = next_opcode(common, cc); | |
1444 | } | |
1445 | ||
1446 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1447 | while (cc < ccend) | while (cc < ccend) |
1448 | switch(*cc) | switch(*cc) |
# | Line 1239 while (cc < ccend) | Line 1451 while (cc < ccend) |
1451 | SLJIT_ASSERT(common->has_set_som); | SLJIT_ASSERT(common->has_set_som); |
1452 | if (!setsom_found) | if (!setsom_found) |
1453 | { | { |
1454 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0)); |
1455 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); |
1456 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1457 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
# | Line 1250 while (cc < ccend) | Line 1462 while (cc < ccend) |
1462 | break; | break; |
1463 | ||
1464 | case OP_MARK: | case OP_MARK: |
1465 | case OP_PRUNE_ARG: | |
1466 | case OP_THEN_ARG: | |
1467 | SLJIT_ASSERT(common->mark_ptr != 0); | SLJIT_ASSERT(common->mark_ptr != 0); |
1468 | if (!setmark_found) | if (!setmark_found) |
1469 | { | { |
1470 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->mark_ptr); |
1471 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); |
1472 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1473 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
# | Line 1266 while (cc < ccend) | Line 1480 while (cc < ccend) |
1480 | case OP_RECURSE: | case OP_RECURSE: |
1481 | if (common->has_set_som && !setsom_found) | if (common->has_set_som && !setsom_found) |
1482 | { | { |
1483 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0)); |
1484 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); |
1485 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1486 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
# | Line 1275 while (cc < ccend) | Line 1489 while (cc < ccend) |
1489 | } | } |
1490 | if (common->mark_ptr != 0 && !setmark_found) | if (common->mark_ptr != 0 && !setmark_found) |
1491 | { | { |
1492 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->mark_ptr); |
1493 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); |
1494 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1495 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
# | Line 1284 while (cc < ccend) | Line 1498 while (cc < ccend) |
1498 | } | } |
1499 | if (common->capture_last_ptr != 0 && !capture_last_found) | if (common->capture_last_ptr != 0 && !capture_last_found) |
1500 | { | { |
1501 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr); |
1502 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); |
1503 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1504 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
# | Line 1300 while (cc < ccend) | Line 1514 while (cc < ccend) |
1514 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1515 | if (common->capture_last_ptr != 0 && !capture_last_found) | if (common->capture_last_ptr != 0 && !capture_last_found) |
1516 | { | { |
1517 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr); |
1518 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); |
1519 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1520 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
# | Line 1310 while (cc < ccend) | Line 1524 while (cc < ccend) |
1524 | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; |
1525 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); |
1526 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1527 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset)); |
1528 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1)); |
1529 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1530 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1531 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); |
# | Line 1330 OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st | Line 1544 OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st |
1544 | SLJIT_ASSERT(stackpos == STACK(stacktop)); | SLJIT_ASSERT(stackpos == STACK(stacktop)); |
1545 | } | } |
1546 | ||
1547 | static SLJIT_INLINE int get_private_data_length_for_copy(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | static SLJIT_INLINE int get_private_data_copy_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL needs_control_head) |
1548 | { | { |
1549 | int private_data_length = 2; | int private_data_length = needs_control_head ? 3 : 2; |
1550 | int size; | int size; |
1551 | pcre_uchar *alternative; | pcre_uchar *alternative; |
1552 | /* Calculate the sum of the private machine words. */ | /* Calculate the sum of the private machine words. */ |
# | Line 1341 while (cc < ccend) | Line 1555 while (cc < ccend) |
1555 | size = 0; | size = 0; |
1556 | switch(*cc) | switch(*cc) |
1557 | { | { |
1558 | case OP_KET: | |
1559 | if (PRIVATE_DATA(cc) != 0) | |
1560 | { | |
1561 | private_data_length++; | |
1562 | SLJIT_ASSERT(PRIVATE_DATA(cc + 1) != 0); | |
1563 | cc += PRIVATE_DATA(cc + 1); | |
1564 | } | |
1565 | cc += 1 + LINK_SIZE; | |
1566 | break; | |
1567 | ||
1568 | case OP_ASSERT: | case OP_ASSERT: |
1569 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
1570 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
# | Line 1352 while (cc < ccend) | Line 1576 while (cc < ccend) |
1576 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1577 | case OP_SCOND: | case OP_SCOND: |
1578 | private_data_length++; | private_data_length++; |
1579 | SLJIT_ASSERT(PRIVATE_DATA(cc) != 0); | |
1580 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1581 | break; | break; |
1582 | ||
# | Line 1445 return private_data_length; | Line 1670 return private_data_length; |
1670 | } | } |
1671 | ||
1672 | static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, | static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, |
1673 | BOOL save, int stackptr, int stacktop) | BOOL save, int stackptr, int stacktop, BOOL needs_control_head) |
1674 | { | { |
1675 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1676 | int srcw[2]; | int srcw[2]; |
# | Line 1466 stacktop = STACK(stacktop - 1); | Line 1691 stacktop = STACK(stacktop - 1); |
1691 | ||
1692 | if (!save) | if (!save) |
1693 | { | { |
1694 | stackptr += sizeof(sljit_sw); | stackptr += (needs_control_head ? 2 : 1) * sizeof(sljit_sw); |
1695 | if (stackptr < stacktop) | if (stackptr < stacktop) |
1696 | { | { |
1697 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
# | Line 1482 if (!save) | Line 1707 if (!save) |
1707 | /* The tmp1next must be TRUE in either way. */ | /* The tmp1next must be TRUE in either way. */ |
1708 | } | } |
1709 | ||
1710 | while (status != end) | do |
1711 | { | { |
1712 | count = 0; | count = 0; |
1713 | switch(status) | switch(status) |
# | Line 1491 while (status != end) | Line 1716 while (status != end) |
1716 | SLJIT_ASSERT(save && common->recursive_head_ptr != 0); | SLJIT_ASSERT(save && common->recursive_head_ptr != 0); |
1717 | count = 1; | count = 1; |
1718 | srcw[0] = common->recursive_head_ptr; | srcw[0] = common->recursive_head_ptr; |
1719 | if (needs_control_head) | |
1720 | { | |
1721 | SLJIT_ASSERT(common->control_head_ptr != 0); | |
1722 | count = 2; | |
1723 | srcw[1] = common->control_head_ptr; | |
1724 | } | |
1725 | status = loop; | status = loop; |
1726 | break; | break; |
1727 | ||
# | Line 1503 while (status != end) | Line 1734 while (status != end) |
1734 | ||
1735 | switch(*cc) | switch(*cc) |
1736 | { | { |
1737 | case OP_KET: | |
1738 | if (PRIVATE_DATA(cc) != 0) | |
1739 | { | |
1740 | count = 1; | |
1741 | srcw[0] = PRIVATE_DATA(cc); | |
1742 | SLJIT_ASSERT(PRIVATE_DATA(cc + 1) != 0); | |
1743 | cc += PRIVATE_DATA(cc + 1); | |
1744 | } | |
1745 | cc += 1 + LINK_SIZE; | |
1746 | break; | |
1747 | ||
1748 | case OP_ASSERT: | case OP_ASSERT: |
1749 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
1750 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
# | Line 1670 while (status != end) | Line 1912 while (status != end) |
1912 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1913 | stackptr += sizeof(sljit_sw); | stackptr += sizeof(sljit_sw); |
1914 | } | } |
1915 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), srcw[count]); |
1916 | tmp1empty = FALSE; | tmp1empty = FALSE; |
1917 | tmp1next = FALSE; | tmp1next = FALSE; |
1918 | } | } |
# | Line 1681 while (status != end) | Line 1923 while (status != end) |
1923 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1924 | stackptr += sizeof(sljit_sw); | stackptr += sizeof(sljit_sw); |
1925 | } | } |
1926 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), srcw[count]); |
1927 | tmp2empty = FALSE; | tmp2empty = FALSE; |
1928 | tmp1next = TRUE; | tmp1next = TRUE; |
1929 | } | } |
# | Line 1691 while (status != end) | Line 1933 while (status != end) |
1933 | if (tmp1next) | if (tmp1next) |
1934 | { | { |
1935 | SLJIT_ASSERT(!tmp1empty); | SLJIT_ASSERT(!tmp1empty); |
1936 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count], TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), srcw[count], TMP1, 0); |
1937 | tmp1empty = stackptr >= stacktop; | tmp1empty = stackptr >= stacktop; |
1938 | if (!tmp1empty) | if (!tmp1empty) |
1939 | { | { |
# | Line 1703 while (status != end) | Line 1945 while (status != end) |
1945 | else | else |
1946 | { | { |
1947 | SLJIT_ASSERT(!tmp2empty); | SLJIT_ASSERT(!tmp2empty); |
1948 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count], TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), srcw[count], TMP2, 0); |
1949 | tmp2empty = stackptr >= stacktop; | tmp2empty = stackptr >= stacktop; |
1950 | if (!tmp2empty) | if (!tmp2empty) |
1951 | { | { |
# | Line 1715 while (status != end) | Line 1957 while (status != end) |
1957 | } | } |
1958 | } | } |
1959 | } | } |
1960 | while (status != end); | |
1961 | ||
1962 | if (save) | if (save) |
1963 | { | { |
# | Line 1748 if (save) | Line 1991 if (save) |
1991 | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); |
1992 | } | } |
1993 | ||
1994 | static SLJIT_INLINE pcre_uchar *set_then_offsets(compiler_common *common, pcre_uchar *cc, pcre_uint8 *current_offset) | |
1995 | { | |
1996 | pcre_uchar *end = bracketend(cc); | |
1997 | BOOL has_alternatives = cc[GET(cc, 1)] == OP_ALT; | |
1998 | ||
1999 | /* Assert captures then. */ | |
2000 | if (*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) | |
2001 | current_offset = NULL; | |
2002 | /* Conditional block does not. */ | |
2003 | if (*cc == OP_COND || *cc == OP_SCOND) | |
2004 | has_alternatives = FALSE; | |
2005 | ||
2006 | cc = next_opcode(common, cc); | |
2007 | if (has_alternatives) | |
2008 | current_offset = common->then_offsets + (cc - common->start); | |
2009 | ||
2010 | while (cc < end) | |
2011 | { | |
2012 | if ((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)) | |
2013 | cc = set_then_offsets(common, cc, current_offset); | |
2014 | else | |
2015 | { | |
2016 | if (*cc == OP_ALT && has_alternatives) | |
2017 | current_offset = common->then_offsets + (cc + 1 + LINK_SIZE - common->start); | |
2018 | if (*cc >= OP_THEN && *cc <= OP_THEN_ARG && current_offset != NULL) | |
2019 | *current_offset = 1; | |
2020 | cc = next_opcode(common, cc); | |
2021 | } | |
2022 | } | |
2023 | ||
2024 | return end; | |
2025 | } | |
2026 | ||
2027 | #undef CASE_ITERATOR_PRIVATE_DATA_1 | #undef CASE_ITERATOR_PRIVATE_DATA_1 |
2028 | #undef CASE_ITERATOR_PRIVATE_DATA_2A | #undef CASE_ITERATOR_PRIVATE_DATA_2A |
2029 | #undef CASE_ITERATOR_PRIVATE_DATA_2B | #undef CASE_ITERATOR_PRIVATE_DATA_2B |
# | Line 1771 while (list) | Line 2047 while (list) |
2047 | } | } |
2048 | } | } |
2049 | ||
2050 | static SLJIT_INLINE void add_jump(struct sljit_compiler *compiler, jump_list **list, struct sljit_jump* jump) | static SLJIT_INLINE void add_jump(struct sljit_compiler *compiler, jump_list **list, struct sljit_jump *jump) |
2051 | { | { |
2052 | jump_list *list_item = sljit_alloc_memory(compiler, sizeof(jump_list)); | jump_list *list_item = sljit_alloc_memory(compiler, sizeof(jump_list)); |
2053 | if (list_item) | if (list_item) |
# | Line 1782 if (list_item) | Line 2058 if (list_item) |
2058 | } | } |
2059 | } | } |
2060 | ||
2061 | static void add_stub(compiler_common *common, enum stub_types type, int data, struct sljit_jump *start) | static void add_stub(compiler_common *common, struct sljit_jump *start) |
2062 | { | { |
2063 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2064 | stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); | stub_list *list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); |
2065 | ||
2066 | if (list_item) | if (list_item) |
2067 | { | { |
list_item->type = type; | ||
list_item->data = data; | ||
2068 | list_item->start = start; | list_item->start = start; |
2069 | list_item->quit = LABEL(); | list_item->quit = LABEL(); |
2070 | list_item->next = common->stubs; | list_item->next = common->stubs; |
# | Line 1801 if (list_item) | Line 2075 if (list_item) |
2075 | static void flush_stubs(compiler_common *common) | static void flush_stubs(compiler_common *common) |
2076 | { | { |
2077 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2078 | stub_list* list_item = common->stubs; | stub_list *list_item = common->stubs; |
2079 | ||
2080 | while (list_item) | while (list_item) |
2081 | { | { |
2082 | JUMPHERE(list_item->start); | JUMPHERE(list_item->start); |
2083 | switch(list_item->type) | add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); |
{ | ||
case stack_alloc: | ||
add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); | ||
break; | ||
} | ||
