Parent Directory
|
Revision Log
|
Patch
revision 958 by zherczeg, Wed Apr 11 10:19:10 2012 UTC | revision 1252 by zherczeg, Fri Feb 22 11:13:38 2013 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. */ | /* Defines for debugging purposes. */ |
#define LOCAL_SPACE_SIZE 32768 | ||
69 | ||
70 | /* 1 - Use unoptimized capturing brackets. | |
71 | 2 - Enable capture_last_ptr (includes option 1). */ | |
72 | /* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */ | |
73 | ||
74 | /* Allocate memory for the regex stack on the real machine stack. | |
75 | Fast, but limited size. */ | |
76 | #define MACHINE_STACK_SIZE 32768 | |
77 | ||
78 | /* Growth rate for stack allocated by the OS. Should be the multiply | |
79 | of page size. */ | |
80 | #define STACK_GROWTH_RATE 8192 | #define STACK_GROWTH_RATE 8192 |
81 | ||
82 | /* 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 91 The code generator follows the recursive |
91 | expressions. The basic blocks of regular expressions are condition checkers | expressions. The basic blocks of regular expressions are condition checkers |
92 | whose execute different commands depending on the result of the condition check. | whose execute different commands depending on the result of the condition check. |
93 | The relationship between the operators can be horizontal (concatenation) and | The relationship between the operators can be horizontal (concatenation) and |
94 | vertical (sub-expression) (See struct fallback_common for more details). | vertical (sub-expression) (See struct backtrack_common for more details). |
95 | ||
96 | 'ab' - 'a' and 'b' regexps are concatenated | 'ab' - 'a' and 'b' regexps are concatenated |
97 | 'a+' - 'a' is the sub-expression of the '+' operator | 'a+' - 'a' is the sub-expression of the '+' operator |
98 | ||
99 | The condition checkers are boolean (true/false) checkers. Machine code is generated | The condition checkers are boolean (true/false) checkers. Machine code is generated |
100 | 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. |
101 | 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 |
102 | 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 |
103 | branches on the hot path. | branches on the matching path. |
104 | ||
105 | Greedy star operator (*) : | Greedy star operator (*) : |
106 | Hot path: match happens. | Matching path: match happens. |
107 | Fallback path: match failed. | Backtrack path: match failed. |
108 | Non-greedy star operator (*?) : | Non-greedy star operator (*?) : |
109 | Hot path: no need to perform a match. | Matching path: no need to perform a match. |
110 | Fallback path: match is required. | Backtrack path: match is required. |
111 | ||
112 | The following example shows how the code generated for a capturing bracket | The following example shows how the code generated for a capturing bracket |
113 | 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 117 we have the following regular expression |
117 | ||
118 | The generated code will be the following: | The generated code will be the following: |
119 | ||
120 | A hot path | A matching path |
121 | '(' hot path (pushing arguments to the stack) | '(' matching path (pushing arguments to the stack) |
122 | B hot path | B matching path |
123 | ')' hot path (pushing arguments to the stack) | ')' matching path (pushing arguments to the stack) |
124 | D hot path | D matching path |
125 | return with successful match | return with successful match |
126 | ||
127 | D fallback path | D backtrack path |
128 | ')' 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") |
129 | B fallback path | B backtrack path |
130 | C expected path | C expected path |
131 | jump to D hot path | jump to D matching path |
132 | C fallback path | C backtrack path |
133 | A fallback path | A backtrack path |
134 | ||
135 | 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 |
136 | 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 |
137 | to the current fallback code path. The fallback code path must check | to the current backtrack code path. The backtrack path must check |
138 | 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 |
139 | 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 |
140 | frame and continue the execution on the fallback code paths. | frame and continue the execution on the backtrack code paths. |
141 | */ | */ |
142 | ||
143 | /* | /* |
144 | Saved stack frames: | Saved stack frames: |
145 | ||
146 | Atomic blocks and asserts require reloading the values of local variables | Atomic blocks and asserts require reloading the values of private data |
147 | when the fallback mechanism performed. Because of OP_RECURSE, the locals | when the backtrack mechanism performed. Because of OP_RECURSE, the data |
148 | 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 |
149 | mechanism. | mechanism. |
150 | ||
151 | 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: |
152 | ([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] | ([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] |
153 | ||
154 | 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. |
155 | */ | */ |
156 | ||
157 | typedef struct jit_arguments { | typedef struct jit_arguments { |
# | Line 154 typedef struct jit_arguments { | Line 163 typedef struct jit_arguments { |
163 | int *offsets; | int *offsets; |
164 | pcre_uchar *uchar_ptr; | pcre_uchar *uchar_ptr; |
165 | pcre_uchar *mark_ptr; | pcre_uchar *mark_ptr; |
166 | void *callout_data; | |
167 | /* Everything else after. */ | /* Everything else after. */ |
168 | int offsetcount; | int real_offset_count; |
169 | int calllimit; | int offset_count; |
170 | int call_limit; | |
171 | pcre_uint8 notbol; | pcre_uint8 notbol; |
172 | pcre_uint8 noteol; | pcre_uint8 noteol; |
173 | pcre_uint8 notempty; | pcre_uint8 notempty; |
# | Line 167 typedef struct executable_functions { | Line 178 typedef struct executable_functions { |
178 | void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; | void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; |
179 | PUBL(jit_callback) callback; | PUBL(jit_callback) callback; |
180 | void *userdata; | void *userdata; |
181 | pcre_uint32 top_bracket; | |
182 | sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; | sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; |
183 | } executable_functions; | } executable_functions; |
184 | ||
# | Line 175 typedef struct jump_list { | Line 187 typedef struct jump_list { |
187 | struct jump_list *next; | struct jump_list *next; |
188 | } jump_list; | } jump_list; |
189 | ||
enum stub_types { stack_alloc }; | ||
190 | typedef struct stub_list { | typedef struct stub_list { |
enum stub_types type; | ||
int data; | ||
191 | struct sljit_jump *start; | struct sljit_jump *start; |
192 | struct sljit_label *leave; | struct sljit_label *quit; |
193 | struct stub_list *next; | struct stub_list *next; |
194 | } stub_list; | } stub_list; |
195 | ||
196 | enum frame_types { no_frame = -1, no_stack = -2 }; | |
197 | ||
198 | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); | typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); |
199 | ||
200 | /* The following structure is the key data type for the recursive | /* The following structure is the key data type for the recursive |
201 | code generator. It is allocated by compile_hotpath, and contains | code generator. It is allocated by compile_matchingpath, and contains |
202 | the aguments for compile_fallbackpath. Must be the first member | the aguments for compile_backtrackingpath. Must be the first member |
203 | of its descendants. */ | of its descendants. */ |
204 | typedef struct fallback_common { | typedef struct backtrack_common { |
205 | /* Concatenation stack. */ | /* Concatenation stack. */ |
206 | struct fallback_common *prev; | struct backtrack_common *prev; |
207 | jump_list *nextfallbacks; | jump_list *nextbacktracks; |
208 | /* Internal stack (for component operators). */ | /* Internal stack (for component operators). */ |
209 | struct fallback_common *top; | struct backtrack_common *top; |
210 | jump_list *topfallbacks; | jump_list *topbacktracks; |
211 | /* Opcode pointer. */ | /* Opcode pointer. */ |
212 | pcre_uchar *cc; | pcre_uchar *cc; |
213 | } fallback_common; | } backtrack_common; |
214 | ||
215 | typedef struct assert_fallback { | typedef struct assert_backtrack { |
216 | fallback_common common; | backtrack_common common; |
217 | jump_list *condfailed; | jump_list *condfailed; |
218 | /* Less than 0 (-1) if a frame is not needed. */ | /* Less than 0 (-1) if a frame is not needed. */ |
219 | int framesize; | int framesize; |
220 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
221 | int localptr; | int private_data_ptr; |
222 | /* For iterators. */ | /* For iterators. */ |
223 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
224 | } assert_fallback; | } assert_backtrack; |
225 | ||
226 | typedef struct bracket_fallback { | typedef struct bracket_backtrack { |
227 | fallback_common common; | backtrack_common common; |
228 | /* Where to coninue if an alternative is successfully matched. */ | /* Where to coninue if an alternative is successfully matched. */ |
229 | struct sljit_label *althotpath; | struct sljit_label *alternative_matchingpath; |
230 | /* For rmin and rmax iterators. */ | /* For rmin and rmax iterators. */ |
231 | struct sljit_label *recursivehotpath; | struct sljit_label *recursive_matchingpath; |
232 | /* For greedy ? operator. */ | /* For greedy ? operator. */ |
233 | struct sljit_label *zerohotpath; | struct sljit_label *zero_matchingpath; |
234 | /* Contains the branches of a failed condition. */ | /* Contains the branches of a failed condition. */ |
235 | union { | union { |
236 | /* Both for OP_COND, OP_SCOND. */ | /* Both for OP_COND, OP_SCOND. */ |
237 | jump_list *condfailed; | jump_list *condfailed; |
238 | assert_fallback *assert; | assert_backtrack *assert; |
239 | /* For OP_ONCE. -1 if not needed. */ | /* For OP_ONCE. -1 if not needed. */ |
240 | int framesize; | int framesize; |
241 | } u; | } u; |
242 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
243 | int localptr; | int private_data_ptr; |
244 | } bracket_fallback; | } bracket_backtrack; |
245 | ||
246 | typedef struct bracketpos_fallback { | typedef struct bracketpos_backtrack { |
247 | fallback_common common; | backtrack_common common; |
248 | /* Points to our private memory word on the stack. */ | /* Points to our private memory word on the stack. */ |
249 | int localptr; | int private_data_ptr; |
250 | /* Reverting stack is needed. */ | /* Reverting stack is needed. */ |
251 | int framesize; | int framesize; |
252 | /* Allocated stack size. */ | /* Allocated stack size. */ |
253 | int stacksize; | int stacksize; |
254 | } bracketpos_fallback; | } bracketpos_backtrack; |
255 | ||
256 | typedef struct braminzero_fallback { | typedef struct braminzero_backtrack { |
257 | fallback_common common; | backtrack_common common; |
258 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
259 | } braminzero_fallback; | } braminzero_backtrack; |
260 | ||
261 | typedef struct iterator_fallback { | typedef struct iterator_backtrack { |
262 | fallback_common common; | backtrack_common common; |
263 | /* Next iteration. */ | /* Next iteration. */ |
264 | struct sljit_label *hotpath; | struct sljit_label *matchingpath; |
265 | } iterator_fallback; | } iterator_backtrack; |
266 | ||
267 | typedef struct recurse_entry { | typedef struct recurse_entry { |
268 | struct recurse_entry *next; | struct recurse_entry *next; |
# | Line 264 typedef struct recurse_entry { | Line 274 typedef struct recurse_entry { |
274 | int start; | int start; |
275 | } recurse_entry; | } recurse_entry; |
276 | ||
277 | typedef struct recurse_fallback { | typedef struct recurse_backtrack { |
