Parent Directory
|
Revision Log
|
Patch
revision 1195 by zherczeg, Thu Nov 1 15:21:27 2012 UTC | revision 1491 by zherczeg, Mon Jul 7 07:11:16 2014 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 65 system files. */ | Line 65 system files. */ |
65 | #error Unsupported architecture | #error Unsupported architecture |
66 | #endif | #endif |
67 | ||
68 | /* Defines for debugging purposes. */ | |
69 | ||
70 | /* 1 - Use unoptimized capturing brackets. | |
71 | 2 - Enable capture_last_ptr (includes option 1). */ | |
72 | /* #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 157 typedef struct jit_arguments { | Line 166 typedef struct jit_arguments { |
166 | int *offsets; | int *offsets; |
167 | pcre_uchar *uchar_ptr; | pcre_uchar *uchar_ptr; |
168 | pcre_uchar *mark_ptr; | pcre_uchar *mark_ptr; |
169 | void *callout_data; | |
170 | /* Everything else after. */ | /* Everything else after. */ |
171 | int offsetcount; | pcre_uint32 limit_match; |
172 | int calllimit; | int real_offset_count; |
173 | int offset_count; | |
174 | pcre_uint8 notbol; | pcre_uint8 notbol; |
175 | pcre_uint8 noteol; | pcre_uint8 noteol; |
176 | pcre_uint8 notempty; | pcre_uint8 notempty; |
# | Line 168 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 | sljit_uw *read_only_data[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 179 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 209 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 230 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 265 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 | /* This read-only data is available during runtime. */ | |
326 | sljit_uw *read_only_data; | |
327 | /* The total size of the read-only data. */ | |
328 | sljit_uw read_only_data_size; | |
329 | /* The next free entry of the read_only_data. */ | |
330 | sljit_uw *read_only_data_ptr; | |
331 | /* Tells whether the capturing bracket is optimized. */ | /* Tells whether the capturing bracket is optimized. */ |
332 | pcre_uint8 *optimized_cbracket; | pcre_uint8 *optimized_cbracket; |
333 | /* Tells whether the starting offset is a target of then. */ | |
334 | pcre_uint8 *then_offsets; | |
335 | /* Current position where a THEN must jump. */ | |
336 | then_trap_backtrack *then_trap; | |
337 | /* Starting offset of private data for capturing brackets. */ | /* Starting offset of private data for capturing brackets. */ |
338 | int cbraptr; | int cbra_ptr; |
339 | /* OVector starting point. Must be divisible by 2. */ | /* Output vector starting point. Must be divisible by 2. */ |
340 | int ovector_start; | int ovector_start; |
341 | /* Last known position of the requested byte. */ | /* Last known position of the requested byte. */ |
342 | int req_char_ptr; | int req_char_ptr; |
343 | /* Head of the last recursion. */ | /* Head of the last recursion. */ |
344 | int recursive_head; | int recursive_head_ptr; |
345 | /* First inspected character for partial matching. */ | /* First inspected character for partial matching. */ |
346 | int start_used_ptr; | int start_used_ptr; |
347 | /* Starting pointer for partial soft matches. */ | /* Starting pointer for partial soft matches. */ |
# | Line 298 typedef struct compiler_common { | Line 350 typedef struct compiler_common { |
350 | int first_line_end; | int first_line_end; |
351 | /* Points to the marked string. */ | /* Points to the marked string. */ |
352 | int mark_ptr; | int mark_ptr; |
353 | /* Recursive control verb management chain. */ | |
354 | int control_head_ptr; | |
355 | /* Points to the last matched capture block index. */ | |
356 | int capture_last_ptr; | |
357 | /* Points to the starting position of the current match. */ | |
358 | int start_ptr; | |
359 | ||
360 | /* Flipped and lower case tables. */ | /* Flipped and lower case tables. */ |
361 | const pcre_uint8 *fcc; | const pcre_uint8 *fcc; |
362 | sljit_sw lcc; | sljit_sw lcc; |
363 | /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */ | /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */ |
364 | int mode; | int mode; |
365 | /* TRUE, when minlength is greater than 0. */ | |
366 | BOOL might_be_empty; | |
367 | /* \K is found in the pattern. */ | |
368 | BOOL has_set_som; | |
369 | /* (*SKIP:arg) is found in the pattern. */ | |
370 | BOOL has_skip_arg; | |
371 | /* (*THEN) is found in the pattern. */ | |
372 | BOOL has_then; | |
373 | /* Needs to know the start position anytime. */ | |
374 | BOOL needs_start_ptr; | |
375 | /* Currently in recurse or negative assert. */ | |
376 | BOOL local_exit; | |
377 | /* Currently in a positive assert. */ | |
378 | BOOL positive_assert; | |
379 | /* Newline control. */ | /* Newline control. */ |
380 | int nltype; | int nltype; |
381 | pcre_uint32 nlmax; | |
382 | pcre_uint32 nlmin; | |
383 | int newline; | int newline; |
384 | int bsr_nltype; | int bsr_nltype; |
385 | pcre_uint32 bsr_nlmax; | |
386 | pcre_uint32 bsr_nlmin; | |
387 | /* Dollar endonly. */ | /* Dollar endonly. */ |
388 | int endonly; | int endonly; |
BOOL has_set_som; | ||
389 | /* Tables. */ | /* Tables. */ |
390 | sljit_sw ctypes; | sljit_sw ctypes; |
int digits[2 + MAX_RANGE_SIZE]; | ||
391 | /* Named capturing brackets. */ | /* Named capturing brackets. */ |
392 | sljit_uw name_table; | pcre_uchar *name_table; |
393 | sljit_sw name_count; | sljit_sw name_count; |
394 | sljit_sw name_entry_size; | sljit_sw name_entry_size; |
395 | ||
396 | /* Labels and jump lists. */ | /* Labels and jump lists. */ |
397 | struct sljit_label *partialmatchlabel; | struct sljit_label *partialmatchlabel; |
398 | struct sljit_label *quitlabel; | struct sljit_label *quit_label; |
399 | struct sljit_label *acceptlabel; | struct sljit_label *forced_quit_label; |
400 | struct sljit_label *accept_label; | |
401 | struct sljit_label *ff_newline_shortcut; | |
402 | stub_list *stubs; | stub_list *stubs; |
403 | label_addr_list *label_addrs; | |
404 | recurse_entry *entries; | recurse_entry *entries; |
405 | recurse_entry *currententry; | recurse_entry *currententry; |
406 | jump_list *partialmatch; | jump_list *partialmatch; |
407 | jump_list *quit; | jump_list *quit; |
408 | jump_list *positive_assert_quit; | |
409 | jump_list *forced_quit; | |
410 | jump_list *accept; | jump_list *accept; |
411 | jump_list *calllimit; | jump_list *calllimit; |
412 | jump_list *stackalloc; | jump_list *stackalloc; |
# | Line 338 typedef struct compiler_common { | Line 417 typedef struct compiler_common { |
417 | jump_list *vspace; | jump_list *vspace; |
418 | jump_list *casefulcmp; | jump_list *casefulcmp; |
419 | jump_list *caselesscmp; | jump_list *caselesscmp; |
420 | jump_list *reset_match; | |
421 | BOOL jscript_compat; | BOOL jscript_compat; |
422 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
423 | BOOL utf; | BOOL utf; |
424 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
425 | BOOL use_ucp; | BOOL use_ucp; |
426 | #endif | #endif |
#ifndef COMPILE_PCRE32 | ||
jump_list *utfreadchar; | ||
#endif | ||
427 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
428 | jump_list *utfreadchar; | |
429 | jump_list *utfreadchar16; | |
430 | jump_list *utfreadtype8; | jump_list *utfreadtype8; |
431 | #endif | #endif |
432 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
# | Line 390 typedef struct compare_context { | Line 469 typedef struct compare_context { |
469 | #endif | #endif |
470 | } compare_context; | } compare_context; |
471 | ||
enum { | ||
frame_end = 0, | ||
frame_setstrbegin = -1, | ||
frame_setmark = -2 | ||
}; | ||
472 | /* Undefine sljit macros. */ | /* Undefine sljit macros. */ |
473 | #undef CMP | #undef CMP |
474 | ||
475 | /* Used for accessing the elements of the stack. */ | /* Used for accessing the elements of the stack. */ |
476 | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_sw)) | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_sw)) |
477 | ||
478 | #define TMP1 SLJIT_TEMPORARY_REG1 | #define TMP1 SLJIT_R0 |
479 | #define TMP2 SLJIT_TEMPORARY_REG3 | #define TMP2 SLJIT_R2 |
480 | #define TMP3 SLJIT_TEMPORARY_EREG2 | #define TMP3 SLJIT_R3 |
481 | #define STR_PTR SLJIT_SAVED_REG1 | #define STR_PTR SLJIT_S0 |
482 | #define STR_END SLJIT_SAVED_REG2 | #define STR_END SLJIT_S1 |
483 | #define STACK_TOP SLJIT_TEMPORARY_REG2 | #define STACK_TOP SLJIT_R1 |
484 | #define STACK_LIMIT SLJIT_SAVED_REG3 | #define STACK_LIMIT SLJIT_S2 |
485 | #define ARGUMENTS SLJIT_SAVED_EREG1 | #define COUNT_MATCH SLJIT_S3 |
486 | #define CALL_COUNT SLJIT_SAVED_EREG2 | #define ARGUMENTS SLJIT_S4 |
487 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 | #define RETURN_ADDR SLJIT_R4 |
488 | ||
489 | /* Local space layout. */ | /* Local space layout. */ |
490 | /* These two locals can be used by the current opcode. */ | /* These two locals can be used by the current opcode. */ |
# | Line 421 enum { | Line 494 enum { |
494 | #define POSSESSIVE0 (2 * sizeof(sljit_sw)) | #define POSSESSIVE0 (2 * sizeof(sljit_sw)) |
495 | #define POSSESSIVE1 (3 * sizeof(sljit_sw)) | #define POSSESSIVE1 (3 * sizeof(sljit_sw)) |
496 | /* Max limit of recursions. */ | /* Max limit of recursions. */ |
497 | #define CALL_LIMIT (4 * sizeof(sljit_sw)) | #define LIMIT_MATCH (4 * sizeof(sljit_sw)) |
498 | /* The output vector is stored on the stack, and contains pointers | /* The output vector is stored on the stack, and contains pointers |
499 | to characters. The vector data is divided into two groups: the first | to characters. The vector data is divided into two groups: the first |
500 | group contains the start / end character pointers, and the second is | group contains the start / end character pointers, and the second is |
501 | 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. */ |
502 | #define OVECTOR_START (common->ovector_start) | #define OVECTOR_START (common->ovector_start) |
503 | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_sw)) | #define OVECTOR(i) (OVECTOR_START + (i) * (sljit_sw)sizeof(sljit_sw)) |
504 | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_sw)) | #define OVECTOR_PRIV(i) (common->cbra_ptr + (i) * (sljit_sw)sizeof(sljit_sw)) |
505 | #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start]) | #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start]) |
506 | ||
507 | #if defined COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
# | Line 459 the start pointers when the end of the c | Line 532 the start pointers when the end of the c |
532 | sljit_set_label(sljit_emit_jump(compiler, (type)), (label)) | sljit_set_label(sljit_emit_jump(compiler, (type)), (label)) |
533 | #define JUMPHERE(jump) \ | #define JUMPHERE(jump) \ |
534 | sljit_set_label((jump), sljit_emit_label(compiler)) | sljit_set_label((jump), sljit_emit_label(compiler)) |
535 | #define SET_LABEL(jump, label) \ | |
536 | sljit_set_label((jump), (label)) | |
537 | #define CMP(type, src1, src1w, src2, src2w) \ | #define CMP(type, src1, src1w, src2, src2w) \ |
538 | sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) | sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) |
539 | #define CMPTO(type, src1, src1w, src2, src2w, label) \ | #define CMPTO(type, src1, src1w, src2, src2w, label) \ |
540 | sljit_set_label(sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)), (label)) | sljit_set_label(sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)), (label)) |
541 | #define COND_VALUE(op, dst, dstw, type) \ | #define OP_FLAGS(op, dst, dstw, src, srcw, type) \ |
542 | sljit_emit_cond_value(compiler, (op), (dst), (dstw), (type)) | sljit_emit_op_flags(compiler, (op), (dst), (dstw), (src), (srcw), (type)) |
543 | #define GET_LOCAL_BASE(dst, dstw, offset) \ | #define GET_LOCAL_BASE(dst, dstw, offset) \ |
544 | sljit_get_local_base(compiler, (dst), (dstw), (offset)) | sljit_get_local_base(compiler, (dst), (dstw), (offset)) |
545 | ||
546 | #define READ_CHAR_MAX 0x7fffffff | |
547 | ||
548 | static pcre_uchar* bracketend(pcre_uchar* cc) | static pcre_uchar* bracketend(pcre_uchar* cc) |
549 | { | { |
550 | 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)); |
# | Line 477 cc += 1 + LINK_SIZE; | Line 554 cc += 1 + LINK_SIZE; |
554 | return cc; | return cc; |
555 | } | } |
556 | ||
557 | static int no_alternatives(pcre_uchar* cc) | |
558 | { | |
559 | int count = 0; | |
560 | SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)); | |
561 | do | |
562 | { | |
563 | cc += GET(cc, 1); | |
564 | count++; | |
565 | } | |
566 | while (*cc == OP_ALT); | |
567 | SLJIT_ASSERT(*cc >= OP_KET && *cc <= OP_KETRPOS); | |
568 | return count; | |
569 | } | |
570 | ||
571 | static int ones_in_half_byte[16] = { | |
572 | /* 0 */ 0, 1, 1, 2, /* 4 */ 1, 2, 2, 3, | |
573 | /* 8 */ 1, 2, 2, 3, /* 12 */ 2, 3, 3, 4 | |
574 | }; | |
575 | ||
576 | /* Functions whose might need modification for all new supported opcodes: | /* Functions whose might need modification for all new supported opcodes: |
577 | next_opcode | next_opcode |
578 | get_private_data_length | check_opcode_types |
579 | set_private_data_ptrs | set_private_data_ptrs |
580 | get_framesize | get_framesize |
581 | init_frame | init_frame |
582 | get_private_data_length_for_copy | get_private_data_copy_length |
583 | copy_private_data | copy_private_data |
584 | compile_matchingpath | compile_matchingpath |
585 | compile_backtrackingpath | compile_backtrackingpath |
# | Line 507 switch(*cc) | Line 603 switch(*cc) |
603 | case OP_WORDCHAR: | case OP_WORDCHAR: |
604 | case OP_ANY: | case OP_ANY: |
605 | case OP_ALLANY: | case OP_ALLANY: |
606 | case OP_NOTPROP: | |
607 | case OP_PROP: | |
608 | case OP_ANYNL: | case OP_ANYNL: |
609 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
610 | case OP_HSPACE: | case OP_HSPACE: |
# | Line 519 switch(*cc) | Line 617 switch(*cc) |
617 | case OP_CIRCM: | case OP_CIRCM: |
618 | case OP_DOLL: | case OP_DOLL: |
619 | 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: | ||
620 | case OP_CRSTAR: | case OP_CRSTAR: |
621 | case OP_CRMINSTAR: | case OP_CRMINSTAR: |
622 | case OP_CRPLUS: | case OP_CRPLUS: |
623 | case OP_CRMINPLUS: | case OP_CRMINPLUS: |
624 | case OP_CRQUERY: | case OP_CRQUERY: |
625 | case OP_CRMINQUERY: | case OP_CRMINQUERY: |
626 | case OP_CRRANGE: | |
627 | case OP_CRMINRANGE: | |
628 | case OP_CRPOSSTAR: | |
629 | case OP_CRPOSPLUS: | |
630 | case OP_CRPOSQUERY: | |
631 | case OP_CRPOSRANGE: | |
632 | case OP_CLASS: | |
633 | case OP_NCLASS: | |
634 | case OP_REF: | |
635 | case OP_REFI: | |
636 | case OP_DNREF: | |
637 | case OP_DNREFI: | |
638 | case OP_RECURSE: | |
639 | case OP_CALLOUT: | |
640 | case OP_ALT: | |
641 | case OP_KET: | |
642 | case OP_KETRMAX: | |
643 | case OP_KETRMIN: | |
644 | case OP_KETRPOS: | |
645 | case OP_REVERSE: | |
646 | case OP_ASSERT: | |
647 | case OP_ASSERT_NOT: | |
648 | case OP_ASSERTBACK: | |
649 | case OP_ASSERTBACK_NOT: | |
650 | case OP_ONCE: | |
651 | case OP_ONCE_NC: | |
652 | case OP_BRA: | |
653 | case OP_BRAPOS: | |
654 | case OP_CBRA: | |
655 | case OP_CBRAPOS: | |
656 | case OP_COND: | |
657 | case OP_SBRA: | |
658 | case OP_SBRAPOS: | |
659 | case OP_SCBRA: | |
660 | case OP_SCBRAPOS: | |
661 | case OP_SCOND: | |
662 | case OP_CREF: | |
663 | case OP_DNCREF: | |
664 | case OP_RREF: | |
665 | case OP_DNRREF: | |
666 | case OP_DEF: | case OP_DEF: |
667 | case OP_BRAZERO: | case OP_BRAZERO: |
668 | case OP_BRAMINZERO: | case OP_BRAMINZERO: |
669 | case OP_BRAPOSZERO: | case OP_BRAPOSZERO: |
670 | case OP_PRUNE: | |
671 | case OP_SKIP: | |
672 | case OP_THEN: | |
673 | case OP_COMMIT: | case OP_COMMIT: |
674 | case OP_FAIL: | case OP_FAIL: |
675 | case OP_ACCEPT: | case OP_ACCEPT: |
676 | case OP_ASSERT_ACCEPT: | case OP_ASSERT_ACCEPT: |
677 | case OP_CLOSE: | |
678 | case OP_SKIPZERO: | case OP_SKIPZERO: |
679 | return cc + 1; | return cc + PRIV(OP_lengths)[*cc]; |
case OP_ANYBYTE: | ||
#ifdef SUPPORT_UTF | ||
if (common->utf) return NULL; | ||
#endif | ||
return cc + 1; | ||
680 | ||
681 | case OP_CHAR: | case OP_CHAR: |
682 | case OP_CHARI: | case OP_CHARI: |
# | Line 561 switch(*cc) | Line 688 switch(*cc) |
688 | case OP_MINPLUS: | case OP_MINPLUS: |
689 | case OP_QUERY: | case OP_QUERY: |
690 | case OP_MINQUERY: | case OP_MINQUERY: |
691 | case OP_UPTO: | |
692 | case OP_MINUPTO: | |
693 | case OP_EXACT: | |
694 | case OP_POSSTAR: | case OP_POSSTAR: |
695 | case OP_POSPLUS: | case OP_POSPLUS: |
696 | case OP_POSQUERY: | case OP_POSQUERY: |
697 | case OP_POSUPTO: | |
698 | case OP_STARI: | case OP_STARI: |
699 | case OP_MINSTARI: | case OP_MINSTARI: |
700 | case OP_PLUSI: | case OP_PLUSI: |
701 | case OP_MINPLUSI: | case OP_MINPLUSI: |
702 | case OP_QUERYI: | case OP_QUERYI: |
703 | case OP_MINQUERYI: | case OP_MINQUERYI: |
704 | case OP_UPTOI: | |
705 | case OP_MINUPTOI: | |
706 | case OP_EXACTI: | |
707 | case OP_POSSTARI: | case OP_POSSTARI: |
708 | case OP_POSPLUSI: | case OP_POSPLUSI: |
709 | case OP_POSQUERYI: | case OP_POSQUERYI: |
710 | case OP_POSUPTOI: | |
711 | case OP_NOTSTAR: | case OP_NOTSTAR: |
712 | case OP_NOTMINSTAR: | case OP_NOTMINSTAR: |
713 | case OP_NOTPLUS: | case OP_NOTPLUS: |
714 | case OP_NOTMINPLUS: | case OP_NOTMINPLUS: |
715 | case OP_NOTQUERY: | case OP_NOTQUERY: |
716 | case OP_NOTMINQUERY: | case OP_NOTMINQUERY: |
717 | case OP_NOTUPTO: | |
718 | case OP_NOTMINUPTO: | |
719 | case OP_NOTEXACT: | |
720 | case OP_NOTPOSSTAR: | case OP_NOTPOSSTAR: |
721 | case OP_NOTPOSPLUS: | case OP_NOTPOSPLUS: |
722 | case OP_NOTPOSQUERY: | case OP_NOTPOSQUERY: |
723 | case OP_NOTPOSUPTO: | |
724 | case OP_NOTSTARI: | case OP_NOTSTARI: |
725 | case OP_NOTMINSTARI: | case OP_NOTMINSTARI: |
726 | case OP_NOTPLUSI: | case OP_NOTPLUSI: |
727 | case OP_NOTMINPLUSI: | case OP_NOTMINPLUSI: |
728 | case OP_NOTQUERYI: | case OP_NOTQUERYI: |
729 | 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: | ||
730 | case OP_NOTUPTOI: | case OP_NOTUPTOI: |
731 | case OP_NOTMINUPTOI: | case OP_NOTMINUPTOI: |
732 | case OP_NOTEXACTI: | case OP_NOTEXACTI: |
733 | case OP_NOTPOSSTARI: | |
734 | case OP_NOTPOSPLUSI: | |
735 | case OP_NOTPOSQUERYI: | |
736 | case OP_NOTPOSUPTOI: | case OP_NOTPOSUPTOI: |