2084 | JUMPTO(SLJIT_JUMP, list_item->quit); | JUMPTO(SLJIT_JUMP, list_item->quit); |
2085 | list_item = list_item->next; | list_item = list_item->next; |
2086 | } | } |
2087 | common->stubs = NULL; | common->stubs = NULL; |
2088 | } | } |
2089 | ||
2090 | static SLJIT_INLINE void decrease_call_count(compiler_common *common) | static void add_label_addr(compiler_common *common, sljit_uw *update_addr) |
2091 | { | |
2092 | DEFINE_COMPILER; | |
2093 | label_addr_list *label_addr; | |
2094 | ||
2095 | label_addr = sljit_alloc_memory(compiler, sizeof(label_addr_list)); | |
2096 | if (label_addr == NULL) | |
2097 | return; | |
2098 | label_addr->label = LABEL(); | |
2099 | label_addr->update_addr = update_addr; | |
2100 | label_addr->next = common->label_addrs; | |
2101 | common->label_addrs = label_addr; | |
2102 | } | |
2103 | ||
2104 | static SLJIT_INLINE void count_match(compiler_common *common) | |
2105 | { | { |
2106 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2107 | ||
2108 | OP2(SLJIT_SUB | SLJIT_SET_E, CALL_COUNT, 0, CALL_COUNT, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, COUNT_MATCH, 0, COUNT_MATCH, 0, SLJIT_IMM, 1); |
2109 | add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO)); | add_jump(compiler, &common->calllimit, JUMP(SLJIT_ZERO)); |
2110 | } | } |
2111 | ||
2112 | static SLJIT_INLINE void allocate_stack(compiler_common *common, int size) | static SLJIT_INLINE void allocate_stack(compiler_common *common, int size) |
# | Line 1831 static SLJIT_INLINE void allocate_stack( | Line 2114 static SLJIT_INLINE void allocate_stack( |
2114 | /* May destroy all locals and registers except TMP2. */ | /* May destroy all locals and registers except TMP2. */ |
2115 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2116 | ||
2117 | SLJIT_ASSERT(size > 0); | |
2118 | OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw)); | OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw)); |
2119 | #ifdef DESTROY_REGISTERS | #ifdef DESTROY_REGISTERS |
2120 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); |
2121 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
2122 | OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); | OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); |
2123 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS0, TMP1, 0); |
2124 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS1, TMP1, 0); |
2125 | #endif | #endif |
2126 | add_stub(common, stack_alloc, 0, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); | add_stub(common, CMP(SLJIT_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); |
2127 | } | } |
2128 | ||
2129 | static SLJIT_INLINE void free_stack(compiler_common *common, int size) | static SLJIT_INLINE void free_stack(compiler_common *common, int size) |
2130 | { | { |
2131 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2132 | ||
2133 | SLJIT_ASSERT(size > 0); | |
2134 | OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw)); | OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw)); |
2135 | } | } |
2136 | ||
2137 | static sljit_uw * allocate_read_only_data(compiler_common *common, sljit_uw size) | |
2138 | { | |
2139 | DEFINE_COMPILER; | |
2140 | sljit_uw *result; | |
2141 | ||
2142 | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) | |
2143 | return NULL; | |
2144 | ||
2145 | result = (sljit_uw *)SLJIT_MALLOC(size + sizeof(sljit_uw), compiler->allocator_data); | |
2146 | if (SLJIT_UNLIKELY(result == NULL)) | |
2147 | { | |
2148 | sljit_set_compiler_memory_error(compiler); | |
2149 | return NULL; | |
2150 | } | |
2151 | ||
2152 | *(void**)result = common->read_only_data_head; | |
2153 | common->read_only_data_head = (void *)result; | |
2154 | return result + 1; | |
2155 | } | |
2156 | ||
2157 | static void free_read_only_data(void *current, void *allocator_data) | |
2158 | { | |
2159 | void *next; | |
2160 | ||
2161 | SLJIT_UNUSED_ARG(allocator_data); | |
2162 | ||
2163 | while (current != NULL) | |
2164 | { | |
2165 | next = *(void**)current; | |
2166 | SLJIT_FREE(current, allocator_data); | |
2167 | current = next; | |
2168 | } | |
2169 | } | |
2170 | ||
2171 | static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) | static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) |
2172 | { | { |
2173 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2174 | struct sljit_label *loop; | struct sljit_label *loop; |
2175 | int i; | int i; |
2176 | ||
2177 | /* At this point we can freely use all temporary registers. */ | /* At this point we can freely use all temporary registers. */ |
2178 | SLJIT_ASSERT(length > 1); | |
2179 | /* TMP1 returns with begin - 1. */ | /* TMP1 returns with begin - 1. */ |
2180 | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_SUB, SLJIT_R0, 0, SLJIT_MEM1(SLJIT_S0), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1)); |
2181 | if (length < 8) | if (length < 8) |
2182 | { | { |
2183 | for (i = 0; i < length; i++) | for (i = 1; i < length; i++) |
2184 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_SCRATCH_REG1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(i), SLJIT_R0, 0); |
2185 | } | } |
2186 | else | else |
2187 | { | { |
2188 | GET_LOCAL_BASE(SLJIT_SCRATCH_REG2, 0, OVECTOR_START - sizeof(sljit_sw)); | GET_LOCAL_BASE(SLJIT_R1, 0, OVECTOR_START); |
2189 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, length); | OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_IMM, length - 1); |
2190 | loop = LABEL(); | loop = LABEL(); |
2191 | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(sljit_sw), SLJIT_SCRATCH_REG1, 0); | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_R1), sizeof(sljit_sw), SLJIT_R0, 0); |
2192 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, 1); |
2193 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_NOT_ZERO, loop); |
2194 | } | |
2195 | } | |
2196 | ||
2197 | static SLJIT_INLINE void do_reset_match(compiler_common *common, int length) | |
2198 | { | |
2199 | DEFINE_COMPILER; | |
2200 | struct sljit_label *loop; | |
2201 | int i; | |
2202 | ||
2203 | SLJIT_ASSERT(length > 1); | |
2204 | /* OVECTOR(1) contains the "string begin - 1" constant. */ | |
2205 | if (length > 2) | |
2206 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1)); | |
2207 | if (length < 8) | |
2208 | { | |
2209 | for (i = 2; i < length; i++) | |
2210 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(i), TMP1, 0); | |
2211 | } | |
2212 | else | |
2213 | { | |
2214 | GET_LOCAL_BASE(TMP2, 0, OVECTOR_START + sizeof(sljit_sw)); | |
2215 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, length - 2); | |
2216 | loop = LABEL(); | |
2217 | OP1(SLJIT_MOVU, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0); | |
2218 | OP2(SLJIT_SUB | SLJIT_SET_E, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 1); | |
2219 | JUMPTO(SLJIT_NOT_ZERO, loop); | |
2220 | } | |
2221 | ||
2222 | OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0); | |
2223 | if (common->mark_ptr != 0) | |
2224 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->mark_ptr, SLJIT_IMM, 0); | |
2225 | if (common->control_head_ptr != 0) | |
2226 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_IMM, 0); | |
2227 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack)); | |
2228 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->start_ptr); | |
2229 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, base)); | |
2230 | } | |
2231 | ||
2232 | static sljit_sw SLJIT_CALL do_search_mark(sljit_sw *current, const pcre_uchar *skip_arg) | |
2233 | { | |
2234 | while (current != NULL) | |
2235 | { | |
2236 | switch (current[-2]) | |
2237 | { | |
2238 | case type_then_trap: | |
2239 | break; | |
2240 | ||
2241 | case type_mark: | |
2242 | if (STRCMP_UC_UC(skip_arg, (pcre_uchar *)current[-3]) == 0) | |
2243 | return current[-4]; | |
2244 | break; | |
2245 | ||
2246 | default: | |
2247 | SLJIT_ASSERT_STOP(); | |
2248 | break; | |
2249 | } | |
2250 | SLJIT_ASSERT(current > (sljit_sw*)current[-1]); | |
2251 | current = (sljit_sw*)current[-1]; | |
2252 | } | } |
2253 | return -1; | |
2254 | } | } |
2255 | ||
2256 | static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket) | static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket) |
# | Line 1879 struct sljit_label *loop; | Line 2260 struct sljit_label *loop; |
2260 | struct sljit_jump *early_quit; | struct sljit_jump *early_quit; |
2261 | ||
2262 | /* At this point we can freely use all registers. */ | /* At this point we can freely use all registers. */ |
2263 | OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); | OP1(SLJIT_MOV, SLJIT_S2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1)); |
2264 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1), STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(1), STR_PTR, 0); |
2265 | ||
2266 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_R0, 0, ARGUMENTS, 0); |
2267 | if (common->mark_ptr != 0) | if (common->mark_ptr != 0) |
2268 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_SP), common->mark_ptr); |
2269 | OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offset_count)); | OP1(SLJIT_MOV_SI, SLJIT_R1, 0, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, offset_count)); |
2270 | if (common->mark_ptr != 0) | if (common->mark_ptr != 0) |
2271 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_SCRATCH_REG3, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_R2, 0); |
2272 | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); | OP2(SLJIT_SUB, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); |
2273 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, SLJIT_R0, 0, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, begin)); |
2274 | GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START); | GET_LOCAL_BASE(SLJIT_S0, 0, OVECTOR_START); |
2275 | /* Unlikely, but possible */ | /* Unlikely, but possible */ |
2276 | early_quit = CMP(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 0); | early_quit = CMP(SLJIT_EQUAL, SLJIT_R1, 0, SLJIT_IMM, 0); |
2277 | loop = LABEL(); | loop = LABEL(); |
2278 | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_SCRATCH_REG1, 0); | OP2(SLJIT_SUB, SLJIT_S1, 0, SLJIT_MEM1(SLJIT_S0), 0, SLJIT_R0, 0); |
2279 | OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_SAVED_REG1, 0, SLJIT_IMM, sizeof(sljit_sw)); | OP2(SLJIT_ADD, SLJIT_S0, 0, SLJIT_S0, 0, SLJIT_IMM, sizeof(sljit_sw)); |
2280 | /* Copy the integer value to the output buffer */ | /* Copy the integer value to the output buffer */ |
2281 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2282 | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_ASHR, SLJIT_S1, 0, SLJIT_S1, 0, SLJIT_IMM, UCHAR_SHIFT); |
2283 | #endif | #endif |
2284 | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_R2), sizeof(int), SLJIT_S1, 0); |
2285 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_R1, 0, SLJIT_R1, 0, SLJIT_IMM, 1); |
2286 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_NOT_ZERO, loop); |
2287 | JUMPHERE(early_quit); | JUMPHERE(early_quit); |
2288 | ||
2289 | /* Calculate the return value, which is the maximum ovector value. */ | /* Calculate the return value, which is the maximum ovector value. */ |
2290 | if (topbracket > 1) | if (topbracket > 1) |
2291 | { | { |
2292 | GET_LOCAL_BASE(SLJIT_SCRATCH_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw)); | GET_LOCAL_BASE(SLJIT_R0, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw)); |
2293 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, topbracket + 1); | OP1(SLJIT_MOV, SLJIT_R1, 0, SLJIT_IMM, topbracket + 1); |
2294 | ||
2295 | /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ | /* OVECTOR(0) is never equal to SLJIT_S2. */ |
2296 | loop = LABEL(); | loop = LABEL(); |
2297 | OP1(SLJIT_MOVU, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), -(2 * (sljit_sw)sizeof(sljit_sw))); | OP1(SLJIT_MOVU, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_R0), -(2 * (sljit_sw)sizeof(sljit_sw))); |
2298 | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, SLJIT_R1, 0, SLJIT_R1, 0, SLJIT_IMM, 1); |
2299 | CMPTO(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG3, 0, loop); | CMPTO(SLJIT_EQUAL, SLJIT_R2, 0, SLJIT_S2, 0, loop); |
2300 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_SCRATCH_REG2, 0); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_R1, 0); |
2301 | } | } |
2302 | else | else |
2303 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); |
# | Line 1925 else | Line 2306 else |
2306 | static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit) | static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit) |
2307 | { | { |
2308 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2309 | struct sljit_jump *jump; | |
2310 | ||
2311 | SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2); | SLJIT_COMPILE_ASSERT(STR_END == SLJIT_S1, str_end_must_be_saved_reg2); |
2312 | SLJIT_ASSERT(common->start_used_ptr != 0 && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); | SLJIT_ASSERT(common->start_used_ptr != 0 && common->start_ptr != 0 |
2313 | && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); | |
2314 | ||
2315 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_R1, 0, ARGUMENTS, 0); |
2316 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); |
2317 | OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offset_count)); | OP1(SLJIT_MOV_SI, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_R1), SLJIT_OFFSETOF(jit_arguments, real_offset_count)); |
