Parent Directory
|
Revision Log
|
Patch
revision 958 by zherczeg, Wed Apr 11 10:19:10 2012 UTC | revision 1084 by chpe, Tue Oct 16 15:55:28 2012 UTC | |
---|---|---|
# | Line 46 POSSIBILITY OF SUCH DAMAGE. | Line 46 POSSIBILITY OF SUCH DAMAGE. |
46 | ||
47 | #include "pcre_internal.h" | #include "pcre_internal.h" |
48 | ||
49 | #ifdef SUPPORT_JIT | #if defined SUPPORT_JIT |
50 | ||
51 | /* All-in-one: Since we use the JIT compiler only from here, | /* All-in-one: Since we use the JIT compiler only from here, |
52 | we just include it. This way we don't need to touch the build | we just include it. This way we don't need to touch the build |
# | Line 65 system files. */ | Line 65 system files. */ |
65 | #error Unsupported architecture | #error Unsupported architecture |
66 | #endif | #endif |
67 | ||
68 | /* Allocate memory on the stack. Fast, but limited size. */ | /* Allocate memory for the regex stack on the real machine stack. |
69 | #define LOCAL_SPACE_SIZE 32768 | Fast, but limited size. */ |
70 | #define MACHINE_STACK_SIZE 32768 | |
71 | ||
72 | /* Growth rate for stack allocated by the OS. Should be the multiply | |
73 | of page size. */ | |
74 | #define STACK_GROWTH_RATE 8192 | #define STACK_GROWTH_RATE 8192 |
75 | ||
76 | /* Enable to check that the allocation could destroy temporaries. */ | /* Enable to check that the allocation could destroy temporaries. */ |
# | Line 82 The code generator follows the recursive | Line 85 The code generator follows the recursive |
85 | expressions. The basic blocks of regular expressions are condition checkers | expressions. The basic blocks of regular expressions are condition checkers |
86 | whose execute different commands depending on the result of the condition check. | whose execute different commands depending on the result of the condition check. |
87 | The relationship between the operators can be horizontal (concatenation) and | The relationship between the operators can be horizontal (concatenation) and |
88 | vertical (sub-expression) (See struct fallback_common for more details). | vertical (sub-expression) (See struct backtrack_common for more details). |
89 | ||
90 | 'ab' - 'a' and 'b' regexps are concatenated | 'ab' - 'a' and 'b' regexps are concatenated |
91 | 'a+' - 'a' is the sub-expression of the '+' operator | 'a+' - 'a' is the sub-expression of the '+' operator |
92 | ||
93 | The condition checkers are boolean (true/false) checkers. Machine code is generated | The condition checkers are boolean (true/false) checkers. Machine code is generated |
94 | for the checker itself and for the actions depending on the result of the checker. | for the checker itself and for the actions depending on the result of the checker. |
95 | The 'true' case is called as the hot path (expected path), and the other is called as | The 'true' case is called as the matching path (expected path), and the other is called as |
96 | the 'fallback' path. Branch instructions are expesive for all CPUs, so we avoid taken | the 'backtrack' path. Branch instructions are expesive for all CPUs, so we avoid taken |
97 | branches on the hot path. | branches on the matching path. |
98 | ||
99 | Greedy star operator (*) : | Greedy star operator (*) : |
100 | Hot path: match happens. | Matching path: match happens. |
101 | Fallback path: match failed. | Backtrack path: match failed. |
102 | Non-greedy star operator (*?) : | Non-greedy star operator (*?) : |
103 | Hot path: no need to perform a match. | Matching path: no need to perform a match. |
104 | Fallback path: match is required. | Backtrack path: match is required. |
105 | ||
106 | The following example shows how the code generated for a capturing bracket | The following example shows how the code generated for a capturing bracket |
107 | with two alternatives. Let A, B, C, D are arbirary regular expressions, and | with two alternatives. Let A, B, C, D are arbirary regular expressions, and |
# | Line 108 we have the following regular expression | Line 111 we have the following regular expression |
111 | ||
112 | The generated code will be the following: | The generated code will be the following: |
113 | ||
114 | A hot path | A matching path |
115 | '(' hot path (pushing arguments to the stack) | '(' matching path (pushing arguments to the stack) |
116 | B hot path | B matching path |
117 | ')' hot path (pushing arguments to the stack) | ')' matching path (pushing arguments to the stack) |
118 | D hot path | D matching path |
119 | return with successful match | return with successful match |
120 | ||
121 | D fallback path | D backtrack path |
122 | ')' fallback path (If we arrived from "C" jump to the fallback of "C") | ')' backtrack path (If we arrived from "C" jump to the backtrack of "C") |
123 | B fallback path | B backtrack path |
124 | C expected path | C expected path |
125 | jump to D hot path | jump to D matching path |
126 | C fallback path | C backtrack path |
127 | A fallback path | A backtrack path |
128 | ||
129 | Notice, that the order of fallback code paths are the opposite of the fast | Notice, that the order of backtrack code paths are the opposite of the fast |
130 | code paths. In this way the topmost value on the stack is always belong | code paths. In this way the topmost value on the stack is always belong |
131 | to the current fallback code path. The fallback code path must check | to the current backtrack code path. The backtrack path must check |
132 | whether there is a next alternative. If so, it needs to jump back to | whether there is a next alternative. If so, it needs to jump back to |
133 | the hot path eventually. Otherwise it needs to clear out its own stack | the matching path eventually. Otherwise it needs to clear out its own stack |
134 | frame and continue the execution on the fallback code paths. | frame and continue the execution on the backtrack code paths. |
135 | */ | */ |
136 | ||
137 | /* | /* |
138 | Saved stack frames: | Saved stack frames: |
139 | ||
140 | Atomic blocks and asserts require reloading the values of local variables | Atomic blocks and asserts require reloading the values of private data |
141 | when the fallback mechanism performed. Because of OP_RECURSE, the locals | when the backtrack mechanism performed. Because of OP_RECURSE, the data |
142 | are not necessarly known in compile time, thus we need a dynamic restore | are not necessarly known in compile time, thus we need a dynamic restore |
143 | mechanism. | mechanism. |
144 | ||
145 | The stack frames are stored in a chain list, and have the following format: | The stack frames are stored in a chain list, and have the following format: |
146 | ([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] | ([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] |
147 | ||
148 | Thus we can restore the locals to a particular point in the stack. | Thus we can restore the private data to a particular point in the stack. |
149 | */ | */ |
150 | ||
151 | typedef struct jit_arguments { | typedef struct jit_arguments { |
# | Line 181 typedef struct stub_list { | Line 184 typedef struct stub_list { |
184 | enum stub_types type; | enum stub_types type; |
185 | int data; | int data; |
186 | struct sljit_jump *start; | struct sljit_jump *start; |
187 | struct sljit_label *leave; | struct sljit_label *quit; |
188 | struct stub_list *next; | struct stub_list *next; |
189 | } stub_list; | } stub_list; |
190 | ||
191 | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); |
192 | ||
193 | /* The following structure is the key data type for the recursive | /* The following structure is the key data type for the recursive |
194 | code generator. It is allocated by compile_hotpath, and contains | code generator. It is allocated by compile_matchingpath, and contains |
195 | the aguments for compile_fallbackpath. Must be the first member | the aguments for compile_backtrackingpath. Must be the first member |
196 | of its descendants. */ | of its descendants. */ |
197 | typedef struct fallback_common { | typedef struct backtrack_common { |
198 | /* Concatenation stack. */ | /* Concatenation stack. */ |
199 | struct fallback_common *prev; | struct backtrack_common *prev; |
200 | jump_list *nextfallbacks; | jump_list *nextbacktracks; |
201 | /* Internal stack (for component operators). */ | /* Internal stack (for component operators). */ |
202 | struct fallback_common *top; | struct backtrack_common *top; |
203 | jump_list *topfallbacks; | jump_list *topbacktracks; |
204 | /* Opcode pointer. */ | /* Opcode pointer. */ |
205 | pcre_uchar *cc; | pcre_uchar *cc; |
206 | } fallback_common; | } backtrack_common; |
207 | ||
208 | typedef struct assert_fallback { | typedef struct assert_backtrack { |
209 | fallback_common common; | backtrack_common common; |
210 | jump_list *condfailed; | jump_list *condfailed; |
211 | /* Less than 0 (-1) if a frame is not needed. */ | /* Less than 0 (-1) if a frame is not needed. */ |
212 | int framesize; | int framesize; |
213 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
214 | int localptr; | int private_data_ptr; |
215 | /* For iterators. */ | /* For iterators. */ |
216 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
217 | } assert_fallback; | } assert_backtrack; |
218 | ||
219 | typedef struct bracket_fallback { | typedef struct bracket_backtrack { |
220 | fallback_common common; | backtrack_common common; |
221 | /* Where to coninue if an alternative is successfully matched. */ | /* Where to coninue if an alternative is successfully matched. */ |
222 | struct sljit_label *althotpath; | struct sljit_label *alternative_matchingpath; |
223 | /* For rmin and rmax iterators. */ | /* For rmin and rmax iterators. */ |
224 | struct sljit_label *recursivehotpath; | struct sljit_label *recursive_matchingpath; |
225 | /* For greedy ? operator. */ | /* For greedy ? operator. */ |
226 | struct sljit_label *zerohotpath; | struct sljit_label *zero_matchingpath; |
227 | /* Contains the branches of a failed condition. */ | /* Contains the branches of a failed condition. */ |
228 | union { | union { |
229 | /* Both for OP_COND, OP_SCOND. */ | /* Both for OP_COND, OP_SCOND. */ |
230 | jump_list *condfailed; | jump_list *condfailed; |
231 | assert_fallback *assert; | assert_backtrack *assert; |
232 | /* For OP_ONCE. -1 if not needed. */ | /* For OP_ONCE. -1 if not needed. */ |
233 | int framesize; | int framesize; |
234 | } u; | } u; |
235 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
236 | int localptr; | int private_data_ptr; |
237 | } bracket_fallback; | } bracket_backtrack; |
238 | ||
239 | typedef struct bracketpos_fallback { | typedef struct bracketpos_backtrack { |
240 | fallback_common common; | backtrack_common common; |
241 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
242 | int localptr; | int private_data_ptr; |
243 | /* Reverting stack is needed. */ | /* Reverting stack is needed. */ |
244 | int framesize; | int framesize; |
245 | /* Allocated stack size. */ | /* Allocated stack size. */ |
246 | int stacksize; | int stacksize; |
247 | } bracketpos_fallback; | } bracketpos_backtrack; |
248 | ||
249 | typedef struct braminzero_fallback { | typedef struct braminzero_backtrack { |
250 | fallback_common common; | backtrack_common common; |
251 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
252 | } braminzero_fallback; | } braminzero_backtrack; |
253 | ||
254 | typedef struct iterator_fallback { | typedef struct iterator_backtrack { |
255 | fallback_common common; | backtrack_common common; |
256 | /* Next iteration. */ | /* Next iteration. */ |
257 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
258 | } iterator_fallback; | } iterator_backtrack; |
259 | ||
260 | typedef struct recurse_entry { | typedef struct recurse_entry { |
261 | struct recurse_entry *next; | struct recurse_entry *next; |
# | Line 264 typedef struct recurse_entry { | Line 267 typedef struct recurse_entry { |
267 | int start; | int start; |
268 | } recurse_entry; | } recurse_entry; |
269 | ||
270 | typedef struct recurse_fallback { | typedef struct recurse_backtrack { |
271 | fallback_common common; | backtrack_common common; |
272 | } recurse_fallback; | } recurse_backtrack; |
273 | ||
274 | #define MAX_RANGE_SIZE 6 | |
275 | ||
276 | typedef struct compiler_common { | typedef struct compiler_common { |
277 | struct sljit_compiler *compiler; | struct sljit_compiler *compiler; |
278 | pcre_uchar *start; | pcre_uchar *start; |
279 | ||
280 | /* Opcode local area direct map. */ | /* Maps private data offset to each opcode. */ |
281 | int *localptrs; | int *private_data_ptrs; |
282 | /* Tells whether the capturing bracket is optimized. */ | |
283 | pcre_uint8 *optimized_cbracket; | |
284 | /* Starting offset of private data for capturing brackets. */ | |
285 | int cbraptr; | int cbraptr; |
286 | /* OVector starting point. Must be divisible by 2. */ | /* OVector starting point. Must be divisible by 2. */ |
287 | int ovector_start; | int ovector_start; |
# | Line 290 typedef struct compiler_common { | Line 298 typedef struct compiler_common { |
298 | /* Points to the marked string. */ | /* Points to the marked string. */ |
299 | int mark_ptr; | int mark_ptr; |
300 | ||
301 | /* Other */ | /* Flipped and lower case tables. */ |
302 | const pcre_uint8 *fcc; | const pcre_uint8 *fcc; |
303 | sljit_w lcc; | sljit_w lcc; |
304 | /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */ | |
305 | int mode; | int mode; |
306 | /* Newline control. */ | |
307 | int nltype; | int nltype; |
308 | int newline; | int newline; |
309 | int bsr_nltype; | int bsr_nltype; |
310 | /* Dollar endonly. */ | |
311 | int endonly; | int endonly; |
312 | BOOL has_set_som; | BOOL has_set_som; |
313 | /* Tables. */ | |
314 | sljit_w ctypes; | sljit_w ctypes; |
315 | int digits[2 + MAX_RANGE_SIZE]; | |
316 | /* Named capturing brackets. */ | |
317 | sljit_uw name_table; | sljit_uw name_table; |
318 | sljit_w name_count; | sljit_w name_count; |
319 | sljit_w name_entry_size; | sljit_w name_entry_size; |
320 | ||
321 | /* Labels and jump lists. */ | /* Labels and jump lists. */ |
322 | struct sljit_label *partialmatchlabel; | struct sljit_label *partialmatchlabel; |
323 | struct sljit_label *leavelabel; | struct sljit_label *quitlabel; |
324 | struct sljit_label *acceptlabel; | struct sljit_label *acceptlabel; |
325 | stub_list *stubs; | stub_list *stubs; |
326 | recurse_entry *entries; | recurse_entry *entries; |
327 | recurse_entry *currententry; | recurse_entry *currententry; |
328 | jump_list *partialmatch; | jump_list *partialmatch; |
329 | jump_list *leave; | jump_list *quit; |
330 | jump_list *accept; | jump_list *accept; |
331 | jump_list *calllimit; | jump_list *calllimit; |
332 | jump_list *stackalloc; | jump_list *stackalloc; |
# | Line 329 typedef struct compiler_common { | Line 343 typedef struct compiler_common { |
343 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
344 | BOOL use_ucp; | BOOL use_ucp; |
345 | #endif | #endif |
346 | #ifndef COMPILE_PCRE32 | |
347 | jump_list *utfreadchar; | jump_list *utfreadchar; |
348 | #endif | |
349 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
350 | jump_list *utfreadtype8; | jump_list *utfreadtype8; |
351 | #endif | #endif |
# | Line 349 typedef struct compare_context { | Line 365 typedef struct compare_context { |
365 | union { | union { |
366 | sljit_i asint; | sljit_i asint; |
367 | sljit_uh asushort; | sljit_uh asushort; |
368 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
369 | sljit_ub asbyte; | sljit_ub asbyte; |
370 | sljit_ub asuchars[4]; | sljit_ub asuchars[4]; |
371 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
372 | sljit_uh asuchars[2]; | sljit_uh asuchars[2]; |
373 | #endif | #elif defined COMPILE_PCRE32 |
374 | sljit_ui asuchars[1]; | |
375 | #endif | #endif |
376 | } c; | } c; |
377 | union { | union { |
378 | sljit_i asint; | sljit_i asint; |
379 | sljit_uh asushort; | sljit_uh asushort; |
380 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
381 | sljit_ub asbyte; | sljit_ub asbyte; |
382 | sljit_ub asuchars[4]; | sljit_ub asuchars[4]; |
383 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
384 | sljit_uh asuchars[2]; | sljit_uh asuchars[2]; |
385 | #endif | #elif defined COMPILE_PCRE32 |
386 | sljit_ui asuchars[1]; | |
387 | #endif | #endif |
388 | } oc; | } oc; |
389 | #endif | #endif |
# | Line 396 enum { | Line 412 enum { |
412 | #define CALL_COUNT SLJIT_SAVED_EREG2 | #define CALL_COUNT SLJIT_SAVED_EREG2 |
413 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 |
414 | ||
415 | /* Locals layout. */ | /* Local space layout. */ |
416 | /* These two locals can be used by the current opcode. */ | /* These two locals can be used by the current opcode. */ |
417 | #define LOCALS0 (0 * sizeof(sljit_w)) | #define LOCALS0 (0 * sizeof(sljit_w)) |
418 | #define LOCALS1 (1 * sizeof(sljit_w)) | #define LOCALS1 (1 * sizeof(sljit_w)) |
# | Line 412 the start pointers when the end of the c | Line 428 the start pointers when the end of the c |
428 | #define OVECTOR_START (common->ovector_start) | #define OVECTOR_START (common->ovector_start) |
429 | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_w)) | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_w)) |
430 | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_w)) | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_w)) |
431 | #define PRIV_DATA(cc) (common->localptrs[(cc) - common->start]) | #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start]) |
432 | ||
433 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
434 | #define MOV_UCHAR SLJIT_MOV_UB | #define MOV_UCHAR SLJIT_MOV_UB |
435 | #define MOVU_UCHAR SLJIT_MOVU_UB | #define MOVU_UCHAR SLJIT_MOVU_UB |
436 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
437 | #define MOV_UCHAR SLJIT_MOV_UH | #define MOV_UCHAR SLJIT_MOV_UH |
438 | #define MOVU_UCHAR SLJIT_MOVU_UH | #define MOVU_UCHAR SLJIT_MOVU_UH |
439 | #elif defined COMPILE_PCRE32 | |
440 | #define MOV_UCHAR SLJIT_MOV_UI | |
441 | #define MOVU_UCHAR SLJIT_MOVU_UI | |
442 | #else | #else |
443 | #error Unsupported compiling mode | #error Unsupported compiling mode |
444 | #endif | #endif |
#endif | ||
445 | ||
446 | /* Shortcuts. */ | /* Shortcuts. */ |
447 | #define DEFINE_COMPILER \ | #define DEFINE_COMPILER \ |
# | Line 461 return cc; | Line 478 return cc; |
478 | ||
479 | /* Functions whose might need modification for all new supported opcodes: | /* Functions whose might need modification for all new supported opcodes: |
480 | next_opcode | next_opcode |
481 | get_localspace | get_private_data_length |
482 | set_localptrs | set_private_data_ptrs |
483 | get_framesize | get_framesize |
484 | init_frame | init_frame |
485 | get_localsize | get_private_data_length_for_copy |
486 | copy_locals | copy_private_data |
487 | compile_hotpath | compile_matchingpath |
488 | compile_fallbackpath | compile_backtrackingpath |
489 | */ | */ |
490 | ||
491 | static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) | static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) |
# | Line 574 switch(*cc) | Line 591 switch(*cc) |
591 | case OP_NOTPOSPLUSI: | case OP_NOTPOSPLUSI: |
592 | case OP_NOTPOSQUERYI: | case OP_NOTPOSQUERYI: |
593 | cc += 2; | cc += 2; |
594 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
595 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
596 | #endif | #endif |
597 | return cc; | return cc; |
# | Line 596 switch(*cc) | Line 613 switch(*cc) |
613 | case OP_NOTEXACTI: | case OP_NOTEXACTI: |
614 | case OP_NOTPOSUPTOI: | case OP_NOTPOSUPTOI: |
615 | cc += 2 + IMM2_SIZE; | cc += 2 + IMM2_SIZE; |
616 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
617 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
618 | #endif | #endif |
619 | return cc; | return cc; |
# | Line 667 switch(*cc) | Line 684 switch(*cc) |
684 | } | } |
685 | } | } |
686 | ||
687 | static int get_localspace(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | #define CASE_ITERATOR_PRIVATE_DATA_1 \ |
688 | case OP_MINSTAR: \ | |
689 | case OP_MINPLUS: \ | |
690 | case OP_QUERY: \ | |
691 | case OP_MINQUERY: \ | |
692 | case OP_MINSTARI: \ | |
693 | case OP_MINPLUSI: \ | |
694 | case OP_QUERYI: \ | |