737 | cc += 2 + IMM2_SIZE; | cc += PRIV(OP_lengths)[*cc]; |
738 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
739 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
740 | #endif | #endif |
741 | return cc; | return cc; |
742 | ||
743 | case OP_NOTPROP: | /* Special cases. */ |
744 | case OP_PROP: | case OP_TYPESTAR: |
745 | return cc + 1 + 2; | case OP_TYPEMINSTAR: |
746 | case OP_TYPEPLUS: | |
747 | case OP_TYPEMINPLUS: | |
748 | case OP_TYPEQUERY: | |
749 | case OP_TYPEMINQUERY: | |
750 | case OP_TYPEUPTO: | case OP_TYPEUPTO: |
751 | case OP_TYPEMINUPTO: | case OP_TYPEMINUPTO: |
752 | case OP_TYPEEXACT: | case OP_TYPEEXACT: |
753 | case OP_TYPEPOSSTAR: | |
754 | case OP_TYPEPOSPLUS: | |
755 | case OP_TYPEPOSQUERY: | |
756 | case OP_TYPEPOSUPTO: | case OP_TYPEPOSUPTO: |
757 | 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; | ||
758 | ||
759 | case OP_CRRANGE: | case OP_ANYBYTE: |
760 | case OP_CRMINRANGE: | #ifdef SUPPORT_UTF |
761 | return cc + 1 + 2 * IMM2_SIZE; | if (common->utf) return NULL; |
762 | #endif | |
763 | case OP_CLASS: | return cc + 1; |
case OP_NCLASS: | ||
return cc + 1 + 32 / sizeof(pcre_uchar); | ||
764 | ||
765 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
766 | case OP_XCLASS: | case OP_XCLASS: |
767 | return cc + GET(cc, 1); | return cc + GET(cc, 1); |
768 | #endif | #endif |
769 | ||
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; | ||
770 | case OP_MARK: | case OP_MARK: |
771 | case OP_PRUNE_ARG: | |
772 | case OP_SKIP_ARG: | |
773 | case OP_THEN_ARG: | |
774 | return cc + 1 + 2 + cc[1]; | return cc + 1 + 2 + cc[1]; |
775 | ||
776 | default: | default: |
777 | /* All opcodes are supported now! */ | |
778 | SLJIT_ASSERT_STOP(); | |
779 | return NULL; | return NULL; |
780 | } | } |
781 | } | } |
782 | ||
783 | #define CASE_ITERATOR_PRIVATE_DATA_1 \ | static BOOL check_opcode_types(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) |
784 | case OP_MINSTAR: \ | { |
785 | case OP_MINPLUS: \ | int count; |
786 | case OP_QUERY: \ | pcre_uchar *slot; |
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: | ||
787 | ||
788 | #define CASE_ITERATOR_PRIVATE_DATA_2A \ | /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ |
789 | case OP_STAR: \ | while (cc < ccend) |
790 | case OP_PLUS: \ | { |
791 | case OP_STARI: \ | switch(*cc) |
792 | case OP_PLUSI: \ | { |
793 | case OP_NOTSTAR: \ | case OP_SET_SOM: |
794 | case OP_NOTPLUS: \ | common->has_set_som = TRUE; |
795 | case OP_NOTSTARI: \ | common->might_be_empty = TRUE; |
796 | case OP_NOTPLUSI: | cc += 1; |
797 | break; | |
798 | ||
799 | #define CASE_ITERATOR_PRIVATE_DATA_2B \ | case OP_REF: |
800 | case OP_UPTO: \ | case OP_REFI: |
801 | case OP_MINUPTO: \ | common->optimized_cbracket[GET2(cc, 1)] = 0; |
802 | case OP_UPTOI: \ | cc += 1 + IMM2_SIZE; |
803 | case OP_MINUPTOI: \ | break; |
case OP_NOTUPTO: \ | ||
case OP_NOTMINUPTO: \ | ||
case OP_NOTUPTOI: \ | ||
case OP_NOTMINUPTOI: | ||
804 | ||
805 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \ | case OP_BRA: |
806 | case OP_TYPEMINSTAR: \ | case OP_CBRA: |
807 | case OP_TYPEMINPLUS: \ | case OP_SBRA: |
808 | case OP_TYPEQUERY: \ | case OP_SCBRA: |
809 | case OP_TYPEMINQUERY: | count = no_alternatives(cc); |
810 | if (count > 4) | |
811 | common->read_only_data_size += count * sizeof(sljit_uw); | |
812 | cc += 1 + LINK_SIZE + (*cc == OP_CBRA || *cc == OP_SCBRA ? IMM2_SIZE : 0); | |
813 | break; | |
814 | ||
815 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \ | case OP_CBRAPOS: |
816 | case OP_TYPESTAR: \ | case OP_SCBRAPOS: |
817 | case OP_TYPEPLUS: | common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0; |
818 | cc += 1 + LINK_SIZE + IMM2_SIZE; | |
819 | break; | |
820 | ||
821 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \ | case OP_COND: |
822 | case OP_TYPEUPTO: \ | case OP_SCOND: |
823 | case OP_TYPEMINUPTO: | /* Only AUTO_CALLOUT can insert this opcode. We do |
824 | not intend to support this case. */ | |
825 | if (cc[1 + LINK_SIZE] == OP_CALLOUT) | |
826 | return FALSE; | |
827 | cc += 1 + LINK_SIZE; | |
828 | break; | |
829 | ||
830 | static int get_class_iterator_size(pcre_uchar *cc) | case OP_CREF: |
831 | { | common->optimized_cbracket[GET2(cc, 1)] = 0; |
832 | cc += 1 + IMM2_SIZE; | |
833 | break; | |
834 | ||
835 | case OP_DNREF: | |
836 | case OP_DNREFI: | |
837 | case OP_DNCREF: | |
838 | count = GET2(cc, 1 + IMM2_SIZE); | |
839 | slot = common->name_table + GET2(cc, 1) * common->name_entry_size; | |
840 | while (count-- > 0) | |
841 | { | |
842 | common->optimized_cbracket[GET2(slot, 0)] = 0; | |
843 | slot += common->name_entry_size; | |
844 | } | |
845 | cc += 1 + 2 * IMM2_SIZE; | |
846 | break; | |
847 | ||
848 | case OP_RECURSE: | |
849 | /* Set its value only once. */ | |
850 | if (common->recursive_head_ptr == 0) | |
851 | { | |
852 | common->recursive_head_ptr = common->ovector_start; | |
853 | common->ovector_start += sizeof(sljit_sw); | |
854 | } | |
855 | cc += 1 + LINK_SIZE; | |
856 | break; | |
857 | ||
858 | case OP_CALLOUT: | |
859 | if (common->capture_last_ptr == 0) | |
860 | { | |
861 | common->capture_last_ptr = common->ovector_start; | |
862 | common->ovector_start += sizeof(sljit_sw); | |
863 | } | |
864 | cc += 2 + 2 * LINK_SIZE; | |
865 | break; | |
866 | ||
867 | case OP_THEN_ARG: | |
868 | common->has_then = TRUE; | |
869 | common->control_head_ptr = 1; | |
870 | /* Fall through. */ | |
871 | ||
872 | case OP_PRUNE_ARG: | |
873 | common->needs_start_ptr = TRUE; | |
874 | /* Fall through. */ | |
875 | ||
876 | case OP_MARK: | |
877 | if (common->mark_ptr == 0) | |
878 | { | |
879 | common->mark_ptr = common->ovector_start; | |
880 | common->ovector_start += sizeof(sljit_sw); | |
881 | } | |
882 | cc += 1 + 2 + cc[1]; | |
883 | break; | |
884 | ||
885 | case OP_THEN: | |
886 | common->has_then = TRUE; | |
887 | common->control_head_ptr = 1; | |
888 | /* Fall through. */ | |
889 | ||
890 | case OP_PRUNE: | |
891 | case OP_SKIP: | |
892 | common->needs_start_ptr = TRUE; | |
893 | cc += 1; | |
894 | break; | |
895 | ||
896 | case OP_SKIP_ARG: | |
897 | common->control_head_ptr = 1; | |
898 | common->has_skip_arg = TRUE; | |
899 | cc += 1 + 2 + cc[1]; | |
900 | break; | |
901 | ||
902 | default: | |
903 | cc = next_opcode(common, cc); | |
904 | if (cc == NULL) | |
905 | return FALSE; | |
906 | break; | |
907 | } | |
908 | } | |
909 | return TRUE; | |
910 | } | |
911 | ||
912 | static int get_class_iterator_size(pcre_uchar *cc) | |
913 | { | |
914 | switch(*cc) | switch(*cc) |
915 | { | { |
916 | case OP_CRSTAR: | case OP_CRSTAR: |
# | Line 762 switch(*cc) | Line 934 switch(*cc) |
934 | } | } |
935 | } | } |
936 | ||
937 | static int get_private_data_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | static BOOL detect_repeat(compiler_common *common, pcre_uchar *begin) |
938 | { | |
939 | pcre_uchar *end = bracketend(begin); | |
940 | pcre_uchar *next; | |
941 | pcre_uchar *next_end; | |
942 | pcre_uchar *max_end; | |
943 | pcre_uchar type; | |
944 | sljit_sw length = end - begin; | |
945 | int min, max, i; | |
946 | ||
947 | /* Detect fixed iterations first. */ | |
948 | if (end[-(1 + LINK_SIZE)] != OP_KET) | |
949 | return FALSE; | |
950 | ||
951 | /* Already detected repeat. */ | |
952 | if (common->private_data_ptrs[end - common->start - LINK_SIZE] != 0) | |
953 | return TRUE; | |
954 | ||
955 | next = end; | |
956 | min = 1; | |
957 | while (1) | |
958 | { | |
959 | if (*next != *begin) | |
960 | break; | |
961 | next_end = bracketend(next); | |
962 | if (next_end - next != length || memcmp(begin, next, IN_UCHARS(length)) != 0) | |
963 | break; | |
964 | next = next_end; | |
965 | min++; | |
966 | } | |
967 | ||
968 | if (min == 2) | |
969 | return FALSE; | |
970 | ||
971 | max = 0; | |
972 | max_end = next; | |
973 | if (*next == OP_BRAZERO || *next == OP_BRAMINZERO) | |
974 | { | |
975 | type = *next; | |
976 | while (1) | |
977 | { | |
978 | if (next[0] != type || next[1] != OP_BRA || next[2 + LINK_SIZE] != *begin) | |
979 | break; | |
980 | next_end = bracketend(next + 2 + LINK_SIZE); | |
981 | if (next_end - next != (length + 2 + LINK_SIZE) || memcmp(begin, next + 2 + LINK_SIZE, IN_UCHARS(length)) != 0) | |
982 | break; | |
983 | next = next_end; | |
984 | max++; | |
985 | } | |
986 | ||
987 | if (next[0] == type && next[1] == *begin && max >= 1) | |
988 | { | |
989 | next_end = bracketend(next + 1); | |
990 | if (next_end - next == (length + 1) && memcmp(begin, next + 1, IN_UCHARS(length)) == 0) | |
991 | { | |
992 | for (i = 0; i < max; i++, next_end += 1 + LINK_SIZE) | |
993 | if (*next_end != OP_KET) | |
994 | break; | |
995 | ||
996 | if (i == max) | |
997 | { | |
998 | common->private_data_ptrs[max_end - common->start - LINK_SIZE] = next_end - max_end; | |
999 | common->private_data_ptrs[max_end - common->start - LINK_SIZE + 1] = (type == OP_BRAZERO) ? OP_UPTO : OP_MINUPTO; | |
1000 | /* +2 the original and the last. */ | |
1001 | common->private_data_ptrs[max_end - common->start - LINK_SIZE + 2] = max + 2; | |
1002 | if (min == 1) | |
1003 | return TRUE; | |
1004 | min--; | |
1005 | max_end -= (1 + LINK_SIZE) + GET(max_end, -LINK_SIZE); | |
1006 | } | |
1007 | } | |
1008 | } | |
1009 | } | |
1010 | ||
1011 | if (min >= 3) | |
1012 | { | |
1013 | common->private_data_ptrs[end - common->start - LINK_SIZE] = max_end - end; | |
1014 | common->private_data_ptrs[end - common->start - LINK_SIZE + 1] = OP_EXACT; | |
1015 | common->private_data_ptrs[end - common->start - LINK_SIZE + 2] = min; | |
1016 | return TRUE; | |
1017 | } | |
1018 | ||
1019 | return FALSE; | |
1020 | } | |
1021 | ||
1022 | #define CASE_ITERATOR_PRIVATE_DATA_1 \ | |
1023 | case OP_MINSTAR: \ | |
1024 | case OP_MINPLUS: \ | |
1025 | case OP_QUERY: \ | |
1026 | case OP_MINQUERY: \ | |
1027 | case OP_MINSTARI: \ | |
1028 | case OP_MINPLUSI: \ | |
1029 | case OP_QUERYI: \ | |
1030 | case OP_MINQUERYI: \ | |
1031 | case OP_NOTMINSTAR: \ | |
1032 | case OP_NOTMINPLUS: \ | |
1033 | case OP_NOTQUERY: \ | |
1034 | case OP_NOTMINQUERY: \ | |
1035 | case OP_NOTMINSTARI: \ | |
1036 | case OP_NOTMINPLUSI: \ | |
1037 | case OP_NOTQUERYI: \ | |
1038 | case OP_NOTMINQUERYI: | |
1039 | ||
1040 | #define CASE_ITERATOR_PRIVATE_DATA_2A \ | |
1041 | case OP_STAR: \ | |
1042 | case OP_PLUS: \ | |
1043 | case OP_STARI: \ | |
1044 | case OP_PLUSI: \ | |
1045 | case OP_NOTSTAR: \ | |
1046 | case OP_NOTPLUS: \ | |
1047 | case OP_NOTSTARI: \ | |
1048 | case OP_NOTPLUSI: | |
1049 | ||
1050 | #define CASE_ITERATOR_PRIVATE_DATA_2B \ | |
1051 | case OP_UPTO: \ | |
1052 | case OP_MINUPTO: \ | |
1053 | case OP_UPTOI: \ | |
1054 | case OP_MINUPTOI: \ | |
1055 | case OP_NOTUPTO: \ | |
1056 | case OP_NOTMINUPTO: \ | |
1057 | case OP_NOTUPTOI: \ | |
1058 | case OP_NOTMINUPTOI: | |
1059 | ||
1060 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \ | |
1061 | case OP_TYPEMINSTAR: \ | |
1062 | case OP_TYPEMINPLUS: \ | |
1063 | case OP_TYPEQUERY: \ | |
1064 | case OP_TYPEMINQUERY: | |
1065 | ||
1066 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \ | |
1067 | case OP_TYPESTAR: \ | |
1068 | case OP_TYPEPLUS: | |
1069 | ||
1070 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \ | |
1071 | case OP_TYPEUPTO: \ | |
1072 | case OP_TYPEMINUPTO: | |
1073 | ||
1074 | static void set_private_data_ptrs(compiler_common *common, int *private_data_start, pcre_uchar *ccend) | |
1075 | { | { |
1076 | int private_data_length = 0; | pcre_uchar *cc = common->start; |
1077 | pcre_uchar *alternative; | pcre_uchar *alternative; |
pcre_uchar *name; | ||
1078 | pcre_uchar *end = NULL; | pcre_uchar *end = NULL; |
1079 | int space, size, i; | int private_data_ptr = *private_data_start; |
1080 | pcre_uint32 bracketlen; | int space, size, bracketlen; |
1081 | ||
/* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ | ||
1082 | while (cc < ccend) | while (cc < ccend) |
1083 | { | { |
1084 | space = 0; | space = 0; |
1085 | size = 0; | size = 0; |
1086 | bracketlen = 0; | bracketlen = 0; |
1087 | if (private_data_ptr > SLJIT_MAX_LOCAL_SIZE) | |
1088 | return; | |
1089 | ||
1090 | if (*cc == OP_ONCE || *cc == OP_ONCE_NC || *cc == OP_BRA || *cc == OP_CBRA || *cc == OP_COND) | |
1091 | if (detect_repeat(common, cc)) | |
1092 | { | |
1093 | /* These brackets are converted to repeats, so no global | |
1094 | based single character repeat is allowed. */ | |
1095 | if (cc >= end) | |
1096 | end = bracketend(cc); | |
1097 | } | |
1098 | ||
1099 | switch(*cc) | switch(*cc) |
1100 | { | { |
1101 | case OP_SET_SOM: | case OP_KET: |
1102 | common->has_set_som = TRUE; | if (common->private_data_ptrs[cc + 1 - common->start] != 0) |
1103 | cc += 1; | { |
1104 | break; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
1105 | private_data_ptr += sizeof(sljit_sw); | |
1106 | case OP_REF: | cc += common->private_data_ptrs[cc + 1 - common->start]; |
1107 | case OP_REFI: | } |
1108 | common->optimized_cbracket[GET2(cc, 1)] = 0; | cc += 1 + LINK_SIZE; |
cc += 1 + IMM2_SIZE; | ||
1109 | break; | break; |
1110 | ||
1111 | case OP_ASSERT: | case OP_ASSERT: |
# | Line 799 while (cc < ccend) | Line 1117 while (cc < ccend) |
1117 | case OP_BRAPOS: | case OP_BRAPOS: |
1118 | case OP_SBRA: | case OP_SBRA: |
1119 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1120 | private_data_length += sizeof(sljit_sw); | case OP_SCOND: |
1121 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | |
1122 | private_data_ptr += sizeof(sljit_sw); | |
1123 | bracketlen = 1 + LINK_SIZE; | bracketlen = 1 + LINK_SIZE; |
1124 | break; | break; |
1125 | ||
1126 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1127 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1128 | private_data_length += sizeof(sljit_sw); | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
1129 | common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0; | private_data_ptr += sizeof(sljit_sw); |
1130 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; |
1131 | break; | break; |
1132 | ||
1133 | case OP_COND: | case OP_COND: |
1134 | case OP_SCOND: | /* Might be a hidden SCOND. */ |
1135 | bracketlen = cc[1 + LINK_SIZE]; | alternative = cc + GET(cc, 1); |
1136 | if (bracketlen == OP_CREF) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
{ | ||
bracketlen = GET2(cc, 1 + LINK_SIZE + 1); | ||
common->optimized_cbracket[bracketlen] = 0; | ||
} | ||
else if (bracketlen == OP_NCREF) | ||
{ | ||
bracketlen = GET2(cc, 1 + LINK_SIZE + 1); | ||
name = (pcre_uchar *)common->name_table; | ||
alternative = name; | ||
for (i = 0; i < common->name_count; i++) | ||
{ | ||
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++) | ||
{ | ||
if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0) | ||
common->optimized_cbracket[GET2(alternative, 0)] = 0; | ||
alternative += common->name_entry_size; | ||
} | ||
} | ||
if (*cc == OP_COND) | ||
1137 | { | { |
1138 | /* Might be a hidden SCOND. */ | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
1139 | alternative = cc + GET(cc, 1); | private_data_ptr += sizeof(sljit_sw); |
if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | ||
private_data_length += sizeof(sljit_sw); | ||
1140 | } | } |
else | ||
private_data_length += sizeof(sljit_sw); | ||
1141 | bracketlen = 1 + LINK_SIZE; | bracketlen = 1 + LINK_SIZE; |
1142 | break; | break; |
1143 | ||
# | Line 904 while (cc < ccend) | Line 1195 while (cc < ccend) |
1195 | break; | break; |
1196 | #endif | #endif |
1197 | ||
case OP_RECURSE: | ||
/* Set its value only once. */ | ||
if (common->recursive_head == 0) | ||
{ | ||
common->recursive_head = common->ovector_start; | ||
common->ovector_start += sizeof(sljit_sw); | ||
} | ||
cc += 1 + LINK_SIZE; | ||
break; | ||
case OP_MARK: | ||
if (common->mark_ptr == 0) | ||
{ | ||
common->mark_ptr = common->ovector_start; | ||
common->ovector_start += sizeof(sljit_sw); | ||
} | ||
cc += 1 + 2 + cc[1]; | ||
break; | ||
1198 | default: | default: |
1199 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1200 | if (cc == NULL) | SLJIT_ASSERT(cc != NULL); |
return -1; | ||
1201 | break; | break; |
1202 | } | } |
1203 | ||
1204 | /* Character iterators, which are not inside a repeated bracket, | |
1205 | gets a private slot instead of allocating it on the stack. */ | |
1206 | if (space > 0 && cc >= end) | if (space > 0 && cc >= end) |
1207 | private_data_length += sizeof(sljit_sw) * space; | { |
1208 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | |
1209 | private_data_ptr += sizeof(sljit_sw) * space; | |
1210 | } | |
1211 | ||
1212 | if (size != 0) | if (size != 0) |
1213 | { | { |
# | Line 946 while (cc < ccend) | Line 1222 while (cc < ccend) |
1222 | cc += size; | cc += size; |
1223 | } | } |
1224 | ||
1225 | if (bracketlen != 0) | if (bracketlen > 0) |
1226 | { | { |
1227 | if (cc >= end) | if (cc >= end) |
1228 | { | { |
# | Line 957 while (cc < ccend) | Line 1233 while (cc < ccend) |
1233 | cc += bracketlen; | cc += bracketlen; |
1234 | } | } |
1235 | } | } |
1236 | return private_data_length; | *private_data_start = private_data_ptr; |
1237 | } | } |
1238 | ||
1239 | static void set_private_data_ptrs(compiler_common *common, int private_data_ptr, pcre_uchar *ccend) | /* Returns with a frame_types (always < 0) if no need for frame. */ |
1240 | static int get_framesize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL recursive, BOOL* needs_control_head) | |
1241 | { | { |
1242 | pcre_uchar *cc = common->start; | int length = 0; |
1243 | pcre_uchar *alternative; | int possessive = 0; |
1244 | pcre_uchar *end = NULL; | BOOL stack_restore = FALSE; |
1245 | int space, size, bracketlen; | BOOL setsom_found = recursive; |
1246 | BOOL setmark_found = recursive; | |
1247 | /* The last capture is a local variable even for recursions. */ | |
1248 | BOOL capture_last_found = FALSE; | |
1249 | ||
1250 | while (cc < ccend) | #if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD |
1251 | SLJIT_ASSERT(common->control_head_ptr != 0); | |
1252 | *needs_control_head = TRUE; | |
1253 | #else | |
1254 | *needs_control_head = FALSE; | |
1255 | #endif | |
1256 | ||