2318 | CMPTO(SLJIT_C_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 2, quit); | CMPTO(SLJIT_SIG_LESS, SLJIT_R2, 0, SLJIT_IMM, 2, quit); |
2319 | ||
2320 | /* Store match begin and end. */ | /* Store match begin and end. */ |
2321 | OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, SLJIT_S0, 0, SLJIT_MEM1(SLJIT_R1), SLJIT_OFFSETOF(jit_arguments, begin)); |
2322 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); | OP1(SLJIT_MOV, SLJIT_R1, 0, SLJIT_MEM1(SLJIT_R1), SLJIT_OFFSETOF(jit_arguments, offsets)); |
2323 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start); | |
2324 | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0); | jump = CMP(SLJIT_SIG_LESS, SLJIT_R2, 0, SLJIT_IMM, 3); |
2325 | OP2(SLJIT_SUB, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_SP), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_ptr : (common->hit_start + (int)sizeof(sljit_sw)), SLJIT_S0, 0); | |
2326 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | |
2327 | OP2(SLJIT_ASHR, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, UCHAR_SHIFT); | |
2328 | #endif | |
2329 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_R1), 2 * sizeof(int), SLJIT_R2, 0); | |
2330 | JUMPHERE(jump); | |
2331 | ||
2332 | OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_SP), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start); | |
2333 | OP2(SLJIT_SUB, SLJIT_S1, 0, STR_END, 0, SLJIT_S0, 0); | |
2334 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2335 | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_ASHR, SLJIT_S1, 0, SLJIT_S1, 0, SLJIT_IMM, UCHAR_SHIFT); |
2336 | #endif | #endif |
2337 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_R1), sizeof(int), SLJIT_S1, 0); |
2338 | ||
2339 | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG1, 0); | OP2(SLJIT_SUB, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_S0, 0); |
2340 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2341 | OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_ASHR, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, UCHAR_SHIFT); |
2342 | #endif | #endif |
2343 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 0, SLJIT_SCRATCH_REG3, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_R1), 0, SLJIT_R2, 0); |
2344 | ||
2345 | JUMPTO(SLJIT_JUMP, quit); | JUMPTO(SLJIT_JUMP, quit); |
2346 | } | } |
# | Line 1962 struct sljit_jump *jump; | Line 2354 struct sljit_jump *jump; |
2354 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2355 | { | { |
2356 | /* The value of -1 must be kept for start_used_ptr! */ | /* The value of -1 must be kept for start_used_ptr! */ |
2357 | OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, 1); | OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, SLJIT_IMM, 1); |
2358 | /* Jumps if start_used_ptr < STR_PTR, or start_used_ptr == -1. Although overwriting | /* Jumps if start_used_ptr < STR_PTR, or start_used_ptr == -1. Although overwriting |
2359 | is not necessary if start_used_ptr == STR_PTR, it does not hurt as well. */ | is not necessary if start_used_ptr == STR_PTR, it does not hurt as well. */ |
2360 | jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_PTR, 0); | jump = CMP(SLJIT_LESS_EQUAL, TMP1, 0, STR_PTR, 0); |
2361 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0); |
2362 | JUMPHERE(jump); | JUMPHERE(jump); |
2363 | } | } |
2364 | else if (common->mode == JIT_PARTIAL_HARD_COMPILE) | else if (common->mode == JIT_PARTIAL_HARD_COMPILE) |
2365 | { | { |
2366 | jump = CMP(SLJIT_C_LESS_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | jump = CMP(SLJIT_LESS_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0); |
2367 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0); |
2368 | JUMPHERE(jump); | JUMPHERE(jump); |
2369 | } | } |
2370 | } | } |
2371 | ||
2372 | static SLJIT_INLINE BOOL char_has_othercase(compiler_common *common, pcre_uchar* cc) | static SLJIT_INLINE BOOL char_has_othercase(compiler_common *common, pcre_uchar *cc) |
2373 | { | { |
2374 | /* Detects if the character has an othercase. */ | /* Detects if the character has an othercase. */ |
2375 | unsigned int c; | unsigned int c; |
# | Line 2020 if (common->utf && c > 127) | Line 2412 if (common->utf && c > 127) |
2412 | return TABLE_GET(c, common->fcc, c); | return TABLE_GET(c, common->fcc, c); |
2413 | } | } |
2414 | ||
2415 | static unsigned int char_get_othercase_bit(compiler_common *common, pcre_uchar* cc) | static unsigned int char_get_othercase_bit(compiler_common *common, pcre_uchar *cc) |
2416 | { | { |
2417 | /* Detects if the character and its othercase has only 1 bit difference. */ | /* Detects if the character and its othercase has only 1 bit difference. */ |
2418 | unsigned int c, oc, bit; | unsigned int c, oc, bit; |
# | Line 2098 return (bit < 256) ? ((0 << 8) | bit) : | Line 2490 return (bit < 256) ? ((0 << 8) | bit) : |
2490 | ||
2491 | static void check_partial(compiler_common *common, BOOL force) | static void check_partial(compiler_common *common, BOOL force) |
2492 | { | { |
2493 | /* Checks whether a partial matching is occured. Does not modify registers. */ | /* Checks whether a partial matching is occurred. Does not modify registers. */ |
2494 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2495 | struct sljit_jump *jump = NULL; | struct sljit_jump *jump = NULL; |
2496 | ||
# | Line 2108 if (common->mode == JIT_COMPILE) | Line 2500 if (common->mode == JIT_COMPILE) |
2500 | return; | return; |
2501 | ||
2502 | if (!force) | if (!force) |
2503 | jump = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | jump = CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0); |
2504 | else if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | else if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2505 | jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1); | jump = CMP(SLJIT_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, SLJIT_IMM, -1); |
2506 | ||
2507 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2508 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, 0); |
2509 | else | else |
2510 | { | { |
2511 | if (common->partialmatchlabel != NULL) | if (common->partialmatchlabel != NULL) |
# | Line 2126 if (jump != NULL) | Line 2518 if (jump != NULL) |
2518 | JUMPHERE(jump); | JUMPHERE(jump); |
2519 | } | } |
2520 | ||
2521 | static struct sljit_jump *check_str_end(compiler_common *common) | static void check_str_end(compiler_common *common, jump_list **end_reached) |
2522 | { | { |
2523 | /* Does not affect registers. Usually used in a tight spot. */ | /* Does not affect registers. Usually used in a tight spot. */ |
2524 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2525 | struct sljit_jump *jump; | struct sljit_jump *jump; |
struct sljit_jump *nohit; | ||
struct sljit_jump *return_value; | ||
2526 | ||
2527 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
2528 | return CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | { |
2529 | add_jump(compiler, end_reached, CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | |
2530 | return; | |
2531 | } | |
2532 | ||
2533 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump = CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0); |
2534 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2535 | { | { |
2536 | nohit = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | add_jump(compiler, end_reached, CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0)); |
2537 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, 0); |
2538 | JUMPHERE(nohit); | add_jump(compiler, end_reached, JUMP(SLJIT_JUMP)); |
return_value = JUMP(SLJIT_JUMP); | ||
2539 | } | } |
2540 | else | else |
2541 | { | { |
2542 | return_value = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | add_jump(compiler, end_reached, CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0)); |
2543 | if (common->partialmatchlabel != NULL) | if (common->partialmatchlabel != NULL) |
2544 | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); |
2545 | else | else |
2546 | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); |
2547 | } | } |
2548 | JUMPHERE(jump); | JUMPHERE(jump); |
return return_value; | ||
2549 | } | } |
2550 | ||
2551 | static void detect_partial_match(compiler_common *common, jump_list **backtracks) | static void detect_partial_match(compiler_common *common, jump_list **backtracks) |
# | Line 2164 struct sljit_jump *jump; | Line 2555 struct sljit_jump *jump; |
2555 | ||
2556 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
2557 | { | { |
2558 | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); |
2559 | return; | return; |
2560 | } | } |
2561 | ||
2562 | /* Partial matching mode. */ | /* Partial matching mode. */ |
2563 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump = CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0); |
2564 | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0)); |
2565 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2566 | { | { |
2567 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, 0); |
2568 | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
2569 | } | } |
2570 | else | else |
# | Line 2186 else | Line 2577 else |
2577 | JUMPHERE(jump); | JUMPHERE(jump); |
2578 | } | } |
2579 | ||
2580 | static void read_char(compiler_common *common) | static void peek_char(compiler_common *common, pcre_uint32 max) |
2581 | { | { |
2582 | /* Reads the character into TMP1, updates STR_PTR. | /* Reads the character into TMP1, keeps STR_PTR. |
2583 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2584 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2585 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2586 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2587 | #endif | #endif |
2588 | ||
2589 | SLJIT_UNUSED_ARG(max); | |
2590 | ||
2591 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2592 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 |
2593 | if (common->utf) | if (common->utf) |
2594 | { | { |
2595 | #if defined COMPILE_PCRE8 | if (max < 128) return; |
2596 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | |
2597 | #elif defined COMPILE_PCRE16 | jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2598 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
#endif /* COMPILE_PCRE[8|16] */ | ||
2599 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2600 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | |
2601 | JUMPHERE(jump); | JUMPHERE(jump); |
2602 | } | } |
2603 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2604 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2605 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
2606 | if (common->utf) | |
2607 | { | |
2608 | if (max < 0xd800) return; | |
2609 | ||
2610 | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
2611 | jump = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1); | |
2612 | /* TMP2 contains the high surrogate. */ | |
2613 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2614 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40); | |
2615 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10); | |
2616 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff); | |
2617 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2618 | JUMPHERE(jump); | |
2619 | } | |
2620 | #endif | |
2621 | } | } |
2622 | ||
2623 | static void peek_char(compiler_common *common) | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 |
2624 | ||
2625 | static BOOL is_char7_bitset(const pcre_uint8 *bitset, BOOL nclass) | |
2626 | { | { |
2627 | /* Reads the character into TMP1, keeps STR_PTR. | /* Tells whether the character codes below 128 are enough |
2628 | Does not check STR_END. TMP2 Destroyed. */ | to determine a match. */ |
2629 | const pcre_uint8 value = nclass ? 0xff : 0; | |
2630 | const pcre_uint8 *end = bitset + 32; | |
2631 | ||
2632 | bitset += 16; | |
2633 | do | |
2634 | { | |
2635 | if (*bitset++ != value) | |
2636 | return FALSE; | |
2637 | } | |
2638 | while (bitset < end); | |
2639 | return TRUE; | |
2640 | } | |
2641 | ||
2642 | static void read_char7_type(compiler_common *common, BOOL full_read) | |
2643 | { | |
2644 | /* Reads the precise character type of a character into TMP1, if the character | |
2645 | is less than 128. Otherwise it returns with zero. Does not check STR_END. The | |
2646 | full_read argument tells whether characters above max are accepted or not. */ | |
2647 | DEFINE_COMPILER; | DEFINE_COMPILER; |
#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | ||
2648 | struct sljit_jump *jump; | struct sljit_jump *jump; |
#endif | ||
2649 | ||
2650 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | SLJIT_ASSERT(common->utf); |
2651 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
2652 | if (common->utf) | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2653 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2654 | ||
2655 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
2656 | ||
2657 | if (full_read) | |
2658 | { | { |
2659 | #if defined COMPILE_PCRE8 | jump = CMP(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0xc0); |
2660 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0); |
2661 | #elif defined COMPILE_PCRE16 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | ||
#endif /* COMPILE_PCRE[8|16] */ | ||
add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | ||
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | ||