278 | fallback_common common; | backtrack_common common; |
279 | } recurse_fallback; | BOOL inlined_pattern; |
280 | } recurse_backtrack; | |
281 | ||
282 | #define MAX_RANGE_SIZE 6 | |
283 | ||
284 | typedef struct compiler_common { | typedef struct compiler_common { |
285 | struct sljit_compiler *compiler; | struct sljit_compiler *compiler; |
286 | pcre_uchar *start; | pcre_uchar *start; |
287 | ||
288 | /* Opcode local area direct map. */ | /* Maps private data offset to each opcode. */ |
289 | int *localptrs; | int *private_data_ptrs; |
290 | /* Tells whether the capturing bracket is optimized. */ | |
291 | pcre_uint8 *optimized_cbracket; | |
292 | /* Starting offset of private data for capturing brackets. */ | |
293 | int cbraptr; | int cbraptr; |
294 | /* OVector starting point. Must be divisible by 2. */ | /* OVector starting point. Must be divisible by 2. */ |
295 | int ovector_start; | int ovector_start; |
296 | /* Last known position of the requested byte. */ | /* Last known position of the requested byte. */ |
297 | int req_char_ptr; | int req_char_ptr; |
298 | /* Head of the last recursion. */ | /* Head of the last recursion. */ |
299 | int recursive_head; | int recursive_head_ptr; |
300 | /* First inspected character for partial matching. */ | /* First inspected character for partial matching. */ |
301 | int start_used_ptr; | int start_used_ptr; |
302 | /* Starting pointer for partial soft matches. */ | /* Starting pointer for partial soft matches. */ |
# | Line 289 typedef struct compiler_common { | Line 305 typedef struct compiler_common { |
305 | int first_line_end; | int first_line_end; |
306 | /* Points to the marked string. */ | /* Points to the marked string. */ |
307 | int mark_ptr; | int mark_ptr; |
308 | /* Points to the last matched capture block index. */ | |
309 | int capture_last_ptr; | |
310 | ||
311 | /* Other */ | /* Flipped and lower case tables. */ |
312 | const pcre_uint8 *fcc; | const pcre_uint8 *fcc; |
313 | sljit_w lcc; | sljit_sw lcc; |
314 | /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */ | |
315 | int mode; | int mode; |
316 | /* Newline control. */ | |
317 | int nltype; | int nltype; |
318 | int newline; | int newline; |
319 | int bsr_nltype; | int bsr_nltype; |
320 | /* Dollar endonly. */ | |
321 | int endonly; | int endonly; |
322 | BOOL has_set_som; | BOOL has_set_som; |
323 | sljit_w ctypes; | /* Tables. */ |
324 | sljit_sw ctypes; | |
325 | int digits[2 + MAX_RANGE_SIZE]; | |
326 | /* Named capturing brackets. */ | |
327 | sljit_uw name_table; | sljit_uw name_table; |
328 | sljit_w name_count; | sljit_sw name_count; |
329 | sljit_w name_entry_size; | sljit_sw name_entry_size; |
330 | ||
331 | /* Labels and jump lists. */ | /* Labels and jump lists. */ |
332 | struct sljit_label *partialmatchlabel; | struct sljit_label *partialmatchlabel; |
333 | struct sljit_label *leavelabel; | struct sljit_label *quit_label; |
334 | struct sljit_label *acceptlabel; | struct sljit_label *forced_quit_label; |
335 | struct sljit_label *accept_label; | |
336 | stub_list *stubs; | stub_list *stubs; |
337 | recurse_entry *entries; | recurse_entry *entries; |
338 | recurse_entry *currententry; | recurse_entry *currententry; |
339 | jump_list *partialmatch; | jump_list *partialmatch; |
340 | jump_list *leave; | jump_list *quit; |
341 | jump_list *forced_quit; | |
342 | jump_list *accept; | jump_list *accept; |
343 | jump_list *calllimit; | jump_list *calllimit; |
344 | jump_list *stackalloc; | jump_list *stackalloc; |
# | Line 329 typedef struct compiler_common { | Line 355 typedef struct compiler_common { |
355 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
356 | BOOL use_ucp; | BOOL use_ucp; |
357 | #endif | #endif |
358 | #ifndef COMPILE_PCRE32 | |
359 | jump_list *utfreadchar; | jump_list *utfreadchar; |
360 | #endif | |
361 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
362 | jump_list *utfreadtype8; | jump_list *utfreadtype8; |
363 | #endif | #endif |
# | Line 347 typedef struct compare_context { | Line 375 typedef struct compare_context { |
375 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
376 | int ucharptr; | int ucharptr; |
377 | union { | union { |
378 | sljit_i asint; | sljit_si 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 | } c; | } c; |
389 | union { | union { |
390 | sljit_i asint; | sljit_si asint; |
391 | sljit_uh asushort; | sljit_uh asushort; |
392 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
393 | sljit_ub asbyte; | sljit_ub asbyte; |
394 | sljit_ub asuchars[4]; | sljit_ub asuchars[4]; |
395 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
396 | sljit_uh asuchars[2]; | sljit_uh asuchars[2]; |
397 | #endif | #elif defined COMPILE_PCRE32 |
398 | sljit_ui asuchars[1]; | |
399 | #endif | #endif |
400 | } oc; | } oc; |
401 | #endif | #endif |
402 | } compare_context; | } compare_context; |
403 | ||
enum { | ||
frame_end = 0, | ||
frame_setstrbegin = -1, | ||
frame_setmark = -2 | ||
}; | ||
404 | /* Undefine sljit macros. */ | /* Undefine sljit macros. */ |
405 | #undef CMP | #undef CMP |
406 | ||
407 | /* Used for accessing the elements of the stack. */ | /* Used for accessing the elements of the stack. */ |
408 | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_w)) | #define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_sw)) |
409 | ||
410 | #define TMP1 SLJIT_TEMPORARY_REG1 | #define TMP1 SLJIT_SCRATCH_REG1 |
411 | #define TMP2 SLJIT_TEMPORARY_REG3 | #define TMP2 SLJIT_SCRATCH_REG3 |
412 | #define TMP3 SLJIT_TEMPORARY_EREG2 | #define TMP3 SLJIT_TEMPORARY_EREG2 |
413 | #define STR_PTR SLJIT_SAVED_REG1 | #define STR_PTR SLJIT_SAVED_REG1 |
414 | #define STR_END SLJIT_SAVED_REG2 | #define STR_END SLJIT_SAVED_REG2 |
415 | #define STACK_TOP SLJIT_TEMPORARY_REG2 | #define STACK_TOP SLJIT_SCRATCH_REG2 |
416 | #define STACK_LIMIT SLJIT_SAVED_REG3 | #define STACK_LIMIT SLJIT_SAVED_REG3 |
417 | #define ARGUMENTS SLJIT_SAVED_EREG1 | #define ARGUMENTS SLJIT_SAVED_EREG1 |
418 | #define CALL_COUNT SLJIT_SAVED_EREG2 | #define CALL_COUNT SLJIT_SAVED_EREG2 |
419 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 | #define RETURN_ADDR SLJIT_TEMPORARY_EREG1 |
420 | ||
421 | /* Locals layout. */ | /* Local space layout. */ |
422 | /* These two locals can be used by the current opcode. */ | /* These two locals can be used by the current opcode. */ |
423 | #define LOCALS0 (0 * sizeof(sljit_w)) | #define LOCALS0 (0 * sizeof(sljit_sw)) |
424 | #define LOCALS1 (1 * sizeof(sljit_w)) | #define LOCALS1 (1 * sizeof(sljit_sw)) |
425 | /* Two local variables for possessive quantifiers (char1 cannot use them). */ | /* Two local variables for possessive quantifiers (char1 cannot use them). */ |
426 | #define POSSESSIVE0 (2 * sizeof(sljit_w)) | #define POSSESSIVE0 (2 * sizeof(sljit_sw)) |
427 | #define POSSESSIVE1 (3 * sizeof(sljit_w)) | #define POSSESSIVE1 (3 * sizeof(sljit_sw)) |
428 | /* Max limit of recursions. */ | /* Max limit of recursions. */ |
429 | #define CALL_LIMIT (4 * sizeof(sljit_w)) | #define CALL_LIMIT (4 * sizeof(sljit_sw)) |
430 | /* The output vector is stored on the stack, and contains pointers | /* The output vector is stored on the stack, and contains pointers |
431 | to characters. The vector data is divided into two groups: the first | to characters. The vector data is divided into two groups: the first |
432 | group contains the start / end character pointers, and the second is | group contains the start / end character pointers, and the second is |
433 | the start pointers when the end of the capturing group has not yet reached. */ | the start pointers when the end of the capturing group has not yet reached. */ |
434 | #define OVECTOR_START (common->ovector_start) | #define OVECTOR_START (common->ovector_start) |
435 | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_w)) | #define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_sw)) |
436 | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_w)) | #define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_sw)) |
437 | #define PRIV_DATA(cc) (common->localptrs[(cc) - common->start]) | #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start]) |
438 | ||
439 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
440 | #define MOV_UCHAR SLJIT_MOV_UB | #define MOV_UCHAR SLJIT_MOV_UB |
441 | #define MOVU_UCHAR SLJIT_MOVU_UB | #define MOVU_UCHAR SLJIT_MOVU_UB |
442 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
443 | #define MOV_UCHAR SLJIT_MOV_UH | #define MOV_UCHAR SLJIT_MOV_UH |
444 | #define MOVU_UCHAR SLJIT_MOVU_UH | #define MOVU_UCHAR SLJIT_MOVU_UH |
445 | #elif defined COMPILE_PCRE32 | |
446 | #define MOV_UCHAR SLJIT_MOV_UI | |
447 | #define MOVU_UCHAR SLJIT_MOVU_UI | |
448 | #else | #else |
449 | #error Unsupported compiling mode | #error Unsupported compiling mode |
450 | #endif | #endif |
#endif | ||
451 | ||
452 | /* Shortcuts. */ | /* Shortcuts. */ |
453 | #define DEFINE_COMPILER \ | #define DEFINE_COMPILER \ |
# | Line 441 the start pointers when the end of the c | Line 464 the start pointers when the end of the c |
464 | sljit_set_label(sljit_emit_jump(compiler, (type)), (label)) | sljit_set_label(sljit_emit_jump(compiler, (type)), (label)) |
465 | #define JUMPHERE(jump) \ | #define JUMPHERE(jump) \ |
466 | sljit_set_label((jump), sljit_emit_label(compiler)) | sljit_set_label((jump), sljit_emit_label(compiler)) |
467 | #define SET_LABEL(jump, label) \ | |
468 | sljit_set_label((jump), (label)) | |
469 | #define CMP(type, src1, src1w, src2, src2w) \ | #define CMP(type, src1, src1w, src2, src2w) \ |
470 | sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) | sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) |
471 | #define CMPTO(type, src1, src1w, src2, src2w, label) \ | #define CMPTO(type, src1, src1w, src2, src2w, label) \ |
472 | sljit_set_label(sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)), (label)) | sljit_set_label(sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)), (label)) |
473 | #define COND_VALUE(op, dst, dstw, type) \ | #define OP_FLAGS(op, dst, dstw, src, srcw, type) \ |
474 | sljit_emit_cond_value(compiler, (op), (dst), (dstw), (type)) | sljit_emit_op_flags(compiler, (op), (dst), (dstw), (src), (srcw), (type)) |
475 | #define GET_LOCAL_BASE(dst, dstw, offset) \ | #define GET_LOCAL_BASE(dst, dstw, offset) \ |
476 | sljit_get_local_base(compiler, (dst), (dstw), (offset)) | sljit_get_local_base(compiler, (dst), (dstw), (offset)) |
477 | ||
# | Line 461 return cc; | Line 486 return cc; |
486 | ||
487 | /* Functions whose might need modification for all new supported opcodes: | /* Functions whose might need modification for all new supported opcodes: |
488 | next_opcode | next_opcode |
489 | get_localspace | get_private_data_length |
490 | set_localptrs | set_private_data_ptrs |
491 | get_framesize | get_framesize |
492 | init_frame | init_frame |
493 | get_localsize | get_private_data_length_for_copy |
494 | copy_locals | copy_private_data |
495 | compile_hotpath | compile_matchingpath |
496 | compile_fallbackpath | compile_backtrackingpath |
497 | */ | */ |
498 | ||
499 | static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) | static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) |
# | Line 489 switch(*cc) | Line 514 switch(*cc) |
514 | case OP_WORDCHAR: | case OP_WORDCHAR: |
515 | case OP_ANY: | case OP_ANY: |
516 | case OP_ALLANY: | case OP_ALLANY: |
517 | case OP_NOTPROP: | |
518 | case OP_PROP: | |
519 | case OP_ANYNL: | case OP_ANYNL: |
520 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
521 | case OP_HSPACE: | case OP_HSPACE: |
# | Line 501 switch(*cc) | Line 528 switch(*cc) |
528 | case OP_CIRCM: | case OP_CIRCM: |
529 | case OP_DOLL: | case OP_DOLL: |
530 | case OP_DOLLM: | case OP_DOLLM: |
case OP_TYPESTAR: | ||
case OP_TYPEMINSTAR: | ||