1257 | if (ccend == NULL) | |
1258 | { | { |
1259 | space = 0; | ccend = bracketend(cc) - (1 + LINK_SIZE); |
1260 | size = 0; | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) |
1261 | bracketlen = 0; | { |
1262 | possessive = length = (common->capture_last_ptr != 0) ? 5 : 3; | |
1263 | /* This is correct regardless of common->capture_last_ptr. */ | |
1264 | capture_last_found = TRUE; | |
1265 | } | |
1266 | cc = next_opcode(common, cc); | |
1267 | } | |
1268 | ||
1269 | SLJIT_ASSERT(cc != NULL); | |
1270 | while (cc < ccend) | |
1271 | switch(*cc) | switch(*cc) |
1272 | { | { |
1273 | case OP_ASSERT: | case OP_SET_SOM: |
1274 | case OP_ASSERT_NOT: | SLJIT_ASSERT(common->has_set_som); |
1275 | case OP_ASSERTBACK: | stack_restore = TRUE; |
1276 | case OP_ASSERTBACK_NOT: | if (!setsom_found) |
1277 | case OP_ONCE: | { |
1278 | case OP_ONCE_NC: | length += 2; |
1279 | case OP_BRAPOS: | setsom_found = TRUE; |
1280 | case OP_SBRA: | } |
1281 | case OP_SBRAPOS: | cc += 1; |
case OP_SCOND: | ||
common->private_data_ptrs[cc - common->start] = private_data_ptr; | ||
private_data_ptr += sizeof(sljit_sw); | ||
bracketlen = 1 + LINK_SIZE; | ||
break; | ||
case OP_CBRAPOS: | ||
case OP_SCBRAPOS: | ||
common->private_data_ptrs[cc - common->start] = private_data_ptr; | ||
private_data_ptr += sizeof(sljit_sw); | ||
bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | ||
break; | ||
case OP_COND: | ||
/* Might be a hidden SCOND. */ | ||
alternative = cc + GET(cc, 1); | ||
if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | ||
{ | ||
common->private_data_ptrs[cc - common->start] = private_data_ptr; | ||
private_data_ptr += sizeof(sljit_sw); | ||
} | ||
bracketlen = 1 + LINK_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); | ||
break; | ||
#endif | ||
default: | ||
cc = next_opcode(common, cc); | ||
SLJIT_ASSERT(cc != NULL); | ||
break; | ||
} | ||
if (space > 0 && cc >= end) | ||
{ | ||
common->private_data_ptrs[cc - common->start] = private_data_ptr; | ||
private_data_ptr += sizeof(sljit_sw) * space; | ||
} | ||
if (size != 0) | ||
{ | ||
if (size < 0) | ||
{ | ||
cc += -size; | ||
#ifdef SUPPORT_UTF | ||
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | ||
#endif | ||
} | ||
else | ||
cc += size; | ||
} | ||
if (bracketlen > 0) | ||
{ | ||
if (cc >= end) | ||
{ | ||
end = bracketend(cc); | ||
if (end[-1 - LINK_SIZE] == OP_KET) | ||
end = NULL; | ||
} | ||
cc += bracketlen; | ||
} | ||
} | ||
} | ||
/* Returns with -1 if no need for frame. */ | ||
static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive) | ||
{ | ||
pcre_uchar *ccend = bracketend(cc); | ||
int length = 0; | ||
BOOL possessive = FALSE; | ||
BOOL setsom_found = recursive; | ||
BOOL setmark_found = recursive; | ||
if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) | ||
{ | ||
length = 3; | ||
possessive = TRUE; | ||
} | ||
cc = next_opcode(common, cc); | ||
SLJIT_ASSERT(cc != NULL); | ||
while (cc < ccend) | ||
switch(*cc) | ||
{ | ||
case OP_SET_SOM: | ||
SLJIT_ASSERT(common->has_set_som); | ||
if (!setsom_found) | ||
{ | ||
length += 2; | ||
setsom_found = TRUE; | ||
} | ||
cc += 1; | ||
1282 | break; | break; |
1283 | ||
1284 | case OP_MARK: | case OP_MARK: |
1285 | case OP_PRUNE_ARG: | |
1286 | case OP_THEN_ARG: | |
1287 | SLJIT_ASSERT(common->mark_ptr != 0); | SLJIT_ASSERT(common->mark_ptr != 0); |
1288 | stack_restore = TRUE; | |
1289 | if (!setmark_found) | if (!setmark_found) |
1290 | { | { |
1291 | length += 2; | length += 2; |
1292 | setmark_found = TRUE; | setmark_found = TRUE; |
1293 | } | } |
1294 | if (common->control_head_ptr != 0) | |
1295 | *needs_control_head = TRUE; | |
1296 | cc += 1 + 2 + cc[1]; | cc += 1 + 2 + cc[1]; |
1297 | break; | break; |
1298 | ||
1299 | case OP_RECURSE: | case OP_RECURSE: |
1300 | stack_restore = TRUE; | |
1301 | if (common->has_set_som && !setsom_found) | if (common->has_set_som && !setsom_found) |
1302 | { | { |
1303 | length += 2; | length += 2; |
# | Line 1150 while (cc < ccend) | Line 1308 while (cc < ccend) |
1308 | length += 2; | length += 2; |
1309 | setmark_found = TRUE; | setmark_found = TRUE; |
1310 | } | } |
1311 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1312 | { | |
1313 | length += 2; | |
1314 | capture_last_found = TRUE; | |
1315 | } | |
1316 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1317 | break; | break; |
1318 | ||
# | Line 1157 while (cc < ccend) | Line 1320 while (cc < ccend) |
1320 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1321 | case OP_SCBRA: | case OP_SCBRA: |
1322 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1323 | stack_restore = TRUE; | |
1324 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1325 | { | |
1326 | length += 2; | |
1327 | capture_last_found = TRUE; | |
1328 | } | |
1329 | length += 3; | length += 3; |
1330 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1331 | break; | break; |
1332 | ||
1333 | default: | default: |
1334 | stack_restore = TRUE; | |
1335 | /* Fall through. */ | |
1336 | ||
1337 | case OP_NOT_WORD_BOUNDARY: | |
1338 | case OP_WORD_BOUNDARY: | |
1339 | case OP_NOT_DIGIT: | |
1340 | case OP_DIGIT: | |
1341 | case OP_NOT_WHITESPACE: | |
1342 | case OP_WHITESPACE: | |
1343 | case OP_NOT_WORDCHAR: | |
1344 | case OP_WORDCHAR: | |
1345 | case OP_ANY: | |
1346 | case OP_ALLANY: | |
1347 | case OP_ANYBYTE: | |
1348 | case OP_NOTPROP: | |
1349 | case OP_PROP: | |
1350 | case OP_ANYNL: | |
1351 | case OP_NOT_HSPACE: | |
1352 | case OP_HSPACE: | |
1353 | case OP_NOT_VSPACE: | |
1354 | case OP_VSPACE: | |
1355 | case OP_EXTUNI: | |
1356 | case OP_EODN: | |
1357 | case OP_EOD: | |
1358 | case OP_CIRC: | |
1359 | case OP_CIRCM: | |
1360 | case OP_DOLL: | |
1361 | case OP_DOLLM: | |
1362 | case OP_CHAR: | |
1363 | case OP_CHARI: | |
1364 | case OP_NOT: | |
1365 | case OP_NOTI: | |
1366 | ||
1367 | case OP_EXACT: | |
1368 | case OP_POSSTAR: | |
1369 | case OP_POSPLUS: | |
1370 | case OP_POSQUERY: | |
1371 | case OP_POSUPTO: | |
1372 | ||
1373 | case OP_EXACTI: | |
1374 | case OP_POSSTARI: | |
1375 | case OP_POSPLUSI: | |
1376 | case OP_POSQUERYI: | |
1377 | case OP_POSUPTOI: | |
1378 | ||
1379 | case OP_NOTEXACT: | |
1380 | case OP_NOTPOSSTAR: | |
1381 | case OP_NOTPOSPLUS: | |
1382 | case OP_NOTPOSQUERY: | |
1383 | case OP_NOTPOSUPTO: | |
1384 | ||
1385 | case OP_NOTEXACTI: | |
1386 | case OP_NOTPOSSTARI: | |
1387 | case OP_NOTPOSPLUSI: | |
1388 | case OP_NOTPOSQUERYI: | |
1389 | case OP_NOTPOSUPTOI: | |
1390 | ||
1391 | case OP_TYPEEXACT: | |
1392 | case OP_TYPEPOSSTAR: | |
1393 | case OP_TYPEPOSPLUS: | |
1394 | case OP_TYPEPOSQUERY: | |
1395 | case OP_TYPEPOSUPTO: | |
1396 | ||
1397 | case OP_CLASS: | |
1398 | case OP_NCLASS: | |
1399 | case OP_XCLASS: | |
1400 | ||
1401 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1402 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1403 | break; | break; |
1404 | } | } |
1405 | ||
1406 | /* Possessive quantifiers can use a special case. */ | /* Possessive quantifiers can use a special case. */ |
1407 | if (SLJIT_UNLIKELY(possessive) && length == 3) | if (SLJIT_UNLIKELY(possessive == length)) |
1408 | return -1; | return stack_restore ? no_frame : no_stack; |
1409 | ||
1410 | if (length > 0) | if (length > 0) |
1411 | return length + 1; | return length + 1; |
1412 | return -1; | return stack_restore ? no_frame : no_stack; |
1413 | } | } |
1414 | ||
1415 | 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) |
1416 | { | { |
1417 | DEFINE_COMPILER; | DEFINE_COMPILER; |
pcre_uchar *ccend = bracketend(cc); | ||
1418 | BOOL setsom_found = recursive; | BOOL setsom_found = recursive; |
1419 | BOOL setmark_found = recursive; | BOOL setmark_found = recursive; |
1420 | /* The last capture is a local variable even for recursions. */ | |
1421 | BOOL capture_last_found = FALSE; | |
1422 | int offset; | int offset; |
1423 | ||
1424 | /* >= 1 + shortest item size (2) */ | /* >= 1 + shortest item size (2) */ |
# | Line 1189 SLJIT_UNUSED_ARG(stacktop); | Line 1426 SLJIT_UNUSED_ARG(stacktop); |
1426 | SLJIT_ASSERT(stackpos >= stacktop + 2); | SLJIT_ASSERT(stackpos >= stacktop + 2); |
1427 | ||
1428 | stackpos = STACK(stackpos); | stackpos = STACK(stackpos); |
1429 | if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS)) | if (ccend == NULL) |
1430 | cc = next_opcode(common, cc); | { |
1431 | ccend = bracketend(cc) - (1 + LINK_SIZE); | |
1432 | if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS)) | |
1433 | cc = next_opcode(common, cc); | |
1434 | } | |
1435 | ||
1436 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1437 | while (cc < ccend) | while (cc < ccend) |
1438 | switch(*cc) | switch(*cc) |
# | Line 1199 while (cc < ccend) | Line 1441 while (cc < ccend) |
1441 | SLJIT_ASSERT(common->has_set_som); | SLJIT_ASSERT(common->has_set_som); |
1442 | if (!setsom_found) | if (!setsom_found) |
1443 | { | { |
1444 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0)); |
1445 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); |
1446 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1447 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1448 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
# | Line 1210 while (cc < ccend) | Line 1452 while (cc < ccend) |
1452 | break; | break; |
1453 | ||
1454 | case OP_MARK: | case OP_MARK: |
1455 | case OP_PRUNE_ARG: | |
1456 | case OP_THEN_ARG: | |
1457 | SLJIT_ASSERT(common->mark_ptr != 0); | SLJIT_ASSERT(common->mark_ptr != 0); |
1458 | if (!setmark_found) | if (!setmark_found) |
1459 | { | { |
1460 | 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); |
1461 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); |
1462 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1463 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1464 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
# | Line 1226 while (cc < ccend) | Line 1470 while (cc < ccend) |
1470 | case OP_RECURSE: | case OP_RECURSE: |
1471 | if (common->has_set_som && !setsom_found) | if (common->has_set_som && !setsom_found) |
1472 | { | { |
1473 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0)); |
1474 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); |
1475 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1476 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1477 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
# | Line 1235 while (cc < ccend) | Line 1479 while (cc < ccend) |
1479 | } | } |
1480 | if (common->mark_ptr != 0 && !setmark_found) | if (common->mark_ptr != 0 && !setmark_found) |
1481 | { | { |
1482 | 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); |
1483 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); |
1484 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1485 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1486 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1487 | setmark_found = TRUE; | setmark_found = TRUE; |
1488 | } | } |
1489 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1490 | { | |
1491 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr); | |
1492 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); | |
1493 | stackpos += (int)sizeof(sljit_sw); | |
1494 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1495 | stackpos += (int)sizeof(sljit_sw); | |
1496 | capture_last_found = TRUE; | |
1497 | } | |
1498 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1499 | break; | break; |
1500 | ||
# | Line 1249 while (cc < ccend) | Line 1502 while (cc < ccend) |
1502 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1503 | case OP_SCBRA: | case OP_SCBRA: |
1504 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1505 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1506 | { | |
1507 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr); | |
1508 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); | |
1509 | stackpos += (int)sizeof(sljit_sw); | |
1510 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1511 | stackpos += (int)sizeof(sljit_sw); | |
1512 | capture_last_found = TRUE; | |
1513 | } | |
1514 | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; |
1515 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); |
1516 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1517 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset)); |
1518 | 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)); |
1519 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1520 | stackpos += (int)sizeof(sljit_sw); | stackpos += (int)sizeof(sljit_sw); |
1521 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); |
# | Line 1268 while (cc < ccend) | Line 1530 while (cc < ccend) |
1530 | break; | break; |
1531 | } | } |
1532 | ||
1533 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_end); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, 0); |
1534 | SLJIT_ASSERT(stackpos == STACK(stacktop)); | SLJIT_ASSERT(stackpos == STACK(stacktop)); |
1535 | } | } |
1536 | ||
1537 | 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) |
1538 | { | { |
1539 | int private_data_length = 2; | int private_data_length = needs_control_head ? 3 : 2; |
1540 | int size; | int size; |
1541 | pcre_uchar *alternative; | pcre_uchar *alternative; |
1542 | /* Calculate the sum of the private machine words. */ | /* Calculate the sum of the private machine words. */ |
# | Line 1283 while (cc < ccend) | Line 1545 while (cc < ccend) |
1545 | size = 0; | size = 0; |
1546 | switch(*cc) | switch(*cc) |
1547 | { | { |
1548 | case OP_KET: | |
1549 | if (PRIVATE_DATA(cc) != 0) | |
1550 | private_data_length++; | |
1551 | cc += 1 + LINK_SIZE; | |
1552 | break; | |
1553 | ||
1554 | case OP_ASSERT: | case OP_ASSERT: |
1555 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
1556 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
# | Line 1387 return private_data_length; | Line 1655 return private_data_length; |
1655 | } | } |
1656 | ||
1657 | 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, |
1658 | BOOL save, int stackptr, int stacktop) | BOOL save, int stackptr, int stacktop, BOOL needs_control_head) |
1659 | { | { |
1660 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1661 | int srcw[2]; | int srcw[2]; |
# | Line 1408 stacktop = STACK(stacktop - 1); | Line 1676 stacktop = STACK(stacktop - 1); |
1676 | ||
1677 | if (!save) | if (!save) |
1678 | { | { |
1679 | stackptr += sizeof(sljit_sw); | stackptr += (needs_control_head ? 2 : 1) * sizeof(sljit_sw); |
1680 | if (stackptr < stacktop) | if (stackptr < stacktop) |
1681 | { | { |
1682 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
# | Line 1424 if (!save) | Line 1692 if (!save) |
1692 | /* The tmp1next must be TRUE in either way. */ | /* The tmp1next must be TRUE in either way. */ |
1693 | } | } |
1694 | ||
1695 | while (status != end) | do |
1696 | { | { |
1697 | count = 0; | count = 0; |
1698 | switch(status) | switch(status) |
1699 | { | { |
1700 | case start: | case start: |
1701 | SLJIT_ASSERT(save && common->recursive_head != 0); | SLJIT_ASSERT(save && common->recursive_head_ptr != 0); |
1702 | count = 1; | count = 1; |
1703 | srcw[0] = common->recursive_head; | srcw[0] = common->recursive_head_ptr; |
1704 | if (needs_control_head) | |
1705 | { | |
1706 | SLJIT_ASSERT(common->control_head_ptr != 0); | |
1707 | count = 2; | |
1708 | srcw[1] = common->control_head_ptr; | |
1709 | } | |
1710 | status = loop; | status = loop; |
1711 | break; | break; |
1712 | ||
# | Line 1445 while (status != end) | Line 1719 while (status != end) |
1719 | ||
1720 | switch(*cc) | switch(*cc) |
1721 | { | { |
1722 | case OP_KET: | |
1723 | if (PRIVATE_DATA(cc) != 0) | |
1724 | { | |
1725 | count = 1; | |
1726 | srcw[0] = PRIVATE_DATA(cc); | |
1727 | } | |
1728 | cc += 1 + LINK_SIZE; | |
1729 | break; | |
1730 | ||
1731 | case OP_ASSERT: | case OP_ASSERT: |
1732 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
1733 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
# | Line 1612 while (status != end) | Line 1895 while (status != end) |
1895 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1896 | stackptr += sizeof(sljit_sw); | stackptr += sizeof(sljit_sw); |
1897 | } | } |
1898 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), srcw[count]); |
1899 | tmp1empty = FALSE; | tmp1empty = FALSE; |
1900 | tmp1next = FALSE; | tmp1next = FALSE; |
1901 | } | } |
# | Line 1623 while (status != end) | Line 1906 while (status != end) |
1906 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1907 | stackptr += sizeof(sljit_sw); | stackptr += sizeof(sljit_sw); |
1908 | } | } |
1909 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), srcw[count]); |
1910 | tmp2empty = FALSE; | tmp2empty = FALSE; |
1911 | tmp1next = TRUE; | tmp1next = TRUE; |
1912 | } | } |
# | Line 1633 while (status != end) | Line 1916 while (status != end) |
1916 | if (tmp1next) | if (tmp1next) |
1917 | { | { |
1918 | SLJIT_ASSERT(!tmp1empty); | SLJIT_ASSERT(!tmp1empty); |
1919 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count], TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), srcw[count], TMP1, 0); |
1920 | tmp1empty = stackptr >= stacktop; | tmp1empty = stackptr >= stacktop; |
1921 | if (!tmp1empty) | if (!tmp1empty) |
1922 | { | { |
# | Line 1645 while (status != end) | Line 1928 while (status != end) |
1928 | else | else |
1929 | { | { |
1930 | SLJIT_ASSERT(!tmp2empty); | SLJIT_ASSERT(!tmp2empty); |
1931 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count], TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), srcw[count], TMP2, 0); |
1932 | tmp2empty = stackptr >= stacktop; | tmp2empty = stackptr >= stacktop; |
1933 | if (!tmp2empty) | if (!tmp2empty) |
1934 | { | { |
# | Line 1657 while (status != end) | Line 1940 while (status != end) |
1940 | } | } |
1941 | } | } |
1942 | } | } |
1943 | while (status != end); | |
1944 | ||
1945 | if (save) | if (save) |
1946 | { | { |
# | Line 1690 if (save) | Line 1974 if (save) |
1974 | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); |
1975 | } | } |
1976 | ||
1977 | static SLJIT_INLINE pcre_uchar *set_then_offsets(compiler_common *common, pcre_uchar *cc, pcre_uint8 *current_offset) | |