2662 | JUMPHERE(jump); | JUMPHERE(jump); |
2663 | } | } |
#endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | ||
2664 | } | } |
2665 | ||
2666 | static void read_char8_type(compiler_common *common) | #endif /* SUPPORT_UTF && COMPILE_PCRE8 */ |
2667 | ||
2668 | static void read_char_range(compiler_common *common, pcre_uint32 min, pcre_uint32 max, BOOL update_str_ptr) | |
2669 | { | { |
2670 | /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */ | /* Reads the precise value of a character into TMP1, if the character is |
2671 | between min and max (c >= min && c <= max). Otherwise it returns with a value | |
2672 | outside the range. Does not check STR_END. */ | |
2673 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2674 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2675 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2676 | #endif | #endif |
2677 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
2678 | struct sljit_jump *jump2; | |
2679 | #endif | |
2680 | ||
2681 | #ifdef SUPPORT_UTF | SLJIT_UNUSED_ARG(update_str_ptr); |
2682 | SLJIT_UNUSED_ARG(min); | |
2683 | SLJIT_UNUSED_ARG(max); | |
2684 | SLJIT_ASSERT(min <= max); | |
2685 | ||
2686 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2687 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2688 | ||
2689 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
2690 | if (common->utf) | if (common->utf) |
2691 | { | { |
2692 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | if (max < 128 && !update_str_ptr) return; |
2693 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2694 | #if defined COMPILE_PCRE8 | jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2695 | /* This can be an extra read in some situations, but hopefully | if (min >= 0x10000) |
2696 | it is needed in most cases. */ | { |
2697 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xf0); |
2698 | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); | if (update_str_ptr) |
2699 | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); | OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); |
2700 | JUMPHERE(jump); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
2701 | #elif defined COMPILE_PCRE16 | jump2 = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0x7); |
2702 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); |
2703 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); |
2704 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2705 | JUMPHERE(jump); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
2706 | /* Skip low surrogate if necessary. */ | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); |
2707 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
2708 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2709 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); |
2710 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | if (!update_str_ptr) |
2711 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3)); |
2712 | #elif defined COMPILE_PCRE32 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); |
2713 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
2714 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2715 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | JUMPHERE(jump2); |
2716 | if (update_str_ptr) | |
2717 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0); | |
2718 | } | |
2719 | else if (min >= 0x800 && max <= 0xffff) | |
2720 | { | |
2721 | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xe0); | |
2722 | if (update_str_ptr) | |
2723 | OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
2724 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2725 | jump2 = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0xf); | |
2726 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | |
2727 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | |
2728 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2729 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2730 | if (!update_str_ptr) | |
2731 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | |
2732 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2733 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | |
2734 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2735 | JUMPHERE(jump2); | |
2736 | if (update_str_ptr) | |
2737 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0); | |
2738 | } | |
2739 | else if (max >= 0x800) | |
2740 | add_jump(compiler, (max < 0x10000) ? &common->utfreadchar16 : &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | |
2741 | else if (max < 128) | |
2742 | { | |
2743 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
2744 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | |
2745 | } | |
2746 | else | |
2747 | { | |
2748 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2749 | if (!update_str_ptr) | |
2750 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2751 | else | |
2752 | OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
2753 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | |
2754 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2755 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | |
2756 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2757 | if (update_str_ptr) | |
2758 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0); | |
2759 | } | |
2760 | JUMPHERE(jump); | JUMPHERE(jump); |
#endif /* COMPILE_PCRE[8|16|32] */ | ||
return; | ||
2761 | } | } |
2762 | #endif /* SUPPORT_UTF */ | #endif |
2763 | ||
2764 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
2765 | if (common->utf) | |
2766 | { | |
2767 | if (max >= 0x10000) | |
2768 | { | |
2769 | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
2770 | jump = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1); | |
2771 | /* TMP2 contains the high surrogate. */ | |
2772 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2773 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40); | |
2774 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10); | |
2775 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2776 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff); | |
2777 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2778 | JUMPHERE(jump); | |
2779 | return; | |
2780 | } | |
2781 | ||
2782 | if (max < 0xd800 && !update_str_ptr) return; | |
2783 | ||
2784 | /* Skip low surrogate if necessary. */ | |
2785 | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
2786 | jump = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1); | |
2787 | if (update_str_ptr) | |
2788 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2789 | if (max >= 0xd800) | |
2790 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0x10000); | |
2791 | JUMPHERE(jump); | |
2792 | } | |
2793 | #endif | |
2794 | } | |
2795 | ||
2796 | static SLJIT_INLINE void read_char(compiler_common *common) | |
2797 | { | |
2798 | read_char_range(common, 0, READ_CHAR_MAX, TRUE); | |
2799 | } | |
2800 | ||
2801 | static void read_char8_type(compiler_common *common, BOOL update_str_ptr) | |
2802 | { | |
2803 | /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */ | |
2804 | DEFINE_COMPILER; | |
2805 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
2806 | struct sljit_jump *jump; | |
2807 | #endif | |
2808 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
2809 | struct sljit_jump *jump2; | |
2810 | #endif | |
2811 | ||
2812 | SLJIT_UNUSED_ARG(update_str_ptr); | |
2813 | ||
2814 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2815 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
2816 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | |
2817 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
2818 | if (common->utf) | |
2819 | { | |
2820 | /* This can be an extra read in some situations, but hopefully | |
2821 | it is needed in most cases. */ | |
2822 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
2823 | jump = CMP(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0xc0); | |
2824 | if (!update_str_ptr) | |
2825 | { | |
2826 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2827 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2828 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | |
2829 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | |
2830 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | |
2831 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0); | |
2832 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | |
2833 | jump2 = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 255); | |
2834 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
2835 | JUMPHERE(jump2); | |
2836 | } | |
2837 | else | |
2838 | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); | |
2839 | JUMPHERE(jump); | |
2840 | return; | |
2841 | } | |
2842 | #endif /* SUPPORT_UTF && COMPILE_PCRE8 */ | |
2843 | ||
2844 | #if !defined COMPILE_PCRE8 | |
2845 | /* The ctypes array contains only 256 values. */ | /* The ctypes array contains only 256 values. */ |
2846 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2847 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2848 | #endif | #endif |
2849 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2850 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if !defined COMPILE_PCRE8 |
2851 | JUMPHERE(jump); | JUMPHERE(jump); |
2852 | #endif | #endif |
2853 | ||
2854 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
2855 | if (common->utf && update_str_ptr) | |
2856 | { | |
2857 | /* Skip low surrogate if necessary. */ | |
2858 | OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xd800); | |
2859 | jump = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1); | |
2860 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2861 | JUMPHERE(jump); | |
2862 | } | |
2863 | #endif /* SUPPORT_UTF && COMPILE_PCRE16 */ | |
2864 | } | } |
2865 | ||
2866 | static void skip_char_back(compiler_common *common) | static void skip_char_back(compiler_common *common) |
# | Line 2303 if (common->utf) | Line 2877 if (common->utf) |
2877 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); |
2878 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
2879 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0); |
2880 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); | CMPTO(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); |
2881 | return; | return; |
2882 | } | } |
2883 | #elif defined COMPILE_PCRE16 | #elif defined COMPILE_PCRE16 |
# | Line 2314 if (common->utf) | Line 2888 if (common->utf) |
2888 | /* Skip low surrogate if necessary. */ | /* Skip low surrogate if necessary. */ |
2889 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
2890 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00); |
2891 | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_EQUAL); |
2892 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
2893 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2894 | return; | return; |
# | Line 2324 if (common->utf) | Line 2898 if (common->utf) |
2898 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
2899 | } | } |
2900 | ||
2901 | static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpiftrue) | static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpifmatch) |
2902 | { | { |
2903 | /* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */ | /* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */ |
2904 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2905 | struct sljit_jump *jump; | |
2906 | ||
2907 | if (nltype == NLTYPE_ANY) | if (nltype == NLTYPE_ANY) |
2908 | { | { |
2909 | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); |
2910 | add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(jumpifmatch ? SLJIT_NOT_ZERO : SLJIT_ZERO)); |
2911 | } | } |
2912 | else if (nltype == NLTYPE_ANYCRLF) | else if (nltype == NLTYPE_ANYCRLF) |
2913 | { | { |
2914 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR); | if (jumpifmatch) |
2915 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | { |
2916 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); | add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR)); |
2917 | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); | add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); |
2918 | add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | } |
2919 | else | |
2920 | { | |
2921 | jump = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); | |
2922 | add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); | |
2923 | JUMPHERE(jump); | |
2924 | } | |
2925 | } | } |
2926 | else | else |
2927 | { | { |
2928 | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); |
2929 | add_jump(compiler, backtracks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); | add_jump(compiler, backtracks, CMP(jumpifmatch ? SLJIT_EQUAL : SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); |
2930 | } | } |
2931 | } | } |
2932 | ||
# | Line 2355 else | Line 2936 else |
2936 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2937 | { | { |
2938 | /* Fast decoding a UTF-8 character. TMP1 contains the first byte | /* Fast decoding a UTF-8 character. TMP1 contains the first byte |
2939 | of the character (>= 0xc0). Return char value in TMP1, length - 1 in TMP2. */ | of the character (>= 0xc0). Return char value in TMP1, length in TMP2. */ |
2940 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2941 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2942 | ||