case OP_TYPEPLUS: | ||
case OP_TYPEMINPLUS: | ||
case OP_TYPEQUERY: | ||
case OP_TYPEMINQUERY: | ||
case OP_TYPEPOSSTAR: | ||
case OP_TYPEPOSPLUS: | ||
case OP_TYPEPOSQUERY: | ||
531 | case OP_CRSTAR: | case OP_CRSTAR: |
532 | case OP_CRMINSTAR: | case OP_CRMINSTAR: |
533 | case OP_CRPLUS: | case OP_CRPLUS: |
534 | case OP_CRMINPLUS: | case OP_CRMINPLUS: |
535 | case OP_CRQUERY: | case OP_CRQUERY: |
536 | case OP_CRMINQUERY: | case OP_CRMINQUERY: |
537 | case OP_CRRANGE: | |
538 | case OP_CRMINRANGE: | |
539 | case OP_CLASS: | |
540 | case OP_NCLASS: | |
541 | case OP_REF: | |
542 | case OP_REFI: | |
543 | case OP_RECURSE: | |
544 | case OP_CALLOUT: | |
545 | case OP_ALT: | |
546 | case OP_KET: | |
547 | case OP_KETRMAX: | |
548 | case OP_KETRMIN: | |
549 | case OP_KETRPOS: | |
550 | case OP_REVERSE: | |
551 | case OP_ASSERT: | |
552 | case OP_ASSERT_NOT: | |
553 | case OP_ASSERTBACK: | |
554 | case OP_ASSERTBACK_NOT: | |
555 | case OP_ONCE: | |
556 | case OP_ONCE_NC: | |
557 | case OP_BRA: | |
558 | case OP_BRAPOS: | |
559 | case OP_CBRA: | |
560 | case OP_CBRAPOS: | |
561 | case OP_COND: | |
562 | case OP_SBRA: | |
563 | case OP_SBRAPOS: | |
564 | case OP_SCBRA: | |
565 | case OP_SCBRAPOS: | |
566 | case OP_SCOND: | |
567 | case OP_CREF: | |
568 | case OP_NCREF: | |
569 | case OP_RREF: | |
570 | case OP_NRREF: | |
571 | case OP_DEF: | case OP_DEF: |
572 | case OP_BRAZERO: | case OP_BRAZERO: |
573 | case OP_BRAMINZERO: | case OP_BRAMINZERO: |
# | Line 524 switch(*cc) | Line 576 switch(*cc) |
576 | case OP_FAIL: | case OP_FAIL: |
577 | case OP_ACCEPT: | case OP_ACCEPT: |
578 | case OP_ASSERT_ACCEPT: | case OP_ASSERT_ACCEPT: |
579 | case OP_CLOSE: | |
580 | case OP_SKIPZERO: | case OP_SKIPZERO: |
581 | return cc + 1; | return cc + PRIV(OP_lengths)[*cc]; |
case OP_ANYBYTE: | ||
#ifdef SUPPORT_UTF | ||
if (common->utf) return NULL; | ||
#endif | ||
return cc + 1; | ||
582 | ||
583 | case OP_CHAR: | case OP_CHAR: |
584 | case OP_CHARI: | case OP_CHARI: |
# | Line 543 switch(*cc) | Line 590 switch(*cc) |
590 | case OP_MINPLUS: | case OP_MINPLUS: |
591 | case OP_QUERY: | case OP_QUERY: |
592 | case OP_MINQUERY: | case OP_MINQUERY: |
593 | case OP_UPTO: | |
594 | case OP_MINUPTO: | |
595 | case OP_EXACT: | |
596 | case OP_POSSTAR: | case OP_POSSTAR: |
597 | case OP_POSPLUS: | case OP_POSPLUS: |
598 | case OP_POSQUERY: | case OP_POSQUERY: |
599 | case OP_POSUPTO: | |
600 | case OP_STARI: | case OP_STARI: |
601 | case OP_MINSTARI: | case OP_MINSTARI: |
602 | case OP_PLUSI: | case OP_PLUSI: |
603 | case OP_MINPLUSI: | case OP_MINPLUSI: |
604 | case OP_QUERYI: | case OP_QUERYI: |
605 | case OP_MINQUERYI: | case OP_MINQUERYI: |
606 | case OP_UPTOI: | |
607 | case OP_MINUPTOI: | |
608 | case OP_EXACTI: | |
609 | case OP_POSSTARI: | case OP_POSSTARI: |
610 | case OP_POSPLUSI: | case OP_POSPLUSI: |
611 | case OP_POSQUERYI: | case OP_POSQUERYI: |
612 | case OP_POSUPTOI: | |
613 | case OP_NOTSTAR: | case OP_NOTSTAR: |
614 | case OP_NOTMINSTAR: | case OP_NOTMINSTAR: |
615 | case OP_NOTPLUS: | case OP_NOTPLUS: |
616 | case OP_NOTMINPLUS: | case OP_NOTMINPLUS: |
617 | case OP_NOTQUERY: | case OP_NOTQUERY: |
618 | case OP_NOTMINQUERY: | case OP_NOTMINQUERY: |
619 | case OP_NOTUPTO: | |
620 | case OP_NOTMINUPTO: | |
621 | case OP_NOTEXACT: | |
622 | case OP_NOTPOSSTAR: | case OP_NOTPOSSTAR: |
623 | case OP_NOTPOSPLUS: | case OP_NOTPOSPLUS: |
624 | case OP_NOTPOSQUERY: | case OP_NOTPOSQUERY: |
625 | case OP_NOTPOSUPTO: | |
626 | case OP_NOTSTARI: | case OP_NOTSTARI: |
627 | case OP_NOTMINSTARI: | case OP_NOTMINSTARI: |
628 | case OP_NOTPLUSI: | case OP_NOTPLUSI: |
629 | case OP_NOTMINPLUSI: | case OP_NOTMINPLUSI: |
630 | case OP_NOTQUERYI: | case OP_NOTQUERYI: |
631 | case OP_NOTMINQUERYI: | case OP_NOTMINQUERYI: |
case OP_NOTPOSSTARI: | ||
case OP_NOTPOSPLUSI: | ||
case OP_NOTPOSQUERYI: | ||
cc += 2; | ||
#ifdef SUPPORT_UTF | ||
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | ||
#endif | ||
return cc; | ||
case OP_UPTO: | ||
case OP_MINUPTO: | ||
case OP_EXACT: | ||
case OP_POSUPTO: | ||
case OP_UPTOI: | ||
case OP_MINUPTOI: | ||
case OP_EXACTI: | ||
case OP_POSUPTOI: | ||
case OP_NOTUPTO: | ||
case OP_NOTMINUPTO: | ||
case OP_NOTEXACT: | ||
case OP_NOTPOSUPTO: | ||
632 | case OP_NOTUPTOI: | case OP_NOTUPTOI: |
633 | case OP_NOTMINUPTOI: | case OP_NOTMINUPTOI: |
634 | case OP_NOTEXACTI: | case OP_NOTEXACTI: |
635 | case OP_NOTPOSSTARI: | |
636 | case OP_NOTPOSPLUSI: | |
637 | case OP_NOTPOSQUERYI: | |
638 | case OP_NOTPOSUPTOI: | case OP_NOTPOSUPTOI: |
639 | cc += 2 + IMM2_SIZE; | cc += PRIV(OP_lengths)[*cc]; |
640 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
641 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); |
642 | #endif | #endif |
643 | return cc; | return cc; |
644 | ||
645 | case OP_NOTPROP: | /* Special cases. */ |
646 | case OP_PROP: | case OP_TYPESTAR: |
647 | return cc + 1 + 2; | case OP_TYPEMINSTAR: |
648 | case OP_TYPEPLUS: | |
649 | case OP_TYPEMINPLUS: | |
650 | case OP_TYPEQUERY: | |
651 | case OP_TYPEMINQUERY: | |
652 | case OP_TYPEUPTO: | case OP_TYPEUPTO: |
653 | case OP_TYPEMINUPTO: | case OP_TYPEMINUPTO: |
654 | case OP_TYPEEXACT: | case OP_TYPEEXACT: |
655 | case OP_TYPEPOSSTAR: | |
656 | case OP_TYPEPOSPLUS: | |
657 | case OP_TYPEPOSQUERY: | |
658 | case OP_TYPEPOSUPTO: | case OP_TYPEPOSUPTO: |
659 | case OP_REF: | return cc + PRIV(OP_lengths)[*cc] - 1; |
case OP_REFI: | ||
case OP_CREF: | ||
case OP_NCREF: | ||
case OP_RREF: | ||
case OP_NRREF: | ||
case OP_CLOSE: | ||
cc += 1 + IMM2_SIZE; | ||
return cc; | ||
case OP_CRRANGE: | ||
case OP_CRMINRANGE: | ||
return cc + 1 + 2 * IMM2_SIZE; | ||
660 | ||
661 | case OP_CLASS: | case OP_ANYBYTE: |
662 | case OP_NCLASS: | #ifdef SUPPORT_UTF |
663 | return cc + 1 + 32 / sizeof(pcre_uchar); | if (common->utf) return NULL; |
664 | #endif | |
665 | return cc + 1; | |
666 | ||
667 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
668 | case OP_XCLASS: | case OP_XCLASS: |
669 | return cc + GET(cc, 1); | return cc + GET(cc, 1); |
670 | #endif | #endif |
671 | ||
case OP_RECURSE: | ||
case OP_ASSERT: | ||
case OP_ASSERT_NOT: | ||
case OP_ASSERTBACK: | ||
case OP_ASSERTBACK_NOT: | ||
case OP_REVERSE: | ||
case OP_ONCE: | ||
case OP_ONCE_NC: | ||
case OP_BRA: | ||
case OP_BRAPOS: | ||
case OP_COND: | ||
case OP_SBRA: | ||
case OP_SBRAPOS: | ||
case OP_SCOND: | ||
case OP_ALT: | ||
case OP_KET: | ||
case OP_KETRMAX: | ||
case OP_KETRMIN: | ||
case OP_KETRPOS: | ||
return cc + 1 + LINK_SIZE; | ||
case OP_CBRA: | ||
case OP_CBRAPOS: | ||
case OP_SCBRA: | ||
case OP_SCBRAPOS: | ||
return cc + 1 + LINK_SIZE + IMM2_SIZE; | ||
672 | case OP_MARK: | case OP_MARK: |
673 | return cc + 1 + 2 + cc[1]; | return cc + 1 + 2 + cc[1]; |
674 | ||
# | Line 667 switch(*cc) | Line 677 switch(*cc) |
677 | } | } |
678 | } | } |
679 | ||
680 | static int get_localspace(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | #define CASE_ITERATOR_PRIVATE_DATA_1 \ |
681 | case OP_MINSTAR: \ | |
682 | case OP_MINPLUS: \ | |
683 | case OP_QUERY: \ | |
684 | case OP_MINQUERY: \ | |
685 | case OP_MINSTARI: \ | |
686 | case OP_MINPLUSI: \ | |
687 | case OP_QUERYI: \ | |
688 | case OP_MINQUERYI: \ | |
689 | case OP_NOTMINSTAR: \ | |
690 | case OP_NOTMINPLUS: \ | |
691 | case OP_NOTQUERY: \ | |
692 | case OP_NOTMINQUERY: \ | |
693 | case OP_NOTMINSTARI: \ | |
694 | case OP_NOTMINPLUSI: \ | |
695 | case OP_NOTQUERYI: \ | |
696 | case OP_NOTMINQUERYI: | |
697 | ||
698 | #define CASE_ITERATOR_PRIVATE_DATA_2A \ | |
699 | case OP_STAR: \ | |
700 | case OP_PLUS: \ | |
701 | case OP_STARI: \ | |
702 | case OP_PLUSI: \ | |
703 | case OP_NOTSTAR: \ | |
704 | case OP_NOTPLUS: \ | |
705 | case OP_NOTSTARI: \ | |
706 | case OP_NOTPLUSI: | |
707 | ||
708 | #define CASE_ITERATOR_PRIVATE_DATA_2B \ | |
709 | case OP_UPTO: \ | |
710 | case OP_MINUPTO: \ | |
711 | case OP_UPTOI: \ | |
712 | case OP_MINUPTOI: \ | |
713 | case OP_NOTUPTO: \ | |
714 | case OP_NOTMINUPTO: \ | |
715 | case OP_NOTUPTOI: \ | |
716 | case OP_NOTMINUPTOI: | |
717 | ||
718 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \ | |
719 | case OP_TYPEMINSTAR: \ | |
720 | case OP_TYPEMINPLUS: \ | |
721 | case OP_TYPEQUERY: \ | |
722 | case OP_TYPEMINQUERY: | |
723 | ||
724 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \ | |
725 | case OP_TYPESTAR: \ | |
726 | case OP_TYPEPLUS: | |
727 | ||
728 | #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \ | |
729 | case OP_TYPEUPTO: \ | |
730 | case OP_TYPEMINUPTO: | |
731 | ||
732 | static int get_class_iterator_size(pcre_uchar *cc) | |
733 | { | { |
734 | int localspace = 0; | switch(*cc) |
735 | { | |
736 | case OP_CRSTAR: | |
737 | case OP_CRPLUS: | |
738 | return 2; | |
739 | ||
740 | case OP_CRMINSTAR: | |
741 | case OP_CRMINPLUS: | |
742 | case OP_CRQUERY: | |
743 | case OP_CRMINQUERY: | |
744 | return 1; | |
745 | ||
746 | case OP_CRRANGE: | |
747 | case OP_CRMINRANGE: | |
748 | if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE)) | |
749 | return 0; | |
750 | return 2; | |
751 | ||
752 | default: | |
753 | return 0; | |
754 | } | |
755 | } | |
756 | ||
757 | static int get_private_data_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) | |
758 | { | |
759 | int private_data_length = 0; | |
760 | pcre_uchar *alternative; | pcre_uchar *alternative; |
761 | pcre_uchar *name; | |
762 | pcre_uchar *end = NULL; | |
763 | int space, size, i; | |
764 | pcre_uint32 bracketlen; | |
765 | ||
766 | /* 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. */ |
767 | while (cc < ccend) | while (cc < ccend) |
768 | { | { |
769 | space = 0; | |
770 | size = 0; | |
771 | bracketlen = 0; | |
772 | switch(*cc) | switch(*cc) |
773 | { | { |
774 | case OP_SET_SOM: | case OP_SET_SOM: |
# | Line 681 while (cc < ccend) | Line 776 while (cc < ccend) |
776 | cc += 1; | cc += 1; |
777 | break; | break; |
778 | ||
779 | case OP_REF: | |
780 | case OP_REFI: | |
781 | common->optimized_cbracket[GET2(cc, 1)] = 0; | |
782 | cc += 1 + IMM2_SIZE; | |
783 | break; | |
784 | ||
785 | case OP_ASSERT: | case OP_ASSERT: |
786 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
787 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
# | Line 690 while (cc < ccend) | Line 791 while (cc < ccend) |
791 | case OP_BRAPOS: | case OP_BRAPOS: |
792 | case OP_SBRA: | case OP_SBRA: |
793 | case OP_SBRAPOS: | case OP_SBRAPOS: |
794 | case OP_SCOND: | private_data_length += sizeof(sljit_sw); |
795 | localspace += sizeof(sljit_w); | bracketlen = 1 + LINK_SIZE; |
cc += 1 + LINK_SIZE; | ||
796 | break; | break; |
797 | ||
798 | case OP_CBRAPOS: | case OP_CBRAPOS: |
799 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
800 | localspace += sizeof(sljit_w); | private_data_length += sizeof(sljit_sw); |
801 | cc += 1 + LINK_SIZE + IMM2_SIZE; | common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0; |
802 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
803 | break; | break; |
804 | ||
805 | case OP_COND: | case OP_COND: |
806 | /* Might be a hidden SCOND. */ | case OP_SCOND: |
807 | alternative = cc + GET(cc, 1); | /* Only AUTO_CALLOUT can insert this opcode. We do |
808 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | not intend to support this case. */ |
809 | localspace += sizeof(sljit_w); | if (cc[1 + LINK_SIZE] == OP_CALLOUT) |
810 | cc += 1 + LINK_SIZE; | return -1; |
811 | ||
812 | if (*cc == OP_COND) | |
813 | { | |
814 | /* Might be a hidden SCOND. */ | |
815 | alternative = cc + GET(cc, 1); | |
816 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | |
817 | private_data_length += sizeof(sljit_sw); | |
818 | } | |
819 | else | |
820 | private_data_length += sizeof(sljit_sw); | |
821 | bracketlen = 1 + LINK_SIZE; | |
822 | break; | |
823 | ||
824 | case OP_CREF: | |
825 | i = GET2(cc, 1); | |
826 | common->optimized_cbracket[i] = 0; | |