1978 | { | |
1979 | pcre_uchar *end = bracketend(cc); | |
1980 | BOOL has_alternatives = cc[GET(cc, 1)] == OP_ALT; | |
1981 | ||
1982 | /* Assert captures then. */ | |
1983 | if (*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) | |
1984 | current_offset = NULL; | |
1985 | /* Conditional block does not. */ | |
1986 | if (*cc == OP_COND || *cc == OP_SCOND) | |
1987 | has_alternatives = FALSE; | |
1988 | ||
1989 | cc = next_opcode(common, cc); | |
1990 | if (has_alternatives) | |
1991 | current_offset = common->then_offsets + (cc - common->start); | |
1992 | ||
1993 | while (cc < end) | |
1994 | { | |
1995 | if ((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)) | |
1996 | cc = set_then_offsets(common, cc, current_offset); | |
1997 | else | |
1998 | { | |
1999 | if (*cc == OP_ALT && has_alternatives) | |
2000 | current_offset = common->then_offsets + (cc + 1 + LINK_SIZE - common->start); | |
2001 | if (*cc >= OP_THEN && *cc <= OP_THEN_ARG && current_offset != NULL) | |
2002 | *current_offset = 1; | |
2003 | cc = next_opcode(common, cc); | |
2004 | } | |
2005 | } | |
2006 | ||
2007 | return end; | |
2008 | } | |
2009 | ||
2010 | #undef CASE_ITERATOR_PRIVATE_DATA_1 | #undef CASE_ITERATOR_PRIVATE_DATA_1 |
2011 | #undef CASE_ITERATOR_PRIVATE_DATA_2A | #undef CASE_ITERATOR_PRIVATE_DATA_2A |
2012 | #undef CASE_ITERATOR_PRIVATE_DATA_2B | #undef CASE_ITERATOR_PRIVATE_DATA_2B |
# | Line 1708 while (list) | Line 2025 while (list) |
2025 | { | { |
2026 | /* sljit_set_label is clever enough to do nothing | /* sljit_set_label is clever enough to do nothing |
2027 | if either the jump or the label is NULL. */ | if either the jump or the label is NULL. */ |
2028 | sljit_set_label(list->jump, label); | SET_LABEL(list->jump, label); |
2029 | list = list->next; | list = list->next; |
2030 | } | } |
2031 | } | } |
# | Line 1724 if (list_item) | Line 2041 if (list_item) |
2041 | } | } |
2042 | } | } |
2043 | ||
2044 | 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) |
2045 | { | { |
2046 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2047 | stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); | stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); |
2048 | ||
2049 | if (list_item) | if (list_item) |
2050 | { | { |
list_item->type = type; | ||
list_item->data = data; | ||
2051 | list_item->start = start; | list_item->start = start; |
2052 | list_item->quit = LABEL(); | list_item->quit = LABEL(); |
2053 | list_item->next = common->stubs; | list_item->next = common->stubs; |
# | Line 1748 stub_list* list_item = common->stubs; | Line 2063 stub_list* list_item = common->stubs; |
2063 | while (list_item) | while (list_item) |
2064 | { | { |
2065 | JUMPHERE(list_item->start); | JUMPHERE(list_item->start); |
2066 | 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; | ||
} | ||
2067 | JUMPTO(SLJIT_JUMP, list_item->quit); | JUMPTO(SLJIT_JUMP, list_item->quit); |
2068 | list_item = list_item->next; | list_item = list_item->next; |
2069 | } | } |
2070 | common->stubs = NULL; | common->stubs = NULL; |
2071 | } | } |
2072 | ||
2073 | static SLJIT_INLINE void decrease_call_count(compiler_common *common) | static void add_label_addr(compiler_common *common, sljit_uw *update_addr) |
2074 | { | { |
2075 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2076 | label_addr_list *label_addr; | |
2077 | ||
2078 | OP2(SLJIT_SUB | SLJIT_SET_E, CALL_COUNT, 0, CALL_COUNT, 0, SLJIT_IMM, 1); | label_addr = sljit_alloc_memory(compiler, sizeof(label_addr_list)); |
2079 | add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO)); | if (label_addr == NULL) |
2080 | return; | |
2081 | label_addr->label = LABEL(); | |
2082 | label_addr->update_addr = update_addr; | |
2083 | label_addr->next = common->label_addrs; | |
2084 | common->label_addrs = label_addr; | |
2085 | } | |
2086 | ||
2087 | static SLJIT_INLINE void count_match(compiler_common *common) | |
2088 | { | |
2089 | DEFINE_COMPILER; | |
2090 | ||
2091 | OP2(SLJIT_SUB | SLJIT_SET_E, COUNT_MATCH, 0, COUNT_MATCH, 0, SLJIT_IMM, 1); | |
2092 | add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO)); | |
2093 | } | } |
2094 | ||
2095 | static SLJIT_INLINE void allocate_stack(compiler_common *common, int size) | static SLJIT_INLINE void allocate_stack(compiler_common *common, int size) |
# | Line 1778 OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, | Line 2102 OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, |
2102 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); |
2103 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
2104 | OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); | OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); |
2105 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS0, TMP1, 0); |
2106 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS1, TMP1, 0); |
2107 | #endif | #endif |
2108 | add_stub(common, stack_alloc, 0, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); | add_stub(common, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); |
2109 | } | } |
2110 | ||
2111 | static SLJIT_INLINE void free_stack(compiler_common *common, int size) | static SLJIT_INLINE void free_stack(compiler_common *common, int size) |
# | Line 1795 static SLJIT_INLINE void reset_ovector(c | Line 2119 static SLJIT_INLINE void reset_ovector(c |
2119 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2120 | struct sljit_label *loop; | struct sljit_label *loop; |
2121 | int i; | int i; |
2122 | ||
2123 | /* At this point we can freely use all temporary registers. */ | /* At this point we can freely use all temporary registers. */ |
2124 | SLJIT_ASSERT(length > 1); | |
2125 | /* TMP1 returns with begin - 1. */ | /* TMP1 returns with begin - 1. */ |
2126 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_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)); |
2127 | if (length < 8) | |
2128 | { | |
2129 | for (i = 1; i < length; i++) | |
2130 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(i), SLJIT_R0, 0); | |
2131 | } | |
2132 | else | |
2133 | { | |
2134 | GET_LOCAL_BASE(SLJIT_R1, 0, OVECTOR_START); | |
2135 | OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_IMM, length - 1); | |
2136 | loop = LABEL(); | |
2137 | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_R1), sizeof(sljit_sw), SLJIT_R0, 0); | |
2138 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, 1); | |
2139 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | |
2140 | } | |
2141 | } | |
2142 | ||
2143 | static SLJIT_INLINE void do_reset_match(compiler_common *common, int length) | |
2144 | { | |
2145 | DEFINE_COMPILER; | |
2146 | struct sljit_label *loop; | |
2147 | int i; | |
2148 | ||
2149 | SLJIT_ASSERT(length > 1); | |
2150 | /* OVECTOR(1) contains the "string begin - 1" constant. */ | |
2151 | if (length > 2) | |
2152 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1)); | |
2153 | if (length < 8) | if (length < 8) |
2154 | { | { |
2155 | for (i = 0; i < length; i++) | for (i = 2; i < length; i++) |
2156 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_TEMPORARY_REG1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(i), TMP1, 0); |
2157 | } | } |
2158 | else | else |
2159 | { | { |
2160 | GET_LOCAL_BASE(SLJIT_TEMPORARY_REG2, 0, OVECTOR_START - sizeof(sljit_sw)); | GET_LOCAL_BASE(TMP2, 0, OVECTOR_START + sizeof(sljit_sw)); |
2161 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, length); | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, length - 2); |
2162 | loop = LABEL(); | loop = LABEL(); |
2163 | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(sljit_sw), SLJIT_TEMPORARY_REG1, 0); | OP1(SLJIT_MOVU, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0); |
2164 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 1); |
2165 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_C_NOT_ZERO, loop); |
2166 | } | } |
2167 | ||
2168 | OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0); | |
2169 | if (common->mark_ptr != 0) | |
2170 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->mark_ptr, SLJIT_IMM, 0); | |
2171 | if (common->control_head_ptr != 0) | |
2172 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_IMM, 0); | |
2173 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack)); | |
2174 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->start_ptr); | |
2175 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, base)); | |
2176 | } | |
2177 | ||
2178 | static sljit_sw SLJIT_CALL do_search_mark(sljit_sw *current, const pcre_uchar *skip_arg) | |
2179 | { | |
2180 | while (current != NULL) | |
2181 | { | |
2182 | switch (current[-2]) | |
2183 | { | |
2184 | case type_then_trap: | |
2185 | break; | |
2186 | ||
2187 | case type_mark: | |
2188 | if (STRCMP_UC_UC(skip_arg, (pcre_uchar *)current[-3]) == 0) | |
2189 | return current[-4]; | |
2190 | break; | |
2191 | ||
2192 | default: | |
2193 | SLJIT_ASSERT_STOP(); | |
2194 | break; | |
2195 | } | |
2196 | current = (sljit_sw*)current[-1]; | |
2197 | } | |
2198 | return -1; | |
2199 | } | } |
2200 | ||
2201 | static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket) | static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket) |
2202 | { | { |
2203 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2204 | struct sljit_label *loop; | struct sljit_label *loop; |
2205 | struct sljit_jump *earlyexit; | struct sljit_jump *early_quit; |
2206 | ||
2207 | /* At this point we can freely use all registers. */ | /* At this point we can freely use all registers. */ |
2208 | 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)); |
2209 | 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); |
2210 | ||
2211 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_R0, 0, ARGUMENTS, 0); |
2212 | if (common->mark_ptr != 0) | if (common->mark_ptr != 0) |
2213 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_SP), common->mark_ptr); |
2214 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | OP1(SLJIT_MOV_SI, SLJIT_R1, 0, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, offset_count)); |
2215 | if (common->mark_ptr != 0) | if (common->mark_ptr != 0) |
2216 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_TEMPORARY_REG3, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_R2, 0); |
2217 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_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)); |
2218 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, SLJIT_R0, 0, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, begin)); |
2219 | GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START); | GET_LOCAL_BASE(SLJIT_S0, 0, OVECTOR_START); |
2220 | /* Unlikely, but possible */ | /* Unlikely, but possible */ |
2221 | earlyexit = CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 0); | early_quit = CMP(SLJIT_C_EQUAL, SLJIT_R1, 0, SLJIT_IMM, 0); |
2222 | loop = LABEL(); | loop = LABEL(); |
2223 | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_TEMPORARY_REG1, 0); | OP2(SLJIT_SUB, SLJIT_S1, 0, SLJIT_MEM1(SLJIT_S0), 0, SLJIT_R0, 0); |
2224 | 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)); |
2225 | /* Copy the integer value to the output buffer */ | /* Copy the integer value to the output buffer */ |
2226 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2227 | 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); |
2228 | #endif | #endif |
2229 | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_R2), sizeof(int), SLJIT_S1, 0); |
2230 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_R1, 0, SLJIT_R1, 0, SLJIT_IMM, 1); |
2231 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_C_NOT_ZERO, loop); |
2232 | JUMPHERE(earlyexit); | JUMPHERE(early_quit); |
2233 | ||
2234 | /* Calculate the return value, which is the maximum ovector value. */ | /* Calculate the return value, which is the maximum ovector value. */ |
2235 | if (topbracket > 1) | if (topbracket > 1) |
2236 | { | { |
2237 | GET_LOCAL_BASE(SLJIT_TEMPORARY_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw)); | GET_LOCAL_BASE(SLJIT_R0, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw)); |
2238 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, topbracket + 1); | OP1(SLJIT_MOV, SLJIT_R1, 0, SLJIT_IMM, topbracket + 1); |
2239 | ||
2240 | /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ | /* OVECTOR(0) is never equal to SLJIT_S2. */ |
2241 | loop = LABEL(); | loop = LABEL(); |
2242 | OP1(SLJIT_MOVU, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), -(2 * (sljit_sw)sizeof(sljit_sw))); | OP1(SLJIT_MOVU, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_R0), -(2 * (sljit_sw)sizeof(sljit_sw))); |
2243 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, SLJIT_R1, 0, SLJIT_R1, 0, SLJIT_IMM, 1); |
2244 | CMPTO(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG3, 0, loop); | CMPTO(SLJIT_C_EQUAL, SLJIT_R2, 0, SLJIT_S2, 0, loop); |
2245 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_TEMPORARY_REG2, 0); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_R1, 0); |
2246 | } | } |
2247 | else | else |
2248 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); |
# | Line 1867 else | Line 2251 else |
2251 | 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) |
2252 | { | { |
2253 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2254 | struct sljit_jump *jump; | |
2255 | ||
2256 | 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); |
2257 | 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 |
2258 | && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); | |
2259 | ||
2260 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_R1, 0, ARGUMENTS, 0); |
2261 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); |
2262 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | OP1(SLJIT_MOV_SI, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_R1), SLJIT_OFFSETOF(jit_arguments, real_offset_count)); |
2263 | CMPTO(SLJIT_C_LESS, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 2, quit); | CMPTO(SLJIT_C_SIG_LESS, SLJIT_R2, 0, SLJIT_IMM, 2, quit); |
2264 | ||
2265 | /* Store match begin and end. */ | /* Store match begin and end. */ |
2266 | OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, SLJIT_S0, 0, SLJIT_MEM1(SLJIT_R1), SLJIT_OFFSETOF(jit_arguments, begin)); |
2267 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); | OP1(SLJIT_MOV, SLJIT_R1, 0, SLJIT_MEM1(SLJIT_R1), SLJIT_OFFSETOF(jit_arguments, offsets)); |
2268 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start); | |
2269 | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0); | jump = CMP(SLJIT_C_SIG_LESS, SLJIT_R2, 0, SLJIT_IMM, 3); |
2270 | 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); | |
2271 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | |
2272 | OP2(SLJIT_ASHR, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, UCHAR_SHIFT); | |
2273 | #endif | |
2274 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_R1), 2 * sizeof(int), SLJIT_R2, 0); | |
2275 | JUMPHERE(jump); | |
2276 | ||
2277 | OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_SP), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start); | |
2278 | OP2(SLJIT_SUB, SLJIT_S1, 0, STR_END, 0, SLJIT_S0, 0); | |
2279 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2280 | 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); |
2281 | #endif | #endif |
2282 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_R1), sizeof(int), SLJIT_S1, 0); |
2283 | ||
2284 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG1, 0); | OP2(SLJIT_SUB, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_S0, 0); |
2285 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2286 | OP2(SLJIT_ASHR, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_ASHR, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, UCHAR_SHIFT); |
2287 | #endif | #endif |
2288 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), 0, SLJIT_TEMPORARY_REG3, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_R1), 0, SLJIT_R2, 0); |
2289 | ||
2290 | JUMPTO(SLJIT_JUMP, quit); | JUMPTO(SLJIT_JUMP, quit); |
2291 | } | } |
# | Line 1904 struct sljit_jump *jump; | Line 2299 struct sljit_jump *jump; |
2299 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2300 | { | { |
2301 | /* The value of -1 must be kept for start_used_ptr! */ | /* The value of -1 must be kept for start_used_ptr! */ |
2302 | 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); |
2303 | /* 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 |
2304 | 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. */ |
2305 | jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_PTR, 0); | jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_PTR, 0); |
2306 | 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); |
2307 | JUMPHERE(jump); | JUMPHERE(jump); |
2308 | } | } |
2309 | else if (common->mode == JIT_PARTIAL_HARD_COMPILE) | else if (common->mode == JIT_PARTIAL_HARD_COMPILE) |
2310 | { | { |
2311 | jump = CMP(SLJIT_C_LESS_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | jump = CMP(SLJIT_C_LESS_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0); |
2312 | 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); |
2313 | JUMPHERE(jump); | JUMPHERE(jump); |
2314 | } | } |
2315 | } | } |
# | Line 2040 return (bit < 256) ? ((0 << 8) | bit) : | Line 2435 return (bit < 256) ? ((0 << 8) | bit) : |
2435 | ||
2436 | static void check_partial(compiler_common *common, BOOL force) | static void check_partial(compiler_common *common, BOOL force) |
2437 | { | { |
2438 | /* Checks whether a partial matching is occured. Does not modify registers. */ | /* Checks whether a partial matching is occurred. Does not modify registers. */ |
2439 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2440 | struct sljit_jump *jump = NULL; | struct sljit_jump *jump = NULL; |
2441 | ||
# | Line 2050 if (common->mode == JIT_COMPILE) | Line 2445 if (common->mode == JIT_COMPILE) |
2445 | return; | return; |
2446 | ||
2447 | if (!force) | if (!force) |