2943 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2944 | /* Searching for the first zero. */ | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
2945 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); |
jump = JUMP(SLJIT_C_NOT_ZERO); | ||
/* Two byte sequence. */ | ||
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | ||
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1f); | ||
2946 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); |
2947 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
2948 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2949 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2950 | /* Searching for the first zero. */ | |
2951 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); | |
2952 | jump = JUMP(SLJIT_NOT_ZERO); | |
2953 | /* Two byte sequence. */ | |
2954 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2955 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2)); | |
2956 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
JUMPHERE(jump); | ||
2957 | ||
2958 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10); | JUMPHERE(jump); |
jump = JUMP(SLJIT_C_NOT_ZERO); | ||
/* Three byte sequence. */ | ||
2959 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
2960 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0f); | OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800); |
2961 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 12); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); |
2962 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | ||
2963 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2964 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); | |
2965 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10000); | |
2966 | jump = JUMP(SLJIT_NOT_ZERO); | |
2967 | /* Three byte sequence. */ | |
2968 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
2969 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3)); |
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | ||
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2)); | ||
2970 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
JUMPHERE(jump); | ||
2971 | ||
2972 | /* Four byte sequence. */ | /* Four byte sequence. */ |
2973 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | JUMPHERE(jump); |
2974 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x07); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); |
2975 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 18); | OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); |
2976 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2977 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3)); | |
2978 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12); | ||
2979 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2980 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(4)); |
2981 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | |
2982 | } | |
2983 | ||
2984 | static void do_utfreadchar16(compiler_common *common) | |
2985 | { | |
2986 | /* Fast decoding a UTF-8 character. TMP1 contains the first byte | |
2987 | of the character (>= 0xc0). Return value in TMP1. */ | |
2988 | DEFINE_COMPILER; | |
2989 | struct sljit_jump *jump; | |
2990 | ||
2991 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); | |
2992 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2993 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | |
2994 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2995 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | ||
2996 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2997 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(3)); | |
2998 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3)); | /* Searching for the first zero. */ |
2999 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); | |
3000 | jump = JUMP(SLJIT_NOT_ZERO); | |
3001 | /* Two byte sequence. */ | |
3002 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
3003 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | |
3004 | ||
3005 | JUMPHERE(jump); | |
3006 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x400); | |
3007 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_NOT_ZERO); | |
3008 | /* This code runs only in 8 bit mode. No need to shift the value. */ | |
3009 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | |
3010 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
3011 | OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800); | |
3012 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
3013 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
3014 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
3015 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3)); | /* Three byte sequence. */ |
3016 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | |
3017 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3018 | } | } |
3019 | ||
# | Line 2421 struct sljit_jump *compare; | Line 3028 struct sljit_jump *compare; |
3028 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3029 | ||
3030 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x20); |
3031 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_NOT_ZERO); |
3032 | /* Two byte sequence. */ | /* Two byte sequence. */ |
3033 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
3034 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
3035 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f); |
3036 | /* The upper 5 bits are known at this point. */ | |
3037 | compare = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0x3); | |
3038 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); |
3039 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); |
3040 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0); | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0); |
compare = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | ||
3041 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
3042 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3043 | ||
3044 | JUMPHERE(compare); | JUMPHERE(compare); |
3045 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
3046 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
JUMPHERE(jump); | ||
3047 | ||
3048 | /* We only have types for characters less than 256. */ | /* We only have types for characters less than 256. */ |
OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0); | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | ||
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | ||
sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | ||
} | ||
#elif defined COMPILE_PCRE16 | ||
static void do_utfreadchar(compiler_common *common) | ||
{ | ||
/* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char | ||
of the character (>= 0xd800). Return char value in TMP1, length - 1 in TMP2. */ | ||
DEFINE_COMPILER; | ||
struct sljit_jump *jump; | ||
sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); | ||
jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00); | ||
/* Do nothing, only return. */ | ||
sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | ||
3049 | JUMPHERE(jump); | JUMPHERE(jump); |
3050 | /* Combine two 16 bit characters. */ | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0); |
3051 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
3052 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff); | ||
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 10); | ||
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3ff); | ||
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | ||
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); | ||
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); | ||
3053 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3054 | } | } |
3055 | ||
3056 | #endif /* COMPILE_PCRE[8|16] */ | #endif /* COMPILE_PCRE8 */ |
3057 | ||
3058 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
3059 | ||
# | Line 2533 if (firstline) | Line 3114 if (firstline) |
3114 | { | { |
3115 | mainloop = LABEL(); | mainloop = LABEL(); |
3116 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
3117 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3118 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
3119 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
3120 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); | CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); |
3121 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); | CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); |
3122 | JUMPHERE(end); | JUMPHERE(end); |
3123 | OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_SP), common->first_line_end, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
3124 | } | } |
3125 | else | else |
3126 | { | { |
3127 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3128 | mainloop = LABEL(); | mainloop = LABEL(); |
3129 | /* Continual stores does not cause data dependency. */ | /* Continual stores does not cause data dependency. */ |
3130 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->first_line_end, STR_PTR, 0); |
3131 | read_char(common); | read_char_range(common, common->nlmin, common->nlmax, TRUE); |
3132 | check_newlinechar(common, common->nltype, &newline, TRUE); | check_newlinechar(common, common->nltype, &newline, TRUE); |
3133 | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); | CMPTO(SLJIT_LESS, STR_PTR, 0, STR_END, 0, mainloop); |
3134 | JUMPHERE(end); | JUMPHERE(end); |
3135 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->first_line_end, STR_PTR, 0); |
3136 | set_jumps(newline, LABEL()); | set_jumps(newline, LABEL()); |
3137 | } | } |
3138 | ||
# | Line 2564 if (newlinecheck) | Line 3145 if (newlinecheck) |
3145 | { | { |
3146 | newlinelabel = LABEL(); | newlinelabel = LABEL(); |
3147 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
3148 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3149 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3150 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, common->newline & 0xff); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, common->newline & 0xff); |
3151 | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_EQUAL); |
3152 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3153 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
3154 | #endif | #endif |
# | Line 2587 if (readuchar) | Line 3168 if (readuchar) |
3168 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3169 | ||
3170 | if (newlinecheck) | if (newlinecheck) |
3171 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); | CMPTO(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); |
3172 | ||
3173 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
3174 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
3175 | #if defined COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3176 | if (common->utf) | if (common->utf) |
3177 | { | { |
3178 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | singlechar = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
3179 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); |
3180 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3181 | JUMPHERE(singlechar); | JUMPHERE(singlechar); |
# | Line 2602 if (common->utf) | Line 3183 if (common->utf) |
3183 | #elif defined COMPILE_PCRE16 | #elif defined COMPILE_PCRE16 |
3184 | if (common->utf) | if (common->utf) |
3185 | { | { |
3186 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | singlechar = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
3187 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
3188 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); |
3189 | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_EQUAL); |
3190 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
3191 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3192 | JUMPHERE(singlechar); | JUMPHERE(singlechar); |
# | Line 2623 if (newlinecheck) | Line 3204 if (newlinecheck) |
3204 | return mainloop; | return mainloop; |
3205 | } | } |
3206 | ||
3207 | #define MAX_N_CHARS 3 | #define MAX_N_CHARS 16 |
3208 | #define MAX_N_BYTES 8 | |
3209 | ||
3210 | static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline) | static SLJIT_INLINE void add_prefix_byte(pcre_uint8 byte, pcre_uint8 *bytes) |
3211 | { | { |
3212 | DEFINE_COMPILER; | pcre_uint8 len = bytes[0]; |
3213 | struct sljit_label *start; | int i; |
struct sljit_jump *quit; | ||
pcre_uint32 chars[MAX_N_CHARS * 2]; | ||
pcre_uchar *cc = common->start + 1 + LINK_SIZE; | ||
int location = 0; | ||
pcre_int32 len, c, bit, caseless; | ||
int must_stop; | ||
3214 | ||
3215 | /* We do not support alternatives now. */ | if (len == 255) |
3216 | if (*(common->start + GET(common->start, 1)) == OP_ALT) | return; |
3217 | return FALSE; | |
3218 | if (len == 0) | |
3219 | { | |
3220 | bytes[0] = 1; | |
3221 | bytes[1] = byte; | |
3222 | return; | |
3223 | } | |
3224 | ||
3225 | for (i = len; i > 0; i--) | |
3226 | if (bytes[i] == byte) | |
3227 | return; | |
3228 | ||
3229 | if (len >= MAX_N_BYTES - 1) | |
3230 | { | |
3231 | bytes[0] = 255; | |
3232 | return; | |
3233 | } | |
3234 | ||
3235 | len++; | |
3236 | bytes[len] = byte; | |
3237 | bytes[0] = len; | |
3238 | } | |
3239 | ||
3240 | static int scan_prefix(compiler_common *common, pcre_uchar *cc, pcre_uint32 *chars, pcre_uint8 *bytes, int max_chars, pcre_uint32 *rec_count) | |
3241 | { | |
3242 | /* Recursive function, which scans prefix literals. */ | |
3243 | BOOL last, any, caseless; | |