827 | cc += 1 + IMM2_SIZE; | |
828 | break; | |
829 | ||
830 | case OP_NCREF: | |
831 | bracketlen = GET2(cc, 1); | |
832 | name = (pcre_uchar *)common->name_table; | |
833 | alternative = name; | |
834 | for (i = 0; i < common->name_count; i++) | |
835 | { | |
836 | if (GET2(name, 0) == bracketlen) break; | |
837 | name += common->name_entry_size; | |
838 | } | |
839 | SLJIT_ASSERT(i != common->name_count); | |
840 | ||
841 | for (i = 0; i < common->name_count; i++) | |
842 | { | |
843 | if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0) | |
844 | common->optimized_cbracket[GET2(alternative, 0)] = 0; | |
845 | alternative += common->name_entry_size; | |
846 | } | |
847 | bracketlen = 0; | |
848 | cc += 1 + IMM2_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; | |
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; | 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_ptr == 0) |
908 | { | { |
909 | common->recursive_head = common->ovector_start; | common->recursive_head_ptr = common->ovector_start; |
910 | common->ovector_start += sizeof(sljit_w); | common->ovector_start += sizeof(sljit_sw); |
911 | } | } |
912 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
913 | break; | break; |
914 | ||
915 | case OP_CALLOUT: | |
916 | if (common->capture_last_ptr == 0) | |
917 | { | |
918 | common->capture_last_ptr = common->ovector_start; | |
919 | common->ovector_start += sizeof(sljit_sw); | |
920 | } | |
921 | cc += 2 + 2 * LINK_SIZE; | |
922 | break; | |
923 | ||
924 | case OP_MARK: | case OP_MARK: |
925 | if (common->mark_ptr == 0) | if (common->mark_ptr == 0) |
926 | { | { |
927 | common->mark_ptr = common->ovector_start; | common->mark_ptr = common->ovector_start; |
928 | common->ovector_start += sizeof(sljit_w); | common->ovector_start += sizeof(sljit_sw); |
929 | } | } |
930 | cc += 1 + 2 + cc[1]; | cc += 1 + 2 + cc[1]; |
931 | break; | break; |
# | Line 734 while (cc < ccend) | Line 936 while (cc < ccend) |
936 | return -1; | return -1; |
937 | break; | break; |
938 | } | } |
939 | ||
940 | if (space > 0 && cc >= end) | |
941 | private_data_length += sizeof(sljit_sw) * space; | |
942 | ||
943 | if (size != 0) | |
944 | { | |
945 | if (size < 0) | |
946 | { | |
947 | cc += -size; | |
948 | #ifdef SUPPORT_UTF | |
949 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
950 | #endif | |
951 | } | |
952 | else | |
953 | cc += size; | |
954 | } | |
955 | ||
956 | if (bracketlen != 0) | |
957 | { | |
958 | if (cc >= end) | |
959 | { | |
960 | end = bracketend(cc); | |
961 | if (end[-1 - LINK_SIZE] == OP_KET) | |
962 | end = NULL; | |
963 | } | |
964 | cc += bracketlen; | |
965 | } | |
966 | } | } |
967 | return localspace; | return private_data_length; |
968 | } | } |
969 | ||
970 | 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) |
971 | { | { |
972 | pcre_uchar *cc = common->start; | pcre_uchar *cc = common->start; |
973 | pcre_uchar *alternative; | pcre_uchar *alternative; |
974 | pcre_uchar *end = NULL; | |
975 | int space, size, bracketlen; | |
976 | ||
977 | while (cc < ccend) | while (cc < ccend) |
978 | { | { |
979 | space = 0; | |
980 | size = 0; | |
981 | bracketlen = 0; | |
982 | switch(*cc) | switch(*cc) |
983 | { | { |
984 | case OP_ASSERT: | case OP_ASSERT: |
# | Line 756 while (cc < ccend) | Line 991 while (cc < ccend) |
991 | case OP_SBRA: | case OP_SBRA: |
992 | case OP_SBRAPOS: | case OP_SBRAPOS: |
993 | case OP_SCOND: | case OP_SCOND: |
994 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
995 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_sw); |
996 | cc += 1 + LINK_SIZE; | bracketlen = 1 + LINK_SIZE; |
997 | break; | break; |
998 | ||
999 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1000 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1001 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
1002 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_sw); |
1003 | cc += 1 + LINK_SIZE + IMM2_SIZE; | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; |
1004 | break; | break; |
1005 | ||
1006 | case OP_COND: | case OP_COND: |
# | Line 773 while (cc < ccend) | Line 1008 while (cc < ccend) |
1008 | alternative = cc + GET(cc, 1); | alternative = cc + GET(cc, 1); |
1009 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1010 | { | { |
1011 | common->localptrs[cc - common->start] = localptr; | common->private_data_ptrs[cc - common->start] = private_data_ptr; |
1012 | localptr += sizeof(sljit_w); | private_data_ptr += sizeof(sljit_sw); |
1013 | } | } |
1014 | cc += 1 + LINK_SIZE; | bracketlen = 1 + LINK_SIZE; |
1015 | break; | |
1016 | ||
1017 | case OP_BRA: | |
1018 | bracketlen = 1 + LINK_SIZE; | |
1019 | break; | |
1020 | ||
1021 | case OP_CBRA: | |
1022 | case OP_SCBRA: | |
1023 | bracketlen = 1 + LINK_SIZE + IMM2_SIZE; | |
1024 | break; | |
1025 | ||
1026 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1027 | space = 1; | |
1028 | size = -2; | |
1029 | break; | |
1030 | ||
1031 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1032 | space = 2; | |
1033 | size = -2; | |
1034 | break; | |
1035 | ||
1036 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1037 | space = 2; | |
1038 | size = -(2 + IMM2_SIZE); | |
1039 | break; | |
1040 | ||
1041 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1042 | space = 1; | |
1043 | size = 1; | |
1044 | break; | |
1045 | ||
1046 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1047 | if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI) | |
1048 | space = 2; | |
1049 | size = 1; | |
1050 | break; | |
1051 | ||
1052 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1053 | if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) | |
1054 | space = 2; | |
1055 | size = 1 + IMM2_SIZE; | |
1056 | break; | |
1057 | ||
1058 | case OP_CLASS: | |
1059 | case OP_NCLASS: | |
1060 | size += 1 + 32 / sizeof(pcre_uchar); | |
1061 | space = get_class_iterator_size(cc + size); | |
1062 | break; | |
1063 | ||
1064 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1065 | case OP_XCLASS: | |
1066 | size = GET(cc, 1); | |
1067 | space = get_class_iterator_size(cc + size); | |
1068 | break; | break; |
1069 | #endif | |
1070 | ||
1071 | default: | default: |
1072 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1073 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1074 | break; | break; |
1075 | } | } |
1076 | ||
1077 | if (space > 0 && cc >= end) | |
1078 | { | |
1079 | common->private_data_ptrs[cc - common->start] = private_data_ptr; | |
1080 | private_data_ptr += sizeof(sljit_sw) * space; | |
1081 | } | |
1082 | ||
1083 | if (size != 0) | |
1084 | { | |
1085 | if (size < 0) | |
1086 | { | |
1087 | cc += -size; | |
1088 | #ifdef SUPPORT_UTF | |
1089 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1090 | #endif | |
1091 | } | |
1092 | else | |
1093 | cc += size; | |
1094 | } | |
1095 | ||
1096 | if (bracketlen > 0) | |
1097 | { | |
1098 | if (cc >= end) | |
1099 | { | |
1100 | end = bracketend(cc); | |
1101 | if (end[-1 - LINK_SIZE] == OP_KET) | |
1102 | end = NULL; | |
1103 | } | |
1104 | cc += bracketlen; | |
1105 | } | |
1106 | } | } |
1107 | } | } |
1108 | ||
1109 | /* Returns with -1 if no need for frame. */ | /* Returns with a frame_types (always < 0) if no need for frame. */ |
1110 | static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive) | static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive) |
1111 | { | { |
1112 | pcre_uchar *ccend = bracketend(cc); | pcre_uchar *ccend = bracketend(cc) - (1 + LINK_SIZE); |
1113 | int length = 0; | int length = 0; |
1114 | BOOL possessive = FALSE; | int possessive = 0; |
1115 | BOOL stack_restore = FALSE; | |
1116 | BOOL setsom_found = recursive; | BOOL setsom_found = recursive; |
1117 | BOOL setmark_found = recursive; | BOOL setmark_found = recursive; |
1118 | /* The last capture is a local variable even for recursions. */ | |
1119 | BOOL capture_last_found = FALSE; | |
1120 | ||
1121 | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) | if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) |
1122 | { | { |
1123 | length = 3; | possessive = length = (common->capture_last_ptr != 0) ? 5 : 3; |
1124 | possessive = TRUE; | /* This is correct regardless of common->capture_last_ptr. */ |
1125 | capture_last_found = TRUE; | |
1126 | } | } |
1127 | ||
1128 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
# | Line 809 while (cc < ccend) | Line 1132 while (cc < ccend) |
1132 | { | { |
1133 | case OP_SET_SOM: | case OP_SET_SOM: |
1134 | SLJIT_ASSERT(common->has_set_som); | SLJIT_ASSERT(common->has_set_som); |
1135 | stack_restore = TRUE; | |
1136 | if (!setsom_found) | if (!setsom_found) |
1137 | { | { |
1138 | length += 2; | length += 2; |
# | Line 819 while (cc < ccend) | Line 1143 while (cc < ccend) |
1143 | ||
1144 | case OP_MARK: | case OP_MARK: |
1145 | SLJIT_ASSERT(common->mark_ptr != 0); | SLJIT_ASSERT(common->mark_ptr != 0); |
1146 | stack_restore = TRUE; | |
1147 | if (!setmark_found) | if (!setmark_found) |
1148 | { | { |
1149 | length += 2; | length += 2; |
# | Line 828 while (cc < ccend) | Line 1153 while (cc < ccend) |
1153 | break; | break; |
1154 | ||
1155 | case OP_RECURSE: | case OP_RECURSE: |
1156 | stack_restore = TRUE; | |
1157 | if (common->has_set_som && !setsom_found) | if (common->has_set_som && !setsom_found) |
1158 | { | { |
1159 | length += 2; | length += 2; |
# | Line 838 while (cc < ccend) | Line 1164 while (cc < ccend) |
1164 | length += 2; | length += 2; |
1165 | setmark_found = TRUE; | setmark_found = TRUE; |
1166 | } | } |
1167 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1168 | { | |
1169 | length += 2; | |
1170 | capture_last_found = TRUE; | |
1171 | } | |
1172 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1173 | break; | break; |
1174 | ||
1175 | case OP_CBRA: | case OP_CBRA: |
1176 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1177 | case OP_SCBRA: | case OP_SCBRA: |
1178 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1179 | length += 3; | stack_restore = TRUE; |
1180 | cc += 1 + LINK_SIZE + IMM2_SIZE; | if (common->capture_last_ptr != 0 && !capture_last_found) |
1181 | break; | { |
1182 | length += 2; | |
1183 | capture_last_found = TRUE; | |
1184 | } | |
1185 | length += 3; | |
1186 | cc += 1 + LINK_SIZE + IMM2_SIZE; | |
1187 | break; | |
1188 | ||
1189 | default: | |
1190 | stack_restore = TRUE; | |
1191 | /* Fall through. */ | |
1192 | ||
1193 | case OP_NOT_WORD_BOUNDARY: | |
1194 | case OP_WORD_BOUNDARY: | |
1195 | case OP_NOT_DIGIT: | |
1196 | case OP_DIGIT: | |
1197 | case OP_NOT_WHITESPACE: | |
1198 | case OP_WHITESPACE: | |
1199 | case OP_NOT_WORDCHAR: | |
1200 | case OP_WORDCHAR: | |
1201 | case OP_ANY: | |
1202 | case OP_ALLANY: | |
1203 | case OP_ANYBYTE: | |
1204 | case OP_NOTPROP: | |
1205 | case OP_PROP: | |
1206 | case OP_ANYNL: | |
1207 | case OP_NOT_HSPACE: | |
1208 | case OP_HSPACE: | |
1209 | case OP_NOT_VSPACE: | |
1210 | case OP_VSPACE: | |
1211 | case OP_EXTUNI: | |
1212 | case OP_EODN: | |
1213 | case OP_EOD: | |
1214 | case OP_CIRC: | |
1215 | case OP_CIRCM: | |
1216 | case OP_DOLL: | |
1217 | case OP_DOLLM: | |
1218 | case OP_CHAR: | |
1219 | case OP_CHARI: | |
1220 | case OP_NOT: | |
1221 | case OP_NOTI: | |
1222 | ||
1223 | case OP_EXACT: | |
1224 | case OP_POSSTAR: | |
1225 | case OP_POSPLUS: | |
1226 | case OP_POSQUERY: | |
1227 | case OP_POSUPTO: | |
1228 | ||
1229 | case OP_EXACTI: | |
1230 | case OP_POSSTARI: | |
1231 | case OP_POSPLUSI: | |
1232 | case OP_POSQUERYI: | |
1233 | case OP_POSUPTOI: | |
1234 | ||
1235 | case OP_NOTEXACT: | |
1236 | case OP_NOTPOSSTAR: | |
1237 | case OP_NOTPOSPLUS: | |
1238 | case OP_NOTPOSQUERY: | |
1239 | case OP_NOTPOSUPTO: | |
1240 | ||
1241 | case OP_NOTEXACTI: | |
1242 | case OP_NOTPOSSTARI: | |
1243 | case OP_NOTPOSPLUSI: | |
1244 | case OP_NOTPOSQUERYI: | |