2448 | jump = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); | jump = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0); |
2449 | else if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | else if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2450 | jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1); | jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, SLJIT_IMM, -1); |
2451 | ||
2452 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2453 | 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); |
2454 | else | else |
2455 | { | { |
2456 | if (common->partialmatchlabel != NULL) | if (common->partialmatchlabel != NULL) |
# | Line 2068 if (jump != NULL) | Line 2463 if (jump != NULL) |
2463 | JUMPHERE(jump); | JUMPHERE(jump); |
2464 | } | } |
2465 | ||
2466 | static struct sljit_jump *check_str_end(compiler_common *common) | static void check_str_end(compiler_common *common, jump_list **end_reached) |
2467 | { | { |
2468 | /* Does not affect registers. Usually used in a tight spot. */ | /* Does not affect registers. Usually used in a tight spot. */ |
2469 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2470 | struct sljit_jump *jump; | struct sljit_jump *jump; |
struct sljit_jump *nohit; | ||
struct sljit_jump *return_value; | ||
2471 | ||
2472 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
2473 | return CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | { |
2474 | add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | |
2475 | return; | |
2476 | } | |
2477 | ||
2478 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); |
2479 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2480 | { | { |
2481 | 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_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0)); |
2482 | 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); |
2483 | JUMPHERE(nohit); | add_jump(compiler, end_reached, JUMP(SLJIT_JUMP)); |
return_value = JUMP(SLJIT_JUMP); | ||
2484 | } | } |
2485 | else | else |
2486 | { | { |
2487 | 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_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0)); |
2488 | if (common->partialmatchlabel != NULL) | if (common->partialmatchlabel != NULL) |
2489 | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); | JUMPTO(SLJIT_JUMP, common->partialmatchlabel); |
2490 | else | else |
2491 | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); | add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); |
2492 | } | } |
2493 | JUMPHERE(jump); | JUMPHERE(jump); |
return return_value; | ||
2494 | } | } |
2495 | ||
2496 | static void detect_partial_match(compiler_common *common, jump_list **backtracks) | static void detect_partial_match(compiler_common *common, jump_list **backtracks) |
# | Line 2112 if (common->mode == JIT_COMPILE) | Line 2506 if (common->mode == JIT_COMPILE) |
2506 | ||
2507 | /* Partial matching mode. */ | /* Partial matching mode. */ |
2508 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); |
2509 | 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_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0)); |
2510 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2511 | { | { |
2512 | 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); |
2513 | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
2514 | } | } |
2515 | else | else |
# | Line 2128 else | Line 2522 else |
2522 | JUMPHERE(jump); | JUMPHERE(jump); |
2523 | } | } |
2524 | ||
2525 | static void read_char(compiler_common *common) | static void peek_char(compiler_common *common, pcre_uint32 max) |
2526 | { | { |
2527 | /* Reads the character into TMP1, updates STR_PTR. | /* Reads the character into TMP1, keeps STR_PTR. |
2528 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2529 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2530 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2531 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2532 | #endif | #endif |
2533 | ||
2534 | SLJIT_UNUSED_ARG(max); | |
2535 | ||
2536 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2537 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 |
2538 | if (common->utf) | if (common->utf) |
2539 | { | { |
2540 | #if defined COMPILE_PCRE8 | if (max < 128) return; |
2541 | ||
2542 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2543 | #elif defined COMPILE_PCRE16 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | ||
#endif /* COMPILE_PCRE[8|16] */ | ||
2544 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2545 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | |
2546 | JUMPHERE(jump); | JUMPHERE(jump); |
2547 | } | } |
2548 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2549 | ||
2550 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
2551 | if (common->utf) | |
2552 | { | |
2553 | if (max < 0xd800) return; | |
2554 | ||
2555 | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
2556 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1); | |
2557 | /* TMP2 contains the high surrogate. */ | |
2558 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2559 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40); | |
2560 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10); | |
2561 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff); | |
2562 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2563 | JUMPHERE(jump); | |
2564 | } | |
2565 | #endif | |
2566 | } | |
2567 | ||
2568 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
2569 | ||
2570 | static BOOL is_char7_bitset(const pcre_uint8 *bitset, BOOL nclass) | |
2571 | { | |
2572 | /* Tells whether the character codes below 128 are enough | |
2573 | to determine a match. */ | |
2574 | const pcre_uint8 value = nclass ? 0xff : 0; | |
2575 | const pcre_uint8* end = bitset + 32; | |
2576 | ||
2577 | bitset += 16; | |
2578 | do | |
2579 | { | |
2580 | if (*bitset++ != value) | |
2581 | return FALSE; | |
2582 | } | |
2583 | while (bitset < end); | |
2584 | return TRUE; | |
2585 | } | |
2586 | ||
2587 | static void read_char7_type(compiler_common *common, BOOL full_read) | |
2588 | { | |
2589 | /* Reads the precise character type of a character into TMP1, if the character | |
2590 | is less than 128. Otherwise it returns with zero. Does not check STR_END. The | |
2591 | full_read argument tells whether characters above max are accepted or not. */ | |
2592 | DEFINE_COMPILER; | |
2593 | struct sljit_jump *jump; | |
2594 | ||
2595 | SLJIT_ASSERT(common->utf); | |
2596 | ||
2597 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | |
2598 | 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)); |
2599 | ||
2600 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
2601 | ||
2602 | if (full_read) | |
2603 | { | |
2604 | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); | |
2605 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
2606 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | |
2607 | JUMPHERE(jump); | |
2608 | } | |
2609 | } | } |
2610 | ||
2611 | static void peek_char(compiler_common *common) | #endif /* SUPPORT_UTF && COMPILE_PCRE8 */ |
2612 | ||
2613 | static void read_char_range(compiler_common *common, pcre_uint32 min, pcre_uint32 max, BOOL update_str_ptr) | |
2614 | { | { |
2615 | /* Reads the character into TMP1, keeps STR_PTR. | /* Reads the precise value of a character into TMP1, if the character is |
2616 | Does not check STR_END. TMP2 Destroyed. */ | between min and max (c >= min && c <= max). Otherwise it returns with a value |
2617 | outside the range. Does not check STR_END. */ | |
2618 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2619 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2620 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2621 | #endif | #endif |
2622 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
2623 | struct sljit_jump *jump2; | |
2624 | #endif | |
2625 | ||
2626 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | SLJIT_UNUSED_ARG(update_str_ptr); |
2627 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | SLJIT_UNUSED_ARG(min); |
2628 | SLJIT_UNUSED_ARG(max); | |
2629 | SLJIT_ASSERT(min <= max); | |
2630 | ||
2631 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2632 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2633 | ||
2634 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
2635 | if (common->utf) | if (common->utf) |
2636 | { | { |
2637 | #if defined COMPILE_PCRE8 | if (max < 128 && !update_str_ptr) return; |
2638 | ||
2639 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2640 | #elif defined COMPILE_PCRE16 | if (min >= 0x10000) |
2641 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | { |
2642 | #endif /* COMPILE_PCRE[8|16] */ | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xf0); |
2643 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | if (update_str_ptr) |
2644 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); |
2645 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2646 | jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0x7); | |
2647 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | |
2648 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | |
2649 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2650 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2651 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2652 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | |
2653 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2654 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); | |
2655 | if (!update_str_ptr) | |
2656 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3)); | |
2657 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2658 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | |
2659 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2660 | JUMPHERE(jump2); | |
2661 | if (update_str_ptr) | |
2662 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0); | |
2663 | } | |
2664 | else if (min >= 0x800 && max <= 0xffff) | |
2665 | { | |
2666 | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xe0); | |
2667 | if (update_str_ptr) | |
2668 | OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
2669 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2670 | jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xf); | |
2671 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | |
2672 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | |
2673 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2674 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2675 | if (!update_str_ptr) | |
2676 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | |
2677 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2678 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | |
2679 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2680 | JUMPHERE(jump2); | |
2681 | if (update_str_ptr) | |
2682 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0); | |
2683 | } | |
2684 | else if (max >= 0x800) | |
2685 | add_jump(compiler, (max < 0x10000) ? &common->utfreadchar16 : &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | |
2686 | else if (max < 128) | |
2687 | { | |
2688 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
2689 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | |
2690 | } | |
2691 | else | |
2692 | { | |
2693 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2694 | if (!update_str_ptr) | |
2695 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2696 | else | |
2697 | OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
2698 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | |
2699 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2700 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | |
2701 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2702 | if (update_str_ptr) | |
2703 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0); | |
2704 | } | |
2705 | JUMPHERE(jump); | JUMPHERE(jump); |
2706 | } | } |
2707 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | #endif |
2708 | ||
2709 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
2710 | if (common->utf) | |
2711 | { | |
2712 | if (max >= 0x10000) | |
2713 | { | |
2714 | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
2715 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1); | |
2716 | /* TMP2 contains the high surrogate. */ | |
2717 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2718 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40); | |
2719 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10); | |
2720 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2721 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff); | |
2722 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2723 | JUMPHERE(jump); | |
2724 | return; | |
2725 | } | |
2726 | ||
2727 | if (max < 0xd800 && !update_str_ptr) return; | |
2728 | ||
2729 | /* Skip low surrogate if necessary. */ | |
2730 | OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
2731 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1); | |
2732 | if (update_str_ptr) | |
2733 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2734 | if (max >= 0xd800) | |
2735 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0x10000); | |
2736 | JUMPHERE(jump); | |
2737 | } | |
2738 | #endif | |
2739 | } | |
2740 | ||
2741 | static SLJIT_INLINE void read_char(compiler_common *common) | |
2742 | { | |
2743 | read_char_range(common, 0, READ_CHAR_MAX, TRUE); | |
2744 | } | } |
2745 | ||
2746 | static void read_char8_type(compiler_common *common) | static void read_char8_type(compiler_common *common, BOOL update_str_ptr) |
2747 | { | { |
2748 | /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */ | /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */ |
2749 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2750 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
2751 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2752 | #endif | #endif |
2753 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
2754 | struct sljit_jump *jump2; | |
2755 | #endif | |
2756 | ||
2757 | #ifdef SUPPORT_UTF | SLJIT_UNUSED_ARG(update_str_ptr); |
2758 | ||
2759 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | |
2760 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2761 | ||
2762 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
2763 | if (common->utf) | if (common->utf) |
2764 | { | { |
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | ||
#if defined COMPILE_PCRE8 | ||
2765 | /* This can be an extra read in some situations, but hopefully | /* This can be an extra read in some situations, but hopefully |
2766 | it is needed in most cases. */ | it is needed in most cases. */ |
2767 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2768 | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); |
2769 | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); | if (!update_str_ptr) |
2770 | JUMPHERE(jump); | { |
2771 | #elif defined COMPILE_PCRE16 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
2772 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
2773 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
2774 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); |
2775 | JUMPHERE(jump); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); |
2776 | /* Skip low surrogate if necessary. */ | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0); |
2777 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2778 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800); | jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2779 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2780 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | JUMPHERE(jump2); |
2781 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | } |
2782 | #elif defined COMPILE_PCRE32 | else |
2783 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); |
jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | ||
OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | ||
2784 | JUMPHERE(jump); | JUMPHERE(jump); |
#endif /* COMPILE_PCRE[8|16|32] */ | ||
2785 | return; | return; |
2786 | } | } |
2787 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF && COMPILE_PCRE8 */ |
2788 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | |
2789 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | #if !defined COMPILE_PCRE8 |
#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | ||
2790 | /* The ctypes array contains only 256 values. */ | /* The ctypes array contains only 256 values. */ |
2791 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2792 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2793 | #endif | #endif |
2794 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2795 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if !defined COMPILE_PCRE8 |
2796 | JUMPHERE(jump); | JUMPHERE(jump); |
2797 | #endif | #endif |
2798 | ||
2799 | #if defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
2800 | if (common->utf && update_str_ptr) | |
2801 | { | |
2802 | /* Skip low surrogate if necessary. */ | |
2803 | OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xd800); | |
2804 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1); | |
2805 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2806 | JUMPHERE(jump); | |
2807 | } | |
2808 | #endif /* SUPPORT_UTF && COMPILE_PCRE16 */ | |
2809 | } | } |
2810 | ||
2811 | static void skip_char_back(compiler_common *common) | static void skip_char_back(compiler_common *common) |
# | Line 2256 if (common->utf) | Line 2833 if (common->utf) |
2833 | /* Skip low surrogate if necessary. */ | /* Skip low surrogate if necessary. */ |
2834 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
2835 | 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); |
2836 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2837 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
2838 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2839 | return; | return; |