3244 | int len, repeat, len_save, consumed = 0; | |
3245 | pcre_uint32 chr, mask; | |
3246 | pcre_uchar *alternative, *cc_save, *oc; | |
3247 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
3248 | pcre_uchar othercase[8]; | |
3249 | #elif defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
3250 | pcre_uchar othercase[2]; | |
3251 | #else | |
3252 | pcre_uchar othercase[1]; | |
3253 | #endif | |
3254 | ||
3255 | repeat = 1; | |
3256 | while (TRUE) | while (TRUE) |
3257 | { | { |
3258 | caseless = 0; | if (*rec_count == 0) |
3259 | must_stop = 1; | return 0; |
3260 | switch(*cc) | (*rec_count)--; |
{ | ||
case OP_CHAR: | ||
must_stop = 0; | ||
cc++; | ||
break; | ||
3261 | ||
3262 | last = TRUE; | |
3263 | any = FALSE; | |
3264 | caseless = FALSE; | |
3265 | ||
3266 | switch (*cc) | |
3267 | { | |
3268 | case OP_CHARI: | case OP_CHARI: |
3269 | caseless = 1; | caseless = TRUE; |
3270 | must_stop = 0; | case OP_CHAR: |
3271 | last = FALSE; | |
3272 | cc++; | cc++; |
3273 | break; | break; |
3274 | ||
# | Line 2672 while (TRUE) | Line 3287 while (TRUE) |
3287 | cc++; | cc++; |
3288 | continue; | continue; |
3289 | ||
3290 | case OP_ASSERT: | |
3291 | case OP_ASSERT_NOT: | |
3292 | case OP_ASSERTBACK: | |
3293 | case OP_ASSERTBACK_NOT: | |
3294 | cc = bracketend(cc); | |
3295 | continue; | |
3296 | ||
3297 | case OP_PLUSI: | |
3298 | case OP_MINPLUSI: | |
3299 | case OP_POSPLUSI: | |
3300 | caseless = TRUE; | |
3301 | case OP_PLUS: | case OP_PLUS: |
3302 | case OP_MINPLUS: | case OP_MINPLUS: |
3303 | case OP_POSPLUS: | case OP_POSPLUS: |
3304 | cc++; | cc++; |
3305 | break; | break; |
3306 | ||
3307 | case OP_EXACTI: | |
3308 | caseless = TRUE; | |
3309 | case OP_EXACT: | case OP_EXACT: |
3310 | repeat = GET2(cc, 1); | |
3311 | last = FALSE; | |
3312 | cc += 1 + IMM2_SIZE; | cc += 1 + IMM2_SIZE; |
3313 | break; | break; |
3314 | ||
3315 | case OP_PLUSI: | case OP_QUERYI: |
3316 | case OP_MINPLUSI: | case OP_MINQUERYI: |
3317 | case OP_POSPLUSI: | case OP_POSQUERYI: |
3318 | caseless = 1; | caseless = TRUE; |
3319 | case OP_QUERY: | |
3320 | case OP_MINQUERY: | |
3321 | case OP_POSQUERY: | |
3322 | len = 1; | |
3323 | cc++; | cc++; |
3324 | #ifdef SUPPORT_UTF | |
3325 | if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc); | |
3326 | #endif | |
3327 | max_chars = scan_prefix(common, cc + len, chars, bytes, max_chars, rec_count); | |
3328 | if (max_chars == 0) | |
3329 | return consumed; | |
3330 | last = FALSE; | |
3331 | break; | break; |
3332 | ||
3333 | case OP_EXACTI: | case OP_KET: |
3334 | caseless = 1; | cc += 1 + LINK_SIZE; |
3335 | cc += 1 + IMM2_SIZE; | continue; |
break; | ||
3336 | ||
3337 | default: | case OP_ALT: |
3338 | must_stop = 2; | cc += GET(cc, 1); |
3339 | break; | continue; |
} | ||
3340 | ||
3341 | if (must_stop == 2) | case OP_ONCE: |
3342 | break; | case OP_ONCE_NC: |
3343 | case OP_BRA: | |
3344 | case OP_BRAPOS: | |
3345 | case OP_CBRA: | |
3346 | case OP_CBRAPOS: | |
3347 | alternative = cc + GET(cc, 1); | |
3348 | while (*alternative == OP_ALT) | |
3349 | { | |
3350 | max_chars = scan_prefix(common, alternative + 1 + LINK_SIZE, chars, bytes, max_chars, rec_count); | |
3351 | if (max_chars == 0) | |
3352 | return consumed; | |
3353 | alternative += GET(alternative, 1); | |
3354 | } | |
3355 | ||
3356 | len = 1; | if (*cc == OP_CBRA || *cc == OP_CBRAPOS) |
3357 | #ifdef SUPPORT_UTF | cc += IMM2_SIZE; |
3358 | if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]); | cc += 1 + LINK_SIZE; |
3359 | continue; | |
3360 | ||
3361 | case OP_CLASS: | |
3362 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
3363 | if (common->utf && !is_char7_bitset((const pcre_uint8 *)(cc + 1), FALSE)) return consumed; | |
3364 | #endif | #endif |
3365 | any = TRUE; | |
3366 | cc += 1 + 32 / sizeof(pcre_uchar); | |
3367 | break; | |
3368 | ||
3369 | if (caseless && char_has_othercase(common, cc)) | case OP_NCLASS: |
3370 | { | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
3371 | caseless = char_get_othercase_bit(common, cc); | if (common->utf) return consumed; |
if (caseless == 0) | ||
return FALSE; | ||
#ifdef COMPILE_PCRE8 | ||
caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8)); | ||
#else | ||
if ((caseless & 0x100) != 0) | ||
caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9)); | ||
else | ||
caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9)); | ||
3372 | #endif | #endif |
3373 | } | any = TRUE; |
3374 | else | cc += 1 + 32 / sizeof(pcre_uchar); |
3375 | caseless = 0; | break; |
3376 | ||
3377 | while (len > 0 && location < MAX_N_CHARS * 2) | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
3378 | { | case OP_XCLASS: |
3379 | c = *cc; | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
3380 | bit = 0; | if (common->utf) return consumed; |
3381 | if (len == (caseless & 0xff)) | #endif |
3382 | { | any = TRUE; |
3383 | bit = caseless >> 8; | cc += GET(cc, 1); |
3384 | c |= bit; | break; |
3385 | } | #endif |
3386 | ||
3387 | case OP_DIGIT: | |
3388 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
3389 | if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_digit, FALSE)) | |
3390 | return consumed; | |
3391 | #endif | |
3392 | any = TRUE; | |
3393 | cc++; | |
3394 | break; | |
3395 | ||
3396 | case OP_WHITESPACE: | |
3397 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
3398 | if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_space, FALSE)) | |
3399 | return consumed; | |
3400 | #endif | |
3401 | any = TRUE; | |
3402 | cc++; | |
3403 | break; | |
3404 | ||
3405 | chars[location] = c; | case OP_WORDCHAR: |
3406 | chars[location + 1] = bit; | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 |
3407 | if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_word, FALSE)) | |
3408 | return consumed; | |
3409 | #endif | |
3410 | any = TRUE; | |
3411 | cc++; | |
3412 | break; | |
3413 | ||
3414 | len--; | case OP_NOT: |
3415 | location += 2; | case OP_NOTI: |
3416 | cc++; | |
3417 | /* Fall through. */ | |
3418 | case OP_NOT_DIGIT: | |
3419 | case OP_NOT_WHITESPACE: | |
3420 | case OP_NOT_WORDCHAR: | |
3421 | case OP_ANY: | |
3422 | case OP_ALLANY: | |
3423 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
3424 | if (common->utf) return consumed; | |
3425 | #endif | |
3426 | any = TRUE; | |
3427 | cc++; | cc++; |
3428 | break; | |
3429 | ||
3430 | #ifdef SUPPORT_UCP | |
3431 | case OP_NOTPROP: | |
3432 | case OP_PROP: | |
3433 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
3434 | if (common->utf) return consumed; | |
3435 | #endif | |
3436 | any = TRUE; | |
3437 | cc += 1 + 2; | |
3438 | break; | |
3439 | #endif | |
3440 | ||
3441 | case OP_TYPEEXACT: | |
3442 | repeat = GET2(cc, 1); | |
3443 | cc += 1 + IMM2_SIZE; | |
3444 | continue; | |
3445 | ||
3446 | case OP_NOTEXACT: | |
3447 | case OP_NOTEXACTI: | |
3448 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
3449 | if (common->utf) return consumed; | |
3450 | #endif | |
3451 | any = TRUE; | |
3452 | repeat = GET2(cc, 1); | |
3453 | cc += 1 + IMM2_SIZE + 1; | |
3454 | break; | |
3455 | ||
3456 | default: | |
3457 | return consumed; | |
3458 | } | |
3459 | ||
3460 | if (any) | |
3461 | { | |
3462 | #if defined COMPILE_PCRE8 | |
3463 | mask = 0xff; | |
3464 | #elif defined COMPILE_PCRE16 | |
3465 | mask = 0xffff; | |
3466 | #elif defined COMPILE_PCRE32 | |
3467 | mask = 0xffffffff; | |
3468 | #else | |
3469 | SLJIT_ASSERT_STOP(); | |
3470 | #endif | |
3471 | ||
3472 | do | |
3473 | { | |
3474 | chars[0] = mask; | |
3475 | chars[1] = mask; | |
3476 | bytes[0] = 255; | |
3477 | ||
3478 | consumed++; | |
3479 | if (--max_chars == 0) | |
3480 | return consumed; | |
3481 | chars += 2; | |
3482 | bytes += MAX_N_BYTES; | |
3483 | } | |
3484 | while (--repeat > 0); | |
3485 | ||
3486 | repeat = 1; | |
3487 | continue; | |
3488 | } | |
3489 | ||
3490 | len = 1; | |
3491 | #ifdef SUPPORT_UTF | |
3492 | if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc); | |
3493 | #endif | |
3494 | ||
3495 | if (caseless && char_has_othercase(common, cc)) | |
3496 | { | |
3497 | #ifdef SUPPORT_UTF | |
3498 | if (common->utf) | |
3499 | { | |
3500 | GETCHAR(chr, cc); | |
3501 | if ((int)PRIV(ord2utf)(char_othercase(common, chr), othercase) != len) | |
3502 | return consumed; | |
3503 | } | |
3504 | else | |
3505 | #endif | |
3506 | { | |
3507 | chr = *cc; | |
3508 | othercase[0] = TABLE_GET(chr, common->fcc, chr); | |
3509 | } | |
3510 | } | |
3511 | else | |
3512 | caseless = FALSE; | |
3513 | ||
3514 | len_save = len; | |
3515 | cc_save = cc; | |
3516 | while (TRUE) | |
3517 | { | |
3518 | oc = othercase; | |
3519 | do | |
3520 | { | |
3521 | chr = *cc; | |
3522 | #ifdef COMPILE_PCRE32 | |
3523 | if (SLJIT_UNLIKELY(chr == NOTACHAR)) | |
3524 | return consumed; | |
3525 | #endif | |
3526 | add_prefix_byte((pcre_uint8)chr, bytes); | |
3527 | ||
3528 | mask = 0; | |
3529 | if (caseless) | |
3530 | { | |
3531 | add_prefix_byte((pcre_uint8)*oc, bytes); | |
3532 | mask = *cc ^ *oc; | |
3533 | chr |= mask; | |
3534 | } | |
3535 | ||
3536 | #ifdef COMPILE_PCRE32 | |
3537 | if (chars[0] == NOTACHAR && chars[1] == 0) | |
3538 | #else | |
3539 | if (chars[0] == NOTACHAR) | |
3540 | #endif | |
3541 | { | |
3542 | chars[0] = chr; | |
3543 | chars[1] = mask; | |
3544 | } | |
3545 | else | |
3546 | { | |
3547 | mask |= chars[0] ^ chr; | |
3548 | chr |= mask; | |
3549 | chars[0] = chr; | |
3550 | chars[1] |= mask; | |
3551 | } | |
3552 | ||
3553 | len--; | |
3554 | consumed++; | |
3555 | if (--max_chars == 0) | |
3556 | return consumed; | |
3557 | chars += 2; | |
3558 | bytes += MAX_N_BYTES; | |
3559 | cc++; | |
3560 | oc++; | |
3561 | } | |
3562 | while (len > 0); | |
3563 | ||
3564 | if (--repeat == 0) | |
3565 | break; | |
3566 | ||
3567 | len = len_save; | |
3568 | cc = cc_save; | |
3569 | } | |
3570 | ||
3571 | repeat = 1; | |
3572 | if (last) | |
3573 | return consumed; | |
3574 | } | |
3575 | } | |
3576 | ||
3577 | static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline) | |
3578 | { | |
3579 | DEFINE_COMPILER; | |
3580 | struct sljit_label *start; | |
3581 | struct sljit_jump *quit; | |
3582 | pcre_uint32 chars[MAX_N_CHARS * 2]; | |
3583 | pcre_uint8 bytes[MAX_N_CHARS * MAX_N_BYTES]; | |
3584 | pcre_uint8 ones[MAX_N_CHARS]; | |
3585 | int offsets[3]; | |
3586 | pcre_uint32 mask; | |
3587 | pcre_uint8 *byte_set, *byte_set_end; | |
3588 | int i, max, from; | |
3589 | int range_right = -1, range_len = 3 - 1; | |
3590 | sljit_ub *update_table = NULL; | |
3591 | BOOL in_range; | |
3592 | pcre_uint32 rec_count; | |
3593 | ||
3594 | for (i = 0; i < MAX_N_CHARS; i++) | |
3595 | { | |
3596 | chars[i << 1] = NOTACHAR; | |
3597 | chars[(i << 1) + 1] = 0; | |
3598 | bytes[i * MAX_N_BYTES] = 0; | |
3599 | } | |
3600 | ||
3601 | rec_count = 10000; | |
3602 | max = scan_prefix(common, common->start, chars, bytes, MAX_N_CHARS, &rec_count); | |
3603 | ||
3604 | if (max <= 1) | |
3605 | return FALSE; | |
3606 | ||
3607 | for (i = 0; i < max; i++) | |
3608 | { | |
3609 | mask = chars[(i << 1) + 1]; | |
3610 | ones[i] = ones_in_half_byte[mask & 0xf]; | |
3611 | mask >>= 4; | |
3612 | while (mask != 0) | |
3613 | { | |
3614 | ones[i] += ones_in_half_byte[mask & 0xf]; | |
3615 | mask >>= 4; | |
3616 | } | |
3617 | } | |
3618 | ||
3619 | in_range = FALSE; | |
3620 | from = 0; /* Prevent compiler "uninitialized" warning */ | |
3621 | for (i = 0; i <= max; i++) | |
3622 | { | |
3623 | if (in_range && (i - from) > range_len && (bytes[(i - 1) * MAX_N_BYTES] <= 4)) | |
3624 | { | |
3625 | range_len = i - from; | |
3626 | range_right = i - 1; | |
3627 | } | |
3628 | ||
3629 | if (i < max && bytes[i * MAX_N_BYTES] < 255) | |
3630 | { | |
3631 | if (!in_range) | |
3632 | { | |
3633 | in_range = TRUE; | |
3634 | from = i; | |
3635 | } | |
3636 | } | |
3637 | else if (in_range) | |
3638 | in_range = FALSE; | |
3639 | } | |
3640 | ||
3641 | if (range_right >= 0) | |
3642 | { | |
3643 | update_table = (sljit_ub *)allocate_read_only_data(common, 256); | |
3644 | if (update_table == NULL) | |
3645 | return TRUE; | |
3646 | memset(update_table, IN_UCHARS(range_len), 256); | |
3647 | ||
3648 | for (i = 0; i < range_len; i++) | |
3649 | { | |
3650 | byte_set = bytes + ((range_right - i) * MAX_N_BYTES); | |
3651 | SLJIT_ASSERT(byte_set[0] > 0 && byte_set[0] < 255); | |
3652 | byte_set_end = byte_set + byte_set[0]; | |
3653 | byte_set++; | |
3654 | while (byte_set <= byte_set_end) | |
3655 | { | |
3656 | if (update_table[*byte_set] > IN_UCHARS(i)) | |
3657 | update_table[*byte_set] = IN_UCHARS(i); | |
3658 | byte_set++; | |
3659 | } | |
3660 | } | } |
3661 | } | |
3662 | ||
3663 | if (location >= MAX_N_CHARS * 2 || must_stop != 0) | offsets[0] = -1; |
3664 | /* Scan forward. */ | |
3665 | for (i = 0; i < max; i++) | |
3666 | if (ones[i] <= 2) { | |
3667 | offsets[0] = i; | |
3668 | break; | break; |
3669 | } | } |
3670 | ||
3671 | /* At least two characters are required. */ | if (offsets[0] < 0 && range_right < 0) |
3672 | if (location < 2 * 2) | return FALSE; |
3673 | ||
3674 | if (offsets[0] >= 0) | |
3675 | { | |
3676 | /* Scan backward. */ | |
3677 | offsets[1] = -1; | |
3678 | for (i = max - 1; i > offsets[0]; i--) | |
3679 | if (ones[i] <= 2 && i != range_right) | |
3680 | { | |
3681 | offsets[1] = i; | |
3682 | break; | |
3683 | } | |
3684 | ||
3685 | /* This case is handled better by fast_forward_first_char. */ | |
3686 | if (offsets[1] == -1 && offsets[0] == 0 && range_right < 0) | |
3687 | return FALSE; | return FALSE; |