1245 | case OP_NOTPOSUPTOI: | |
1246 | ||
1247 | case OP_TYPEEXACT: | |
1248 | case OP_TYPEPOSSTAR: | |
1249 | case OP_TYPEPOSPLUS: | |
1250 | case OP_TYPEPOSQUERY: | |
1251 | case OP_TYPEPOSUPTO: | |
1252 | ||
1253 | case OP_CLASS: | |
1254 | case OP_NCLASS: | |
1255 | case OP_XCLASS: | |
1256 | ||
default: | ||
1257 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1258 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
1259 | break; | break; |
1260 | } | } |
1261 | ||
1262 | /* Possessive quantifiers can use a special case. */ | /* Possessive quantifiers can use a special case. */ |
1263 | if (SLJIT_UNLIKELY(possessive) && length == 3) | if (SLJIT_UNLIKELY(possessive == length)) |
1264 | return -1; | return stack_restore ? no_frame : no_stack; |
1265 | ||
1266 | if (length > 0) | if (length > 0) |
1267 | return length + 1; | return length + 1; |
1268 | return -1; | return stack_restore ? no_frame : no_stack; |
1269 | } | } |
1270 | ||
1271 | static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive) | static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive) |
1272 | { | { |
1273 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1274 | pcre_uchar *ccend = bracketend(cc); | pcre_uchar *ccend = bracketend(cc) - (1 + LINK_SIZE); |
1275 | BOOL setsom_found = recursive; | BOOL setsom_found = recursive; |
1276 | BOOL setmark_found = recursive; | BOOL setmark_found = recursive; |
1277 | /* The last capture is a local variable even for recursions. */ | |
1278 | BOOL capture_last_found = FALSE; | |
1279 | int offset; | int offset; |
1280 | ||
1281 | /* >= 1 + shortest item size (2) */ | /* >= 1 + shortest item size (2) */ |
# | Line 888 while (cc < ccend) | Line 1294 while (cc < ccend) |
1294 | if (!setsom_found) | if (!setsom_found) |
1295 | { | { |
1296 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); |
1297 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); |
1298 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1299 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1300 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1301 | setsom_found = TRUE; | setsom_found = TRUE; |
1302 | } | } |
1303 | cc += 1; | cc += 1; |
# | Line 902 while (cc < ccend) | Line 1308 while (cc < ccend) |
1308 | if (!setmark_found) | if (!setmark_found) |
1309 | { | { |
1310 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); |
1311 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); |
1312 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1313 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1314 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1315 | setmark_found = TRUE; | setmark_found = TRUE; |
1316 | } | } |
1317 | cc += 1 + 2 + cc[1]; | cc += 1 + 2 + cc[1]; |
# | Line 915 while (cc < ccend) | Line 1321 while (cc < ccend) |
1321 | if (common->has_set_som && !setsom_found) | if (common->has_set_som && !setsom_found) |
1322 | { | { |
1323 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); |
1324 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); |
1325 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1326 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1327 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1328 | setsom_found = TRUE; | setsom_found = TRUE; |
1329 | } | } |
1330 | if (common->mark_ptr != 0 && !setmark_found) | if (common->mark_ptr != 0 && !setmark_found) |
1331 | { | { |
1332 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); |
1333 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); |
1334 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1335 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1336 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1337 | setmark_found = TRUE; | setmark_found = TRUE; |
1338 | } | } |
1339 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1340 | { | |
1341 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); | |
1342 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); | |
1343 | stackpos += (int)sizeof(sljit_sw); | |
1344 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1345 | stackpos += (int)sizeof(sljit_sw); | |
1346 | capture_last_found = TRUE; | |
1347 | } | |
1348 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1349 | break; | break; |
1350 | ||
# | Line 937 while (cc < ccend) | Line 1352 while (cc < ccend) |
1352 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1353 | case OP_SCBRA: | case OP_SCBRA: |
1354 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1355 | if (common->capture_last_ptr != 0 && !capture_last_found) | |
1356 | { | |
1357 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); | |
1358 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); | |
1359 | stackpos += (int)sizeof(sljit_sw); | |
1360 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | |
1361 | stackpos += (int)sizeof(sljit_sw); | |
1362 | capture_last_found = TRUE; | |
1363 | } | |
1364 | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; | offset = (GET2(cc, 1 + LINK_SIZE)) << 1; |
1365 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); |
1366 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1367 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); |
1368 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); |
1369 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); |
1370 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1371 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); |
1372 | stackpos += (int)sizeof(sljit_w); | stackpos += (int)sizeof(sljit_sw); |
1373 | ||
1374 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1375 | break; | break; |
# | Line 956 while (cc < ccend) | Line 1380 while (cc < ccend) |
1380 | break; | break; |
1381 | } | } |
1382 | ||
1383 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_end); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, 0); |
1384 | SLJIT_ASSERT(stackpos == STACK(stacktop)); | SLJIT_ASSERT(stackpos == STACK(stacktop)); |
1385 | } | } |
1386 | ||
1387 | 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) |
1388 | { | { |
1389 | int localsize = 2; | int private_data_length = 2; |
1390 | int size; | |
1391 | pcre_uchar *alternative; | pcre_uchar *alternative; |
1392 | /* Calculate the sum of the local variables. */ | /* Calculate the sum of the private machine words. */ |
1393 | while (cc < ccend) | while (cc < ccend) |
1394 | { | { |
1395 | size = 0; | |
1396 | switch(*cc) | switch(*cc) |
1397 | { | { |
1398 | case OP_ASSERT: | case OP_ASSERT: |
# | Line 979 while (cc < ccend) | Line 1405 while (cc < ccend) |
1405 | case OP_SBRA: | case OP_SBRA: |
1406 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1407 | case OP_SCOND: | case OP_SCOND: |
1408 | localsize++; | private_data_length++; |
1409 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1410 | break; | break; |
1411 | ||
1412 | case OP_CBRA: | case OP_CBRA: |
1413 | case OP_SCBRA: | case OP_SCBRA: |
1414 | localsize++; | if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) |
1415 | private_data_length++; | |
1416 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1417 | break; | break; |
1418 | ||
1419 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1420 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1421 | localsize += 2; | private_data_length += 2; |
1422 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1423 | break; | break; |
1424 | ||
# | Line 999 while (cc < ccend) | Line 1426 while (cc < ccend) |
1426 | /* Might be a hidden SCOND. */ | /* Might be a hidden SCOND. */ |
1427 | alternative = cc + GET(cc, 1); | alternative = cc + GET(cc, 1); |
1428 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1429 | localsize++; | private_data_length++; |
1430 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1431 | break; | break; |
1432 | ||
1433 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1434 | if (PRIVATE_DATA(cc)) | |
1435 | private_data_length++; | |
1436 | cc += 2; | |
1437 | #ifdef SUPPORT_UTF | |
1438 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1439 | #endif | |
1440 | break; | |
1441 | ||
1442 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1443 | if (PRIVATE_DATA(cc)) | |
1444 | private_data_length += 2; | |
1445 | cc += 2; | |
1446 | #ifdef SUPPORT_UTF | |
1447 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1448 | #endif | |
1449 | break; | |
1450 | ||
1451 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1452 | if (PRIVATE_DATA(cc)) | |
1453 | private_data_length += 2; | |
1454 | cc += 2 + IMM2_SIZE; | |
1455 | #ifdef SUPPORT_UTF | |
1456 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1457 | #endif | |
1458 | break; | |
1459 | ||
1460 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1461 | if (PRIVATE_DATA(cc)) | |
1462 | private_data_length++; | |
1463 | cc += 1; | |
1464 | break; | |
1465 | ||
1466 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1467 | if (PRIVATE_DATA(cc)) | |
1468 | private_data_length += 2; | |
1469 | cc += 1; | |
1470 | break; | |
1471 | ||
1472 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1473 | if (PRIVATE_DATA(cc)) | |
1474 | private_data_length += 2; | |
1475 | cc += 1 + IMM2_SIZE; | |
1476 | break; | |
1477 | ||
1478 | case OP_CLASS: | |
1479 | case OP_NCLASS: | |
1480 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1481 | case OP_XCLASS: | |
1482 | size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); | |
1483 | #else | |
1484 | size = 1 + 32 / (int)sizeof(pcre_uchar); | |
1485 | #endif | |
1486 | if (PRIVATE_DATA(cc)) | |
1487 | private_data_length += get_class_iterator_size(cc + size); | |
1488 | cc += size; | |
1489 | break; | |
1490 | ||
1491 | default: | default: |
1492 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1493 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
# | Line 1010 while (cc < ccend) | Line 1495 while (cc < ccend) |
1495 | } | } |
1496 | } | } |
1497 | SLJIT_ASSERT(cc == ccend); | SLJIT_ASSERT(cc == ccend); |
1498 | return localsize; | return private_data_length; |
1499 | } | } |
1500 | ||
1501 | 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, |
1502 | BOOL save, int stackptr, int stacktop) | BOOL save, int stackptr, int stacktop) |
1503 | { | { |
1504 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1505 | int srcw[2]; | int srcw[2]; |
1506 | int count; | int count, size; |
1507 | BOOL tmp1next = TRUE; | BOOL tmp1next = TRUE; |
1508 | BOOL tmp1empty = TRUE; | BOOL tmp1empty = TRUE; |
1509 | BOOL tmp2empty = TRUE; | BOOL tmp2empty = TRUE; |
# | Line 1035 stacktop = STACK(stacktop - 1); | Line 1520 stacktop = STACK(stacktop - 1); |
1520 | ||
1521 | if (!save) | if (!save) |
1522 | { | { |
1523 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1524 | if (stackptr < stacktop) | if (stackptr < stacktop) |
1525 | { | { |
1526 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1527 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1528 | tmp1empty = FALSE; | tmp1empty = FALSE; |
1529 | } | } |
1530 | if (stackptr < stacktop) | if (stackptr < stacktop) |
1531 | { | { |
1532 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1533 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1534 | tmp2empty = FALSE; | tmp2empty = FALSE; |
1535 | } | } |
1536 | /* The tmp1next must be TRUE in either way. */ | /* The tmp1next must be TRUE in either way. */ |
# | Line 1057 while (status != end) | Line 1542 while (status != end) |
1542 | switch(status) | switch(status) |
1543 | { | { |
1544 | case start: | case start: |
1545 | SLJIT_ASSERT(save && common->recursive_head != 0); | SLJIT_ASSERT(save && common->recursive_head_ptr != 0); |
1546 | count = 1; | count = 1; |
1547 | srcw[0] = common->recursive_head; | srcw[0] = common->recursive_head_ptr; |
1548 | status = loop; | status = loop; |
1549 | break; | break; |
1550 | ||