# | Line 2266 if (common->utf) | Line 2843 if (common->utf) |
2843 | 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)); |
2844 | } | } |
2845 | ||
2846 | 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) |
2847 | { | { |
2848 | /* 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. */ |
2849 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2850 | struct sljit_jump *jump; | |
2851 | ||
2852 | if (nltype == NLTYPE_ANY) | if (nltype == NLTYPE_ANY) |
2853 | { | { |
2854 | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); |
2855 | add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(jumpifmatch ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
2856 | } | } |
2857 | else if (nltype == NLTYPE_ANYCRLF) | else if (nltype == NLTYPE_ANYCRLF) |
2858 | { | { |
2859 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR); | if (jumpifmatch) |
2860 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | { |
2861 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR)); |
2862 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); |
2863 | add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | } |
2864 | else | |
2865 | { | |
2866 | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); | |
2867 | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); | |
2868 | JUMPHERE(jump); | |
2869 | } | |
2870 | } | } |
2871 | else | else |
2872 | { | { |
2873 | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); |
2874 | 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_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); |
2875 | } | } |
2876 | } | } |
2877 | ||
# | Line 2297 else | Line 2881 else |
2881 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2882 | { | { |
2883 | /* Fast decoding a UTF-8 character. TMP1 contains the first byte | /* Fast decoding a UTF-8 character. TMP1 contains the first byte |
2884 | 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. */ |
2885 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2886 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2887 | ||
2888 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2889 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2890 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | |
2891 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2892 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | |
2893 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | |
2894 | ||
2895 | /* Searching for the first zero. */ | /* Searching for the first zero. */ |
2896 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); |
2897 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_C_NOT_ZERO); |
2898 | /* Two byte sequence. */ | /* Two byte sequence. */ |
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | ||
2899 | 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)); |
2900 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1f); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2)); |
2901 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | |
2902 | ||
2903 | JUMPHERE(jump); | |
2904 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2905 | OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800); | |
2906 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); |
2907 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
2908 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); | ||
sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | ||
JUMPHERE(jump); | ||
2909 | ||
2910 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10000); |
2911 | jump = JUMP(SLJIT_C_NOT_ZERO); | jump = JUMP(SLJIT_C_NOT_ZERO); |
2912 | /* Three byte sequence. */ | /* Three byte sequence. */ |
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | ||
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0f); | ||
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 12); | ||
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | ||
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | ||
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | ||
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); | ||
2913 | 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)); |
2914 | 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)); | ||
2915 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
JUMPHERE(jump); | ||
2916 | ||
2917 | /* Four byte sequence. */ | /* Four byte sequence. */ |
2918 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | JUMPHERE(jump); |
2919 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x07); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); |
2920 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 18); | OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); |
2921 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2922 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3)); | |
2923 | 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); | ||
2924 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2925 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(4)); |
2926 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | |
2927 | } | |
2928 | ||
2929 | static void do_utfreadchar16(compiler_common *common) | |
2930 | { | |
2931 | /* Fast decoding a UTF-8 character. TMP1 contains the first byte | |
2932 | of the character (>= 0xc0). Return value in TMP1. */ | |
2933 | DEFINE_COMPILER; | |
2934 | struct sljit_jump *jump; | |
2935 | ||
2936 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); | |
2937 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2938 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | |
2939 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2940 | 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); | ||
2941 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2942 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(3)); | |
2943 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3)); | /* Searching for the first zero. */ |
2944 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800); | |
2945 | jump = JUMP(SLJIT_C_NOT_ZERO); | |
2946 | /* Two byte sequence. */ | |
2947 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2948 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | |
2949 | ||
2950 | JUMPHERE(jump); | |
2951 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x400); | |
2952 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_NOT_ZERO); | |
2953 | /* This code runs only in 8 bit mode. No need to shift the value. */ | |
2954 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | |
2955 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2956 | OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800); | |
2957 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); | |
2958 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); |
2959 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); |
2960 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3)); | /* Three byte sequence. */ |
2961 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | |
2962 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2963 | } | } |
2964 | ||
# | Line 2368 jump = JUMP(SLJIT_C_NOT_ZERO); | Line 2978 jump = JUMP(SLJIT_C_NOT_ZERO); |
2978 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
2979 | 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)); |
2980 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f); |
2981 | /* The upper 5 bits are known at this point. */ | |
2982 | compare = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0x3); | |
2983 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); |
2984 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); |
2985 | 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); | ||
2986 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2987 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2988 | ||
2989 | JUMPHERE(compare); | JUMPHERE(compare); |
2990 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2991 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
JUMPHERE(jump); | ||
2992 | ||
2993 | /* 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); | ||
2994 | JUMPHERE(jump); | JUMPHERE(jump); |
2995 | /* Combine two 16 bit characters. */ | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0); |
2996 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2997 | 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); | ||
2998 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2999 | } | } |
3000 | ||
3001 | #endif /* COMPILE_PCRE[8|16] */ | #endif /* COMPILE_PCRE8 */ |
3002 | ||
3003 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
3004 | ||
# | Line 2481 if (firstline) | Line 3065 if (firstline) |
3065 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); |
3066 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); |
3067 | JUMPHERE(end); | JUMPHERE(end); |
3068 | 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)); |
3069 | } | } |
3070 | else | else |
3071 | { | { |
3072 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3073 | mainloop = LABEL(); | mainloop = LABEL(); |
3074 | /* Continual stores does not cause data dependency. */ | /* Continual stores does not cause data dependency. */ |
3075 | 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); |
3076 | read_char(common); | read_char_range(common, common->nlmin, common->nlmax, TRUE); |
3077 | check_newlinechar(common, common->nltype, &newline, TRUE); | check_newlinechar(common, common->nltype, &newline, TRUE); |
3078 | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); |
3079 | JUMPHERE(end); | JUMPHERE(end); |
3080 | 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); |
3081 | set_jumps(newline, LABEL()); | set_jumps(newline, LABEL()); |
3082 | } | } |
3083 | ||
# | Line 2509 if (newlinecheck) | Line 3093 if (newlinecheck) |
3093 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3094 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3095 | 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); |
3096 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3097 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3098 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
3099 | #endif | #endif |
3100 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3101 | nl = JUMP(SLJIT_JUMP); | nl = JUMP(SLJIT_JUMP); |
3102 | } | } |
3103 | ||
3104 | mainloop = LABEL(); | |
3105 | ||
3106 | /* Increasing the STR_PTR here requires one less jump in the most common case. */ | |
3107 | #ifdef SUPPORT_UTF | |
3108 | if (common->utf) readuchar = TRUE; | |
3109 | #endif | |
3110 | if (newlinecheck) readuchar = TRUE; | |
3111 | ||
3112 | if (readuchar) | |
3113 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | |
3114 | ||
3115 | if (newlinecheck) | |
3116 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); | |
3117 | ||
3118 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
3119 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
3120 | #if defined COMPILE_PCRE8 | |
3121 | if (common->utf) | |
3122 | { | |
3123 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | |
3124 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | |
3125 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
3126 | JUMPHERE(singlechar); | |
3127 | } | |
3128 | #elif defined COMPILE_PCRE16 | |
3129 | if (common->utf) | |
3130 | { | |
3131 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | |
3132 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | |
3133 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | |
3134 | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
3135 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | |
3136 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | |
3137 | JUMPHERE(singlechar); | |
3138 | } | |
3139 | #endif /* COMPILE_PCRE[8|16] */ | |
3140 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | |
3141 | JUMPHERE(start); | |
3142 | ||
3143 | if (newlinecheck) | |
3144 | { | |
3145 | JUMPHERE(end); | |
3146 | JUMPHERE(nl); | |
3147 | } | |
3148 | ||
3149 | return mainloop; | |
3150 | } | |
3151 | ||
3152 | #define MAX_N_CHARS 16 | |
3153 | #define MAX_N_BYTES 8 | |
3154 | ||
3155 | static SLJIT_INLINE void add_prefix_byte(pcre_uint8 byte, pcre_uint8 *bytes) | |
3156 | { | |
3157 | pcre_uint8 len = bytes[0]; | |
3158 | int i; | |
3159 | ||
3160 | if (len == 255) | |
3161 | return; | |
3162 | ||
3163 | if (len == 0) | |
3164 | { | |
3165 | bytes[0] = 1; | |
3166 | bytes[1] = byte; | |
3167 | return; | |
3168 | } | |
3169 | ||
3170 | for (i = len; i > 0; i--) | |
3171 | if (bytes[i] == byte) | |
3172 | return; | |
3173 | ||
3174 | if (len >= MAX_N_BYTES - 1) | |
3175 | { | |
3176 | bytes[0] = 255; | |
3177 | return; | |
3178 | } | |
3179 | ||
3180 | len++; | |
3181 | bytes[len] = byte; | |
3182 | bytes[0] = len; | |
3183 | } | |
3184 | ||
3185 | static int scan_prefix(compiler_common *common, pcre_uchar *cc, pcre_uint32 *chars, pcre_uint8 *bytes, int max_chars) | |
3186 | { | |
3187 | /* Recursive function, which scans prefix literals. */ | |
3188 | BOOL last, any, caseless; | |
3189 | int len, repeat, len_save, consumed = 0; | |
3190 | pcre_uint32 chr, mask; | |
3191 | pcre_uchar *alternative, *cc_save, *oc; | |
3192 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
3193 | pcre_uchar othercase[8]; | |
3194 | #elif defined SUPPORT_UTF && defined COMPILE_PCRE16 | |
3195 | pcre_uchar othercase[2]; | |
3196 | #else | |
3197 | pcre_uchar othercase[1]; | |
3198 | #endif | |
3199 | ||
3200 | repeat = 1; | |
3201 | while (TRUE) | |
3202 | { | |
3203 | last = TRUE; | |
3204 | any = FALSE; | |
3205 | caseless = FALSE; | |
3206 | switch (*cc) | |
3207 | { | |
3208 | case OP_CHARI: | |
3209 | caseless = TRUE; | |
3210 | case OP_CHAR: | |
3211 | last = FALSE; | |
3212 | cc++; | |
3213 | break; | |
3214 | ||
3215 | case OP_SOD: | |
3216 | case OP_SOM: | |
3217 | case OP_SET_SOM: | |
3218 | case OP_NOT_WORD_BOUNDARY: | |
3219 | case OP_WORD_BOUNDARY: | |
3220 | case OP_EODN: | |
3221 | case OP_EOD: | |
3222 | case OP_CIRC: | |
3223 | case OP_CIRCM: | |
3224 | case OP_DOLL: | |
3225 | case OP_DOLLM: | |
3226 | /* Zero width assertions. */ | |
3227 | cc++; | |
3228 | continue; | |
3229 | ||
3230 | case OP_ASSERT: | |
3231 | case OP_ASSERT_NOT: | |
3232 | case OP_ASSERTBACK: | |
3233 | case OP_ASSERTBACK_NOT: | |
3234 | cc = bracketend(cc); | |
3235 | continue; | |
3236 | ||
3237 | case OP_PLUSI: | |
3238 | case OP_MINPLUSI: | |
3239 | case OP_POSPLUSI: | |
3240 | caseless = TRUE; | |
3241 | case OP_PLUS: | |
3242 | case OP_MINPLUS: | |
3243 | case OP_POSPLUS: | |
3244 | cc++; | |
3245 | break; | |
3246 | ||
3247 | case OP_EXACTI: | |
3248 | caseless = TRUE; | |
3249 | case OP_EXACT: | |
3250 | repeat = GET2(cc, 1); | |
3251 | last = FALSE; | |
3252 | cc += 1 + IMM2_SIZE; | |
3253 | break; | |
3254 | ||
3255 | case OP_QUERYI: | |
3256 | case OP_MINQUERYI: | |
3257 | case OP_POSQUERYI: | |
3258 | caseless = TRUE; | |
3259 | case OP_QUERY: | |
3260 | case OP_MINQUERY: | |
3261 | case OP_POSQUERY: | |
3262 | len = 1; | |
3263 | cc++; | |
3264 | #ifdef SUPPORT_UTF | |
3265 | if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc); | |
3266 | #endif | |
3267 | max_chars = scan_prefix(common, cc + len, chars, bytes, max_chars); | |
3268 | if (max_chars == 0) | |
3269 | return consumed; | |
3270 | last = FALSE; | |
3271 | break; | |
3272 | ||
3273 | case OP_KET: | |
3274 | cc += 1 + LINK_SIZE; | |
3275 | continue; | |
3276 | ||
3277 | case OP_ALT: | |
3278 | cc += GET(cc, 1); | |
3279 | continue; | |
3280 | ||
3281 | case OP_ONCE: | |
3282 | case OP_ONCE_NC: | |
3283 | case OP_BRA: | |
3284 | case OP_BRAPOS: | |
3285 | case OP_CBRA: | |
3286 | case OP_CBRAPOS: | |
3287 | alternative = cc + GET(cc, 1); | |
3288 | while (*alternative == OP_ALT) | |
3289 | { | |
3290 | max_chars = scan_prefix(common, alternative + 1 + LINK_SIZE, chars, bytes, max_chars); | |
3291 | if (max_chars == 0) | |
3292 | return consumed; | |
3293 | alternative += GET(alternative, 1); | |
3294 | } | |
3295 | ||
3296 | if (*cc == OP_CBRA || *cc == OP_CBRAPOS) | |
3297 | cc += IMM2_SIZE; | |
3298 | cc += 1 + LINK_SIZE; | |
3299 | continue; | |
3300 | ||
3301 | case OP_CLASS: | |
3302 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
3303 | if (common->utf && !is_char7_bitset((const pcre_uint8 *)(cc + 1), FALSE)) return consumed; | |
3304 | #endif | |
3305 | any = TRUE; | |
3306 | cc += 1 + 32 / sizeof(pcre_uchar); | |
3307 | break; | |
3308 | ||
3309 | case OP_NCLASS: | |
3310 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
3311 | if (common->utf) return consumed; | |
3312 | #endif | |
3313 | any = TRUE; | |
3314 | cc += 1 + 32 / sizeof(pcre_uchar); | |
3315 | break; | |
3316 | ||
3317 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
3318 | case OP_XCLASS: | |
3319 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
3320 | if (common->utf) return consumed; | |
3321 | #endif | |
3322 | any = TRUE; | |
3323 | cc += GET(cc, 1); | |
3324 | break; | |
3325 | #endif | |
3326 | ||
3327 | case OP_DIGIT: | |
3328 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
3329 | if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_digit, FALSE)) | |
3330 | return consumed; | |
3331 | #endif | |
3332 | any = TRUE; | |
3333 | cc++; | |
3334 | break; | |
3335 | ||
3336 | case OP_WHITESPACE: | |
3337 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
3338 | if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_space, FALSE)) | |
3339 | return consumed; | |
3340 | #endif | |
3341 | any = TRUE; | |
3342 | cc++; | |
3343 | break; | |
3344 | ||
3345 | case OP_WORDCHAR: | |
3346 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | |
3347 | if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_word, FALSE)) | |
3348 | return consumed; | |
3349 | #endif | |
3350 | any = TRUE; | |
3351 | cc++; | |
3352 | break; | |
3353 | ||
3354 | case OP_NOT: | |
3355 | case OP_NOTI: | |
3356 | cc++; | |
3357 | /* Fall through. */ | |
3358 | case OP_NOT_DIGIT: | |
3359 | case OP_NOT_WHITESPACE: | |
3360 | case OP_NOT_WORDCHAR: | |
3361 | case OP_ANY: | |
3362 | case OP_ALLANY: | |
3363 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