3688 | ||
3689 | offsets[2] = -1; | |
3690 | /* We only search for a middle character if there is no range check. */ | |
3691 | if (offsets[1] >= 0 && range_right == -1) | |
3692 | { | |
3693 | /* Scan from middle. */ | |
3694 | for (i = (offsets[0] + offsets[1]) / 2 + 1; i < offsets[1]; i++) | |
3695 | if (ones[i] <= 2) | |
3696 | { | |
3697 | offsets[2] = i; | |
3698 | break; | |
3699 | } | |
3700 | ||
3701 | if (offsets[2] == -1) | |
3702 | { | |
3703 | for (i = (offsets[0] + offsets[1]) / 2; i > offsets[0]; i--) | |
3704 | if (ones[i] <= 2) | |
3705 | { | |
3706 | offsets[2] = i; | |
3707 | break; | |
3708 | } | |
3709 | } | |
3710 | } | |
3711 | ||
3712 | SLJIT_ASSERT(offsets[1] == -1 || (offsets[0] < offsets[1])); | |
3713 | SLJIT_ASSERT(offsets[2] == -1 || (offsets[0] < offsets[2] && offsets[1] > offsets[2])); | |
3714 | ||
3715 | chars[0] = chars[offsets[0] << 1]; | |
3716 | chars[1] = chars[(offsets[0] << 1) + 1]; | |
3717 | if (offsets[2] >= 0) | |
3718 | { | |
3719 | chars[2] = chars[offsets[2] << 1]; | |
3720 | chars[3] = chars[(offsets[2] << 1) + 1]; | |
3721 | } | |
3722 | if (offsets[1] >= 0) | |
3723 | { | |
3724 | chars[4] = chars[offsets[1] << 1]; | |
3725 | chars[5] = chars[(offsets[1] << 1) + 1]; | |
3726 | } | |
3727 | } | |
3728 | ||
3729 | max -= 1; | |
3730 | if (firstline) | if (firstline) |
3731 | { | { |
3732 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3733 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->first_line_end); | |
3734 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
3735 | OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max)); |
3736 | quit = CMP(SLJIT_LESS_EQUAL, STR_END, 0, TMP1, 0); | |
3737 | OP1(SLJIT_MOV, STR_END, 0, TMP1, 0); | |
3738 | JUMPHERE(quit); | |
3739 | } | } |
3740 | else | else |
3741 | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max)); |
3742 | ||
3743 | #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32) | |
3744 | if (range_right >= 0) | |
3745 | OP1(SLJIT_MOV, RETURN_ADDR, 0, SLJIT_IMM, (sljit_sw)update_table); | |
3746 | #endif | |
3747 | ||
3748 | start = LABEL(); | start = LABEL(); |
3749 | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3750 | ||
3751 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | SLJIT_ASSERT(range_right >= 0 || offsets[0] >= 0); |
3752 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
3753 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | if (range_right >= 0) |
3754 | if (chars[1] != 0) | { |
3755 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]); | #if defined COMPILE_PCRE8 || (defined SLJIT_LITTLE_ENDIAN && SLJIT_LITTLE_ENDIAN) |
3756 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right)); |
3757 | if (location > 2 * 2) | #else |
3758 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right + 1) - 1); |
3759 | if (chars[3] != 0) | #endif |
3760 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]); | |
3761 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start); | #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32) |
3762 | if (location > 2 * 2) | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(RETURN_ADDR, TMP1), 0); |
3763 | { | #else |
3764 | if (chars[5] != 0) | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)update_table); |
3765 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]); | #endif |
3766 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3767 | CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, start); | |
3768 | } | |
3769 | ||
3770 | if (offsets[0] >= 0) | |
3771 | { | |
3772 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[0])); | |
3773 | if (offsets[1] >= 0) | |
3774 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[1])); | |
3775 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
3776 | ||
3777 | if (chars[1] != 0) | |
3778 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]); | |
3779 | CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start); | |
3780 | if (offsets[2] >= 0) | |
3781 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[2] - 1)); | |
3782 | ||
3783 | if (offsets[1] >= 0) | |
3784 | { | |
3785 | if (chars[5] != 0) | |
3786 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[5]); | |
3787 | CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[4], start); | |
3788 | } | |
3789 | ||
3790 | if (offsets[2] >= 0) | |
3791 | { | |
3792 | if (chars[3] != 0) | |
3793 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[3]); | |
3794 | CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[2], start); | |
3795 | } | |
3796 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
3797 | } | } |
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | ||
3798 | ||
3799 | JUMPHERE(quit); | JUMPHERE(quit); |
3800 | ||
3801 | if (firstline) | if (firstline) |
3802 | { | |
3803 | if (range_right >= 0) | |
3804 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->first_line_end); | |
3805 | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
3806 | if (range_right >= 0) | |
3807 | { | |
3808 | quit = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP1, 0); | |
3809 | OP1(SLJIT_MOV, STR_PTR, 0, TMP1, 0); | |
3810 | JUMPHERE(quit); | |
3811 | } | |
3812 | } | |
3813 | else | else |
3814 | OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); | OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max)); |
3815 | return TRUE; | return TRUE; |
3816 | } | } |
3817 | ||
3818 | #undef MAX_N_CHARS | #undef MAX_N_CHARS |
3819 | #undef MAX_N_BYTES | |
3820 | ||
3821 | static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline) | static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline) |
3822 | { | { |
# | Line 2804 if (firstline) | Line 3830 if (firstline) |
3830 | { | { |
3831 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3832 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
3833 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), common->first_line_end); |
3834 | } | } |
3835 | ||
3836 | start = LABEL(); | start = LABEL(); |
3837 | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3838 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3839 | ||
3840 | oc = first_char; | oc = first_char; |
# | Line 2821 if (caseless) | Line 3847 if (caseless) |
3847 | #endif | #endif |
3848 | } | } |
3849 | if (first_char == oc) | if (first_char == oc) |
3850 | found = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, first_char); | found = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, first_char); |
3851 | else | else |
3852 | { | { |
3853 | bit = first_char ^ oc; | bit = first_char ^ oc; |
3854 | if (is_powerof2(bit)) | if (is_powerof2(bit)) |
3855 | { | { |
3856 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); |
3857 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, first_char | bit); | found = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, first_char | bit); |
3858 | } | } |
3859 | else | else |
3860 | { | { |
3861 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, first_char); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, first_char); |
3862 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_EQUAL); |
3863 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); |
3864 | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_EQUAL); |
3865 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_NOT_ZERO); |
3866 | } | } |
3867 | } | } |
3868 | ||
# | Line 2864 if (firstline) | Line 3890 if (firstline) |
3890 | { | { |
3891 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3892 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
3893 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), common->first_line_end); |
3894 | } | } |
3895 | ||
3896 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
3897 | { | { |
3898 | lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3899 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
3900 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); |
3901 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); |
3902 | firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP2, 0); | firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0); |
3903 | ||
3904 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); |
3905 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0); |
3906 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_GREATER_EQUAL); |
3907 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3908 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); |
3909 | #endif | #endif |
# | Line 2885 if (common->nltype == NLTYPE_FIXED && co | Line 3911 if (common->nltype == NLTYPE_FIXED && co |
3911 | ||
3912 | loop = LABEL(); | loop = LABEL(); |
3913 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
3914 | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3915 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); |
3916 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
3917 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); | CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); |
3918 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); | CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); |
3919 | ||
3920 | JUMPHERE(quit); | JUMPHERE(quit); |
3921 | JUMPHERE(firstchar); | JUMPHERE(firstchar); |
3922 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
3923 | ||
3924 | if (firstline) | if (firstline) |
3925 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
3926 | return; | return; |
3927 | } | } |
3928 | ||
3929 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
3930 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); |
3931 | firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP2, 0); | firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0); |
3932 | skip_char_back(common); | skip_char_back(common); |
3933 | ||
3934 | loop = LABEL(); | loop = LABEL(); |
3935 | read_char(common); | common->ff_newline_shortcut = loop; |
3936 | lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
3937 | read_char_range(common, common->nlmin, common->nlmax, TRUE); | |
3938 | lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
3939 | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) |
3940 | foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); | foundcr = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); |
3941 | check_newlinechar(common, common->nltype, &newline, FALSE); | check_newlinechar(common, common->nltype, &newline, FALSE); |
3942 | set_jumps(newline, loop); | set_jumps(newline, loop); |
3943 | ||
# | Line 2917 if (common->nltype == NLTYPE_ANY || comm | Line 3945 if (common->nltype == NLTYPE_ANY || comm |
3945 | { | { |
3946 | quit = JUMP(SLJIT_JUMP); | quit = JUMP(SLJIT_JUMP); |
3947 | JUMPHERE(foundcr); | JUMPHERE(foundcr); |
3948 | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | notfoundnl = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3949 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3950 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); |
3951 | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_EQUAL); |
3952 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3953 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
3954 | #endif | #endif |
# | Line 2935 if (firstline) | Line 3963 if (firstline) |
3963 | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
3964 | } | } |
3965 | ||
3966 | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks); | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks); |
3967 | ||
3968 | static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline) | static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, pcre_uint8 *start_bits, BOOL firstline) |
3969 | { | { |
3970 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3971 | struct sljit_label *start; | struct sljit_label *start; |
3972 | struct sljit_jump *quit; | struct sljit_jump *quit; |
3973 | struct sljit_jump *found = NULL; | struct sljit_jump *found = NULL; |
3974 | jump_list *matches = NULL; | jump_list *matches = NULL; |
pcre_uint8 inverted_start_bits[32]; | ||
int i; | ||
3975 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3976 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3977 | #endif | #endif |
3978 | ||
for (i = 0; i < 32; ++i) | ||
inverted_start_bits[i] = ~(((pcre_uint8*)start_bits)[i]); | ||
3979 | if (firstline) | if (firstline) |
3980 | { | { |
3981 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3982 | OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); | OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); |
3983 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), common->first_line_end); |
3984 | } | } |
3985 | ||
3986 | start = LABEL(); | start = LABEL(); |
3987 | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3988 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3989 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
3990 | if (common->utf) | if (common->utf) |
3991 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
3992 | #endif | #endif |
3993 | ||
3994 | if (!check_class_ranges(common, inverted_start_bits, (inverted_start_bits[31] & 0x80) != 0, &matches)) | if (!check_class_ranges(common, start_bits, (start_bits[31] & 0x80) != 0, TRUE, &matches)) |
3995 | { | { |
3996 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3997 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 255); |
3998 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); |
3999 | JUMPHERE(jump); | JUMPHERE(jump); |