695 | case OP_MINQUERYI: \ | |
696 | case OP_NOTMINSTAR: \ | |
697 | case OP_NOTMINPLUS: \ | |
698 | case OP_NOTQUERY: \ | |
699 | case OP_NOTMINQUERY: \ | |
700 | case OP_NOTMINSTARI: \ | |
701 | case OP_NOTMINPLUSI: \ | |
702 | case OP_NOTQUERYI: \ | |
703 | case OP_NOTMINQUERYI: | |
704 | ||
705 | #define CASE_ITERATOR_PRIVATE_DATA_2A \ | |
706 | case OP_STAR: \ | |
707 | case OP_PLUS: \ | |
708 | case OP_STARI: \ | |
709 | case OP_PLUSI: \ | |
710 | case OP_NOTSTAR: \ | |
711 | case OP_NOTPLUS: \ | |
712 | case OP_NOTSTARI: \ | |
713 | case OP_NOTPLUSI: | |
714 | ||
715 | #define CASE_ITERATOR_PRIVATE_DATA_2B \ | |
716 | case OP_UPTO: \ | |
717 | case OP_MINUPTO: \ | |
718 | case OP_UPTOI: \ | |
719 | case OP_MINUPTOI: \ | |
720 | case OP_NOTUPTO: \ | |
721 | case OP_NOTMINUPTO: \ | |
722 | case OP_NOTUPTOI: \ | |
723 | case OP_NOTMINUPTOI: | |
724 | ||
725 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \ | |
726 | case OP_TYPEMINSTAR: \ | |
727 | case OP_TYPEMINPLUS: \ | |
728 | case OP_TYPEQUERY: \ | |
729 | case OP_TYPEMINQUERY: | |
730 | ||
731 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \ | |
732 | case OP_TYPESTAR: \ | |
733 | case OP_TYPEPLUS: | |
734 | ||
735 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \ | |
736 | case OP_TYPEUPTO: \ | |
737 | case OP_TYPEMINUPTO: | |
738 | ||
739 | static int get_class_iterator_size(pcre_uchar *cc) | |
740 | { | { |
741 | int localspace = 0; | switch(*cc) |
742 | { | |
743 | case OP_CRSTAR: | |
744 | case OP_CRPLUS: | |
745 | return 2; | |
746 | ||
747 | case OP_CRMINSTAR: | |
748 | case OP_CRMINPLUS: | |
749 | case OP_CRQUERY: | |
750 | case OP_CRMINQUERY: | |
751 | return 1; | |
752 | ||
753 | case OP_CRRANGE: | |
754 | case OP_CRMINRANGE: | |
755 | if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE)) | |
756 | return 0; | |
757 | return 2; | |
758 | ||
759 | default: | |
760 | return 0; | |
761 | } | |
762 | } | |
763 | ||
764 | static int get_private_data_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | |
765 | { | |
766 | int private_data_length = 0; | |
767 | pcre_uchar *alternative; | pcre_uchar *alternative; |
768 | pcre_uchar *name; | |
769 | pcre_uchar *end = NULL; | |
770 | int space, size, bracketlen, i; | |
771 | ||
772 | /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ | /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ |
773 | while (cc < ccend) | while (cc < ccend) |
774 | { | { |
775 | space = 0; | |
776 | size = 0; | |
777 | bracketlen = 0; | |
778 | switch(*cc) | switch(*cc) |
779 | { | { |
780 | case OP_SET_SOM: | case OP_SET_SOM: |
# | Line 681 while (cc < ccend) | Line 782 while (cc < ccend) |
782 | cc += 1; | cc += 1; |
783 | break; | break; |
784 | ||
785 | case OP_REF: | |
786 | case OP_REFI: | |
787 | common->optimized_cbracket[GET2(cc, 1)] = 0; | |
788 | cc += 1 + IMM2_SIZE; | |
789 | break; | |
790 | ||
791 | case OP_ASSERT: | case OP_ASSERT: |
792 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
793 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
# | Line 690 while (cc < ccend) | Line 797 while (cc < ccend) |
797 | case OP_BRAPOS: | case OP_BRAPOS: |
798 | case OP_SBRA: | case OP_SBRA: |
799 | case OP_SBRAPOS: | case OP_SBRAPOS: |
800 | case OP_SCOND: | private_data_length += sizeof(sljit_w); |
801 | localspace += sizeof(sljit_w); | bracketlen = 1 + LINK_SIZE; |
cc += 1 + LINK_SIZE; | ||
802 | break; | break; |
803 | ||
804 | case OP_CBRAPOS: | case OP_CBRAPOS: |
805 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
806 | localspace += sizeof(sljit_w); | private_data_length += sizeof(sljit_w); |
807 | cc += 1 + LINK_SIZE + IMM2_SIZE; | common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0; |
808 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
809 | break; | break; |
810 | ||
811 | case OP_COND: | case OP_COND: |
812 | /* Might be a hidden SCOND. */ | case OP_SCOND: |
813 | alternative = cc + GET(cc, 1); | bracketlen = cc[1 + LINK_SIZE]; |
814 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (bracketlen == OP_CREF) |
815 | localspace += sizeof(sljit_w); | { |
816 | cc += 1 + LINK_SIZE; | bracketlen = GET2(cc, 1 + LINK_SIZE + 1); |
817 | common->optimized_cbracket[bracketlen] = 0; | |
818 | } | |
819 | else if (bracketlen == OP_NCREF) | |
820 | { | |
821 | bracketlen = GET2(cc, 1 + LINK_SIZE + 1); | |
822 | name = (pcre_uchar *)common->name_table; | |
823 | alternative = name; | |
824 | for (i = 0; i < common->name_count; i++) | |
825 | { | |
826 | if (GET2(name, 0) == bracketlen) break; | |
827 | name += common->name_entry_size; | |
828 | } | |
829 | SLJIT_ASSERT(i != common->name_count); | |
830 | ||
831 | for (i = 0; i < common->name_count; i++) | |
832 | { | |
833 | if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0) | |
834 | common->optimized_cbracket[GET2(alternative, 0)] = 0; | |
835 | alternative += common->name_entry_size; | |
836 | } | |
837 | } | |
838 | ||
839 | if (*cc == OP_COND) | |
840 | { | |
841 | /* Might be a hidden SCOND. */ | |
842 | alternative = cc + GET(cc, 1); | |
843 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | |
844 | private_data_length += sizeof(sljit_w); | |
845 | } | |
846 | else | |
847 | private_data_length += sizeof(sljit_w); | |
848 | bracketlen = 1 + LINK_SIZE; | |
849 | break; | |
850 | ||
851 | case OP_BRA: | |
852 | bracketlen = 1 + LINK_SIZE; | |
853 | break; | |
854 | ||
855 | case OP_CBRA: | |
856 | case OP_SCBRA: | |
857 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
858 | break; | |
859 | ||
860 | CASE_ITERATOR_PRIVATE_DATA_1 | |
861 | space = 1; | |
862 | size = -2; | |
863 | break; | |
864 | ||
865 | CASE_ITERATOR_PRIVATE_DATA_2A | |
866 | space = 2; | |
867 | size = -2; | |
868 | break; | |
869 | ||
870 | CASE_ITERATOR_PRIVATE_DATA_2B | |
871 | space = 2; | |
872 | size = -(2 + IMM2_SIZE); | |
873 | break; | break; |
874 | ||
875 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
876 | space = 1; | |
877 | size = 1; | |
878 | break; | |
879 | ||
880 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
881 | if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI) | |
882 | space = 2; | |
883 | size = 1; | |
884 | break; | |
885 | ||
886 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
887 | if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) | |
888 | space = 2; | |
889 | size = 1 + IMM2_SIZE; | |
890 | break; | |
891 | ||
892 | case OP_CLASS: | |
893 | case OP_NCLASS: | |
894 | size += 1 + 32 / sizeof(pcre_uchar); | |
895 | space = get_class_iterator_size(cc + size); | |
896 | break; | |
897 | ||
898 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
899 | case OP_XCLASS: | |
900 | size = GET(cc, 1); | |
901 | space = get_class_iterator_size(cc + size); | |
902 | break; | |
903 | #endif | |
904 | ||
905 | case OP_RECURSE: | case OP_RECURSE: |
906 | /* Set its value only once. */ | /* Set its value only once. */ |
907 | if (common->recursive_head == 0) | if (common->recursive_head == 0) |
# | Line 734 while (cc < ccend) | Line 927 while (cc < ccend) |
927 | return -1; | return -1; |
928 | break; | break; |
929 | } | } |
930 | ||
931 | if (space > 0 && cc >= end) | |
932 | private_data_length += sizeof(sljit_w) * space; | |
933 | ||
934 | if (size != 0) | |
935 | { | |
936 | if (size < 0) | |
937 | { | |
938 | cc += -size; | |
939 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
940 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
941 | #endif | |
942 | } | |
943 | else | |
944 | cc += size; | |
945 | } | |
946 | ||
947 | if (bracketlen > 0) | |
948 | { | |
949 | if (cc >= end) | |
950 | { | |
951 | end = bracketend(cc); | |
952 | if (end[-1 - LINK_SIZE] == OP_KET) | |
953 | end = NULL; | |
954 | } | |
955 | cc += bracketlen; | |
956 | } | |
957 | } | } |
958 | return localspace; | return private_data_length; |
959 | } | } |
960 | ||
961 | static void set_localptrs(compiler_common *common, int localptr, pcre_uchar *ccend) | static void set_private_data_ptrs(compiler_common *common, int private_data_ptr, pcre_uchar *ccend) |
962 | { | { |
963 | pcre_uchar *cc = common->start; | pcre_uchar *cc = common->start; |
964 | pcre_uchar *alternative; | pcre_uchar *alternative; |
965 | pcre_uchar *end = NULL; | |
966 | int space, size, bracketlen; | |
967 | ||
968 | while (cc < ccend) | while (cc < ccend) |
969 | { | { |
970 | space = 0; | |
971 | size = 0; | |
972 | bracketlen = 0; | |
973 | switch(*cc) | switch(*cc) |
974 | { | { |
975 | case OP_ASSERT: | case OP_ASSERT: |
# | Line 756 while (cc < ccend) | Line 982 while (cc < ccend) |
982 | case OP_SBRA: | case OP_SBRA: |
983 | case OP_SBRAPOS: | case OP_SBRAPOS: |
984 | case OP_SCOND: | case OP_SCOND: |
985 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
986 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_w); |
987 | cc += 1 + LINK_SIZE; | bracketlen = 1 + LINK_SIZE; |
988 | break; | break; |
989 | ||
990 | case OP_CBRAPOS: | case OP_CBRAPOS: |
991 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
992 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
993 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_w); |
994 | cc += 1 + LINK_SIZE + IMM2_SIZE; | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; |
995 | break; | break; |
996 | ||
997 | case OP_COND: | case OP_COND: |
# | Line 773 while (cc < ccend) | Line 999 while (cc < ccend) |
999 | alternative = cc + GET(cc, 1); | alternative = cc + GET(cc, 1); |
1000 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1001 | { | { |
1002 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
1003 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_w); |
1004 | } | } |
1005 | cc += 1 + LINK_SIZE; | bracketlen = 1 + LINK_SIZE; |
1006 | break; | |
1007 | ||
1008 | case OP_BRA: | |
1009 | bracketlen = 1 + LINK_SIZE; | |
1010 | break; | |
1011 | ||
1012 | case OP_CBRA: | |
1013 | case OP_SCBRA: | |
1014 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
1015 | break; | |
1016 | ||
1017 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1018 | space = 1; | |
1019 | size = -2; | |
1020 | break; | break; |
1021 | ||
1022 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1023 | space = 2; | |
1024 | size = -2; | |
1025 | break; | |
1026 | ||
1027 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1028 | space = 2; | |
1029 | size = -(2 + IMM2_SIZE); | |
1030 | break; | |
1031 | ||
1032 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1033 | space = 1; | |
1034 | size = 1; | |
1035 | break; | |
1036 | ||
1037 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1038 | if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI) | |
1039 | space = 2; | |
1040 | size = 1; | |
1041 | break; | |
1042 | ||
1043 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1044 | if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) | |
1045 | space = 2; | |
1046 | size = 1 + IMM2_SIZE; | |
1047 | break; | |
1048 | ||
1049 | case OP_CLASS: | |
1050 | case OP_NCLASS: | |
1051 | size += 1 + 32 / sizeof(pcre_uchar); | |
1052 | space = get_class_iterator_size(cc + size); | |
1053 | break; | |
1054 | ||
1055 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1056 | case OP_XCLASS: | |
1057 | size = GET(cc, 1); | |
1058 | space = get_class_iterator_size(cc + size); | |
1059 | break; | |
1060 | #endif | |
1061 | ||
1062 | default: | default: |
1063 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1064 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1065 | break; | break; |
1066 | } | } |
1067 | ||
1068 | if (space > 0 && cc >= end) | |
1069 | { | |
1070 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | |
1071 | private_data_ptr += sizeof(sljit_w) * space; | |
1072 | } | |
1073 | ||
1074 | if (size != 0) | |
1075 | { | |
1076 | if (size < 0) | |
1077 | { | |
1078 | cc += -size; | |
1079 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
1080 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1081 | #endif | |
1082 | } | |
1083 | else | |
1084 | cc += size; | |
1085 | } | |
1086 | ||
1087 | if (bracketlen > 0) | |
1088 | { | |
1089 | if (cc >= end) | |
1090 | { | |
1091 | end = bracketend(cc); | |
1092 | if (end[-1 - LINK_SIZE] == OP_KET) | |
1093 | end = NULL; | |
1094 | } | |
1095 | cc += bracketlen; | |
1096 | } | |
1097 | } | } |
1098 | } | } |
1099 | ||
# | Line 960 OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st | Line 1270 OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st |
1270 | SLJIT_ASSERT(stackpos == STACK(stacktop)); | SLJIT_ASSERT(stackpos == STACK(stacktop)); |
1271 | } | } |
1272 | ||
1273 | static SLJIT_INLINE int get_localsize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | static SLJIT_INLINE int get_private_data_length_for_copy(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) |
1274 | { | { |
1275 | int localsize = 2; | int private_data_length = 2; |
1276 | int size; | |
1277 | pcre_uchar *alternative; | pcre_uchar *alternative; |
1278 | /* Calculate the sum of the local variables. */ | /* Calculate the sum of the private machine words. */ |
1279 | while (cc < ccend) | while (cc < ccend) |
1280 | { | { |
1281 | size = 0; | |
1282 | switch(*cc) | switch(*cc) |
1283 | { | { |
1284 | case OP_ASSERT: | case OP_ASSERT: |
# | Line 979 while (cc < ccend) | Line 1291 while (cc < ccend) |
1291 | case OP_SBRA: | case OP_SBRA: |
1292 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1293 | case OP_SCOND: | case OP_SCOND: |
1294 | localsize++; | private_data_length++; |
1295 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1296 | break; | break; |
1297 | ||
1298 | case OP_CBRA: | case OP_CBRA: |
1299 | case OP_SCBRA: | case OP_SCBRA: |
1300 | localsize++; | if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) |
1301 | private_data_length++; | |
1302 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1303 | break; | break; |
1304 | ||
1305 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1306 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1307 | localsize += 2; | private_data_length += 2; |
1308 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1309 | break; | break; |
1310 | ||
# | Line 999 while (cc < ccend) | Line 1312 while (cc < ccend) |
1312 | /* Might be a hidden SCOND. */ | /* Might be a hidden SCOND. */ |
1313 | alternative = cc + GET(cc, 1); | alternative = cc + GET(cc, 1); |
1314 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1315 | localsize++; | private_data_length++; |
1316 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1317 | break; | break; |
1318 | ||
1319 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1320 | if (PRIVATE_DATA(cc)) | |
1321 | private_data_length++; | |
1322 | cc += 2; | |
1323 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
1324 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1325 | #endif | |
1326 | break; | |
1327 | ||
1328 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1329 | if (PRIVATE_DATA(cc)) | |
1330 | private_data_length += 2; | |
1331 | cc += 2; | |
1332 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
1333 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1334 | #endif | |
1335 | break; | |
1336 | ||
1337 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1338 | if (PRIVATE_DATA(cc)) | |
1339 | private_data_length += 2; | |
1340 | cc += 2 + IMM2_SIZE; | |
1341 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
1342 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1343 | #endif | |
1344 | break; | |
1345 | ||
1346 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1347 | if (PRIVATE_DATA(cc)) | |
1348 | private_data_length++; | |
1349 | cc += 1; | |
1350 | break; | |
1351 | ||
1352 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1353 | if (PRIVATE_DATA(cc)) | |
1354 | private_data_length += 2; | |
1355 | cc += 1; | |
1356 | break; | |
1357 | ||
1358 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1359 | if (PRIVATE_DATA(cc)) | |
1360 | private_data_length += 2; | |
1361 | cc += 1 + IMM2_SIZE; | |
1362 | break; | |
1363 | ||
1364 | case OP_CLASS: | |
1365 | case OP_NCLASS: | |
1366 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1367 | case OP_XCLASS: | |
1368 | size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); | |
1369 | #else | |
1370 | size = 1 + 32 / (int)sizeof(pcre_uchar); | |
1371 | #endif | |
1372 | if (PRIVATE_DATA(cc)) | |
1373 | private_data_length += get_class_iterator_size(cc + size); | |
1374 | cc += size; | |
1375 | break; | |
1376 | ||
1377 | default: | default: |
1378 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1379 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
# | Line 1010 while (cc < ccend) | Line 1381 while (cc < ccend) |
1381 | } | } |
1382 | } | } |
1383 | SLJIT_ASSERT(cc == ccend); | SLJIT_ASSERT(cc == ccend); |
1384 | return localsize; | return private_data_length; |
1385 | } | } |
1386 | ||
1387 | static void copy_locals(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, | static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, |
1388 | BOOL save, int stackptr, int stacktop) | BOOL save, int stackptr, int stacktop) |
1389 | { | { |
1390 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1391 | int srcw[2]; | int srcw[2]; |
1392 | int count; | int count, size; |
1393 | BOOL tmp1next = TRUE; | BOOL tmp1next = TRUE; |
1394 | BOOL tmp1empty = TRUE; | BOOL tmp1empty = TRUE; |
1395 | BOOL tmp2empty = TRUE; | BOOL tmp2empty = TRUE; |
# | Line 1083 while (status != end) | Line 1454 while (status != end) |
1454 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1455 | case OP_SCOND: | case OP_SCOND: |
1456 | count = 1; | count = 1; |
1457 | srcw[0] = PRIV_DATA(cc); | srcw[0] = PRIVATE_DATA(cc); |
1458 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
1459 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1460 | break; | break; |
1461 | ||
1462 | case OP_CBRA: | case OP_CBRA: |
1463 | case OP_SCBRA: | case OP_SCBRA: |
1464 | count = 1; | if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) |
1465 | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | { |
1466 | count = 1; | |
1467 | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | |
1468 | } | |
1469 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1470 | break; | break; |
1471 | ||
1472 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1473 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1474 | count = 2; | count = 2; |
1475 | srcw[0] = PRIVATE_DATA(cc); | |
1476 | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); |
1477 | srcw[0] = PRIV_DATA(cc); | SLJIT_ASSERT(srcw[0] != 0 && srcw[1] != 0); |
SLJIT_ASSERT(srcw[0] != 0); | ||
1478 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1479 | break; | break; |
1480 | ||
# | Line 1110 while (status != end) | Line 1484 while (status != end) |
1484 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1485 | { | { |
1486 | count = 1; | count = 1; |
1487 | srcw[0] = PRIV_DATA(cc); | srcw[0] = PRIVATE_DATA(cc); |
1488 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
1489 | } | } |
1490 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1491 | break; | break; |
1492 | ||
1493 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1494 | if (PRIVATE_DATA(cc)) | |
1495 | { | |
1496 | count = 1; | |
1497 | srcw[0] = PRIVATE_DATA(cc); | |
1498 | } | |
1499 | cc += 2; | |
1500 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
1501 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1502 | #endif | |
1503 | break; | |
1504 | ||
1505 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1506 | if (PRIVATE_DATA(cc)) | |
1507 | { | |
1508 | count = 2; | |
1509 | srcw[0] = PRIVATE_DATA(cc); | |
1510 | srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_w); | |
1511 | } | |
1512 | cc += 2; | |
1513 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
1514 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1515 | #endif | |
1516 | break; | |
1517 | ||
1518 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1519 | if (PRIVATE_DATA(cc)) | |
1520 | { | |
1521 | count = 2; | |
1522 | srcw[0] = PRIVATE_DATA(cc); | |
1523 | srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_w); | |
1524 | } | |
1525 | cc += 2 + IMM2_SIZE; | |
1526 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
1527 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1528 | #endif | |