3364 | if (common->utf) return consumed; | |
3365 | #endif | |
3366 | any = TRUE; | |
3367 | cc++; | |
3368 | break; | |
3369 | ||
3370 | #ifdef SUPPORT_UCP | |
3371 | case OP_NOTPROP: | |
3372 | case OP_PROP: | |
3373 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
3374 | if (common->utf) return consumed; | |
3375 | #endif | |
3376 | any = TRUE; | |
3377 | cc += 1 + 2; | |
3378 | break; | |
3379 | #endif | |
3380 | ||
3381 | case OP_TYPEEXACT: | |
3382 | repeat = GET2(cc, 1); | |
3383 | cc += 1 + IMM2_SIZE; | |
3384 | continue; | |
3385 | ||
3386 | case OP_NOTEXACT: | |
3387 | case OP_NOTEXACTI: | |
3388 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
3389 | if (common->utf) return consumed; | |
3390 | #endif | |
3391 | any = TRUE; | |
3392 | repeat = GET2(cc, 1); | |
3393 | cc += 1 + IMM2_SIZE + 1; | |
3394 | break; | |
3395 | ||
3396 | default: | |
3397 | return consumed; | |
3398 | } | |
3399 | ||
3400 | if (any) | |
3401 | { | |
3402 | #if defined COMPILE_PCRE8 | |
3403 | mask = 0xff; | |
3404 | #elif defined COMPILE_PCRE16 | |
3405 | mask = 0xffff; | |
3406 | #elif defined COMPILE_PCRE32 | |
3407 | mask = 0xffffffff; | |
3408 | #else | |
3409 | SLJIT_ASSERT_STOP(); | |
3410 | #endif | |
3411 | ||
3412 | do | |
3413 | { | |
3414 | chars[0] = mask; | |
3415 | chars[1] = mask; | |
3416 | bytes[0] = 255; | |
3417 | ||
3418 | consumed++; | |
3419 | if (--max_chars == 0) | |
3420 | return consumed; | |
3421 | chars += 2; | |
3422 | bytes += MAX_N_BYTES; | |
3423 | } | |
3424 | while (--repeat > 0); | |
3425 | ||
3426 | repeat = 1; | |
3427 | continue; | |
3428 | } | |
3429 | ||
3430 | len = 1; | |
3431 | #ifdef SUPPORT_UTF | |
3432 | if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc); | |
3433 | #endif | |
3434 | ||
3435 | if (caseless && char_has_othercase(common, cc)) | |
3436 | { | |
3437 | #ifdef SUPPORT_UTF | |
3438 | if (common->utf) | |
3439 | { | |
3440 | GETCHAR(chr, cc); | |
3441 | if ((int)PRIV(ord2utf)(char_othercase(common, chr), othercase) != len) | |
3442 | return consumed; | |
3443 | } | |
3444 | else | |
3445 | #endif | |
3446 | { | |
3447 | chr = *cc; | |
3448 | othercase[0] = TABLE_GET(chr, common->fcc, chr); | |
3449 | } | |
3450 | } | |
3451 | else | |
3452 | caseless = FALSE; | |
3453 | ||
3454 | mainloop = LABEL(); | len_save = len; |
3455 | cc_save = cc; | |
3456 | while (TRUE) | |
3457 | { | |
3458 | oc = othercase; | |
3459 | do | |
3460 | { | |
3461 | chr = *cc; | |
3462 | #ifdef COMPILE_PCRE32 | |
3463 | if (SLJIT_UNLIKELY(chr == NOTACHAR)) | |
3464 | return consumed; | |
3465 | #endif | |
3466 | add_prefix_byte((pcre_uint8)chr, bytes); | |
3467 | ||
3468 | /* Increasing the STR_PTR here requires one less jump in the most common case. */ | mask = 0; |
3469 | #ifdef SUPPORT_UTF | if (caseless) |
3470 | if (common->utf) readuchar = TRUE; | { |
3471 | add_prefix_byte((pcre_uint8)*oc, bytes); | |
3472 | mask = *cc ^ *oc; | |
3473 | chr |= mask; | |
3474 | } | |
3475 | ||
3476 | #ifdef COMPILE_PCRE32 | |
3477 | if (chars[0] == NOTACHAR && chars[1] == 0) | |
3478 | #else | |
3479 | if (chars[0] == NOTACHAR) | |
3480 | #endif | #endif |
3481 | if (newlinecheck) readuchar = TRUE; | { |
3482 | chars[0] = chr; | |
3483 | chars[1] = mask; | |
3484 | } | |
3485 | else | |
3486 | { | |
3487 | mask |= chars[0] ^ chr; | |
3488 | chr |= mask; | |
3489 | chars[0] = chr; | |
3490 | chars[1] |= mask; | |
3491 | } | |
3492 | ||
3493 | if (readuchar) | len--; |
3494 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | consumed++; |
3495 | if (--max_chars == 0) | |
3496 | return consumed; | |
3497 | chars += 2; | |
3498 | bytes += MAX_N_BYTES; | |
3499 | cc++; | |
3500 | oc++; | |
3501 | } | |
3502 | while (len > 0); | |
3503 | ||
3504 | if (newlinecheck) | if (--repeat == 0) |
3505 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); | break; |
3506 | ||
3507 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | len = len_save; |
3508 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | cc = cc_save; |
3509 | #if defined COMPILE_PCRE8 | } |
if (common->utf) | ||
{ | ||
singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | ||
OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | ||
JUMPHERE(singlechar); | ||
} | ||
#elif defined COMPILE_PCRE16 | ||
if (common->utf) | ||
{ | ||
singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | ||
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | ||
OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); | ||
COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | ||
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | ||
JUMPHERE(singlechar); | ||
} | ||
#endif /* COMPILE_PCRE[8|16] */ | ||
#endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | ||
JUMPHERE(start); | ||
3510 | ||
3511 | if (newlinecheck) | repeat = 1; |
3512 | { | if (last) |
3513 | JUMPHERE(end); | return consumed; |
JUMPHERE(nl); | ||
3514 | } | } |
return mainloop; | ||
3515 | } | } |
3516 | ||
#define MAX_N_CHARS 3 | ||
3517 | static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline) | static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline) |
3518 | { | { |
3519 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3520 | struct sljit_label *start; | struct sljit_label *start; |
3521 | struct sljit_jump *quit; | struct sljit_jump *quit; |
3522 | pcre_uint32 chars[MAX_N_CHARS * 2]; | pcre_uint32 chars[MAX_N_CHARS * 2]; |
3523 | pcre_uchar *cc = common->start + 1 + IMM2_SIZE; | pcre_uint8 bytes[MAX_N_CHARS * MAX_N_BYTES]; |
3524 | int location = 0; | pcre_uint8 ones[MAX_N_CHARS]; |
3525 | pcre_int32 len, c, bit, caseless; | int offsets[3]; |
3526 | int must_stop; | pcre_uint32 mask; |
3527 | pcre_uint8 *byte_set, *byte_set_end; | |
3528 | int i, max, from; | |
3529 | int range_right = -1, range_len = 3 - 1; | |
3530 | sljit_ub *update_table = NULL; | |
3531 | BOOL in_range; | |
3532 | ||
3533 | /* This is even TRUE, if both are NULL. */ | |
3534 | SLJIT_ASSERT(common->read_only_data_ptr == common->read_only_data); | |
3535 | ||
3536 | for (i = 0; i < MAX_N_CHARS; i++) | |
3537 | { | |
3538 | chars[i << 1] = NOTACHAR; | |
3539 | chars[(i << 1) + 1] = 0; | |
3540 | bytes[i * MAX_N_BYTES] = 0; | |
3541 | } | |
3542 | ||
3543 | max = scan_prefix(common, common->start, chars, bytes, MAX_N_CHARS); | |
3544 | ||
3545 | /* We do not support alternatives now. */ | if (max <= 1) |
if (*(common->start + GET(common->start, 1)) == OP_ALT) | ||
3546 | return FALSE; | return FALSE; |
3547 | ||
3548 | while (TRUE) | for (i = 0; i < max; i++) |
3549 | { | { |
3550 | caseless = 0; | mask = chars[(i << 1) + 1]; |
3551 | must_stop = 1; | ones[i] = ones_in_half_byte[mask & 0xf]; |
3552 | switch(*cc) | mask >>= 4; |
3553 | while (mask != 0) | |
3554 | { | { |
3555 | case OP_CHAR: | ones[i] += ones_in_half_byte[mask & 0xf]; |
3556 | must_stop = 0; | mask >>= 4; |
3557 | cc++; | } |
3558 | break; | } |
3559 | ||
3560 | case OP_CHARI: | in_range = FALSE; |
3561 | caseless = 1; | from = 0; /* Prevent compiler "uninitialized" warning */ |
3562 | must_stop = 0; | for (i = 0; i <= max; i++) |
3563 | cc++; | { |
3564 | break; | if (in_range && (i - from) > range_len && (bytes[(i - 1) * MAX_N_BYTES] <= 4)) |
3565 | { | |
3566 | range_len = i - from; | |
3567 | range_right = i - 1; | |
3568 | } | |
3569 | ||
3570 | case OP_SOD: | if (i < max && bytes[i * MAX_N_BYTES] < 255) |
3571 | case OP_SOM: | { |
3572 | case OP_SET_SOM: | if (!in_range) |
3573 | case OP_NOT_WORD_BOUNDARY: | { |
3574 | case OP_WORD_BOUNDARY: | in_range = TRUE; |
3575 | case OP_EODN: | from = i; |
3576 | case OP_EOD: | } |
3577 | case OP_CIRC: | } |
3578 | case OP_CIRCM: | else if (in_range) |
3579 | case OP_DOLL: | in_range = FALSE; |
3580 | case OP_DOLLM: | } |
/* Zero width assertions. */ | ||
cc++; | ||
continue; | ||
3581 | ||
3582 | case OP_PLUS: | if (range_right >= 0) |
3583 | case OP_MINPLUS: | { |
3584 | case OP_POSPLUS: | /* Since no data is consumed (see the assert in the beginning |
3585 | cc++; | of this function), this space can be reallocated. */ |
3586 | break; | if (common->read_only_data) |
3587 | SLJIT_FREE(common->read_only_data); | |
3588 | ||
3589 | case OP_EXACT: | common->read_only_data_size += 256; |
3590 | cc += 1 + IMM2_SIZE; | common->read_only_data = (sljit_uw *)SLJIT_MALLOC(common->read_only_data_size); |
3591 | break; | if (common->read_only_data == NULL) |
3592 | return TRUE; | |
3593 | ||
3594 | case OP_PLUSI: | update_table = (sljit_ub *)common->read_only_data; |
3595 | case OP_MINPLUSI: | common->read_only_data_ptr = (sljit_uw *)(update_table + 256); |
3596 | case OP_POSPLUSI: | memset(update_table, IN_UCHARS(range_len), 256); |
caseless = 1; | ||
cc++; | ||
break; | ||
3597 | ||
3598 | case OP_EXACTI: | for (i = 0; i < range_len; i++) |
3599 | caseless = 1; | { |
3600 | cc += 1 + IMM2_SIZE; | byte_set = bytes + ((range_right - i) * MAX_N_BYTES); |
3601 | break; | SLJIT_ASSERT(byte_set[0] > 0 && byte_set[0] < 255); |
3602 | byte_set_end = byte_set + byte_set[0]; | |
3603 | byte_set++; | |
3604 | while (byte_set <= byte_set_end) | |
3605 | { | |
3606 | if (update_table[*byte_set] > IN_UCHARS(i)) | |
3607 | update_table[*byte_set] = IN_UCHARS(i); | |
3608 | byte_set++; | |
3609 | } | |
3610 | } | |
3611 | } | |
3612 | ||
3613 | default: | offsets[0] = -1; |
3614 | must_stop = 2; | /* Scan forward. */ |
3615 | for (i = 0; i < max; i++) | |
3616 | if (ones[i] <= 2) { | |
3617 | offsets[0] = i; | |
3618 | break; | break; |
3619 | } | } |
3620 | ||
3621 | if (must_stop == 2) | if (offsets[0] < 0 && range_right < 0) |
3622 | return FALSE; | |
3623 | ||
3624 | if (offsets[0] >= 0) | |
3625 | { | |
3626 | /* Scan backward. */ | |
3627 | offsets[1] = -1; | |
3628 | for (i = max - 1; i > offsets[0]; i--) | |
3629 | if (ones[i] <= 2 && i != range_right) | |
3630 | { | |
3631 | offsets[1] = i; | |
3632 | break; | break; |
3633 | } | |
3634 | ||
3635 | len = 1; | /* This case is handled better by fast_forward_first_char. */ |
3636 | #ifdef SUPPORT_UTF | if (offsets[1] == -1 && offsets[0] == 0 && range_right < 0) |
3637 | if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]); | return FALSE; |
#endif | ||
3638 | ||
3639 | if (caseless && char_has_othercase(common, cc)) | offsets[2] = -1; |
3640 | { | /* We only search for a middle character if there is no range check. */ |
3641 | caseless = char_get_othercase_bit(common, cc); | if (offsets[1] >= 0 && range_right == -1) |
3642 | if (caseless == 0) | { |
3643 | return FALSE; | /* Scan from middle. */ |
3644 | #ifdef COMPILE_PCRE8 | for (i = (offsets[0] + offsets[1]) / 2 + 1; i < offsets[1]; i++) |
3645 | caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8)); | if (ones[i] <= 2) |
3646 | #else | { |
3647 | if ((caseless & 0x100) != 0) | offsets[2] = i; |
3648 | caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9)); | break; |
3649 | else | } |
caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9)); | ||
#endif | ||
} | ||
else | ||
caseless = 0; | ||
3650 | ||
3651 | while (len > 0 && location < MAX_N_CHARS * 2) | if (offsets[2] == -1) |
{ | ||
c = *cc; | ||
bit = 0; | ||
if (len == (caseless & 0xff)) | ||
3652 | { | { |
3653 | bit = caseless >> 8; | for (i = (offsets[0] + offsets[1]) / 2; i > offsets[0]; i--) |
3654 | c |= bit; | if (ones[i] <= 2) |
3655 | { | |
3656 | offsets[2] = i; | |
3657 | break; | |
3658 | } | |
3659 | } | } |
3660 | } | |
3661 | ||
3662 | chars[location] = c; | SLJIT_ASSERT(offsets[1] == -1 || (offsets[0] < offsets[1])); |
3663 | chars[location + 1] = bit; | SLJIT_ASSERT(offsets[2] == -1 || (offsets[0] < offsets[2] && offsets[1] > offsets[2])); |
3664 | ||
3665 | len--; | chars[0] = chars[offsets[0] << 1]; |
3666 | location += 2; | chars[1] = chars[(offsets[0] << 1) + 1]; |
3667 | cc++; | if (offsets[2] >= 0) |
3668 | { | |
3669 | chars[2] = chars[offsets[2] << 1]; | |
3670 | chars[3] = chars[(offsets[2] << 1) + 1]; | |
3671 | } | |
3672 | if (offsets[1] >= 0) | |
3673 | { | |
3674 | chars[4] = chars[offsets[1] << 1]; | |
3675 | chars[5] = chars[(offsets[1] << 1) + 1]; | |
3676 | } | } |
if (location >= MAX_N_CHARS * 2 || must_stop != 0) | ||
break; | ||
3677 | } | } |
3678 | ||
3679 | /* At least two characters are required. */ | max -= 1; |
if (location < 2 * 2) | ||
return FALSE; | ||
3680 | if (firstline) | if (firstline) |
3681 | { | { |
3682 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3683 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->first_line_end); | |
3684 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
3685 | OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, (location >> 1) - 1); | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max)); |
3686 | quit = CMP(SLJIT_C_LESS_EQUAL, STR_END, 0, TMP1, 0); | |
3687 | OP1(SLJIT_MOV, STR_END, 0, TMP1, 0); | |
3688 | JUMPHERE(quit); | |
3689 | } | } |
3690 | else | else |
3691 | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, (location >> 1) - 1); | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max)); |
3692 | ||
3693 | #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32) | |
3694 | if (range_right >= 0) | |
3695 | OP1(SLJIT_MOV, RETURN_ADDR, 0, SLJIT_IMM, (sljit_sw)update_table); | |
3696 | #endif | |
3697 | ||
3698 | start = LABEL(); | start = LABEL(); |
3699 | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3700 | ||
3701 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | SLJIT_ASSERT(range_right >= 0 || offsets[0] >= 0); |
3702 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
3703 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | if (range_right >= 0) |
3704 | if (chars[1] != 0) | { |
3705 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]); | #if defined COMPILE_PCRE8 || (defined SLJIT_LITTLE_ENDIAN && SLJIT_LITTLE_ENDIAN) |
3706 | 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)); |
3707 | if (location > 2 * 2) | #else |
3708 | 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); |
3709 | if (chars[3] != 0) | #endif |
3710 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]); | |
3711 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start); | #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32) |
3712 | if (location > 2 * 2) | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(RETURN_ADDR, TMP1), 0); |
3713 | { | #else |
3714 | if (chars[5] != 0) | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)update_table); |
3715 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]); | #endif |
3716 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3717 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, start); | |
3718 | } | |
3719 | ||
3720 | if (offsets[0] >= 0) | |
3721 | { | |
3722 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[0])); | |
3723 | if (offsets[1] >= 0) | |
3724 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[1])); | |
3725 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
3726 | ||
3727 | if (chars[1] != 0) | |
3728 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]); | |
3729 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start); | |
3730 | if (offsets[2] >= 0) | |
3731 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[2] - 1)); | |
3732 | ||
3733 | if (offsets[1] >= 0) | |
3734 | { | |
3735 | if (chars[5] != 0) | |
3736 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[5]); | |
3737 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[4], start); | |
3738 | } | |
3739 | ||
3740 | if (offsets[2] >= 0) | |
3741 | { | |
3742 | if (chars[3] != 0) | |
3743 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[3]); | |
3744 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[2], start); | |
3745 | } | |
3746 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
3747 | } | } |
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | ||
3748 | ||
3749 | JUMPHERE(quit); | JUMPHERE(quit); |
3750 | ||
3751 | if (firstline) | if (firstline) |
3752 | { | |
3753 | if (range_right >= 0) | |
3754 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->first_line_end); | |
3755 | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
3756 | if (range_right >= 0) | |
3757 | { | |
3758 | quit = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); | |
3759 | OP1(SLJIT_MOV, STR_PTR, 0, TMP1, 0); | |
3760 | JUMPHERE(quit); | |
3761 | } | |
3762 | } | |
3763 | else | else |
3764 | OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, (location >> 1) - 1); | OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max)); |
3765 | return TRUE; | return TRUE; |
3766 | } | } |
3767 | ||
3768 | #undef MAX_N_CHARS | #undef MAX_N_CHARS |
3769 | #undef MAX_N_BYTES | |
3770 | ||
3771 | 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) |
3772 | { | { |
# | Line 2746 if (firstline) | Line 3780 if (firstline) |
3780 | { | { |
3781 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3782 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
3783 | 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); |
3784 | } | } |
3785 | ||
3786 | start = LABEL(); | start = LABEL(); |
# | Line 2775 else | Line 3809 else |
3809 | else | else |
3810 | { | { |
3811 | 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); |
3812 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3813 | 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); |
3814 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3815 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_C_NOT_ZERO); |
3816 | } | } |
3817 | } | } |
# | Line 2806 if (firstline) | Line 3840 if (firstline) |
3840 | { | { |
3841 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3842 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); |
3843 | 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); |
3844 | } | } |
3845 | ||
3846 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
# | Line 2819 if (common->nltype == NLTYPE_FIXED && co | Line 3853 if (common->nltype == NLTYPE_FIXED && co |
3853 | ||
3854 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); |
3855 | 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); |