# | Line 1083 while (status != end) | Line 1568 while (status != end) |
1568 | case OP_SBRAPOS: | case OP_SBRAPOS: |
1569 | case OP_SCOND: | case OP_SCOND: |
1570 | count = 1; | count = 1; |
1571 | srcw[0] = PRIV_DATA(cc); | srcw[0] = PRIVATE_DATA(cc); |
1572 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
1573 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1574 | break; | break; |
1575 | ||
1576 | case OP_CBRA: | case OP_CBRA: |
1577 | case OP_SCBRA: | case OP_SCBRA: |
1578 | count = 1; | if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) |
1579 | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | { |
1580 | count = 1; | |
1581 | srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | |
1582 | } | |
1583 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1584 | break; | break; |
1585 | ||
1586 | case OP_CBRAPOS: | case OP_CBRAPOS: |
1587 | case OP_SCBRAPOS: | case OP_SCBRAPOS: |
1588 | count = 2; | count = 2; |
1589 | srcw[0] = PRIVATE_DATA(cc); | |
1590 | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); | srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); |
1591 | srcw[0] = PRIV_DATA(cc); | SLJIT_ASSERT(srcw[0] != 0 && srcw[1] != 0); |
SLJIT_ASSERT(srcw[0] != 0); | ||
1592 | cc += 1 + LINK_SIZE + IMM2_SIZE; | cc += 1 + LINK_SIZE + IMM2_SIZE; |
1593 | break; | break; |
1594 | ||
# | Line 1110 while (status != end) | Line 1598 while (status != end) |
1598 | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) | if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) |
1599 | { | { |
1600 | count = 1; | count = 1; |
1601 | srcw[0] = PRIV_DATA(cc); | srcw[0] = PRIVATE_DATA(cc); |
1602 | SLJIT_ASSERT(srcw[0] != 0); | SLJIT_ASSERT(srcw[0] != 0); |
1603 | } | } |
1604 | cc += 1 + LINK_SIZE; | cc += 1 + LINK_SIZE; |
1605 | break; | break; |
1606 | ||
1607 | CASE_ITERATOR_PRIVATE_DATA_1 | |
1608 | if (PRIVATE_DATA(cc)) | |
1609 | { | |
1610 | count = 1; | |
1611 | srcw[0] = PRIVATE_DATA(cc); | |
1612 | } | |
1613 | cc += 2; | |
1614 | #ifdef SUPPORT_UTF | |
1615 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1616 | #endif | |
1617 | break; | |
1618 | ||
1619 | CASE_ITERATOR_PRIVATE_DATA_2A | |
1620 | if (PRIVATE_DATA(cc)) | |
1621 | { | |
1622 | count = 2; | |
1623 | srcw[0] = PRIVATE_DATA(cc); | |
1624 | srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); | |
1625 | } | |
1626 | cc += 2; | |
1627 | #ifdef SUPPORT_UTF | |
1628 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1629 | #endif | |
1630 | break; | |
1631 | ||
1632 | CASE_ITERATOR_PRIVATE_DATA_2B | |
1633 | if (PRIVATE_DATA(cc)) | |
1634 | { | |
1635 | count = 2; | |
1636 | srcw[0] = PRIVATE_DATA(cc); | |
1637 | srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); | |
1638 | } | |
1639 | cc += 2 + IMM2_SIZE; | |
1640 | #ifdef SUPPORT_UTF | |
1641 | if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); | |
1642 | #endif | |
1643 | break; | |
1644 | ||
1645 | CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1646 | if (PRIVATE_DATA(cc)) | |
1647 | { | |
1648 | count = 1; | |
1649 | srcw[0] = PRIVATE_DATA(cc); | |
1650 | } | |
1651 | cc += 1; | |
1652 | break; | |
1653 | ||
1654 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1655 | if (PRIVATE_DATA(cc)) | |
1656 | { | |
1657 | count = 2; | |
1658 | srcw[0] = PRIVATE_DATA(cc); | |
1659 | srcw[1] = srcw[0] + sizeof(sljit_sw); | |
1660 | } | |
1661 | cc += 1; | |
1662 | break; | |
1663 | ||
1664 | CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1665 | if (PRIVATE_DATA(cc)) | |
1666 | { | |
1667 | count = 2; | |
1668 | srcw[0] = PRIVATE_DATA(cc); | |
1669 | srcw[1] = srcw[0] + sizeof(sljit_sw); | |
1670 | } | |
1671 | cc += 1 + IMM2_SIZE; | |
1672 | break; | |
1673 | ||
1674 | case OP_CLASS: | |
1675 | case OP_NCLASS: | |
1676 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 | |
1677 | case OP_XCLASS: | |
1678 | size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); | |
1679 | #else | |
1680 | size = 1 + 32 / (int)sizeof(pcre_uchar); | |
1681 | #endif | |
1682 | if (PRIVATE_DATA(cc)) | |
1683 | switch(get_class_iterator_size(cc + size)) | |
1684 | { | |
1685 | case 1: | |
1686 | count = 1; | |
1687 | srcw[0] = PRIVATE_DATA(cc); | |
1688 | break; | |
1689 | ||
1690 | case 2: | |
1691 | count = 2; | |
1692 | srcw[0] = PRIVATE_DATA(cc); | |
1693 | srcw[1] = srcw[0] + sizeof(sljit_sw); | |
1694 | break; | |
1695 | ||
1696 | default: | |
1697 | SLJIT_ASSERT_STOP(); | |
1698 | break; | |
1699 | } | |
1700 | cc += size; | |
1701 | break; | |
1702 | ||
1703 | default: | default: |
1704 | cc = next_opcode(common, cc); | cc = next_opcode(common, cc); |
1705 | SLJIT_ASSERT(cc != NULL); | SLJIT_ASSERT(cc != NULL); |
# | Line 1138 while (status != end) | Line 1722 while (status != end) |
1722 | if (!tmp1empty) | if (!tmp1empty) |
1723 | { | { |
1724 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1725 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1726 | } | } |
1727 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); |
1728 | tmp1empty = FALSE; | tmp1empty = FALSE; |
# | Line 1149 while (status != end) | Line 1733 while (status != end) |
1733 | if (!tmp2empty) | if (!tmp2empty) |
1734 | { | { |
1735 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1736 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1737 | } | } |
1738 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); |
1739 | tmp2empty = FALSE; | tmp2empty = FALSE; |
# | Line 1166 while (status != end) | Line 1750 while (status != end) |
1750 | if (!tmp1empty) | if (!tmp1empty) |
1751 | { | { |
1752 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1753 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1754 | } | } |
1755 | tmp1next = FALSE; | tmp1next = FALSE; |
1756 | } | } |
# | Line 1178 while (status != end) | Line 1762 while (status != end) |
1762 | if (!tmp2empty) | if (!tmp2empty) |
1763 | { | { |
1764 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); |
1765 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1766 | } | } |
1767 | tmp1next = TRUE; | tmp1next = TRUE; |
1768 | } | } |
# | Line 1193 if (save) | Line 1777 if (save) |
1777 | if (!tmp1empty) | if (!tmp1empty) |
1778 | { | { |
1779 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1780 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1781 | } | } |
1782 | if (!tmp2empty) | if (!tmp2empty) |
1783 | { | { |
1784 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1785 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1786 | } | } |
1787 | } | } |
1788 | else | else |
# | Line 1206 if (save) | Line 1790 if (save) |
1790 | if (!tmp2empty) | if (!tmp2empty) |
1791 | { | { |
1792 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); |
1793 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1794 | } | } |
1795 | if (!tmp1empty) | if (!tmp1empty) |
1796 | { | { |
1797 | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); |
1798 | stackptr += sizeof(sljit_w); | stackptr += sizeof(sljit_sw); |
1799 | } | } |
1800 | } | } |
1801 | } | } |
1802 | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); | SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); |
1803 | } | } |
1804 | ||
1805 | static SLJIT_INLINE BOOL ispowerof2(unsigned int value) | #undef CASE_ITERATOR_PRIVATE_DATA_1 |
1806 | #undef CASE_ITERATOR_PRIVATE_DATA_2A | |
1807 | #undef CASE_ITERATOR_PRIVATE_DATA_2B | |
1808 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_1 | |
1809 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2A | |
1810 | #undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2B | |
1811 | ||
1812 | static SLJIT_INLINE BOOL is_powerof2(unsigned int value) | |
1813 | { | { |
1814 | return (value & (value - 1)) == 0; | return (value & (value - 1)) == 0; |
1815 | } | } |
# | Line 1228 static SLJIT_INLINE void set_jumps(jump_ | Line 1819 static SLJIT_INLINE void set_jumps(jump_ |
1819 | while (list) | while (list) |
1820 | { | { |
1821 | /* sljit_set_label is clever enough to do nothing | /* sljit_set_label is clever enough to do nothing |
1822 | if either the jump or the label is NULL */ | if either the jump or the label is NULL. */ |
1823 | sljit_set_label(list->jump, label); | SET_LABEL(list->jump, label); |
1824 | list = list->next; | list = list->next; |
1825 | } | } |
1826 | } | } |
# | Line 1245 if (list_item) | Line 1836 if (list_item) |
1836 | } | } |
1837 | } | } |
1838 | ||
1839 | static void add_stub(compiler_common *common, enum stub_types type, int data, struct sljit_jump *start) | static void add_stub(compiler_common *common, struct sljit_jump *start) |
1840 | { | { |
1841 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1842 | stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); | stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); |
1843 | ||
1844 | if (list_item) | if (list_item) |
1845 | { | { |
list_item->type = type; | ||
list_item->data = data; | ||
1846 | list_item->start = start; | list_item->start = start; |
1847 | list_item->leave = LABEL(); | list_item->quit = LABEL(); |
1848 | list_item->next = common->stubs; | list_item->next = common->stubs; |
1849 | common->stubs = list_item; | common->stubs = list_item; |
1850 | } | } |
# | Line 1269 stub_list* list_item = common->stubs; | Line 1858 stub_list* list_item = common->stubs; |
1858 | while (list_item) | while (list_item) |
1859 | { | { |
1860 | JUMPHERE(list_item->start); | JUMPHERE(list_item->start); |
1861 | switch(list_item->type) | add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); |
1862 | { | JUMPTO(SLJIT_JUMP, list_item->quit); |
case stack_alloc: | ||
add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); | ||
break; | ||
} | ||
JUMPTO(SLJIT_JUMP, list_item->leave); | ||
1863 | list_item = list_item->next; | list_item = list_item->next; |
1864 | } | } |
1865 | common->stubs = NULL; | common->stubs = NULL; |
# | Line 1294 static SLJIT_INLINE void allocate_stack( | Line 1878 static SLJIT_INLINE void allocate_stack( |
1878 | /* May destroy all locals and registers except TMP2. */ | /* May destroy all locals and registers except TMP2. */ |
1879 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1880 | ||
1881 | OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_w)); | OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw)); |
1882 | #ifdef DESTROY_REGISTERS | #ifdef DESTROY_REGISTERS |
1883 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); |
1884 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
# | Line 1302 OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); | Line 1886 OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); |
1886 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0); |
1887 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); |
1888 | #endif | #endif |
1889 | add_stub(common, stack_alloc, 0, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); | add_stub(common, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); |
1890 | } | } |
1891 | ||
1892 | static SLJIT_INLINE void free_stack(compiler_common *common, int size) | static SLJIT_INLINE void free_stack(compiler_common *common, int size) |
1893 | { | { |
1894 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1895 | OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_w)); | OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw)); |
1896 | } | } |
1897 | ||
1898 | static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) | static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) |
# | Line 1318 struct sljit_label *loop; | Line 1902 struct sljit_label *loop; |
1902 | int i; | int i; |
1903 | /* At this point we can freely use all temporary registers. */ | /* At this point we can freely use all temporary registers. */ |
1904 | /* TMP1 returns with begin - 1. */ | /* TMP1 returns with begin - 1. */ |