1529 | break; | |
1530 | ||
1531 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1532 | if (PRIVATE_DATA(cc)) | |
1533 | { | |
1534 | count = 1; | |
1535 | srcw[0] = PRIVATE_DATA(cc); | |
1536 | } | |
1537 | cc += 1; | |
1538 | break; | |
1539 | ||
1540 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1541 | if (PRIVATE_DATA(cc)) | |
1542 | { | |
1543 | count = 2; | |
1544 | srcw[0] = PRIVATE_DATA(cc); | |
1545 | srcw[1] = srcw[0] + sizeof(sljit_w); | |
1546 | } | |
1547 | cc += 1; | |
1548 | break; | |
1549 | ||
1550 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1551 | if (PRIVATE_DATA(cc)) | |
1552 | { | |
1553 | count = 2; | |
1554 | srcw[0] = PRIVATE_DATA(cc); | |
1555 | srcw[1] = srcw[0] + sizeof(sljit_w); | |
1556 | } | |
1557 | cc += 1 + IMM2_SIZE; | |
1558 | break; | |
1559 | ||
1560 | case OP_CLASS: | |
1561 | case OP_NCLASS: | |
1562 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1563 | case OP_XCLASS: | |
1564 | size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); | |
1565 | #else | |
1566 | size = 1 + 32 / (int)sizeof(pcre_uchar); | |
1567 | #endif | |
1568 | if (PRIVATE_DATA(cc)) | |
1569 | switch(get_class_iterator_size(cc + size)) | |
1570 | { | |
1571 | case 1: | |
1572 | count = 1; | |
1573 | srcw[0] = PRIVATE_DATA(cc); | |
1574 | break; | |
1575 | ||
1576 | case 2: | |
1577 | count = 2; | |
1578 | srcw[0] = PRIVATE_DATA(cc); | |
1579 | srcw[1] = srcw[0] + sizeof(sljit_w); | |
1580 | break; | |
1581 | ||
1582 | default: | |
1583 | SLJIT_ASSERT_STOP(); | |
1584 | break; | |
1585 | } | |
1586 | cc += size; | |
1587 | break; | |
1588 | ||
1589 | default: | default: |
1590 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1591 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
# | Line 1218 if (save) | Line 1688 if (save) |
1688 | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); |
1689 | } | } |
1690 | ||
1691 | static SLJIT_INLINE BOOL ispowerof2(unsigned int value) | #undef CASE_ITERATOR_PRIVATE_DATA_1 |
1692 | #undef CASE_ITERATOR_PRIVATE_DATA_2A | |
1693 | #undef CASE_ITERATOR_PRIVATE_DATA_2B | |
1694 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1695 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1696 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1697 | ||
1698 | static SLJIT_INLINE BOOL is_powerof2(unsigned int value) | |
1699 | { | { |
1700 | return (value & (value - 1)) == 0; | return (value & (value - 1)) == 0; |
1701 | } | } |
# | Line 1228 static SLJIT_INLINE void set_jumps(jump_ | Line 1705 static SLJIT_INLINE void set_jumps(jump_ |
1705 | while (list) | while (list) |
1706 | { | { |
1707 | /* sljit_set_label is clever enough to do nothing | /* sljit_set_label is clever enough to do nothing |
1708 | if either the jump or the label is NULL */ | if either the jump or the label is NULL. */ |
1709 | sljit_set_label(list->jump, label); | sljit_set_label(list->jump, label); |
1710 | list = list->next; | list = list->next; |
1711 | } | } |
# | Line 1255 if (list_item) | Line 1732 if (list_item) |
1732 | list_item->type = type; | list_item->type = type; |
1733 | list_item->data = data; | list_item->data = data; |
1734 | list_item->start = start; | list_item->start = start; |
1735 | list_item->leave = LABEL(); | list_item->quit = LABEL(); |
1736 | list_item->next = common->stubs; | list_item->next = common->stubs; |
1737 | common->stubs = list_item; | common->stubs = list_item; |
1738 | } | } |
# | Line 1275 while (list_item) | Line 1752 while (list_item) |
1752 | add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); |
1753 | break; | break; |
1754 | } | } |
1755 | JUMPTO(SLJIT_JUMP, list_item->leave); | JUMPTO(SLJIT_JUMP, list_item->quit); |
1756 | list_item = list_item->next; | list_item = list_item->next; |
1757 | } | } |
1758 | common->stubs = NULL; | common->stubs = NULL; |
# | Line 1360 loop = LABEL(); | Line 1837 loop = LABEL(); |
1837 | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_TEMPORARY_REG1, 0); | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_TEMPORARY_REG1, 0); |
1838 | OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_SAVED_REG1, 0, SLJIT_IMM, sizeof(sljit_w)); | OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_SAVED_REG1, 0, SLJIT_IMM, sizeof(sljit_w)); |
1839 | /* Copy the integer value to the output buffer */ | /* Copy the integer value to the output buffer */ |
1840 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
1841 | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT); |
1842 | #endif | #endif |
1843 | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); |
1844 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); |
# | Line 1385 else | Line 1862 else |
1862 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); |
1863 | } | } |
1864 | ||
1865 | static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *leave) | static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit) |
1866 | { | { |
1867 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1868 | ||
# | Line 1395 SLJIT_ASSERT(common->start_used_ptr != 0 | Line 1872 SLJIT_ASSERT(common->start_used_ptr != 0 |
1872 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); |
1873 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); |
1874 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsetcount)); |
1875 | CMPTO(SLJIT_C_LESS, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 2, leave); | CMPTO(SLJIT_C_LESS, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 2, quit); |
1876 | ||
1877 | /* Store match begin and end. */ | /* Store match begin and end. */ |
1878 | OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); |
1879 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); |
1880 | 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); | 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); |
1881 | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0); | OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0); |
1882 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
1883 | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT); |
1884 | #endif | #endif |
1885 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); |
1886 | ||
1887 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG1, 0); | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG1, 0); |
1888 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
1889 | OP2(SLJIT_ASHR, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); | OP2(SLJIT_ASHR, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); |
1890 | #endif | #endif |
1891 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), 0, SLJIT_TEMPORARY_REG3, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), 0, SLJIT_TEMPORARY_REG3, 0); |
1892 | ||
1893 | JUMPTO(SLJIT_JUMP, leave); | JUMPTO(SLJIT_JUMP, quit); |
1894 | } | } |
1895 | ||
1896 | static SLJIT_INLINE void check_start_used_ptr(compiler_common *common) | static SLJIT_INLINE void check_start_used_ptr(compiler_common *common) |
# | Line 1524 if (c <= 127 && bit == 0x20) | Line 2001 if (c <= 127 && bit == 0x20) |
2001 | return (0 << 8) | 0x20; | return (0 << 8) | 0x20; |
2002 | ||
2003 | /* Since c != oc, they must have at least 1 bit difference. */ | /* Since c != oc, they must have at least 1 bit difference. */ |
2004 | if (!ispowerof2(bit)) | if (!is_powerof2(bit)) |
2005 | return 0; | return 0; |
2006 | ||
2007 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2008 | ||
2009 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2010 | if (common->utf && c > 127) | if (common->utf && c > 127) |
# | Line 1543 if (common->utf && c > 127) | Line 2020 if (common->utf && c > 127) |
2020 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2021 | return (0 << 8) | bit; | return (0 << 8) | bit; |
2022 | ||
2023 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2024 | ||
#ifdef COMPILE_PCRE16 | ||
2025 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2026 | if (common->utf && c > 65535) | if (common->utf && c > 65535) |
2027 | { | { |
# | Line 1556 if (common->utf && c > 65535) | Line 2032 if (common->utf && c > 65535) |
2032 | } | } |
2033 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2034 | return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); | return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); |
#endif /* COMPILE_PCRE16 */ | ||
2035 | ||
2036 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE[8|16|32] */ |
2037 | } | } |
2038 | ||
2039 | static void check_partial(compiler_common *common, BOOL force) | static void check_partial(compiler_common *common, BOOL force) |
# | Line 1622 JUMPHERE(jump); | Line 2097 JUMPHERE(jump); |
2097 | return return_value; | return return_value; |
2098 | } | } |
2099 | ||
2100 | static void fallback_at_str_end(compiler_common *common, jump_list **fallbacks) | static void detect_partial_match(compiler_common *common, jump_list **backtracks) |
2101 | { | { |
2102 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2103 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2104 | ||
2105 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
2106 | { | { |
2107 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); |
2108 | return; | return; |
2109 | } | } |
2110 | ||
2111 | /* Partial matching mode. */ | /* Partial matching mode. */ |
2112 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); |
2113 | add_jump(compiler, fallbacks, 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_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); |
2114 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2115 | { | { |
2116 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); |
2117 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
2118 | } | } |
2119 | else | else |
2120 | { | { |
# | Line 1656 static void read_char(compiler_common *c | Line 2131 static void read_char(compiler_common *c |
2131 | /* Reads the character into TMP1, updates STR_PTR. | /* Reads the character into TMP1, updates STR_PTR. |
2132 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2133 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2134 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2135 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2136 | #endif | #endif |
2137 | ||
2138 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2139 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2140 | if (common->utf) | if (common->utf) |
2141 | { | { |
2142 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2143 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2144 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2145 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
2146 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
#endif /* COMPILE_PCRE8 */ | ||
2147 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2148 | JUMPHERE(jump); | JUMPHERE(jump); |
2149 | } | } |
2150 | #endif | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2151 | 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)); |
2152 | } | } |
2153 | ||
# | Line 1683 static void peek_char(compiler_common *c | Line 2156 static void peek_char(compiler_common *c |
2156 | /* Reads the character into TMP1, keeps STR_PTR. | /* Reads the character into TMP1, keeps STR_PTR. |
2157 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2158 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2159 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2160 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2161 | #endif | #endif |
2162 | ||
2163 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2164 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2165 | if (common->utf) | if (common->utf) |
2166 | { | { |
2167 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2168 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2169 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2170 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
2171 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
#endif /* COMPILE_PCRE8 */ | ||
2172 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2173 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2174 | JUMPHERE(jump); | JUMPHERE(jump); |
2175 | } | } |
2176 | #endif | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2177 | } | } |
2178 | ||
2179 | static void read_char8_type(compiler_common *common) | static void read_char8_type(compiler_common *common) |
2180 | { | { |
2181 | /* 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. */ |
2182 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2183 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2184 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2185 | #endif | #endif |
2186 | ||
# | Line 1718 if (common->utf) | Line 2189 if (common->utf) |
2189 | { | { |
2190 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2191 | 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)); |
2192 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2193 | /* This can be an extra read in some situations, but hopefully | /* This can be an extra read in some situations, but hopefully |
2194 | it is needed in most cases. */ | it is needed in most cases. */ |
2195 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2196 | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); |
2197 | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); |
2198 | JUMPHERE(jump); | JUMPHERE(jump); |
2199 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2200 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2201 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2202 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
# | Line 1737 if (common->utf) | Line 2207 if (common->utf) |
2207 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); |
2208 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); |
2209 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2210 | #endif | #elif defined COMPILE_PCRE32 |
2211 | #endif /* COMPILE_PCRE8 */ | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2212 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | |
2213 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
2214 | JUMPHERE(jump); | |
2215 | #endif /* COMPILE_PCRE[8|16|32] */ | |
2216 | return; | return; |
2217 | } | } |
2218 | #endif | #endif /* SUPPORT_UTF */ |
2219 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2220 | 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)); |
2221 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2222 | /* The ctypes array contains only 256 values. */ | /* The ctypes array contains only 256 values. */ |
2223 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2224 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2225 | #endif | #endif |
2226 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2227 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2228 | JUMPHERE(jump); | JUMPHERE(jump); |
2229 | #endif | #endif |
2230 | } | } |
# | Line 1759 static void skip_char_back(compiler_comm | Line 2233 static void skip_char_back(compiler_comm |
2233 | { | { |
2234 | /* Goes one character back. Affects STR_PTR and TMP1. Does not check begin. */ | /* Goes one character back. Affects STR_PTR and TMP1. Does not check begin. */ |
2235 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2236 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2237 | #if defined COMPILE_PCRE8 | |
2238 | struct sljit_label *label; | struct sljit_label *label; |
2239 | ||
2240 | if (common->utf) | if (common->utf) |
# | Line 1771 if (common->utf) | Line 2246 if (common->utf) |
2246 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); |
2247 | return; | return; |
2248 | } | } |
2249 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
2250 | if (common->utf) | if (common->utf) |
2251 | { | { |
2252 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); |
# | Line 1785 if (common->utf) | Line 2259 if (common->utf) |
2259 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2260 | return; | return; |
2261 | } | } |
2262 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
2263 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | |
2264 | 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)); |
2265 | } | } |
2266 | ||
2267 | static void check_newlinechar(compiler_common *common, int nltype, jump_list **fallbacks, BOOL jumpiftrue) | static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpiftrue) |
2268 | { | { |
2269 | /* 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. */ |
2270 | DEFINE_COMPILER; | DEFINE_COMPILER; |
# | Line 1797 DEFINE_COMPILER; | Line 2272 DEFINE_COMPILER; |
2272 | if (nltype == NLTYPE_ANY) | if (nltype == NLTYPE_ANY) |
2273 | { | { |
2274 | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); |
2275 | add_jump(compiler, fallbacks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
2276 | } | } |
2277 | else if (nltype == NLTYPE_ANYCRLF) | else if (nltype == NLTYPE_ANYCRLF) |
2278 | { | { |
# | Line 1805 else if (nltype == NLTYPE_ANYCRLF) | Line 2280 else if (nltype == NLTYPE_ANYCRLF) |
2280 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); |
2281 | 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); |
2282 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
2283 | add_jump(compiler, fallbacks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
2284 | } | } |
2285 | else | else |
2286 | { | { |
2287 | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); |
2288 | add_jump(compiler, fallbacks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); | add_jump(compiler, backtracks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); |
2289 | } | } |
2290 | } | } |
2291 | ||
2292 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2293 | ||
2294 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2295 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2296 | { | { |
2297 | /* Fast decoding a UTF-8 character. TMP1 contains the first byte | /* Fast decoding a UTF-8 character. TMP1 contains the first byte |
# | Line 1910 OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | Line 2385 OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2385 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2386 | } | } |
2387 | ||
2388 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 |
2389 | ||
#ifdef COMPILE_PCRE16 | ||
2390 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2391 | { | { |
2392 | /* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char | /* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char |
# | Line 1937 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UC | Line 2411 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UC |
2411 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); |
2412 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2413 | } | } |
#endif /* COMPILE_PCRE16 */ | ||
2414 | ||
2415 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE[8|16] */ |
2416 | ||
2417 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2418 | ||
# | Line 1994 if (firstline) | Line 2467 if (firstline) |
2467 | { | { |
2468 | /* Search for the end of the first line. */ | /* Search for the end of the first line. */ |
2469 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2470 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STR_PTR, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); |
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_END, 0); | ||
2471 | ||
2472 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
2473 | { | { |
# | Line 2006 if (firstline) | Line 2478 if (firstline) |
2478 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
2479 | 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); |
2480 | 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); |
2481 | JUMPHERE(end); | |
2482 | 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_LOCALS_REG), common->first_line_end, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
2483 | } | } |
2484 | else | else |
# | Line 2017 if (firstline) | Line 2490 if (firstline) |
2490 | read_char(common); | read_char(common); |
2491 | check_newlinechar(common, common->nltype, &newline, TRUE); | check_newlinechar(common, common->nltype, &newline, TRUE); |
2492 | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); |
2493 | JUMPHERE(end); | |