3856 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER_EQUAL); |
3857 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3858 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); |
3859 | #endif | #endif |
# | Line 2838 if (common->nltype == NLTYPE_FIXED && co | Line 3872 if (common->nltype == NLTYPE_FIXED && co |
3872 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
3873 | ||
3874 | if (firstline) | if (firstline) |
3875 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
3876 | return; | return; |
3877 | } | } |
3878 | ||
# | Line 2848 firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_ | Line 3882 firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_ |
3882 | skip_char_back(common); | skip_char_back(common); |
3883 | ||
3884 | loop = LABEL(); | loop = LABEL(); |
3885 | read_char(common); | common->ff_newline_shortcut = loop; |
3886 | ||
3887 | read_char_range(common, common->nlmin, common->nlmax, TRUE); | |
3888 | lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3889 | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) |
3890 | foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); | foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); |
# | Line 2862 if (common->nltype == NLTYPE_ANY || comm | Line 3898 if (common->nltype == NLTYPE_ANY || comm |
3898 | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3899 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3900 | 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); |
3901 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3902 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3903 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
3904 | #endif | #endif |
# | Line 2877 if (firstline) | Line 3913 if (firstline) |
3913 | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
3914 | } | } |
3915 | ||
3916 | static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline) | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks); |
3917 | ||
3918 | static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, pcre_uint8 *start_bits, BOOL firstline) | |
3919 | { | { |
3920 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3921 | struct sljit_label *start; | struct sljit_label *start; |
3922 | struct sljit_jump *quit; | struct sljit_jump *quit; |
3923 | struct sljit_jump *found; | struct sljit_jump *found = NULL; |
3924 | jump_list *matches = NULL; | |
3925 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3926 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3927 | #endif | #endif |
# | Line 2891 if (firstline) | Line 3930 if (firstline) |
3930 | { | { |
3931 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3932 | OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); | OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); |
3933 | 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); |
3934 | } | } |
3935 | ||
3936 | start = LABEL(); | start = LABEL(); |
# | Line 2901 OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_P | Line 3940 OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_P |
3940 | if (common->utf) | if (common->utf) |
3941 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
3942 | #endif | #endif |
3943 | ||
3944 | if (!check_class_ranges(common, start_bits, (start_bits[31] & 0x80) != 0, TRUE, &matches)) | |
3945 | { | |
3946 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3947 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); |
3948 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); |
3949 | JUMPHERE(jump); | JUMPHERE(jump); |
3950 | #endif | #endif |
3951 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
3952 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
3953 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)start_bits); |
3954 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); |
3955 | 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); |
3956 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_C_NOT_ZERO); |
3957 | } | |
3958 | ||
3959 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
3960 | if (common->utf) | if (common->utf) |
# | Line 2932 if (common->utf) | Line 3975 if (common->utf) |
3975 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); |
3976 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
3977 | 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); |
3978 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3979 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
3980 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3981 | } | } |
3982 | #endif /* COMPILE_PCRE[8|16] */ | #endif /* COMPILE_PCRE[8|16] */ |
3983 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
3984 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
3985 | JUMPHERE(found); | if (found != NULL) |
3986 | JUMPHERE(found); | |
3987 | if (matches != NULL) | |
3988 | set_jumps(matches, LABEL()); | |
3989 | JUMPHERE(quit); | JUMPHERE(quit); |
3990 | ||
3991 | if (firstline) | if (firstline) |
# | Line 2958 struct sljit_jump *notfound; | Line 4004 struct sljit_jump *notfound; |
4004 | pcre_uint32 oc, bit; | pcre_uint32 oc, bit; |
4005 | ||
4006 | SLJIT_ASSERT(common->req_char_ptr != 0); | SLJIT_ASSERT(common->req_char_ptr != 0); |
4007 | 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); |
4008 | 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); |
4009 | toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); | toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); |
4010 | alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); | alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); |
# | Line 3003 JUMPTO(SLJIT_JUMP, loop); | Line 4049 JUMPTO(SLJIT_JUMP, loop); |
4049 | JUMPHERE(found); | JUMPHERE(found); |
4050 | if (foundoc) | if (foundoc) |
4051 | JUMPHERE(foundoc); | JUMPHERE(foundoc); |
4052 | 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); |
4053 | JUMPHERE(alreadyfound); | JUMPHERE(alreadyfound); |
4054 | JUMPHERE(toolong); | JUMPHERE(toolong); |
4055 | return notfound; | return notfound; |
# | Line 3022 GET_LOCAL_BASE(TMP3, 0, 0); | Line 4068 GET_LOCAL_BASE(TMP3, 0, 0); |
4068 | /* Drop frames until we reach STACK_TOP. */ | /* Drop frames until we reach STACK_TOP. */ |
4069 | mainloop = LABEL(); | mainloop = LABEL(); |
4070 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); |
4071 | jump = CMP(SLJIT_C_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); | OP2(SLJIT_SUB | SLJIT_SET_S, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0); |
4072 | jump = JUMP(SLJIT_C_SIG_LESS_EQUAL); | |
4073 | ||
4074 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); |
4075 | 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)); |
4076 | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_sw), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_sw)); | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_sw), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_sw)); |
# | Line 3030 OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_I | Line 4078 OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_I |
4078 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
4079 | ||
4080 | JUMPHERE(jump); | JUMPHERE(jump); |
4081 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); | jump = JUMP(SLJIT_C_SIG_LESS); |
4082 | /* End of dropping frames. */ | /* End of dropping frames. */ |
4083 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
4084 | ||
4085 | JUMPHERE(jump); | JUMPHERE(jump); |
4086 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setstrbegin); | OP1(SLJIT_NEG, TMP2, 0, TMP2, 0); |
4087 | /* Set string begin. */ | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); |
4088 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); |
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_sw)); | ||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); | ||
JUMPTO(SLJIT_JUMP, mainloop); | ||
JUMPHERE(jump); | ||
if (common->mark_ptr != 0) | ||
{ | ||
jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setmark); | ||
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); | ||
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_sw)); | ||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP2, 0); | ||
JUMPTO(SLJIT_JUMP, mainloop); | ||
JUMPHERE(jump); | ||
} | ||
/* Unknown command. */ | ||
4089 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_sw)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_sw)); |
4090 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
4091 | } | } |
# | Line 3063 static void check_wordboundary(compiler_ | Line 4094 static void check_wordboundary(compiler_ |
4094 | { | { |
4095 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4096 | struct sljit_jump *skipread; | struct sljit_jump *skipread; |
4097 | jump_list *skipread_list = NULL; | |
4098 | #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF | #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF |
4099 | struct sljit_jump *jump; | struct sljit_jump *jump; |
4100 | #endif | #endif |
4101 | ||
4102 | SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); | SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); |
4103 | ||
4104 | sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_SP), LOCALS0); |
4105 | /* Get type of the previous char, and put it to LOCALS1. */ | /* Get type of the previous char, and put it to LOCALS1. */ |
4106 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
4107 | 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)); |
4108 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS1, SLJIT_IMM, 0); |
4109 | skipread = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); | skipread = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); |
4110 | skip_char_back(common); | skip_char_back(common); |
4111 | check_start_used_ptr(common); | check_start_used_ptr(common); |
# | Line 3088 if (common->use_ucp) | Line 4120 if (common->use_ucp) |
4120 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
4121 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
4122 | 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); |
4123 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
4124 | 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); |
4125 | 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); |
4126 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
4127 | JUMPHERE(jump); | JUMPHERE(jump); |
4128 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS1, TMP2, 0); |
4129 | } | } |
4130 | else | else |
4131 | #endif | #endif |
# | Line 3109 else | Line 4141 else |
4141 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes); |
4142 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 4 /* ctype_word */); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 4 /* ctype_word */); |
4143 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
4144 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCALS1, TMP1, 0); |
4145 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
4146 | JUMPHERE(jump); | JUMPHERE(jump); |
4147 | #elif defined SUPPORT_UTF | #elif defined SUPPORT_UTF |
# | Line 3120 else | Line 4152 else |
4152 | JUMPHERE(skipread); | JUMPHERE(skipread); |
4153 | ||
4154 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); |
4155 | skipread = check_str_end(common); | check_str_end(common, &skipread_list); |
4156 | peek_char(common); | peek_char(common, READ_CHAR_MAX); |
4157 | ||
4158 | /* Testing char type. This is a code duplication. */ | /* Testing char type. This is a code duplication. */ |
4159 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
# | Line 3132 if (common->use_ucp) | Line 4164 if (common->use_ucp) |
4164 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
4165 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
4166 | 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); |
4167 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
4168 | 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); |
4169 | 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); |
4170 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
4171 | JUMPHERE(jump); | JUMPHERE(jump); |
4172 | } | } |
4173 | else | else |
# | Line 3161 else | Line 4193 else |
4193 | JUMPHERE(jump); | JUMPHERE(jump); |
4194 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE8 */ |
4195 | } | } |
4196 | JUMPHERE(skipread); | set_jumps(skipread_list, LABEL()); |
OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); | ||
sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | ||
} | ||
/* | ||
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) | ||
{ | ||
DEFINE_COMPILER; | ||
struct sljit_jump *jump; | ||
if (ranges[0] < 0) | ||
return FALSE; | ||
switch(ranges[0]) | ||
{ | ||
case 1: | ||
if (readch) | ||
read_char(common); | ||
add_jump(compiler, backtracks, CMP(ranges[1] == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | ||
return TRUE; | ||
case 2: | ||
if (readch) | ||
read_char(common); | ||
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]); | ||
add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2])); | ||
return TRUE; | ||
case 4: | ||
if (ranges[2] + 1 == ranges[3] && ranges[4] + 1 == ranges[5]) | ||
{ | ||
if (readch) | ||
read_char(common); | ||
if (ranges[1] != 0) | ||
{ | ||
add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | ||
add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | ||
} | ||
else | ||
{ | ||
jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]); | ||
add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | ||
JUMPHERE(jump); | ||
} | ||
return TRUE; | ||
} | ||
if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && is_powerof2(ranges[4] - ranges[2])) | ||
{ | ||
if (readch) | ||
read_char(common); | ||
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4] - ranges[2]); | ||
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4]); | ||
add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[5] - ranges[4])); | ||
return TRUE; | ||
} | ||
return FALSE; | ||
default: | ||
return FALSE; | ||
} | ||
} | ||
static void get_ctype_ranges(compiler_common *common, int flag, int *ranges) | ||
{ | ||
int i, bit, length; | ||
const pcre_uint8 *ctypes = (const pcre_uint8*)common->ctypes; | ||
bit = ctypes[0] & flag; | ||
ranges[0] = -1; | ||
ranges[1] = bit != 0 ? 1 : 0; | ||
length = 0; | ||
4197 | ||
4198 | for (i = 1; i < 256; i++) | OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_SP), LOCALS1); |
4199 | if ((ctypes[i] & flag) != bit) | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_SP), LOCALS0); |
{ | ||
if (length >= MAX_RANGE_SIZE) | ||
return; | ||
ranges[2 + length] = i; | ||
length++; | ||
bit ^= flag; | ||
} | ||
if (bit != 0) | ||
{ | ||
if (length >= MAX_RANGE_SIZE) | ||
return; | ||
ranges[2 + length] = 256; | ||
length++; | ||
} | ||
ranges[0] = length; | ||
4200 | } | } |
4201 | ||
4202 | 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) |
4203 | { | { |
4204 | int ranges[2 + MAX_RANGE_SIZE]; | DEFINE_COMPILER; |
4205 | int ranges[MAX_RANGE_SIZE]; | |
4206 | pcre_uint8 bit, cbit, all; | pcre_uint8 bit, cbit, all; |
4207 | int i, byte, length = 0; | int i, byte, length = 0; |
4208 | ||
4209 | bit = bits[0] & 0x1; | bit = bits[0] & 0x1; |
4210 | ranges[1] = bit; | /* All bits will be zero or one (since bit is zero or one). */ |
/* Can be 0 or 255. */ | ||
4211 | all = -bit; | all = -bit; |
4212 | ||
4213 | for (i = 0; i < 256; ) | for (i = 0; i < 256; ) |
# | Line 3285 for (i = 0; i < 256; ) | Line 4222 for (i = 0; i < 256; ) |
4222 | { | { |
4223 | if (length >= MAX_RANGE_SIZE) | if (length >= MAX_RANGE_SIZE) |
4224 | return FALSE; | return FALSE; |
4225 | ranges[2 + length] = i; | ranges[length] = i; |
4226 | length++; | length++; |
4227 | bit = cbit; | bit = cbit; |
4228 | all = -cbit; | all = -cbit; |
# | Line 3298 if (((bit == 0) && nclass) || ((bit == 1 | Line 4235 if (((bit == 0) && nclass) || ((bit == 1 |
4235 | { | { |
4236 | if (length >= MAX_RANGE_SIZE) | if (length >= MAX_RANGE_SIZE) |
4237 | return FALSE; | return FALSE; |
4238 | ranges[2 + length] = 256; | ranges[length] = 256; |
4239 | length++; | length++; |
4240 | } | } |
ranges[0] = length; | ||
4241 | ||
4242 | return check_ranges(common, ranges, backtracks, FALSE); | if (length < 0 || length > 4) |
4243 | return FALSE; | |
4244 | ||
4245 | bit = bits[0] & 0x1; | |
4246 | if (invert) bit ^= 0x1; | |
4247 | ||
4248 | /* No character is accepted. */ | |
4249 | if (length == 0 && bit == 0) | |
4250 | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); | |
4251 | ||
4252 | switch(length) | |
4253 | { | |
4254 | case 0: | |
4255 | /* When bit != 0, all characters are accepted. */ | |
4256 | return TRUE; | |
4257 | ||
4258 | case 1: | |
4259 | add_jump(compiler, backtracks, CMP(bit == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0])); | |
4260 | return TRUE; | |
4261 | ||
4262 | case 2: | |
4263 | if (ranges[0] + 1 != ranges[1]) | |
4264 | { | |
4265 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]); | |
4266 | add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0])); | |
4267 | } | |
4268 | else | |
4269 | add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0])); | |
4270 | return TRUE; | |
4271 | ||
4272 | case 3: | |
4273 | if (bit != 0) | |
4274 | { | |
4275 | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | |
4276 | if (ranges[0] + 1 != ranges[1]) | |
4277 | { | |
4278 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]); | |
4279 | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0])); | |
4280 | } | |
4281 | else | |
4282 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0])); | |
4283 | return TRUE; | |
4284 | } | |
4285 | ||
4286 | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[0])); | |
4287 | if (ranges[1] + 1 != ranges[2]) | |
4288 | { | |
4289 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1]); | |
4290 | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1])); | |
4291 | } | |
4292 | else | |
4293 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1])); | |
4294 | return TRUE; | |
4295 | ||