1905 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1)); | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1)); |
1906 | if (length < 8) | if (length < 8) |
1907 | { | { |
1908 | for (i = 0; i < length; i++) | for (i = 0; i < length; i++) |
1909 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_TEMPORARY_REG1, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_SCRATCH_REG1, 0); |
1910 | } | } |
1911 | else | else |
1912 | { | { |
1913 | GET_LOCAL_BASE(SLJIT_TEMPORARY_REG2, 0, OVECTOR_START - sizeof(sljit_w)); | GET_LOCAL_BASE(SLJIT_SCRATCH_REG2, 0, OVECTOR_START - sizeof(sljit_sw)); |
1914 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, length); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, length); |
1915 | loop = LABEL(); | loop = LABEL(); |
1916 | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(sljit_w), SLJIT_TEMPORARY_REG1, 0); | OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(sljit_sw), SLJIT_SCRATCH_REG1, 0); |
1917 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 1); |
1918 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_C_NOT_ZERO, loop); |
1919 | } | } |
1920 | } | } |
# | Line 1339 static SLJIT_INLINE void copy_ovector(co | Line 1923 static SLJIT_INLINE void copy_ovector(co |
1923 | { | { |
1924 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1925 | struct sljit_label *loop; | struct sljit_label *loop; |
1926 | struct sljit_jump *earlyexit; | struct sljit_jump *early_quit; |
1927 | ||
1928 | /* At this point we can freely use all registers. */ | /* At this point we can freely use all registers. */ |
1929 | OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); | OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); |
1930 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1), STR_PTR, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1), STR_PTR, 0); |
1931 | ||
1932 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, ARGUMENTS, 0); |
1933 | if (common->mark_ptr != 0) | if (common->mark_ptr != 0) |
1934 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); |
1935 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offset_count)); |
1936 | if (common->mark_ptr != 0) | if (common->mark_ptr != 0) |
1937 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_TEMPORARY_REG3, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_SCRATCH_REG3, 0); |
1938 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); |
1939 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); |
1940 | GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START); | GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START); |
1941 | /* Unlikely, but possible */ | /* Unlikely, but possible */ |
1942 | earlyexit = CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 0); | early_quit = CMP(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 0); |
1943 | loop = LABEL(); | loop = LABEL(); |
1944 | 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_SCRATCH_REG1, 0); |
1945 | 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_sw)); |
1946 | /* Copy the integer value to the output buffer */ | /* Copy the integer value to the output buffer */ |
1947 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
1948 | 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); |
1949 | #endif | #endif |
1950 | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); |
1951 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1); |
1952 | JUMPTO(SLJIT_C_NOT_ZERO, loop); | JUMPTO(SLJIT_C_NOT_ZERO, loop); |
1953 | JUMPHERE(earlyexit); | JUMPHERE(early_quit); |
1954 | ||
1955 | /* Calculate the return value, which is the maximum ovector value. */ | /* Calculate the return value, which is the maximum ovector value. */ |
1956 | if (topbracket > 1) | if (topbracket > 1) |
1957 | { | { |
1958 | GET_LOCAL_BASE(SLJIT_TEMPORARY_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_w)); | GET_LOCAL_BASE(SLJIT_SCRATCH_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw)); |
1959 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, topbracket + 1); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, topbracket + 1); |
1960 | ||
1961 | /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ | /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ |
1962 | loop = LABEL(); | loop = LABEL(); |
1963 | OP1(SLJIT_MOVU, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), -(2 * (sljit_w)sizeof(sljit_w))); | OP1(SLJIT_MOVU, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), -(2 * (sljit_sw)sizeof(sljit_sw))); |
1964 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1); |
1965 | CMPTO(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG3, 0, loop); | CMPTO(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG3, 0, loop); |
1966 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_TEMPORARY_REG2, 0); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_SCRATCH_REG2, 0); |
1967 | } | } |
1968 | else | else |
1969 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); |
1970 | } | } |
1971 | ||
1972 | 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) |
1973 | { | { |
1974 | DEFINE_COMPILER; | DEFINE_COMPILER; |
1975 | struct sljit_jump *jump; | |
1976 | ||
1977 | SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2); | SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2); |
1978 | SLJIT_ASSERT(common->start_used_ptr != 0 && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); | SLJIT_ASSERT(common->start_used_ptr != 0 && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); |
1979 | ||
1980 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0); |
1981 | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); | OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); |
1982 | OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsetcount)); | OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, real_offset_count)); |
1983 | CMPTO(SLJIT_C_LESS, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 2, leave); | CMPTO(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 2, quit); |
1984 | ||
1985 | /* Store match begin and end. */ | /* Store match begin and end. */ |
1986 | 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_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); |
1987 | OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); |
1988 | 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); | |
1989 | jump = CMP(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 3); | |
1990 | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), (common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start) + sizeof(sljit_sw), SLJIT_SAVED_REG1, 0); | |
1991 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 | |
1992 | OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); | |
1993 | #endif | |
1994 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 2 * sizeof(int), SLJIT_SCRATCH_REG3, 0); | |
1995 | JUMPHERE(jump); | |
1996 | ||
1997 | OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start); | |
1998 | 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); |
1999 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2000 | 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); |
2001 | #endif | #endif |
2002 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); |
2003 | ||
2004 | OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG1, 0); | OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG1, 0); |
2005 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2006 | OP2(SLJIT_ASHR, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); | OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); |
2007 | #endif | #endif |
2008 | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), 0, SLJIT_TEMPORARY_REG3, 0); | OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 0, SLJIT_SCRATCH_REG3, 0); |
2009 | ||
2010 | JUMPTO(SLJIT_JUMP, leave); | JUMPTO(SLJIT_JUMP, quit); |
2011 | } | } |
2012 | ||
2013 | 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 2118 if (c <= 127 && bit == 0x20) |
2118 | return (0 << 8) | 0x20; | return (0 << 8) | 0x20; |
2119 | ||
2120 | /* Since c != oc, they must have at least 1 bit difference. */ | /* Since c != oc, they must have at least 1 bit difference. */ |
2121 | if (!ispowerof2(bit)) | if (!is_powerof2(bit)) |
2122 | return 0; | return 0; |
2123 | ||
2124 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2125 | ||
2126 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2127 | if (common->utf && c > 127) | if (common->utf && c > 127) |
# | Line 1543 if (common->utf && c > 127) | Line 2137 if (common->utf && c > 127) |
2137 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2138 | return (0 << 8) | bit; | return (0 << 8) | bit; |
2139 | ||
2140 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2141 | ||
#ifdef COMPILE_PCRE16 | ||
2142 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2143 | if (common->utf && c > 65535) | if (common->utf && c > 65535) |
2144 | { | { |
# | Line 1556 if (common->utf && c > 65535) | Line 2149 if (common->utf && c > 65535) |
2149 | } | } |
2150 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2151 | return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); | return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); |
#endif /* COMPILE_PCRE16 */ | ||
2152 | ||
2153 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE[8|16|32] */ |
2154 | } | } |
2155 | ||
2156 | static void check_partial(compiler_common *common, BOOL force) | static void check_partial(compiler_common *common, BOOL force) |
# | Line 1622 JUMPHERE(jump); | Line 2214 JUMPHERE(jump); |
2214 | return return_value; | return return_value; |
2215 | } | } |
2216 | ||
2217 | static void fallback_at_str_end(compiler_common *common, jump_list **fallbacks) | static void detect_partial_match(compiler_common *common, jump_list **backtracks) |
2218 | { | { |
2219 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2220 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2221 | ||
2222 | if (common->mode == JIT_COMPILE) | if (common->mode == JIT_COMPILE) |
2223 | { | { |
2224 | 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)); |
2225 | return; | return; |
2226 | } | } |
2227 | ||
2228 | /* Partial matching mode. */ | /* Partial matching mode. */ |
2229 | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); | jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); |
2230 | 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)); |
2231 | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) | if (common->mode == JIT_PARTIAL_SOFT_COMPILE) |
2232 | { | { |
2233 | 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); |
2234 | add_jump(compiler, fallbacks, JUMP(SLJIT_JUMP)); | add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); |
2235 | } | } |
2236 | else | else |
2237 | { | { |
# | Line 1656 static void read_char(compiler_common *c | Line 2248 static void read_char(compiler_common *c |
2248 | /* Reads the character into TMP1, updates STR_PTR. | /* Reads the character into TMP1, updates STR_PTR. |
2249 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2250 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2251 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2252 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2253 | #endif | #endif |
2254 | ||
2255 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2256 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2257 | if (common->utf) | if (common->utf) |
2258 | { | { |
2259 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2260 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2261 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2262 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
2263 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
#endif /* COMPILE_PCRE8 */ | ||
2264 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2265 | JUMPHERE(jump); | JUMPHERE(jump); |