2494 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); |
2495 | set_jumps(newline, LABEL()); | set_jumps(newline, LABEL()); |
2496 | } | } |
2497 | ||
2498 | JUMPHERE(end); | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); |
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | ||
2499 | } | } |
2500 | ||
2501 | start = JUMP(SLJIT_JUMP); | start = JUMP(SLJIT_JUMP); |
# | Line 2035 if (newlinecheck) | Line 2508 if (newlinecheck) |
2508 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2509 | 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); |
2510 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); |
2511 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2512 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
2513 | #endif | #endif |
2514 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2515 | nl = JUMP(SLJIT_JUMP); | nl = JUMP(SLJIT_JUMP); |
# | Line 2057 if (newlinecheck) | Line 2530 if (newlinecheck) |
2530 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); |
2531 | ||
2532 | 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)); |
2533 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2534 | #if defined COMPILE_PCRE8 | |
2535 | if (common->utf) | if (common->utf) |
2536 | { | { |
2537 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
# | Line 2065 if (common->utf) | Line 2539 if (common->utf) |
2539 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2540 | JUMPHERE(singlechar); | JUMPHERE(singlechar); |
2541 | } | } |
2542 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
2543 | if (common->utf) | if (common->utf) |
2544 | { | { |
2545 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
# | Line 2077 if (common->utf) | Line 2550 if (common->utf) |
2550 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2551 | JUMPHERE(singlechar); | JUMPHERE(singlechar); |
2552 | } | } |
2553 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
2554 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | |
2555 | JUMPHERE(start); | JUMPHERE(start); |
2556 | ||
2557 | if (newlinecheck) | if (newlinecheck) |
# | Line 2089 if (newlinecheck) | Line 2563 if (newlinecheck) |
2563 | return mainloop; | return mainloop; |
2564 | } | } |
2565 | ||
2566 | #define MAX_N_CHARS 3 | |
2567 | ||
2568 | static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline) | |
2569 | { | |
2570 | DEFINE_COMPILER; | |
2571 | struct sljit_label *start; | |
2572 | struct sljit_jump *quit; | |
2573 | pcre_uint32 chars[MAX_N_CHARS * 2]; | |
2574 | pcre_uchar *cc = common->start + 1 + IMM2_SIZE; | |
2575 | int location = 0; | |
2576 | pcre_int32 len, c, bit, caseless; | |
2577 | int must_stop; | |
2578 | ||
2579 | /* We do not support alternatives now. */ | |
2580 | if (*(common->start + GET(common->start, 1)) == OP_ALT) | |
2581 | return FALSE; | |
2582 | ||
2583 | while (TRUE) | |
2584 | { | |
2585 | caseless = 0; | |
2586 | must_stop = 1; | |
2587 | switch(*cc) | |
2588 | { | |
2589 | case OP_CHAR: | |
2590 | must_stop = 0; | |
2591 | cc++; | |
2592 | break; | |
2593 | ||
2594 | case OP_CHARI: | |
2595 | caseless = 1; | |
2596 | must_stop = 0; | |
2597 | cc++; | |
2598 | break; | |
2599 | ||
2600 | case OP_SOD: | |
2601 | case OP_SOM: | |
2602 | case OP_SET_SOM: | |
2603 | case OP_NOT_WORD_BOUNDARY: | |
2604 | case OP_WORD_BOUNDARY: | |
2605 | case OP_EODN: | |
2606 | case OP_EOD: | |
2607 | case OP_CIRC: | |
2608 | case OP_CIRCM: | |
2609 | case OP_DOLL: | |
2610 | case OP_DOLLM: | |
2611 | /* Zero width assertions. */ | |
2612 | cc++; | |
2613 | continue; | |
2614 | ||
2615 | case OP_PLUS: | |
2616 | case OP_MINPLUS: | |
2617 | case OP_POSPLUS: | |
2618 | cc++; | |
2619 | break; | |
2620 | ||
2621 | case OP_EXACT: | |
2622 | cc += 1 + IMM2_SIZE; | |
2623 | break; | |
2624 | ||
2625 | case OP_PLUSI: | |
2626 | case OP_MINPLUSI: | |
2627 | case OP_POSPLUSI: | |
2628 | caseless = 1; | |
2629 | cc++; | |
2630 | break; | |
2631 | ||
2632 | case OP_EXACTI: | |
2633 | caseless = 1; | |
2634 | cc += 1 + IMM2_SIZE; | |
2635 | break; | |
2636 | ||
2637 | default: | |
2638 | must_stop = 2; | |
2639 | break; | |
2640 | } | |
2641 | ||
2642 | if (must_stop == 2) | |
2643 | break; | |
2644 | ||
2645 | len = 1; | |
2646 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 | |
2647 | if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]); | |
2648 | #endif | |
2649 | ||
2650 | if (caseless && char_has_othercase(common, cc)) | |
2651 | { | |
2652 | caseless = char_get_othercase_bit(common, cc); | |
2653 | if (caseless == 0) | |
2654 | return FALSE; | |
2655 | #ifdef COMPILE_PCRE8 | |
2656 | caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8)); | |
2657 | #else | |
2658 | if ((caseless & 0x100) != 0) | |
2659 | caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9)); | |
2660 | else | |
2661 | caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9)); | |
2662 | #endif | |
2663 | } | |
2664 | else | |
2665 | caseless = 0; | |
2666 | ||
2667 | while (len > 0 && location < MAX_N_CHARS * 2) | |
2668 | { | |
2669 | c = *cc; | |
2670 | bit = 0; | |
2671 | if (len == (caseless & 0xff)) | |
2672 | { | |
2673 | bit = caseless >> 8; | |
2674 | c |= bit; | |
2675 | } | |
2676 | ||
2677 | chars[location] = c; | |
2678 | chars[location + 1] = bit; | |
2679 | ||
2680 | len--; | |
2681 | location += 2; | |
2682 | cc++; | |
2683 | } | |
2684 | ||
2685 | if (location >= MAX_N_CHARS * 2 || must_stop != 0) | |
2686 | break; | |
2687 | } | |
2688 | ||
2689 | /* At least two characters are required. */ | |
2690 | if (location < 2 * 2) | |
2691 | return FALSE; | |
2692 | ||
2693 | if (firstline) | |
2694 | { | |
2695 | SLJIT_ASSERT(common->first_line_end != 0); | |
2696 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | |
2697 | OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, (location >> 1) - 1); | |
2698 | } | |
2699 | else | |
2700 | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, (location >> 1) - 1); | |
2701 | ||
2702 | start = LABEL(); | |
2703 | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
2704 | ||
2705 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2706 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2707 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2708 | if (chars[1] != 0) | |
2709 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]); | |
2710 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start); | |
2711 | if (location > 2 * 2) | |
2712 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2713 | if (chars[3] != 0) | |
2714 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]); | |
2715 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start); | |
2716 | if (location > 2 * 2) | |
2717 | { | |
2718 | if (chars[5] != 0) | |
2719 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]); | |
2720 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start); | |
2721 | } | |
2722 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2723 | ||
2724 | JUMPHERE(quit); | |
2725 | ||
2726 | if (firstline) | |
2727 | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); | |
2728 | else | |
2729 | OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, (location >> 1) - 1); | |
2730 | return TRUE; | |
2731 | } | |
2732 | ||
2733 | #undef MAX_N_CHARS | |
2734 | ||
2735 | 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) |
2736 | { | { |
2737 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2738 | struct sljit_label *start; | struct sljit_label *start; |
2739 | struct sljit_jump *leave; | struct sljit_jump *quit; |
2740 | struct sljit_jump *found; | struct sljit_jump *found; |
2741 | pcre_uchar oc, bit; | pcre_uchar oc, bit; |
2742 | ||
2743 | if (firstline) | if (firstline) |
2744 | { | { |
2745 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2746 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | |
2747 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); |
2748 | } | } |
2749 | ||
2750 | start = LABEL(); | start = LABEL(); |
2751 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2752 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2753 | ||
2754 | oc = first_char; | oc = first_char; |
# | Line 2121 if (first_char == oc) | Line 2765 if (first_char == oc) |
2765 | else | else |
2766 | { | { |
2767 | bit = first_char ^ oc; | bit = first_char ^ oc; |
2768 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
2769 | { | { |
2770 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); |
2771 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, first_char | bit); | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, first_char | bit); |
# | Line 2137 else | Line 2781 else |
2781 | } | } |
2782 | ||
2783 | 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)); |
#if defined SUPPORT_UTF && defined COMPILE_PCRE8 | ||
if (common->utf) | ||
{ | ||
CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); | ||
OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | ||
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | ||
} | ||
#endif | ||
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
if (common->utf) | ||
{ | ||
CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | ||
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); | ||
} | ||
#endif | ||
2784 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
2785 | JUMPHERE(found); | JUMPHERE(found); |
2786 | JUMPHERE(leave); | JUMPHERE(quit); |
2787 | ||
2788 | if (firstline) | if (firstline) |
2789 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
2790 | } | } |
2791 | ||
2792 | static SLJIT_INLINE void fast_forward_newline(compiler_common *common, BOOL firstline) | static SLJIT_INLINE void fast_forward_newline(compiler_common *common, BOOL firstline) |
# | Line 2170 DEFINE_COMPILER; | Line 2795 DEFINE_COMPILER; |
2795 | struct sljit_label *loop; | struct sljit_label *loop; |
2796 | struct sljit_jump *lastchar; | struct sljit_jump *lastchar; |
2797 | struct sljit_jump *firstchar; | struct sljit_jump *firstchar; |
2798 | struct sljit_jump *leave; | struct sljit_jump *quit; |
2799 | struct sljit_jump *foundcr = NULL; | struct sljit_jump *foundcr = NULL; |
2800 | struct sljit_jump *notfoundnl; | struct sljit_jump *notfoundnl; |
2801 | jump_list *newline = NULL; | jump_list *newline = NULL; |
2802 | ||
2803 | if (firstline) | if (firstline) |
2804 | { | { |
2805 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2806 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | |
2807 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); |
2808 | } | } |
2809 | ||
# | Line 2192 if (common->nltype == NLTYPE_FIXED && co | Line 2818 if (common->nltype == NLTYPE_FIXED && co |
2818 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); |
2819 | 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); |
2820 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL); |
2821 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2822 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); |
2823 | #endif | #endif |
2824 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2825 | ||
2826 | loop = LABEL(); | loop = LABEL(); |
2827 | 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)); |
2828 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2829 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); |
2830 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
2831 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); |
2832 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); |
2833 | ||
2834 | JUMPHERE(leave); | JUMPHERE(quit); |
2835 | JUMPHERE(firstchar); | JUMPHERE(firstchar); |
2836 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
2837 | ||
# | Line 2229 set_jumps(newline, loop); | Line 2855 set_jumps(newline, loop); |
2855 | ||
2856 | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) |
2857 | { | { |
2858 | leave = JUMP(SLJIT_JUMP); | quit = JUMP(SLJIT_JUMP); |
2859 | JUMPHERE(foundcr); | JUMPHERE(foundcr); |
2860 | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2861 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2862 | 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); |
2863 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); |
2864 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2865 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
2866 | #endif | #endif |
2867 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2868 | JUMPHERE(notfoundnl); | JUMPHERE(notfoundnl); |
2869 | JUMPHERE(leave); | JUMPHERE(quit); |
2870 | } | } |
2871 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
2872 | JUMPHERE(firstchar); | JUMPHERE(firstchar); |
2873 | ||
2874 | if (firstline) | if (firstline) |
2875 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
2876 | } | } |
2877 | ||
2878 | static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline) | static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline) |
2879 | { | { |
2880 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2881 | struct sljit_label *start; | struct sljit_label *start; |
2882 | struct sljit_jump *leave; | struct sljit_jump *quit; |
2883 | struct sljit_jump *found; | struct sljit_jump *found; |
2884 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
2885 | struct sljit_jump *jump; | struct sljit_jump *jump; |
# | Line 2261 struct sljit_jump *jump; | Line 2887 struct sljit_jump *jump; |
2887 | ||
2888 | if (firstline) | if (firstline) |
2889 | { | { |
2890 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2891 | OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); | |
2892 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); |
2893 | } | } |
2894 | ||
2895 | start = LABEL(); | start = LABEL(); |
2896 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2897 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2898 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2899 | if (common->utf) | if (common->utf) |
# | Line 2289 if (common->utf) | Line 2916 if (common->utf) |
2916 | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); |
2917 | #endif | #endif |
2918 | 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)); |
2919 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #ifdef SUPPORT_UTF |
2920 | #if defined COMPILE_PCRE8 | |
2921 | if (common->utf) | if (common->utf) |
2922 | { | { |
2923 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); |
2924 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); |
2925 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2926 | } | } |
2927 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
2928 | if (common->utf) | if (common->utf) |
2929 | { | { |
2930 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); |
# | Line 2307 if (common->utf) | Line 2934 if (common->utf) |
2934 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
2935 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2936 | } | } |
2937 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
2938 | #endif /* SUPPORT_UTF */ | |
2939 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
2940 | JUMPHERE(found); | JUMPHERE(found); |
2941 | JUMPHERE(leave); | JUMPHERE(quit); |
2942 | ||
2943 | if (firstline) | if (firstline) |
2944 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, RETURN_ADDR, 0); |
2945 | } | } |
2946 | ||
2947 | static SLJIT_INLINE struct sljit_jump *search_requested_char(compiler_common *common, pcre_uchar req_char, BOOL caseless, BOOL has_firstchar) | static SLJIT_INLINE struct sljit_jump *search_requested_char(compiler_common *common, pcre_uchar req_char, BOOL caseless, BOOL has_firstchar) |
# | Line 2325 struct sljit_jump *alreadyfound; | Line 2953 struct sljit_jump *alreadyfound; |
2953 | struct sljit_jump *found; | struct sljit_jump *found; |
2954 | struct sljit_jump *foundoc = NULL; | struct sljit_jump *foundoc = NULL; |
2955 | struct sljit_jump *notfound; | struct sljit_jump *notfound; |
2956 | pcre_uchar oc, bit; | pcre_uint32 oc, bit; |
2957 | ||
2958 | SLJIT_ASSERT(common->req_char_ptr != 0); | SLJIT_ASSERT(common->req_char_ptr != 0); |
2959 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr); |
# | Line 2356 if (req_char == oc) | Line 2984 if (req_char == oc) |
2984 | else | else |
2985 | { | { |
2986 | bit = req_char ^ oc; | bit = req_char ^ oc; |
2987 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
2988 | { | { |
2989 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); |
2990 | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit); | found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit); |
# | Line 2531 else | Line 3159 else |
3159 | JUMPHERE(jump); | JUMPHERE(jump); |
3160 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE8 */ |
3161 | } | } |
3162 | JUMPHERE(skipread); | JUMPHERE(skipread); |
3163 | ||
3164 | OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); | |
3165 | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | |
3166 | } | |
3167 | ||
3168 | /* | |
3169 | range format: | |
3170 | ||
3171 | ranges[0] = length of the range (max MAX_RANGE_SIZE, -1 means invalid range). | |
3172 | ranges[1] = first bit (0 or 1) | |
3173 | ranges[2-length] = position of the bit change (when the current bit is not equal to the previous) | |
3174 | */ | |
3175 | ||
3176 | static BOOL check_ranges(compiler_common *common, int *ranges, jump_list **backtracks, BOOL readch) | |
3177 | { | |
3178 | DEFINE_COMPILER; | |
3179 | struct sljit_jump *jump; | |
3180 | ||
3181 | if (ranges[0] < 0) | |
3182 | return FALSE; | |
3183 | ||
3184 | switch(ranges[0]) | |
3185 | { | |
3186 | case 1: | |
3187 | if (readch) | |
3188 | read_char(common); | |
3189 | add_jump(compiler, backtracks, CMP(ranges[1] == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | |
3190 | return TRUE; | |
3191 | ||
3192 | case 2: | |
3193 | if (readch) | |
3194 | read_char(common); | |
3195 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]); | |
3196 | add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2])); | |
3197 | return TRUE; | |
3198 | ||
3199 | case 4: | |
3200 | if (ranges[2] + 1 == ranges[3] && ranges[4] + 1 == ranges[5]) | |
3201 | { | |
3202 | if (readch) | |
3203 | read_char(common); | |
3204 | if (ranges[1] != 0) | |
3205 | { | |
3206 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | |
3207 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | |
3208 | } | |
3209 | else | |
3210 | { | |
3211 | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]); | |
3212 | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | |
3213 | JUMPHERE(jump); | |
3214 | } | |
3215 | return TRUE; | |
3216 | } | |
3217 | if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && is_powerof2(ranges[4] - ranges[2])) | |
3218 | { | |
3219 | if (readch) | |
3220 | read_char(common); | |
3221 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4] - ranges[2]); | |
3222 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4]); | |
3223 | add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[5] - ranges[4])); | |
3224 | return TRUE; | |
3225 | } | |
3226 | return FALSE; | |
3227 | ||
3228 | default: | |
3229 | return FALSE; | |
3230 | } | |
3231 | } | |
3232 | ||
3233 | static void get_ctype_ranges(compiler_common *common, int flag, int *ranges) | |
3234 | { | |
3235 | int i, bit, length; | |
3236 | const pcre_uint8 *ctypes = (const pcre_uint8*)common->ctypes; | |
3237 | ||
3238 | bit = ctypes[0] & flag; | |
3239 | ranges[0] = -1; | |
3240 | ranges[1] = bit != 0 ? 1 : 0; | |
3241 | length = 0; | |
3242 | ||
3243 | for (i = 1; i < 256; i++) | |
3244 | if ((ctypes[i] & flag) != bit) | |
3245 | { | |
3246 | if (length >= MAX_RANGE_SIZE) | |
3247 | return; | |
3248 | ranges[2 + length] = i; | |
3249 | length++; | |
3250 | bit ^= flag; | |
3251 | } | |
3252 | ||
3253 | if (bit != 0) | |
3254 | { | |
3255 | if (length >= MAX_RANGE_SIZE) | |
3256 | return; | |
3257 | ranges[2 + length] = 256; | |
3258 | length++; | |
3259 | } | |
3260 | ranges[0] = length; | |
3261 | } | |
3262 | ||
3263 | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks) | |
3264 | { | |
3265 | int ranges[2 + MAX_RANGE_SIZE]; | |
3266 | pcre_uint8 bit, cbit, all; | |
3267 | int i, byte, length = 0; | |
3268 | ||
3269 | bit = bits[0] & 0x1; | |
3270 | ranges[1] = bit; | |
3271 | /* Can be 0 or 255. */ | |
3272 | all = -bit; | |
3273 | ||
3274 | for (i = 0; i < 256; ) | |
3275 | { | |