4000 | #endif | #endif |
4001 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
4002 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
4003 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)start_bits); |
4004 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); |
4005 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); |
4006 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_NOT_ZERO); |
4007 | } | } |
4008 | ||
4009 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
# | Line 2992 OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, S | Line 4015 OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, S |
4015 | #if defined COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
4016 | if (common->utf) | if (common->utf) |
4017 | { | { |
4018 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); | CMPTO(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); |
4019 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); |
4020 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
4021 | } | } |
4022 | #elif defined COMPILE_PCRE16 | #elif defined COMPILE_PCRE16 |
4023 | if (common->utf) | if (common->utf) |
4024 | { | { |
4025 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | CMPTO(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); |
4026 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
4027 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); |
4028 | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_EQUAL); |
4029 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
4030 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
4031 | } | } |
# | Line 3031 struct sljit_jump *notfound; | Line 4054 struct sljit_jump *notfound; |
4054 | pcre_uint32 oc, bit; | pcre_uint32 oc, bit; |
4055 | ||
4056 | SLJIT_ASSERT(common->req_char_ptr != 0); | SLJIT_ASSERT(common->req_char_ptr != 0); |
4057 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->req_char_ptr); |
4058 | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, REQ_BYTE_MAX); | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, REQ_BYTE_MAX); |
4059 | toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); | toolong = CMP(SLJIT_LESS, TMP1, 0, STR_END, 0); |
4060 | alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); | alreadyfound = CMP(SLJIT_LESS, STR_PTR, 0, TMP2, 0); |
4061 | ||
4062 | if (has_firstchar) | if (has_firstchar) |
4063 | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
# | Line 3042 else | Line 4065 else |
4065 | OP1(SLJIT_MOV, TMP1, 0, STR_PTR, 0); | OP1(SLJIT_MOV, TMP1, 0, STR_PTR, 0); |
4066 | ||
4067 | loop = LABEL(); | loop = LABEL(); |
4068 | notfound = CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, STR_END, 0); | notfound = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, STR_END, 0); |
4069 | ||
4070 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(TMP1), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(TMP1), 0); |
4071 | oc = req_char; | oc = req_char; |
# | Line 3055 if (caseless) | Line 4078 if (caseless) |
4078 | #endif | #endif |
4079 | } | } |
4080 | if (req_char == oc) | if (req_char == oc) |
4081 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char); | found = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, req_char); |
4082 | else | else |
4083 | { | { |
4084 | bit = req_char ^ oc; | bit = req_char ^ oc; |
4085 | if (is_powerof2(bit)) | if (is_powerof2(bit)) |
4086 | { | { |
4087 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); |
4088 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit); | found = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit); |
4089 | } | } |
4090 | else | else |
4091 | { | { |
4092 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char); | found = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, req_char); |
4093 | foundoc = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, oc); | foundoc = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, oc); |
4094 | } | } |
4095 | } | } |
4096 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); |
# | Line 3076 JUMPTO(SLJIT_JUMP, loop); | Line 4099 JUMPTO(SLJIT_JUMP, loop); |
4099 | JUMPHERE(found); | JUMPHERE(found); |
4100 | if (foundoc) | if (foundoc) |
4101 | JUMPHERE(foundoc); | JUMPHERE(foundoc); |
4102 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->req_char_ptr, TMP1, 0); |
4103 | JUMPHERE(alreadyfound); | JUMPHERE(alreadyfound); |
4104 | JUMPHERE(toolong); | JUMPHERE(toolong); |
4105 | return notfound; | return notfound; |
# | Line 3096 GET_LOCAL_BASE(TMP3, 0, 0); | Line 4119 GET_LOCAL_BASE(TMP3, 0, 0); |
4119 | mainloop = LABEL(); | mainloop = LABEL(); |
4120 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); |
4121 | OP2(SLJIT_SUB | SLJIT_SET_S, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0); | OP2(SLJIT_SUB | SLJIT_SET_S, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0); |
4122 | jump = JUMP(SLJIT_C_SIG_LESS_EQUAL); | jump = JUMP(SLJIT_SIG_LESS_EQUAL); |
4123 | ||
4124 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); |
4125 | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); |
# | Line 3105 OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_I | Line 4128 OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_I |
4128 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
4129 | ||
4130 | JUMPHERE(jump); | JUMPHERE(jump); |
4131 | jump = JUMP(SLJIT_C_SIG_LESS); | jump = JUMP(SLJIT_SIG_LESS); |
4132 | /* End of dropping frames. */ | /* End of dropping frames. */ |
4133 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
4134 | ||
# | Line 3121 static void check_wordboundary(compiler_ | Line 4144 static void check_wordboundary(compiler_ |
4144 | { | { |
4145 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4146 | struct sljit_jump *skipread; | struct sljit_jump *skipread; |
4147 | jump_list *skipread_list = NULL; | |
4148 | #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF | #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF |
4149 | struct sljit_jump *jump; | struct sljit_jump *jump; |
4150 | #endif | #endif |
4151 | ||
4152 | SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); | SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); |
4153 | ||
4154 | sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_SP), LOCALS0); |
4155 | /* Get type of the previous char, and put it to LOCALS1. */ | /* Get type of the previous char, and put it to LOCALS1. */ |
4156 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
4157 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); |
4158 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS1, SLJIT_IMM, 0); |
4159 | skipread = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); | skipread = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP1, 0); |
4160 | skip_char_back(common); | skip_char_back(common); |
4161 | check_start_used_ptr(common); | check_start_used_ptr(common); |
4162 | read_char(common); | read_char(common); |
# | Line 3142 read_char(common); | Line 4166 read_char(common); |
4166 | if (common->use_ucp) | if (common->use_ucp) |
4167 | { | { |
4168 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); |
4169 | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); | jump = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); |
4170 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
4171 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
4172 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); |
4173 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_LESS_EQUAL); |
4174 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); |
4175 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); |
4176 | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_LESS_EQUAL); |
4177 | JUMPHERE(jump); | JUMPHERE(jump); |
4178 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS1, TMP2, 0); |
4179 | } | } |
4180 | else | else |
4181 | #endif | #endif |
4182 | { | { |
4183 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
4184 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_GREATER, TMP1, 0, SLJIT_IMM, 255); |
4185 | #elif defined SUPPORT_UTF | #elif defined SUPPORT_UTF |
4186 | /* Here LOCALS1 has already been zeroed. */ | /* Here LOCALS1 has already been zeroed. */ |
4187 | jump = NULL; | jump = NULL; |
4188 | if (common->utf) | if (common->utf) |
4189 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_GREATER, TMP1, 0, SLJIT_IMM, 255); |
4190 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE8 */ |
4191 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes); |
4192 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 4 /* ctype_word */); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 4 /* ctype_word */); |
4193 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
4194 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS1, TMP1, 0); |
4195 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
4196 | JUMPHERE(jump); | JUMPHERE(jump); |
4197 | #elif defined SUPPORT_UTF | #elif defined SUPPORT_UTF |
# | Line 3178 else | Line 4202 else |
4202 | JUMPHERE(skipread); | JUMPHERE(skipread); |
4203 | ||
4204 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); |
4205 | skipread = check_str_end(common); | check_str_end(common, &skipread_list); |
4206 | peek_char(common); | peek_char(common, READ_CHAR_MAX); |
4207 | ||
4208 | /* Testing char type. This is a code duplication. */ | /* Testing char type. This is a code duplication. */ |
4209 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
4210 | if (common->use_ucp) | if (common->use_ucp) |
4211 | { | { |
4212 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); |
4213 | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); | jump = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); |
4214 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
4215 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
4216 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); |
4217 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_LESS_EQUAL); |
4218 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); |
4219 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); |
4220 | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_LESS_EQUAL); |
4221 | JUMPHERE(jump); | JUMPHERE(jump); |
4222 | } | } |
4223 | else | else |
# | Line 3202 else | Line 4226 else |
4226 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
4227 | /* TMP2 may be destroyed by peek_char. */ | /* TMP2 may be destroyed by peek_char. */ |
4228 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); |
4229 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_GREATER, TMP1, 0, SLJIT_IMM, 255); |
4230 | #elif defined SUPPORT_UTF | #elif defined SUPPORT_UTF |
4231 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); |
4232 | jump = NULL; | jump = NULL; |
4233 | if (common->utf) | if (common->utf) |
4234 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_GREATER, TMP1, 0, SLJIT_IMM, 255); |
4235 | #endif | #endif |
4236 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), common->ctypes); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), common->ctypes); |
4237 | OP2(SLJIT_LSHR, TMP2, 0, TMP2, 0, SLJIT_IMM, 4 /* ctype_word */); | OP2(SLJIT_LSHR, TMP2, 0, TMP2, 0, SLJIT_IMM, 4 /* ctype_word */); |
# | Line 3219 else | Line 4243 else |
4243 | JUMPHERE(jump); | JUMPHERE(jump); |
4244 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE8 */ |
4245 | } | } |
4246 | JUMPHERE(skipread); | set_jumps(skipread_list, LABEL()); |
4247 | ||
4248 | OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); | OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_SP), LOCALS1); |
4249 | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_SP), LOCALS0); |
4250 | } | } |
4251 | ||
4252 | /* | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks) |
range format: | ||
ranges[0] = length of the range (max MAX_RANGE_SIZE, -1 means invalid range). | ||
ranges[1] = first bit (0 or 1) | ||
ranges[2-length] = position of the bit change (when the current bit is not equal to the previous) | ||
*/ | ||
static BOOL check_ranges(compiler_common *common, int *ranges, jump_list **backtracks, BOOL readch) | ||
4253 | { | { |
4254 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4255 | struct sljit_jump *jump; | int ranges[MAX_RANGE_SIZE]; |
4256 | pcre_uint8 bit, cbit, all; | |
4257 | int i, byte, length = 0; | |
4258 | ||
4259 | bit = bits[0] & 0x1; | |
4260 | /* All bits will be zero or one (since bit is zero or one). */ | |
4261 | all = -bit; | |
4262 | ||
4263 | for (i = 0; i < 256; ) | |
4264 | { | |
4265 | byte = i >> 3; | |
4266 | if ((i & 0x7) == 0 && bits[byte] == all) | |
4267 | i += 8; | |
4268 | else | |
4269 | { | |
4270 | cbit = (bits[byte] >> (i & 0x7)) & 0x1; | |
4271 | if (cbit != bit) | |
4272 | { | |
4273 | if (length >= MAX_RANGE_SIZE) | |
4274 | return FALSE; | |
4275 | ranges[length] = i; | |
4276 | length++; | |
4277 | bit = cbit; | |
4278 | all = -cbit; | |
4279 | } | |
4280 | i++; | |
4281 | } | |
4282 | } | |
4283 | ||
4284 | if (((bit == 0) && nclass) || ((bit == 1) && !nclass)) | |
4285 | { | |
4286 | if (length >= MAX_RANGE_SIZE) | |
4287 | return FALSE; | |
4288 | ranges[length] = 256; | |
4289 | length++; | |
4290 | } | |
4291 | ||
4292 | if (ranges[0] < 0) | if (length < 0 || length > 4) |
4293 | return FALSE; | return FALSE; |
4294 | ||
4295 | switch(ranges[0]) | bit = bits[0] & 0x1; |
4296 | if (invert) bit ^= 0x1; | |
4297 | ||