3276 | byte = i >> 3; | |
3277 | if ((i & 0x7) == 0 && bits[byte] == all) | |
3278 | i += 8; | |
3279 | else | |
3280 | { | |
3281 | cbit = (bits[byte] >> (i & 0x7)) & 0x1; | |
3282 | if (cbit != bit) | |
3283 | { | |
3284 | if (length >= MAX_RANGE_SIZE) | |
3285 | return FALSE; | |
3286 | ranges[2 + length] = i; | |
3287 | length++; | |
3288 | bit = cbit; | |
3289 | all = -cbit; | |
3290 | } | |
3291 | i++; | |
3292 | } | |
3293 | } | |
3294 | ||
3295 | OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); | if (((bit == 0) && nclass) || ((bit == 1) && !nclass)) |
3296 | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | { |
3297 | if (length >= MAX_RANGE_SIZE) | |
3298 | return FALSE; | |
3299 | ranges[2 + length] = 256; | |
3300 | length++; | |
3301 | } | |
3302 | ranges[0] = length; | |
3303 | ||
3304 | return check_ranges(common, ranges, backtracks, FALSE); | |
3305 | } | } |
3306 | ||
3307 | static void check_anynewline(compiler_common *common) | static void check_anynewline(compiler_common *common) |
# | Line 2548 OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_I | Line 3315 OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_I |
3315 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); |
3316 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3317 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); |
3318 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3319 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3320 | if (common->utf) | if (common->utf) |
3321 | { | { |
# | Line 2559 if (common->utf) | Line 3326 if (common->utf) |
3326 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3327 | } | } |
3328 | #endif | #endif |
3329 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3330 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
3331 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3332 | } | } |
# | Line 2576 COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_E | Line 3343 COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_E |
3343 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); |
3344 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); |
3345 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xa0); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xa0); |
3346 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3347 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3348 | if (common->utf) | if (common->utf) |
3349 | { | { |
# | Line 2597 if (common->utf) | Line 3364 if (common->utf) |
3364 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3365 | } | } |
3366 | #endif | #endif |
3367 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3368 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
3369 | ||
3370 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
# | Line 2614 OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_I | Line 3381 OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_I |
3381 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); |
3382 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3383 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); |
3384 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3385 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3386 | if (common->utf) | if (common->utf) |
3387 | { | { |
# | Line 2625 if (common->utf) | Line 3392 if (common->utf) |
3392 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3393 | } | } |
3394 | #endif | #endif |
3395 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3396 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
3397 | ||
3398 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
# | Line 2715 sljit_emit_fast_return(compiler, RETURN_ | Line 3482 sljit_emit_fast_return(compiler, RETURN_ |
3482 | static const pcre_uchar *SLJIT_CALL do_utf_caselesscmp(pcre_uchar *src1, jit_arguments *args, pcre_uchar *end1) | static const pcre_uchar *SLJIT_CALL do_utf_caselesscmp(pcre_uchar *src1, jit_arguments *args, pcre_uchar *end1) |
3483 | { | { |
3484 | /* This function would be ineffective to do in JIT level. */ | /* This function would be ineffective to do in JIT level. */ |
3485 | int c1, c2; | pcre_uint32 c1, c2; |
3486 | const pcre_uchar *src2 = args->uchar_ptr; | const pcre_uchar *src2 = args->uchar_ptr; |
3487 | const pcre_uchar *end2 = args->end; | const pcre_uchar *end2 = args->end; |
3488 | const ucd_record *ur; | |
3489 | const pcre_uint32 *pp; | |
3490 | ||
3491 | while (src1 < end1) | while (src1 < end1) |
3492 | { | { |
# | Line 2725 while (src1 < end1) | Line 3494 while (src1 < end1) |
3494 | return (pcre_uchar*)1; | return (pcre_uchar*)1; |
3495 | GETCHARINC(c1, src1); | GETCHARINC(c1, src1); |
3496 | GETCHARINC(c2, src2); | GETCHARINC(c2, src2); |
3497 | if (c1 != c2 && c1 != UCD_OTHERCASE(c2)) return NULL; | ur = GET_UCD(c2); |
3498 | if (c1 != c2 && c1 != c2 + ur->other_case) | |
3499 | { | |
3500 | pp = PRIV(ucd_caseless_sets) + ur->caseset; | |
3501 | for (;;) | |
3502 | { | |
3503 | if (c1 < *pp) return NULL; | |
3504 | if (c1 == *pp++) break; | |
3505 | } | |
3506 | } | |
3507 | } | } |
3508 | return src2; | return src2; |
3509 | } | } |
# | Line 2733 return src2; | Line 3511 return src2; |
3511 | #endif /* SUPPORT_UTF && SUPPORT_UCP */ | #endif /* SUPPORT_UTF && SUPPORT_UCP */ |
3512 | ||
3513 | static pcre_uchar *byte_sequence_compare(compiler_common *common, BOOL caseless, pcre_uchar *cc, | static pcre_uchar *byte_sequence_compare(compiler_common *common, BOOL caseless, pcre_uchar *cc, |
3514 | compare_context* context, jump_list **fallbacks) | compare_context* context, jump_list **backtracks) |
3515 | { | { |
3516 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3517 | unsigned int othercasebit = 0; | unsigned int othercasebit = 0; |
# | Line 2747 if (caseless && char_has_othercase(commo | Line 3525 if (caseless && char_has_othercase(commo |
3525 | othercasebit = char_get_othercase_bit(common, cc); | othercasebit = char_get_othercase_bit(common, cc); |
3526 | SLJIT_ASSERT(othercasebit); | SLJIT_ASSERT(othercasebit); |
3527 | /* Extracting bit difference info. */ | /* Extracting bit difference info. */ |
3528 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3529 | othercasechar = cc + (othercasebit >> 8); | othercasechar = cc + (othercasebit >> 8); |
3530 | othercasebit &= 0xff; | othercasebit &= 0xff; |
3531 | #else | #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
#ifdef COMPILE_PCRE16 | ||
3532 | othercasechar = cc + (othercasebit >> 9); | othercasechar = cc + (othercasebit >> 9); |
3533 | if ((othercasebit & 0x100) != 0) | if ((othercasebit & 0x100) != 0) |
3534 | othercasebit = (othercasebit & 0xff) << 8; | othercasebit = (othercasebit & 0xff) << 8; |
3535 | else | else |
3536 | othercasebit &= 0xff; | othercasebit &= 0xff; |
3537 | #endif | #endif /* COMPILE_PCRE[8|16|32] */ |
#endif | ||
3538 | } | } |
3539 | ||
3540 | if (context->sourcereg == -1) | if (context->sourcereg == -1) |
3541 | { | { |
3542 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3543 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
3544 | if (context->length >= 4) | if (context->length >= 4) |
3545 | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
# | Line 2772 if (context->sourcereg == -1) | Line 3548 if (context->sourcereg == -1) |
3548 | else | else |
3549 | #endif | #endif |
3550 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3551 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
3552 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
3553 | if (context->length >= 4) | if (context->length >= 4) |
3554 | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3555 | else | else |
3556 | #endif | #endif |
3557 | OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3558 | #endif | #elif defined COMPILE_PCRE32 |
3559 | #endif /* COMPILE_PCRE8 */ | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3560 | #endif /* COMPILE_PCRE[8|16|32] */ | |
3561 | context->sourcereg = TMP2; | context->sourcereg = TMP2; |
3562 | } | } |
3563 | ||
3564 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
3565 | utflength = 1; | utflength = 1; |
3566 | #ifndef COMPILE_PCRE32 | |
3567 | if (common->utf && HAS_EXTRALEN(*cc)) | if (common->utf && HAS_EXTRALEN(*cc)) |
3568 | utflength += GET_EXTRALEN(*cc); | utflength += GET_EXTRALEN(*cc); |
3569 | #endif | |
3570 | ||
3571 | do | do |
3572 | { | { |
# | Line 2810 do | Line 3588 do |
3588 | } | } |
3589 | context->ucharptr++; | context->ucharptr++; |
3590 | ||
3591 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3592 | if (context->ucharptr >= 4 || context->length == 0 || (context->ucharptr == 2 && context->length == 1)) | if (context->ucharptr >= 4 || context->length == 0 || (context->ucharptr == 2 && context->length == 1)) |
3593 | #else | #elif defined COMPILE_PCRE16 |
3594 | if (context->ucharptr >= 2 || context->length == 0) | if (context->ucharptr >= 2 || context->length == 0) |
3595 | #elif defined COMPILE_PCRE32 | |
3596 | if (1 /* context->ucharptr >= 1 || context->length == 0 */) | |
3597 | #endif | #endif |
3598 | { | { |
3599 | #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 | |
3600 | if (context->length >= 4) | if (context->length >= 4) |
3601 | OP1(SLJIT_MOV_SI, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_SI, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3602 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3603 | else if (context->length >= 2) | else if (context->length >= 2) |
3604 | OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3605 | else if (context->length >= 1) | else if (context->length >= 1) |
3606 | OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3607 | #else | #elif defined COMPILE_PCRE16 |
3608 | else if (context->length >= 2) | else if (context->length >= 2) |
3609 | OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3610 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
3611 | #elif defined COMPILE_PCRE32 | |
3612 | OP1(MOV_UCHAR, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | |
3613 | #endif /* COMPILE_PCRE[8|16|32] */ | |
3614 | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; |
3615 | ||
3616 | switch(context->ucharptr) | switch(context->ucharptr) |
# | Line 2834 do | Line 3618 do |
3618 | case 4 / sizeof(pcre_uchar): | case 4 / sizeof(pcre_uchar): |
3619 | if (context->oc.asint != 0) | if (context->oc.asint != 0) |
3620 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asint); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asint); |
3621 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asint | context->oc.asint)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asint | context->oc.asint)); |
3622 | break; | break; |
3623 | ||
3624 | #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 | |
3625 | case 2 / sizeof(pcre_uchar): | case 2 / sizeof(pcre_uchar): |
3626 | if (context->oc.asushort != 0) | if (context->oc.asushort != 0) |
3627 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asushort); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asushort); |
3628 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asushort | context->oc.asushort)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asushort | context->oc.asushort)); |
3629 | break; | break; |
3630 | ||
3631 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3632 | case 1: | case 1: |
3633 | if (context->oc.asbyte != 0) | if (context->oc.asbyte != 0) |
3634 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asbyte); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asbyte); |
3635 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asbyte | context->oc.asbyte)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asbyte | context->oc.asbyte)); |
3636 | break; | break; |
3637 | #endif | #endif |
3638 | ||
3639 | #endif /* COMPILE_PCRE[8|16] */ | |
3640 | ||
3641 | default: | default: |
3642 | SLJIT_ASSERT_STOP(); | SLJIT_ASSERT_STOP(); |
3643 | break; | break; |
# | Line 2861 do | Line 3648 do |
3648 | #else | #else |
3649 | ||
3650 | /* Unaligned read is unsupported. */ | /* Unaligned read is unsupported. */ |
#ifdef COMPILE_PCRE8 | ||
if (context->length > 0) | ||
OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | ||
#else | ||
3651 | if (context->length > 0) | if (context->length > 0) |
3652 | OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(MOV_UCHAR, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3653 | #endif | |
3654 | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; |
3655 | ||
3656 | if (othercasebit != 0 && othercasechar == cc) | if (othercasebit != 0 && othercasechar == cc) |
3657 | { | { |
3658 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); |
3659 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc | othercasebit)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc | othercasebit)); |
3660 | } | } |
3661 | else | else |
3662 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc)); |
3663 | ||
3664 | #endif | #endif |
3665 | ||
# | Line 2912 return cc; | Line 3695 return cc; |
3695 | } \ | } \ |
3696 | charoffset = (value); | charoffset = (value); |
3697 | ||
3698 | static void compile_xclass_hotpath(compiler_common *common, pcre_uchar *cc, jump_list **fallbacks) | static void compile_xclass_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks) |
3699 | { | { |
3700 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3701 | jump_list *found = NULL; | jump_list *found = NULL; |
3702 | jump_list **list = (*cc & XCL_NOT) == 0 ? &found : fallbacks; | jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks; |
3703 | unsigned int c; | pcre_int32 c, charoffset; |
3704 | int compares; | const pcre_uint32 *other_cases; |
3705 | struct sljit_jump *jump = NULL; | struct sljit_jump *jump = NULL; |
3706 | pcre_uchar *ccbegin; | pcre_uchar *ccbegin; |
3707 | int compares, invertcmp, numberofcmps; | |
3708 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
3709 | BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; | BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; |
3710 | BOOL charsaved = FALSE; | BOOL charsaved = FALSE; |
3711 | int typereg = TMP1, scriptreg = TMP1; | int typereg = TMP1, scriptreg = TMP1; |
3712 | unsigned int typeoffset; | pcre_int32 typeoffset; |
3713 | #endif | #endif |
int invertcmp, numberofcmps; | ||
unsigned int charoffset; | ||
3714 | ||
3715 | /* Although SUPPORT_UTF must be defined, we are not necessary in utf mode. */ | /* Although SUPPORT_UTF must be defined, we are |
3716 | fallback_at_str_end(common, fallbacks); | not necessary in utf mode even in 8 bit mode. */ |
3717 | detect_partial_match(common, backtracks); | |
3718 | read_char(common); | read_char(common); |
3719 | ||
3720 | if ((*cc++ & XCL_MAP) != 0) | if ((*cc++ & XCL_MAP) != 0) |
# | Line 2944 if ((*cc++ & XCL_MAP) != 0) | Line 3727 if ((*cc++ & XCL_MAP) != 0) |
3727 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); |
3728 | #endif | #endif |
3729 | ||
3730 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, list)) |
3731 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | { |
3732 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
3733 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
3734 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); |
3735 | add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); |
3736 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | |
3737 | add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); | |
3738 | } | |
3739 | ||
3740 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3741 | JUMPHERE(jump); | JUMPHERE(jump); |
# | Line 2973 while (*cc != XCL_END) | Line 3759 while (*cc != XCL_END) |
3759 | if (*cc == XCL_SINGLE) | if (*cc == XCL_SINGLE) |
3760 | { | { |
3761 | cc += 2; | cc += 2; |
3762 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
3763 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
3764 | #endif | #endif |
3765 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
# | Line 2983 while (*cc != XCL_END) | Line 3769 while (*cc != XCL_END) |
3769 | else if (*cc == XCL_RANGE) | else if (*cc == XCL_RANGE) |
3770 | { | { |
3771 | cc += 2; | cc += 2; |
3772 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
3773 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
3774 | #endif | #endif |
3775 | cc++; | cc++; |
3776 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
3777 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
3778 | #endif | #endif |
3779 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
# | Line 3022 while (*cc != XCL_END) | Line 3808 while (*cc != XCL_END) |
3808 | needschar = TRUE; | needschar = TRUE; |
3809 | break; | break; |
3810 | ||
3811 | case PT_CLIST: | |
3812 | needschar = TRUE; | |
3813 | break; | |
3814 | ||
3815 | default: | default: |
3816 | SLJIT_ASSERT_STOP(); | SLJIT_ASSERT_STOP(); |
3817 | break; | break; |
# | Line 3082 typeoffset = 0; | Line 3872 typeoffset = 0; |
3872 | while (*cc != XCL_END) | while (*cc != XCL_END) |
3873 | { | { |
3874 | compares--; | compares--; |
3875 | invertcmp = (compares == 0 && list != fallbacks); | invertcmp = (compares == 0 && list != backtracks); |
3876 | jump = NULL; | jump = NULL; |
3877 | ||
3878 | if (*cc == XCL_SINGLE) | if (*cc == XCL_SINGLE) |
# | Line 3164 while (*cc != XCL_END) | Line 3954 while (*cc != XCL_END) |
3954 | switch(*cc) | switch(*cc) |
3955 | { | { |
3956 | case PT_ANY: | case PT_ANY: |
3957 | if (list != fallbacks) | if (list != backtracks) |
3958 | { | { |
3959 | if ((cc[-1] == XCL_NOTPROP && compares > 0) || (cc[-1] == XCL_PROP && compares == 0)) | if ((cc[-1] == XCL_NOTPROP && compares > 0) || (cc[-1] == XCL_PROP && compares == 0)) |
3960 | continue; | continue; |
# | Line 3231 while (*cc != XCL_END) | Line 4021 while (*cc != XCL_END) |
4021 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL); |
4022 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
4023 | break; | break; |
4024 | ||
4025 | case PT_CLIST: | |
4026 | other_cases = PRIV(ucd_caseless_sets) + cc[1]; | |
4027 | ||
4028 | /* At least three characters are required. | |
4029 | Otherwise this case would be handled by the normal code path. */ | |
4030 | SLJIT_ASSERT(other_cases[0] != NOTACHAR && other_cases[1] != NOTACHAR && other_cases[2] != NOTACHAR); | |
4031 | SLJIT_ASSERT(other_cases[0] < other_cases[1] && other_cases[1] < other_cases[2]); | |
4032 | ||
4033 | /* Optimizing character pairs, if their difference is power of 2. */ | |
4034 | if (is_powerof2(other_cases[1] ^ other_cases[0])) | |
4035 | { | |
4036 | if (charoffset == 0) | |
4037 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); | |
4038 | else | |
4039 | { | |
4040 | OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_w)charoffset); | |
4041 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); | |
4042 | } | |
4043 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[1]); | |
4044 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | |
4045 | other_cases += 2; | |
4046 | } | |
4047 | else if (is_powerof2(other_cases[2] ^ other_cases[1])) | |
4048 | { | |
4049 | if (charoffset == 0) | |
4050 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[2] ^ other_cases[1]); | |
4051 | else | |
4052 | { | |
4053 | OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_w)charoffset); | |
4054 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); | |
4055 | } | |
4056 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]); | |
4057 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | |
4058 | ||
4059 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, other_cases[0] - charoffset); | |
4060 | COND_VALUE(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, SLJIT_C_EQUAL); | |
4061 | ||
4062 | other_cases += 3; | |
4063 | } | |
4064 | else | |
4065 | { | |
4066 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset); | |
4067 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | |
4068 | } | |
4069 | ||
4070 | while (*other_cases != NOTACHAR) | |
4071 | { | |
4072 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset); | |
4073 | COND_VALUE(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, SLJIT_C_EQUAL); | |
4074 | } | |
4075 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | |
4076 | break; | |
4077 | } | } |
4078 | cc += 2; | cc += 2; |
4079 | } | } |
4080 | #endif | #endif |
4081 | ||
4082 | if (jump != NULL) | if (jump != NULL) |
4083 | add_jump(compiler, compares > 0 ? list : fallbacks, jump); | add_jump(compiler, compares > 0 ? list : backtracks, jump); |
4084 | } | } |
4085 | ||
4086 | if (found != NULL) | if (found != NULL) |
# | Line 3249 if (found != NULL) | Line 4092 if (found != NULL) |
4092 | ||
4093 | #endif | #endif |
4094 | ||
4095 | static pcre_uchar *compile_char1_hotpath(compiler_common *common, pcre_uchar type, pcre_uchar *cc, jump_list **fallbacks) | static pcre_uchar *compile_char1_matchingpath(compiler_common *common, pcre_uchar type, pcre_uchar *cc, jump_list **backtracks) |
4096 | { | { |
4097 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4098 | int length; | int length; |
# | Line 3268 switch(type) | Line 4111 switch(type) |
4111 | case OP_SOD: | case OP_SOD: |
4112 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
4113 | 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)); |
4114 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); |
4115 | return cc; | return cc; |
4116 | ||
4117 | case OP_SOM: | case OP_SOM: |
4118 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
4119 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); |
4120 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); |
4121 | return cc; | return cc; |
4122 | ||
4123 | case OP_NOT_WORD_BOUNDARY: | case OP_NOT_WORD_BOUNDARY: |
4124 | case OP_WORD_BOUNDARY: | case OP_WORD_BOUNDARY: |
4125 | add_jump(compiler, &common->wordboundary, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->wordboundary, JUMP(SLJIT_FAST_CALL)); |
4126 | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_WORD_BOUNDARY ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_NOT_WORD_BOUNDARY ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
4127 | return cc; | return cc; |
4128 | ||
4129 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
4130 | case OP_DIGIT: | case OP_DIGIT: |
4131 | fallback_at_str_end(common, fallbacks); | /* Digits are usually 0-9, so it is worth to optimize them. */ |
4132 | read_char8_type(common); | if (common->digits[0] == -2) |
4133 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); | get_ctype_ranges(common, ctype_digit, common->digits); |
4134 | add_jump(compiler, fallbacks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | detect_partial_match(common, backtracks); |
4135 | /* Flip the starting bit in the negative case. */ | |
4136 | if (type == OP_NOT_DIGIT) | |
4137 | common->digits[1] ^= 1; | |
4138 | if (!check_ranges(common, common->digits, backtracks, TRUE)) | |
4139 | { | |
4140 | read_char8_type(common); | |
4141 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); | |
4142 | add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | |
4143 | } | |
4144 | if (type == OP_NOT_DIGIT) | |
4145 | common->digits[1] ^= 1; | |
4146 | return cc; | return cc; |
4147 | ||
4148 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
4149 | case OP_WHITESPACE: | case OP_WHITESPACE: |
4150 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4151 | read_char8_type(common); | read_char8_type(common); |
4152 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space); |
4153 | add_jump(compiler, fallbacks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); |
4154 | return cc; | return cc; |
4155 | ||
4156 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
4157 | case OP_WORDCHAR: | case OP_WORDCHAR: |
4158 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4159 | read_char8_type(common); | read_char8_type(common); |
4160 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word); | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word); |
4161 | add_jump(compiler, fallbacks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); |
4162 | return cc; | return cc; |
4163 | ||
4164 | case OP_ANY: | case OP_ANY: |
4165 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4166 | read_char(common); | read_char(common); |
4167 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
4168 | { | { |
# | Line 3319 switch(type) | Line 4173 switch(type) |
4173 | jump[1] = check_str_end(common); | jump[1] = check_str_end(common); |
4174 | ||
4175 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
4176 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff)); |
4177 | if (jump[1] != NULL) | if (jump[1] != NULL) |
4178 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4179 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4180 | } | } |
4181 | else | else |
4182 | check_newlinechar(common, common->nltype, fallbacks, TRUE); | check_newlinechar(common, common->nltype, backtracks, TRUE); |
4183 | return cc; | return cc; |
4184 | ||
4185 | case OP_ALLANY: | case OP_ALLANY: |
4186 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4187 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
4188 | if (common->utf) | if (common->utf) |
4189 | { | { |
4190 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
4191 | 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)); |
4192 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 |
4193 | #if defined COMPILE_PCRE8 | |
4194 | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
4195 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); |
4196 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
4197 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
4198 | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
4199 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
4200 | 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); |
4201 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); |
4202 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
4203 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
4204 | #endif /* COMPILE_PCRE16 */ | #endif |
#endif /* COMPILE_PCRE8 */ | ||
4205 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4206 | #endif /* COMPILE_PCRE[8|16] */ | |
4207 | return cc; | return cc; |
4208 | } | } |
4209 | #endif | #endif |
# | Line 3357 switch(type) | Line 4211 switch(type) |
4211 | return cc; | return cc; |
4212 | ||
4213 | case OP_ANYBYTE: | case OP_ANYBYTE: |
4214 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4215 | 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)); |
4216 | return cc; | return cc; |
4217 | ||
# | Line 3370 switch(type) | Line 4224 switch(type) |
4224 | propdata[2] = cc[0]; | propdata[2] = cc[0]; |
4225 | propdata[3] = cc[1]; | propdata[3] = cc[1]; |
4226 | propdata[4] = XCL_END; | propdata[4] = XCL_END; |
4227 | compile_xclass_hotpath(common, propdata, fallbacks); | compile_xclass_matchingpath(common, propdata, backtracks); |
4228 | return cc + 2; | return cc + 2; |
4229 | #endif | #endif |
4230 | #endif | #endif |
4231 | ||
4232 | case OP_ANYNL: | case OP_ANYNL: |
4233 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4234 | read_char(common); | read_char(common); |
4235 | jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); | jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); |
4236 | /* We don't need to handle soft partial matching case. */ | /* We don't need to handle soft partial matching case. */ |
# | Line 3389 switch(type) | Line 4243 switch(type) |
4243 | 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)); |
4244 | jump[3] = JUMP(SLJIT_JUMP); | jump[3] = JUMP(SLJIT_JUMP); |
4245 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4246 | check_newlinechar(common, common->bsr_nltype, fallbacks, FALSE); | check_newlinechar(common, common->bsr_nltype, backtracks, FALSE); |
4247 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4248 | JUMPHERE(jump[2]); | JUMPHERE(jump[2]); |
4249 | JUMPHERE(jump[3]); | JUMPHERE(jump[3]); |
# | Line 3397 switch(type) | Line 4251 switch(type) |
4251 | ||
4252 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
4253 | case OP_HSPACE: | case OP_HSPACE: |
4254 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4255 | read_char(common); | read_char(common); |
4256 | add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL)); |
4257 | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
4258 | return cc; | return cc; |
4259 | ||
4260 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
4261 | case OP_VSPACE: | case OP_VSPACE: |
4262 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4263 | read_char(common); | read_char(common); |
4264 | add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL)); |
4265 | add_jump(compiler, fallbacks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); |
4266 | return cc; | return cc; |
4267 | ||
4268 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
4269 | case OP_EXTUNI: | case OP_EXTUNI: |
4270 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4271 | read_char(common); | read_char(common); |
4272 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
4273 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Mc); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop)); |
4274 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_Mn - ucp_Mc)); | /* Optimize register allocation: use a real register. */ |
4275 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0); | |
4276 | OP1(SLJIT_MOV_UB, STACK_TOP, 0, SLJIT_MEM2(TMP1, TMP2), 3); | |
4277 | ||
4278 | label = LABEL(); | label = LABEL(); |
4279 | jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
4280 | OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); | OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); |
4281 | read_char(common); | read_char(common); |
4282 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
4283 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Mc); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop)); |
4284 | CMPTO(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_Mn - ucp_Mc, label); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM2(TMP1, TMP2), 3); |
4285 | ||
4286 | OP2(SLJIT_SHL, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 2); | |
4287 | OP1(SLJIT_MOV_UI, TMP1, 0, SLJIT_MEM1(STACK_TOP), (sljit_w)PRIV(ucp_gbtable)); | |
4288 | OP1(SLJIT_MOV, STACK_TOP, 0, TMP2, 0); | |
4289 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | |
4290 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | |
4291 | JUMPTO(SLJIT_C_NOT_ZERO, label); | |
4292 | ||
4293 | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); |
4294 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4295 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | |
4296 | ||
4297 | if (common->mode == JIT_PARTIAL_HARD_COMPILE) | if (common->mode == JIT_PARTIAL_HARD_COMPILE) |
4298 | { | { |
4299 | jump[0] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump[0] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); |
# | Line 3447 switch(type) | Line 4312 switch(type) |
4312 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
4313 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
4314 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
4315 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); |
4316 | else | else |
4317 | { | { |
4318 | jump[1] = CMP(SLJIT_C_EQUAL, TMP2, 0, STR_END, 0); | jump[1] = CMP(SLJIT_C_EQUAL, TMP2, 0, STR_END, 0); |
# | Line 3455 switch(type) | Line 4320 switch(type) |
4320 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS); |
4321 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); |
4322 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_NOT_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_NOT_EQUAL); |
4323 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_NOT_EQUAL)); | add_jump(compiler, backtracks, JUMP(SLJIT_C_NOT_EQUAL)); |
4324 | check_partial(common, TRUE); | check_partial(common, TRUE); |
4325 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
4326 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4327 | } | } |
4328 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
4329 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); |
4330 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); |
4331 | } | } |
4332 | else if (common->nltype == NLTYPE_FIXED) | else if (common->nltype == NLTYPE_FIXED) |
4333 | { | { |
4334 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
4335 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
4336 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); |
4337 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); |
4338 | } | } |
4339 | else | else |
4340 | { | { |
# | Line 3478 switch(type) | Line 4343 switch(type) |
4343 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
4344 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); |
4345 | jump[2] = JUMP(SLJIT_C_GREATER); | jump[2] = JUMP(SLJIT_C_GREATER); |
4346 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_LESS)); | add_jump(compiler, backtracks, JUMP(SLJIT_C_LESS)); |
4347 | /* Equal. */ | /* Equal. */ |
4348 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
4349 | jump[3] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); | jump[3] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); |
4350 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
4351 | ||
4352 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4353 | if (common->nltype == NLTYPE_ANYCRLF) | if (common->nltype == NLTYPE_ANYCRLF) |
4354 | { | { |
4355 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); |
4356 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, TMP2, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP2, 0, STR_END, 0)); |
4357 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); |
4358 | } | } |
4359 | else | else |
4360 | { | { |
4361 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0); |
4362 | read_char(common); | read_char(common); |
4363 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0)); |
4364 | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); |
4365 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); |
4366 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); |
4367 | } | } |
4368 | JUMPHERE(jump[2]); | JUMPHERE(jump[2]); |
# | Line 3508 switch(type) | Line 4373 switch(type) |
4373 | return cc; | return cc; |
4374 | ||
4375 | case OP_EOD: | case OP_EOD: |
4376 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); |
4377 | check_partial(common, FALSE); | check_partial(common, FALSE); |
4378 | return cc; | return cc; |
4379 | ||
4380 | case OP_CIRC: | case OP_CIRC: |
4381 | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); |
4382 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); |
4383 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0)); |
4384 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); |
4385 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
4386 | return cc; | return cc; |
4387 | ||
4388 | case OP_CIRCM: | case OP_CIRCM: |
# | Line 3525 switch(type) | Line 4390 switch(type) |
4390 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); |
4391 | jump[1] = CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0); | jump[1] = CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0); |
4392 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); |
4393 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
4394 | jump[0] = JUMP(SLJIT_JUMP); | jump[0] = JUMP(SLJIT_JUMP); |
4395 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4396 | ||
4397 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); |
4398 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
4399 | { | { |
4400 | OP2(SLJIT_SUB, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_SUB, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
4401 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, TMP2, 0, TMP1, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP2, 0, TMP1, 0)); |
4402 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); |
4403 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
4404 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); |
4405 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); |
4406 | } | } |
4407 | else | else |
4408 | { | { |
4409 | skip_char_back(common); | skip_char_back(common); |
4410 | read_char(common); | read_char(common); |
4411 | check_newlinechar(common, common->nltype, fallbacks, FALSE); | check_newlinechar(common, common->nltype, backtracks, FALSE); |
4412 | } | } |
4413 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4414 | return cc; | return cc; |
# | Line 3551 switch(type) | Line 4416 switch(type) |
4416 | case OP_DOLL: | case OP_DOLL: |
4417 | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); |
4418 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); |
4419 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
4420 | ||
4421 | if (!common->endonly) | if (!common->endonly) |
4422 | compile_char1_hotpath(common, OP_EODN, cc, fallbacks); | compile_char1_matchingpath(common, OP_EODN, cc, backtracks); |
4423 | else | else |
4424 | { | { |
4425 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); |
4426 | check_partial(common, FALSE); | check_partial(common, FALSE); |
4427 | } | } |
4428 | return cc; | return cc; |
# | Line 3566 switch(type) | Line 4431 switch(type) |
4431 | jump[1] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump[1] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); |
4432 | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); |
4433 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); |
4434 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
4435 | check_partial(common, FALSE); | check_partial(common, FALSE); |
4436 | jump[0] = JUMP(SLJIT_JUMP); | jump[0] = JUMP(SLJIT_JUMP); |
4437 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
# | Line 3576 switch(type) | Line 4441 switch(type) |
4441 | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); |
4442 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
4443 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
4444 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, TMP2, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, TMP2, 0, STR_END, 0)); |
4445 | else | else |
4446 | { | { |
4447 | jump[1] = CMP(SLJIT_C_LESS_EQUAL, TMP2, 0, STR_END, 0); | jump[1] = CMP(SLJIT_C_LESS_EQUAL, TMP2, 0, STR_END, 0); |
4448 | /* STR_PTR = STR_END - IN_UCHARS(1) */ | /* STR_PTR = STR_END - IN_UCHARS(1) */ |
4449 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); |
4450 | check_partial(common, TRUE); | check_partial(common, TRUE); |
4451 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
4452 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4453 | } | } |
4454 | ||
4455 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); |
4456 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); |
4457 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); |
4458 | } | } |
4459 | else | else |
4460 | { | { |
4461 | peek_char(common); | peek_char(common); |
4462 | check_newlinechar(common, common->nltype, fallbacks, FALSE); | check_newlinechar(common, common->nltype, backtracks, FALSE); |
4463 | } | } |
4464 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4465 | return cc; | return cc; |
# | Line 3602 switch(type) | Line 4467 switch(type) |
4467 | case OP_CHAR: | case OP_CHAR: |
4468 | case OP_CHARI: | case OP_CHARI: |
4469 | length = 1; | length = 1; |
4470 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
4471 | if (common->utf && HAS_EXTRALEN(*cc)) length += GET_EXTRALEN(*cc); | if (common->utf && HAS_EXTRALEN(*cc)) length += GET_EXTRALEN(*cc); |
4472 | #endif | #endif |
4473 | if (common->mode == JIT_COMPILE && (type == OP_CHAR || !char_has_othercase(common, cc) || char_get_othercase_bit(common, cc) != 0)) | if (common->mode == JIT_COMPILE && (type == OP_CHAR || !char_has_othercase(common, cc) || char_get_othercase_bit(common, cc) != 0)) |
4474 | { | { |
4475 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); |
4476 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); |
4477 | ||
4478 | context.length = IN_UCHARS(length); | context.length = IN_UCHARS(length); |
4479 | context.sourcereg = -1; | context.sourcereg = -1; |
4480 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
4481 | context.ucharptr = 0; | context.ucharptr = 0; |
4482 | #endif | #endif |
4483 | return byte_sequence_compare(common, type == OP_CHARI, cc, &context, fallbacks); | return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks); |
4484 | } | } |
4485 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4486 | read_char(common); | read_char(common); |
4487 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
4488 | if (common->utf) | if (common->utf) |
# | Line 3629 switch(type) | Line 4494 switch(type) |
4494 | c = *cc; | c = *cc; |
4495 | if (type == OP_CHAR || !char_has_othercase(common, cc)) | if (type == OP_CHAR || !char_has_othercase(common, cc)) |
4496 | { | { |
4497 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c)); |
4498 | return cc + length; | return cc + length; |
4499 | } | } |
4500 | oc = char_othercase(common, c); | oc = char_othercase(common, c); |
4501 | bit = c ^ oc; | bit = c ^ oc; |
4502 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
4503 | { | { |
4504 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit); |
4505 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit)); |
4506 | return cc + length; | return cc + length; |
4507 | } | } |
4508 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c); |
4509 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); |
4510 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, char_othercase(common, c)); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); |
4511 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
4512 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); |
4513 | return cc + length; | return cc + length; |
4514 | ||
4515 | case OP_NOT: | case OP_NOT: |
4516 | case OP_NOTI: | case OP_NOTI: |
4517 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4518 | length = 1; | length = 1; |
4519 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
4520 | if (common->utf) | if (common->utf) |
# | Line 3660 switch(type) | Line 4525 switch(type) |
4525 | { | { |
4526 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
4527 | if (type == OP_NOT || !char_has_othercase(common, cc)) | if (type == OP_NOT || !char_has_othercase(common, cc)) |
4528 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); |
4529 | else | else |
4530 | { | { |
4531 | /* Since UTF8 code page is fixed, we know that c is in [a-z] or [A-Z] range. */ | /* Since UTF8 code page is fixed, we know that c is in [a-z] or [A-Z] range. */ |
4532 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x20); | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x20); |
4533 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, c | 0x20)); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, c | 0x20)); |
4534 | } | } |
4535 | /* Skip the variable-length character. */ | /* Skip the variable-length character. */ |
4536 | 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)); |
# | Line 3690 switch(type) | Line 4555 switch(type) |
4555 | } | } |
4556 | ||
4557 | if (type == OP_NOT || !char_has_othercase(common, cc)) | if (type == OP_NOT || !char_has_othercase(common, cc)) |
4558 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); |
4559 | else | else |
4560 | { | { |
4561 | oc = char_othercase(common, c); | oc = char_othercase(common, c); |
4562 | bit = c ^ oc; | bit = c ^ oc; |
4563 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
4564 | { | { |
4565 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit); |
4566 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c | bit)); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c | bit)); |
4567 | } | } |
4568 | else | else |
4569 | { | { |
4570 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); |
4571 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, oc)); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, oc)); |
4572 | } | } |
4573 | } | } |
4574 | return cc + length; | return cc + length; |
4575 | ||
4576 | case OP_CLASS: | case OP_CLASS: |
4577 | case OP_NCLASS: | case OP_NCLASS: |
4578 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4579 | read_char(common); | read_char(common); |
4580 | if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, backtracks)) | |
4581 | return cc + 32 / sizeof(pcre_uchar); | |
4582 | ||
4583 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
4584 | jump[0] = NULL; | jump[0] = NULL; |
4585 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
# | Line 3723 switch(type) | Line 4591 switch(type) |
4591 | jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); |
4592 | if (type == OP_CLASS) | if (type == OP_CLASS) |
4593 | { | { |
4594 | add_jump(compiler, fallbacks, jump[0]); | add_jump(compiler, backtracks, jump[0]); |
4595 | jump[0] = NULL; | jump[0] = NULL; |
4596 | } | } |
4597 | } | } |
# | Line 3733 switch(type) | Line 4601 switch(type) |
4601 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); |
4602 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); |
4603 | 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); |
4604 | add_jump(compiler, fallbacks, JUMP(SLJIT_C_ZERO)); | add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); |
4605 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
4606 | if (jump[0] != NULL) | if (jump[0] != NULL) |
4607 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4608 | #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */ | #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */ |
4609 | return cc + 32 / sizeof(pcre_uchar); | return cc + 32 / sizeof(pcre_uchar); |
4610 | ||
4611 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
4612 | case OP_XCLASS: | case OP_XCLASS: |
4613 | compile_xclass_hotpath(common, cc + LINK_SIZE, fallbacks); | compile_xclass_matchingpath(common, cc + LINK_SIZE, backtracks); |
4614 | return cc + GET(cc, 0) - 1; | return cc + GET(cc, 0) - 1; |
4615 | #endif | #endif |
4616 | ||
# | Line 3757 switch(type) | Line 4625 switch(type) |
4625 | OP1(SLJIT_MOV, TMP3, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, TMP3, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); |
4626 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, length); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, length); |
4627 | label = LABEL(); | label = LABEL(); |
4628 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP3, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP3, 0)); |
4629 | skip_char_back(common); | skip_char_back(common); |
4630 | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); |
4631 | JUMPTO(SLJIT_C_NOT_ZERO, label); | JUMPTO(SLJIT_C_NOT_ZERO, label); |
# | Line 3767 switch(type) | Line 4635 switch(type) |
4635 | { | { |
4636 | 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)); |
4637 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); |
4638 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS, STR_PTR, 0, TMP1, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, TMP1, 0)); |
4639 | } | } |
4640 | check_start_used_ptr(common); | check_start_used_ptr(common); |
4641 | return cc + LINK_SIZE; | return cc + LINK_SIZE; |
# | Line 3776 SLJIT_ASSERT_STOP(); | Line 4644 SLJIT_ASSERT_STOP(); |
4644 | return cc; | return cc; |
4645 | } | } |
4646 | ||
4647 | static SLJIT_INLINE pcre_uchar *compile_charn_hotpath(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, jump_list **fallbacks) | static SLJIT_INLINE pcre_uchar *compile_charn_matchingpath(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, jump_list **backtracks) |
4648 | { | { |
4649 | /* This function consumes at least one input character. */ | /* This function consumes at least one input character. */ |
4650 | /* To decrease the number of length checks, we try to concatenate the fixed length character sequences. */ | /* To decrease the number of length checks, we try to concatenate the fixed length character sequences. */ |
# | Line 3794 do | Line 4662 do |
4662 | if (*cc == OP_CHAR) | if (*cc == OP_CHAR) |
4663 | { | { |
4664 | size = 1; | size = 1; |
4665 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
4666 | if (common->utf && HAS_EXTRALEN(cc[1])) | if (common->utf && HAS_EXTRALEN(cc[1])) |
4667 | size += GET_EXTRALEN(cc[1]); | size += GET_EXTRALEN(cc[1]); |
4668 | #endif | #endif |
# | Line 3807 do | Line 4675 do |
4675 | { | { |
4676 | if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) | if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) |
4677 | size = 0; | size = 0; |
4678 | #ifndef COMPILE_PCRE32 | |
4679 | else if (HAS_EXTRALEN(cc[1])) | else if (HAS_EXTRALEN(cc[1])) |
4680 | size += GET_EXTRALEN(cc[1]); | size += GET_EXTRALEN(cc[1]); |
4681 | #endif | |
4682 | } | } |
4683 | else | else |
4684 | #endif | #endif |
# | Line 3828 if (context.length > 0) | Line 4698 if (context.length > 0) |
4698 | { | { |
4699 | /* We have a fixed-length byte sequence. */ | /* We have a fixed-length byte sequence. */ |
4700 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, context.length); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, context.length); |
4701 | add_jump(compiler, fallbacks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); |
4702 | ||
4703 | context.sourcereg = -1; | context.sourcereg = -1; |
4704 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
4705 | context.ucharptr = 0; | context.ucharptr = 0; |
4706 | #endif | #endif |
4707 | do cc = byte_sequence_compare(common, *cc == OP_CHARI, cc + 1, &context, fallbacks); while (context.length > 0); | do cc = byte_sequence_compare(common, *cc == OP_CHARI, cc + 1, &context, backtracks); while (context.length > 0); |
4708 | return cc; | return cc; |
4709 | } | } |
4710 | ||
4711 | /* A non-fixed length character will be checked if length == 0. */ | /* A non-fixed length character will be checked if length == 0. */ |
4712 | return compile_char1_hotpath(common, *cc, cc + 1, fallbacks); | return compile_char1_matchingpath(common, *cc, cc + 1, backtracks); |
4713 | } | } |
4714 | ||
4715 | static struct sljit_jump *compile_ref_checks(compiler_common *common, pcre_uchar *cc, jump_list **fallbacks) | static struct sljit_jump *compile_ref_checks(compiler_common *common, pcre_uchar *cc, jump_list **backtracks) |
4716 | { | { |
4717 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4718 | int offset = GET2(cc, 1) << 1; | int offset = GET2(cc, 1) << 1; |
# | Line 3850 int offset = GET2(cc, 1) << 1; | Line 4720 int offset = GET2(cc, 1) << 1; |
4720 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); |
4721 | if (!common->jscript_compat) | if (!common->jscript_compat) |
4722 | { | { |
4723 | if (fallbacks == NULL) | if (backtracks == NULL) |
4724 | { | { |
4725 | /* OVECTOR(1) contains the "string begin - 1" constant. */ | /* OVECTOR(1) contains the "string begin - 1" constant. */ |
4726 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); |
# | Line 3859 if (!common->jscript_compat) | Line 4729 if (!common->jscript_compat) |
4729 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); |
4730 | return JUMP(SLJIT_C_NOT_ZERO); | return JUMP(SLJIT_C_NOT_ZERO); |
4731 | } | } |
4732 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); |
4733 | } | } |
4734 | return CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); | return CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); |
4735 | } | } |
4736 | ||
4737 | /* Forward definitions. */ | /* Forward definitions. */ |
4738 | static void compile_hotpath(compiler_common *, pcre_uchar *, pcre_uchar *, fallback_common *); | static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *); |
4739 | static void compile_fallbackpath(compiler_common *, struct fallback_common *); | static void compile_backtrackingpath(compiler_common *, struct backtrack_common *); |
4740 | ||
4741 | #define PUSH_FALLBACK(size, ccstart, error) \ | #define PUSH_BACKTRACK(size, ccstart, error) \ |
4742 | do \ | do \ |
4743 | { \ | { \ |
4744 | fallback = sljit_alloc_memory(compiler, (size)); \ | backtrack = sljit_alloc_memory(compiler, (size)); \ |
4745 | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ |
4746 | return error; \ | return error; \ |
4747 | memset(fallback, 0, size); \ | memset(backtrack, 0, size); \ |
4748 | fallback->prev = parent->top; \ | backtrack->prev = parent->top; \ |
4749 | fallback->cc = (ccstart); \ | backtrack->cc = (ccstart); \ |
4750 | parent->top = fallback; \ | parent->top = backtrack; \ |
4751 | } \ | } \ |
4752 | while (0) | while (0) |
4753 | ||
4754 | #define PUSH_FALLBACK_NOVALUE(size, ccstart) \ | #define PUSH_BACKTRACK_NOVALUE(size, ccstart) \ |
4755 | do \ | do \ |
4756 | { \ | { \ |
4757 | fallback = sljit_alloc_memory(compiler, (size)); \ | backtrack = sljit_alloc_memory(compiler, (size)); \ |
4758 | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ |
4759 | return; \ | return; \ |
4760 | memset(fallback, 0, size); \ | memset(backtrack, 0, size); \ |
4761 | fallback->prev = parent->top; \ | backtrack->prev = parent->top; \ |
4762 | fallback->cc = (ccstart); \ | backtrack->cc = (ccstart); \ |
4763 | parent->top = fallback; \ | parent->top = backtrack; \ |
4764 | } \ | } \ |
4765 | while (0) | while (0) |
4766 | ||
4767 | #define FALLBACK_AS(type) ((type *)fallback) | #define BACKTRACK_AS(type) ((type *)backtrack) |
4768 | ||
4769 | static pcre_uchar *compile_ref_hotpath(compiler_common *common, pcre_uchar *cc, jump_list **fallbacks, BOOL withchecks, BOOL emptyfail) | static pcre_uchar *compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail) |
4770 | { | { |
4771 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4772 | int offset = GET2(cc, 1) << 1; | int offset = GET2(cc, 1) << 1; |
# | Line 3907 struct sljit_jump *nopartial; | Line 4777 struct sljit_jump *nopartial; |
4777 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); |
4778 | /* OVECTOR(1) contains the "string begin - 1" constant. */ | /* OVECTOR(1) contains the "string begin - 1" constant. */ |
4779 | if (withchecks && !common->jscript_compat) | if (withchecks && !common->jscript_compat) |
4780 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); |
4781 | ||
4782 | #if defined SUPPORT_UTF && defined SUPPORT_UCP | #if defined SUPPORT_UTF && defined SUPPORT_UCP |
4783 | if (common->utf && *cc == OP_REFI) | if (common->utf && *cc == OP_REFI) |
# | Line 3924 if (common->utf && *cc == OP_REFI) | Line 4794 if (common->utf && *cc == OP_REFI) |
4794 | sljit_emit_ijump(compiler, SLJIT_CALL3, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_utf_caselesscmp)); | sljit_emit_ijump(compiler, SLJIT_CALL3, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_utf_caselesscmp)); |
4795 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); |
4796 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
4797 | add_jump(compiler, fallbacks, CMP(SLJIT_C_LESS_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1)); | add_jump(compiler, backtracks, CMP(SLJIT_C_LESS_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1)); |
4798 | else | else |
4799 | { | { |
4800 | add_jump(compiler, fallbacks, CMP(SLJIT_C_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0)); |
4801 | nopartial = CMP(SLJIT_C_NOT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); | nopartial = CMP(SLJIT_C_NOT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); |
4802 | check_partial(common, FALSE); | check_partial(common, FALSE); |
4803 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
4804 | JUMPHERE(nopartial); | JUMPHERE(nopartial); |
4805 | } | } |
4806 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_RETURN_REG, 0); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_RETURN_REG, 0); |
# | Line 3945 else | Line 4815 else |
4815 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
4816 | partial = CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0); | partial = CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0); |
4817 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
4818 | add_jump(compiler, fallbacks, partial); | add_jump(compiler, backtracks, partial); |
4819 | ||
4820 | add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); |
4821 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
4822 | ||
4823 | if (common->mode != JIT_COMPILE) | if (common->mode != JIT_COMPILE) |
4824 | { | { |
# | Line 3960 else | Line 4830 else |
4830 | partial = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0); | partial = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0); |
4831 | OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0); | OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0); |
4832 | add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); |
4833 | add_jump(compiler, fallbacks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); |
4834 | JUMPHERE(partial); | JUMPHERE(partial); |
4835 | check_partial(common, FALSE); | check_partial(common, FALSE); |
4836 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
4837 | JUMPHERE(nopartial); | JUMPHERE(nopartial); |
4838 | } | } |
4839 | } | } |
# | Line 3971 else | Line 4841 else |
4841 | if (jump != NULL) | if (jump != NULL) |
4842 | { | { |
4843 | if (emptyfail) | if (emptyfail) |
4844 | add_jump(compiler, fallbacks, jump); | add_jump(compiler, backtracks, jump); |
4845 | else | else |
4846 | JUMPHERE(jump); | JUMPHERE(jump); |
4847 | } | } |
4848 | return cc + 1 + IMM2_SIZE; | return cc + 1 + IMM2_SIZE; |
4849 | } | } |
4850 | ||
4851 | static SLJIT_INLINE pcre_uchar *compile_ref_iterator_hotpath(compiler_common *common, pcre_uchar *cc, fallback_common *parent) | static SLJIT_INLINE pcre_uchar *compile_ref_iterator_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) |
4852 | { | { |
4853 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4854 | fallback_common *fallback; | backtrack_common *backtrack; |
4855 | pcre_uchar type; | pcre_uchar type; |
4856 | struct sljit_label *label; | struct sljit_label *label; |
4857 | struct sljit_jump *zerolength; | struct sljit_jump *zerolength; |
# | Line 3990 pcre_uchar *ccbegin = cc; | Line 4860 pcre_uchar *ccbegin = cc; |
4860 | int min = 0, max = 0; | int min = 0, max = 0; |
4861 | BOOL minimize; | BOOL minimize; |
4862 | ||
4863 | PUSH_FALLBACK(sizeof(iterator_fallback), cc, NULL); | PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL); |
4864 | ||
4865 | type = cc[1 + IMM2_SIZE]; | type = cc[1 + IMM2_SIZE]; |
4866 | minimize = (type & 0x1) != 0; | minimize = (type & 0x1) != 0; |
# | Line 4042 if (!minimize) | Line 4912 if (!minimize) |
4912 | { | { |
4913 | allocate_stack(common, 1); | allocate_stack(common, 1); |
4914 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); |
4915 | zerolength = compile_ref_checks(common, ccbegin, &fallback->topfallbacks); | zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks); |
4916 | } | } |
4917 | ||
4918 | if (min > 1 || max > 1) | if (min > 1 || max > 1) |
4919 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0); |
4920 | ||
4921 | label = LABEL(); | label = LABEL(); |
4922 | compile_ref_hotpath(common, ccbegin, &fallback->topfallbacks, FALSE, FALSE); | compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, FALSE, FALSE); |
4923 | ||
4924 | if (min > 1 || max > 1) | if (min > 1 || max > 1) |
4925 | { | { |
# | Line 4077 if (!minimize) | Line 4947 if (!minimize) |
4947 | } | } |
4948 | ||
4949 | JUMPHERE(zerolength); | JUMPHERE(zerolength); |
4950 | FALLBACK_AS(iterator_fallback)->hotpath = LABEL(); | BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL(); |
4951 | ||
4952 | decrease_call_count(common); | decrease_call_count(common); |
4953 | return cc; | return cc; |
# | Line 4095 if (min == 0) | Line 4965 if (min == 0) |
4965 | jump = JUMP(SLJIT_JUMP); | jump = JUMP(SLJIT_JUMP); |
4966 | } | } |
4967 | else | else |
4968 | zerolength = compile_ref_checks(common, ccbegin, &fallback->topfallbacks); | zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks); |
4969 | ||
4970 | FALLBACK_AS(iterator_fallback)->hotpath = LABEL(); | BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL(); |
4971 | if (max > 0) | if (max > 0) |
4972 | add_jump(compiler, &fallback->topfallbacks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, max)); | add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, max)); |
4973 | ||
4974 | compile_ref_hotpath(common, ccbegin, &fallback->topfallbacks, TRUE, TRUE); | compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, TRUE, TRUE); |
4975 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); |
4976 | ||
4977 | if (min > 1) | if (min > 1) |
# | Line 4109 if (min > 1) | Line 4979 if (min > 1) |
4979 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); |
4980 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
4981 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); |
4982 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, min, FALLBACK_AS(iterator_fallback)->hotpath); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, min, BACKTRACK_AS(iterator_backtrack)->matchingpath); |
4983 | } | } |
4984 | else if (max > 0) | else if (max > 0) |
4985 | OP2(SLJIT_ADD, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 1); | OP2(SLJIT_ADD, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 1); |
# | Line 4122 decrease_call_count(common); | Line 4992 decrease_call_count(common); |
4992 | return cc; | return cc; |
4993 | } | } |
4994 | ||
4995 | static SLJIT_INLINE pcre_uchar *compile_recurse_hotpath(compiler_common *common, pcre_uchar *cc, fallback_common *parent) | static SLJIT_INLINE pcre_uchar *compile_recurse_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) |
4996 | { | { |
4997 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4998 | fallback_common *fallback; | backtrack_common *backtrack; |
4999 | recurse_entry *entry = common->entries; | recurse_entry *entry = common->entries; |
5000 | recurse_entry *prev = NULL; | recurse_entry *prev = NULL; |
5001 | int start = GET(cc, 1); | int start = GET(cc, 1); |
5002 | ||
5003 | PUSH_FALLBACK(sizeof(recurse_fallback), cc, NULL); | PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL); |
5004 | while (entry != NULL) | while (entry != NULL) |
5005 | { | { |
5006 | if (entry->start == start) | if (entry->start == start) |
# | Line 4175 if (entry->entry == NULL) | Line 5045 if (entry->entry == NULL) |
5045 | else | else |
5046 | JUMPTO(SLJIT_FAST_CALL, entry->entry); | JUMPTO(SLJIT_FAST_CALL, entry->entry); |
5047 | /* Leave if the match is failed. */ | /* Leave if the match is failed. */ |
5048 | add_jump(compiler, &fallback->topfallbacks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0)); | add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0)); |
5049 | return cc + 1 + LINK_SIZE; | return cc + 1 + LINK_SIZE; |
5050 | } | } |
5051 | ||
5052 | static pcre_uchar *compile_assert_hotpath(compiler_common *common, pcre_uchar *cc, assert_fallback *fallback, BOOL conditional) | static pcre_uchar *compile_assert_matchingpath(compiler_common *common, pcre_uchar *cc, assert_backtrack *backtrack, BOOL conditional) |
5053 | { | { |
5054 | DEFINE_COMPILER; | DEFINE_COMPILER; |
5055 | int framesize; | int framesize; |
5056 | int localptr; | int private_data_ptr; |
5057 | fallback_common altfallback; | backtrack_common altbacktrack; |
5058 | pcre_uchar *ccbegin; | pcre_uchar *ccbegin; |
5059 | pcre_uchar opcode; | pcre_uchar opcode; |
5060 | pcre_uchar bra = OP_BRA; | pcre_uchar bra = OP_BRA; |
5061 | jump_list *tmp = NULL; | jump_list *tmp = NULL; |
5062 | jump_list **target = (conditional) ? &fallback->condfailed : &fallback->common.topfallbacks; | jump_list **target = (conditional) ? &backtrack->condfailed : &backtrack->common.topbacktracks; |
5063 | jump_list **found; | jump_list **found; |
5064 | /* Saving previous accept variables. */ | /* Saving previous accept variables. */ |
5065 | struct sljit_label *save_leavelabel = common->leavelabel; | struct sljit_label *save_quitlabel = common->quitlabel; |
5066 | struct sljit_label *save_acceptlabel = common->acceptlabel; | struct sljit_label *save_acceptlabel = common->acceptlabel; |
5067 | jump_list *save_leave = common->leave; | jump_list *save_quit = common->quit; |
5068 | jump_list *save_accept = common->accept; | jump_list *save_accept = common->accept; |
5069 | struct sljit_jump *jump; | struct sljit_jump *jump; |
5070 | struct sljit_jump *brajump = NULL; | struct sljit_jump *brajump = NULL; |
# | Line 4205 if (*cc == OP_BRAZERO || *cc == OP_BRAMI | Line 5075 if (*cc == OP_BRAZERO || *cc == OP_BRAMI |
5075 | bra = *cc; | bra = *cc; |
5076 | cc++; | cc++; |
5077 | } | } |
5078 | localptr = PRIV_DATA(cc); | private_data_ptr = PRIVATE_DATA(cc); |
5079 | SLJIT_ASSERT(localptr != 0); | SLJIT_ASSERT(private_data_ptr != 0); |
5080 | framesize = get_framesize(common, cc, FALSE); | framesize = get_framesize(common, cc, FALSE); |
5081 | fallback->framesize = framesize; | backtrack->framesize = framesize; |
5082 | fallback->localptr = localptr; | backtrack->private_data_ptr = private_data_ptr; |
5083 | opcode = *cc; | opcode = *cc; |
5084 | SLJIT_ASSERT(opcode >= OP_ASSERT && opcode <= OP_ASSERTBACK_NOT); | SLJIT_ASSERT(opcode >= OP_ASSERT && opcode <= OP_ASSERTBACK_NOT); |
5085 | found = (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) ? &tmp : target; | found = (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) ? &tmp : target; |
# | Line 4218 cc += GET(cc, 1); | Line 5088 cc += GET(cc, 1); |
5088 | ||
5089 | if (bra == OP_BRAMINZERO) | if (bra == OP_BRAMINZERO) |
5090 | { | { |
5091 | /* This is a braminzero fallback path. */ | /* This is a braminzero backtrack path. */ |
5092 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); |
5093 | free_stack(common, 1); | free_stack(common, 1); |
5094 | brajump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); | brajump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); |
# | Line 4226 if (bra == OP_BRAMINZERO) | Line 5096 if (bra == OP_BRAMINZERO) |
5096 | ||
5097 | if (framesize < 0) | if (framesize < 0) |
5098 | { | { |
5099 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STACK_TOP, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STACK_TOP, 0); |
5100 | allocate_stack(common, 1); | allocate_stack(common, 1); |
5101 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); |
5102 | } | } |
5103 | else | else |
5104 | { | { |
5105 | allocate_stack(common, framesize + 2); | allocate_stack(common, framesize + 2); |
5106 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); |
5107 | OP2(SLJIT_SUB, TMP2, 0, STACK_TOP, 0, SLJIT_IMM, -STACK(framesize + 1)); | OP2(SLJIT_SUB, TMP2, 0, STACK_TOP, 0, SLJIT_IMM, -STACK(framesize + 1)); |
5108 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP2, 0); |
5109 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); |
5110 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); |
5111 | init_frame(common, ccbegin, framesize + 1, 2, FALSE); | init_frame(common, ccbegin, framesize + 1, 2, FALSE); |
5112 | } | } |
5113 | ||
5114 | memset(&altfallback, 0, sizeof(fallback_common)); | memset(&altbacktrack, 0, sizeof(backtrack_common)); |
5115 | common->leavelabel = NULL; | common->quitlabel = NULL; |
5116 | common->leave = NULL; | common->quit = NULL; |
5117 | while (1) | while (1) |
5118 | { | { |
5119 | common->acceptlabel = NULL; | common->acceptlabel = NULL; |
5120 | common->accept = NULL; | common->accept = NULL; |
5121 | altfallback.top = NULL; | altbacktrack.top = NULL; |
5122 | altfallback.topfallbacks = NULL; | altbacktrack.topbacktracks = NULL; |
5123 | ||
5124 | if (*ccbegin == OP_ALT) | if (*ccbegin == OP_ALT) |
5125 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); |
5126 | ||
5127 | altfallback.cc = ccbegin; | altbacktrack.cc = ccbegin; |
5128 | compile_hotpath(common, ccbegin + 1 + LINK_SIZE, cc, &altfallback); | compile_matchingpath(common, ccbegin + 1 + LINK_SIZE, cc, &altbacktrack); |
5129 | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) |
5130 | { | { |
5131 | common->leavelabel = save_leavelabel; | common->quitlabel = save_quitlabel; |
5132 | common->acceptlabel = save_acceptlabel; | common->acceptlabel = save_acceptlabel; |
5133 | common->leave = save_leave; | common->quit = save_quit; |
5134 | common->accept = save_accept; | common->accept = save_accept; |
5135 | return NULL; | return NULL; |
5136 | } | } |
# | Line 4270 while (1) | Line 5140 while (1) |
5140 | ||
5141 | /* Reset stack. */ | /* Reset stack. */ |
5142 | if (framesize < 0) | if (framesize < 0) |
5143 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); |
5144 | else { | else { |
5145 | if ((opcode != OP_ASSERT_NOT && opcode != OP_ASSERTBACK_NOT) || conditional) | if ((opcode != OP_ASSERT_NOT && opcode != OP_ASSERTBACK_NOT) || conditional) |
5146 | { | { |
5147 | /* We don't need to keep the STR_PTR, only the previous localptr. */ | /* We don't need to keep the STR_PTR, only the previous private_data_ptr. */ |
5148 | OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_w)); | OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_w)); |
5149 | } | } |
5150 | else | else |
5151 | { | { |
5152 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); |
5153 | add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); |
5154 | } | } |
5155 | } | } |
# | Line 4297 while (1) | Line 5167 while (1) |
5167 | { | { |
5168 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); |
5169 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), (framesize + 1) * sizeof(sljit_w)); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), (framesize + 1) * sizeof(sljit_w)); |
5170 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP1, 0); |
5171 | } | } |
5172 | OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); | OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); |
5173 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); |
# | Line 4305 while (1) | Line 5175 while (1) |
5175 | else if (framesize >= 0) | else if (framesize >= 0) |
5176 | { | { |
5177 | /* For OP_BRA and OP_BRAMINZERO. */ | /* For OP_BRA and OP_BRAMINZERO. */ |
5178 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); |
5179 | } | } |
5180 | } | } |
5181 | add_jump(compiler, found, JUMP(SLJIT_JUMP)); | add_jump(compiler, found, JUMP(SLJIT_JUMP)); |
5182 | ||
5183 | compile_fallbackpath(common, altfallback.top); | compile_backtrackingpath(common, altbacktrack.top); |
5184 | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) | if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) |
5185 | { | { |
5186 | common->leavelabel = save_leavelabel; | common->quitlabel = save_quitlabel; |
5187 | common->acceptlabel = save_acceptlabel; | common->acceptlabel = save_acceptlabel; |
5188 | common->leave = save_leave; | common->quit = save_quit; |
5189 | common->accept = save_accept; | common->accept = save_accept; |
5190 | return NULL; | return NULL; |
5191 | } | } |
5192 | set_jumps(altfallback.topfallbacks, LABEL()); | set_jumps(altbacktrack.topbacktracks, LABEL()); |
5193 | ||
5194 | if (*cc != OP_ALT) | if (*cc != OP_ALT) |
5195 | break; | break; |
# | Line 4328 while (1) | Line 5198 while (1) |
5198 | cc += GET(cc, 1); | cc += GET(cc, 1); |
5199 | } | } |
5200 | /* None of them matched. */ | /* None of them matched. */ |
5201 | if (common->leave != NULL) | if (common->quit != NULL) |
5202 | set_jumps(common->leave, LABEL()); | set_jumps(common->quit, LABEL()); |
5203 | ||
5204 | if (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) | if (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) |
5205 | { | { |
# | Line 4356 if (opcode == OP_ASSERT || opcode == OP_ | Line 5226 if (opcode == OP_ASSERT || opcode == OP_ |
5226 | } | } |
5227 | else | else |
5228 | free_stack(common, framesize + 2); | free_stack(common, framesize + 2); |
5229 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP1, 0); |
5230 | } | } |
5231 | jump = JUMP(SLJIT_JUMP); | jump = JUMP(SLJIT_JUMP); |
5232 | if (bra != OP_BRAZERO) | if (bra != OP_BRAZERO) |
# | Line 4381 if (opcode == OP_ASSERT || opcode == OP_ | Line 5251 if (opcode == OP_ASSERT || opcode == OP_ |
5251 | { | { |
5252 | if (bra == OP_BRA) | if (bra == OP_BRA) |
5253 | { | { |
5254 | /* We don't need to keep the STR_PTR, only the previous localptr. */ | /* We don't need to keep the STR_PTR, only the previous private_data_ptr. */ |
5255 | OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_w)); | OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_w)); |
5256 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); |
5257 | } | } |
5258 | else | else |
5259 | { | { |
5260 | /* We don't need to keep the STR_PTR, only the previous localptr. */ | /* We don't need to keep the STR_PTR, only the previous private_data_ptr. */ |
5261 | OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (framesize + 2) * sizeof(sljit_w)); | OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, (framesize + 2) * sizeof(sljit_w)); |
5262 | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); | OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); |
5263 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), bra == OP_BRAZERO ? STR_PTR : SLJIT_IMM, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), bra == OP_BRAZERO ? STR_PTR : SLJIT_IMM, 0); |
5264 | } | } |
# | Line 4396 if (opcode == OP_ASSERT || opcode == OP_ | Line 5266 if (opcode == OP_ASSERT || opcode == OP_ |
5266 | ||
5267 | if (bra == OP_BRAZERO) | if (bra == OP_BRAZERO) |
5268 | { | { |
5269 | fallback->hotpath = LABEL(); | backtrack->matchingpath = LABEL(); |
5270 | sljit_set_label(jump, fallback->hotpath); | sljit_set_label(jump, backtrack->matchingpath); |
5271 | } | } |
5272 | else if (bra == OP_BRAMINZERO) | else if (bra == OP_BRAMINZERO) |
5273 | { | { |
5274 | JUMPTO(SLJIT_JUMP, fallback->hotpath); | JUMPTO(SLJIT_JUMP, backtrack->matchingpath); |
5275 | JUMPHERE(brajump); | JUMPHERE(brajump); |
5276 | if (framesize >= 0) | if (framesize >= 0) |
5277 | { | { |
5278 | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); | OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); |
5279 | add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); |
5280 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); |
5281 | } | } |
5282 | set_jumps(fallback->common.topfallbacks, LABEL()); | set_jumps(backtrack->common.topbacktracks, LABEL()); |
5283 | } | } |
5284 | } | } |
5285 | else | else |
# | Line 4435 else | Line 5305 else |
5305 | } | } |
5306 | else | else |
5307 | free_stack(common, framesize + 2); | free_stack(common, framesize + 2); |
5308 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP1, 0); |
5309 | } | } |
5310 | ||
5311 | if (bra == OP_BRAZERO) | if (bra == OP_BRAZERO) |
5312 | fallback->hotpath = LABEL(); | backtrack->matchingpath = LABEL(); |
5313 | else if (bra == OP_BRAMINZERO) | else if (bra == OP_BRAMINZERO) |
5314 | { | { |
5315 | JUMPTO(SLJIT_JUMP, fallback->hotpath); | JUMPTO(SLJIT_JUMP, backtrack->matchingpath); |
5316 | JUMPHERE(brajump); | JUMPHERE(brajump); |
5317 | } | } |
5318 | ||
5319 | if (bra != OP_BRA) | if (bra != OP_BRA) |
5320 | { | { |
5321 | SLJIT_ASSERT(found == &fallback->common.topfallbacks); | SLJIT_ASSERT(found == &backtrack->common.topbacktracks); |
5322 | set_jumps(fallback->common.topfallbacks, LABEL()); | set_jumps(backtrack->common.topbacktracks, LABEL()); |
5323 | fallback->common.topfallbacks = NULL; | backtrack->common.topbacktracks = NULL; |
5324 | } | } |
5325 | } | } |
5326 | ||
5327 | common->leavelabel = save_leavelabel; | common->quitlabel = save_quitlabel; |
5328 | common->acceptlabel = save_acceptlabel; | common->acceptlabel = save_acceptlabel; |
5329 | common->leave = save_leave; | common->quit = save_quit; |
5330 | common->accept = save_accept; | common->accept = save_accept; |
5331 | return cc + 1 + LINK_SIZE; | return cc + 1 + LINK_SIZE; |
5332 | } | } |
# | Line 4625 return condition; | Line 5495 return condition; |
5495 | Or nothing, if trace is unnecessary | Or nothing, if trace is unnecessary |
5496 | */ | */ |
5497 | ||
5498 | static pcre_uchar *compile_bracket_hotpath(compiler_common *common, pcre_uchar *cc, fallback_common *parent) | static pcre_uchar *compile_bracket_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) |
5499 | { | { |
5500 | DEFINE_COMPILER; | DEFINE_COMPILER; |
5501 | fallback_common *fallback; | backtrack_common *backtrack; |
5502 | pcre_uchar opcode; | pcre_uchar opcode; |
5503 | int localptr = 0; | int private_data_ptr = 0; |
5504 | int offset = 0; | int offset = 0; |
5505 | int stacksize; | int stacksize; |
5506 | pcre_uchar *ccbegin; | pcre_uchar *ccbegin; |
5507 | pcre_uchar *hotpath; | pcre_uchar *matchingpath; |
5508 | pcre_uchar bra = OP_BRA; | pcre_uchar bra = OP_BRA; |
5509 | pcre_uchar ket; | pcre_uchar ket; |
5510 | assert_fallback *assert; | assert_backtrack *assert; |
5511 | BOOL has_alternatives; | BOOL has_alternatives; |
5512 | struct sljit_jump *jump; | struct sljit_jump *jump; |
5513 | struct sljit_jump *skip; | struct sljit_jump *skip; |
5514 | struct sljit_label *rmaxlabel = NULL; | struct sljit_label *rmaxlabel = NULL; |
5515 | struct sljit_jump *braminzerojump = NULL; | struct sljit_jump *braminzerojump = NULL; |
5516 | ||
5517 | PUSH_FALLBACK(sizeof(bracket_fallback), cc, NULL); | PUSH_BACKTRACK(sizeof(bracket_backtrack), cc, NULL); |
5518 | ||
5519 | if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) | if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) |
5520 | { | { |
# | Line 4655 if (*cc == OP_BRAZERO || *cc == OP_BRAMI | Line 5525 if (*cc == OP_BRAZERO || *cc == OP_BRAMI |
5525 | ||
5526 | opcode = *cc; | opcode = *cc; |
5527 | ccbegin = cc; | ccbegin = cc; |
5528 | hotpath = ccbegin + 1 + LINK_SIZE; | matchingpath = ccbegin + 1 + LINK_SIZE; |
5529 | ||
5530 | if ((opcode == OP_COND || opcode == OP_SCOND) && cc[1 + LINK_SIZE] == OP_DEF) | if ((opcode == OP_COND || opcode == OP_SCOND) && cc[1 + LINK_SIZE] == OP_DEF) |
5531 | { | { |
5532 | /* Drop this bracket_fallback. */ | /* Drop this bracket_backtrack. */ |
5533 | parent->top = fallback->prev; | parent->top = backtrack->prev; |
5534 | return bracketend(cc); | return bracketend(cc); |
5535 | } | } |
5536 | ||
# | Line 4672 cc += GET(cc, 1); | Line 5542 cc += GET(cc, 1); |
5542 | has_alternatives = *cc == OP_ALT; | has_alternatives = *cc == OP_ALT; |
5543 | if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) | if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) |
5544 | { | { |
5545 | has_alternatives = (*hotpath == OP_RREF) ? FALSE : TRUE; | has_alternatives = (*matchingpath == OP_RREF) ? FALSE : TRUE; |
5546 | if (*hotpath == OP_NRREF) | if (*matchingpath == OP_NRREF) |
5547 | { | { |
5548 | stacksize = GET2(hotpath, 1); | stacksize = GET2(matchingpath, 1); |
5549 | if (common->currententry == NULL || stacksize == RREF_ANY) | if (common->currententry == NULL || stacksize == RREF_ANY) |
5550 | has_alternatives = FALSE; | has_alternatives = FALSE; |
5551 | else if (common->currententry->start == 0) | else if (common->currententry->start == 0) |
# | Line 4694 if (opcode == OP_CBRA || opcode == OP_SC | Line 5564 if (opcode == OP_CBRA || opcode == OP_SC |
5564 | { | { |
5565 | /* Capturing brackets has a pre-allocated space. */ | /* Capturing brackets has a pre-allocated space. */ |
5566 | offset = GET2(ccbegin, 1 + LINK_SIZE); | offset = GET2(ccbegin, 1 + LINK_SIZE); |
5567 | localptr = OVECTOR_PRIV(offset); | if (common->optimized_cbracket[offset] == 0) |
5568 | offset <<= 1; | { |
5569 | FALLBACK_AS(bracket_fallback)->localptr = localptr; | private_data_ptr = OVECTOR_PRIV(offset); |
5570 | hotpath += IMM2_SIZE; | offset <<= 1; |
5571 | } | |
5572 | else | |
5573 | { | |
5574 | offset <<= 1; | |
5575 | private_data_ptr = OVECTOR(offset); | |
5576 | } | |
5577 | BACKTRACK_AS(bracket_backtrack)->private_data_ptr = private_data_ptr; | |
5578 | matchingpath += IMM2_SIZE; | |
5579 | } | } |
5580 | else if (opcode == OP_ONCE || opcode == OP_SBRA || opcode == OP_SCOND) | else if (opcode == OP_ONCE || opcode == OP_SBRA || opcode == OP_SCOND) |
5581 | { | { |
5582 | /* Other brackets simply allocate the next entry. */ | /* Other brackets simply allocate the next entry. */ |
5583 | localptr = PRIV_DATA(ccbegin); | private_data_ptr = PRIVATE_DATA(ccbegin); |
5584 | SLJIT_ASSERT(localptr != 0); | SLJIT_ASSERT(private_data_ptr != 0); |
5585 | FALLBACK_AS(bracket_fallback)->localptr = localptr; | BACKTRACK_AS(bracket_backtrack)->private_data_ptr = private_data_ptr; |
5586 | if (opcode == OP_ONCE) | if (opcode == OP_ONCE) |
5587 | FALLBACK_AS(bracket_fallback)->u.framesize = get_framesize(common, ccbegin, FALSE); | BACKTRACK_AS(bracket_backtrack)->u.framesize = get_framesize(common, ccbegin, FALSE); |
5588 | } | } |
5589 | ||
5590 | /* Instructions before the first alternative. */ | /* Instructions before the first alternative. */ |
# | Line 4731 if (bra == OP_BRAZERO) |
|