2266 | } | } |
2267 | #endif | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2268 | 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)); |
2269 | } | } |
2270 | ||
# | Line 1683 static void peek_char(compiler_common *c | Line 2273 static void peek_char(compiler_common *c |
2273 | /* Reads the character into TMP1, keeps STR_PTR. | /* Reads the character into TMP1, keeps STR_PTR. |
2274 | Does not check STR_END. TMP2 Destroyed. */ | Does not check STR_END. TMP2 Destroyed. */ |
2275 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2276 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2277 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2278 | #endif | #endif |
2279 | ||
2280 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2281 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2282 | if (common->utf) | if (common->utf) |
2283 | { | { |
2284 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2285 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2286 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2287 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
2288 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
#endif /* COMPILE_PCRE8 */ | ||
2289 | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); |
2290 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2291 | JUMPHERE(jump); | JUMPHERE(jump); |
2292 | } | } |
2293 | #endif | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ |
2294 | } | } |
2295 | ||
2296 | static void read_char8_type(compiler_common *common) | static void read_char8_type(compiler_common *common) |
2297 | { | { |
2298 | /* 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. */ |
2299 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2300 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2301 | struct sljit_jump *jump; | struct sljit_jump *jump; |
2302 | #endif | #endif |
2303 | ||
# | Line 1718 if (common->utf) | Line 2306 if (common->utf) |
2306 | { | { |
2307 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2308 | 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)); |
2309 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2310 | /* This can be an extra read in some situations, but hopefully | /* This can be an extra read in some situations, but hopefully |
2311 | it is needed in most cases. */ | it is needed in most cases. */ |
2312 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2313 | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); | jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); |
2314 | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); |
2315 | JUMPHERE(jump); | JUMPHERE(jump); |
2316 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
2317 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2318 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2319 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
# | Line 1734 if (common->utf) | Line 2321 if (common->utf) |
2321 | /* Skip low surrogate if necessary. */ | /* Skip low surrogate if necessary. */ |
2322 | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); |
2323 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800); |
2324 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2325 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); |
2326 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2327 | #endif | #elif defined COMPILE_PCRE32 |
2328 | #endif /* COMPILE_PCRE8 */ | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2329 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | |
2330 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | |
2331 | JUMPHERE(jump); | |
2332 | #endif /* COMPILE_PCRE[8|16|32] */ | |
2333 | return; | return; |
2334 | } | } |
2335 | #endif | #endif /* SUPPORT_UTF */ |
2336 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); |
2337 | 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)); |
2338 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2339 | /* The ctypes array contains only 256 values. */ | /* The ctypes array contains only 256 values. */ |
2340 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2341 | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); |
2342 | #endif | #endif |
2343 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); |
2344 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2345 | JUMPHERE(jump); | JUMPHERE(jump); |
2346 | #endif | #endif |
2347 | } | } |
# | Line 1759 static void skip_char_back(compiler_comm | Line 2350 static void skip_char_back(compiler_comm |
2350 | { | { |
2351 | /* 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. */ |
2352 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2353 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2354 | #if defined COMPILE_PCRE8 | |
2355 | struct sljit_label *label; | struct sljit_label *label; |
2356 | ||
2357 | if (common->utf) | if (common->utf) |
# | Line 1771 if (common->utf) | Line 2363 if (common->utf) |
2363 | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); | CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); |
2364 | return; | return; |
2365 | } | } |
2366 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
2367 | if (common->utf) | if (common->utf) |
2368 | { | { |
2369 | 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 1780 if (common->utf) | Line 2371 if (common->utf) |
2371 | /* Skip low surrogate if necessary. */ | /* Skip low surrogate if necessary. */ |
2372 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
2373 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00); |
2374 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2375 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
2376 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2377 | return; | return; |
2378 | } | } |
2379 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
2380 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | |
2381 | 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)); |
2382 | } | } |
2383 | ||
2384 | 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) |
2385 | { | { |
2386 | /* 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. */ |
2387 | DEFINE_COMPILER; | DEFINE_COMPILER; |
# | Line 1797 DEFINE_COMPILER; | Line 2389 DEFINE_COMPILER; |
2389 | if (nltype == NLTYPE_ANY) | if (nltype == NLTYPE_ANY) |
2390 | { | { |
2391 | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); |
2392 | 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)); |
2393 | } | } |
2394 | else if (nltype == NLTYPE_ANYCRLF) | else if (nltype == NLTYPE_ANYCRLF) |
2395 | { | { |
2396 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR); |
2397 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2398 | 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); |
2399 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
2400 | 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)); |
2401 | } | } |
2402 | else | else |
2403 | { | { |
2404 | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); | SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); |
2405 | 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)); |
2406 | } | } |
2407 | } | } |
2408 | ||
2409 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
2410 | ||
2411 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
2412 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2413 | { | { |
2414 | /* Fast decoding a UTF-8 character. TMP1 contains the first byte | /* Fast decoding a UTF-8 character. TMP1 contains the first byte |
# | Line 1904 sljit_emit_fast_return(compiler, RETURN_ | Line 2496 sljit_emit_fast_return(compiler, RETURN_ |
2496 | JUMPHERE(jump); | JUMPHERE(jump); |
2497 | ||
2498 | /* We only have types for characters less than 256. */ | /* We only have types for characters less than 256. */ |
2499 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_w)PRIV(utf8_table4) - 0xc0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0); |
2500 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2501 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); |
2502 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2503 | } | } |
2504 | ||
2505 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 |
2506 | ||
#ifdef COMPILE_PCRE16 | ||
2507 | static void do_utfreadchar(compiler_common *common) | static void do_utfreadchar(compiler_common *common) |
2508 | { | { |
2509 | /* 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 2528 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UC |
2528 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); |
2529 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2530 | } | } |
#endif /* COMPILE_PCRE16 */ | ||
2531 | ||
2532 | #endif /* COMPILE_PCRE8 */ | #endif /* COMPILE_PCRE[8|16] */ |
2533 | ||
2534 | #endif /* SUPPORT_UTF */ | #endif /* SUPPORT_UTF */ |
2535 | ||
# | Line 1959 SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && si | Line 2549 SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && si |
2549 | ||
2550 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
2551 | OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); | OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); |
2552 | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_w)PRIV(ucd_stage1)); | OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_stage1)); |
2553 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); |
2554 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); |
2555 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); |
2556 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_stage2)); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_stage2)); |
2557 | OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); | OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); |
2558 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); |
2559 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); |
2560 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
2561 | } | } |
# | Line 1979 struct sljit_label *newlinelabel = NULL; | Line 2569 struct sljit_label *newlinelabel = NULL; |
2569 | struct sljit_jump *start; | struct sljit_jump *start; |
2570 | struct sljit_jump *end = NULL; | struct sljit_jump *end = NULL; |
2571 | struct sljit_jump *nl = NULL; | struct sljit_jump *nl = NULL; |
2572 | #ifdef SUPPORT_UTF | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2573 | struct sljit_jump *singlechar; | struct sljit_jump *singlechar; |
2574 | #endif | #endif |
2575 | jump_list *newline = NULL; | jump_list *newline = NULL; |
# | Line 1994 if (firstline) | Line 2584 if (firstline) |
2584 | { | { |
2585 | /* Search for the end of the first line. */ | /* Search for the end of the first line. */ |
2586 | SLJIT_ASSERT(common->first_line_end != 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2587 | 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); | ||
2588 | ||
2589 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
2590 | { | { |
# | Line 2006 if (firstline) | Line 2595 if (firstline) |
2595 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); |
2596 | 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); |
2597 | 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); |
2598 | JUMPHERE(end); | |
2599 | 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)); |
2600 | } | } |
2601 | else | else |
# | Line 2017 if (firstline) | Line 2607 if (firstline) |
2607 | read_char(common); | read_char(common); |
2608 | check_newlinechar(common, common->nltype, &newline, TRUE); | check_newlinechar(common, common->nltype, &newline, TRUE); |
2609 | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); | CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); |
2610 | JUMPHERE(end); | |
2611 | 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); |
2612 | set_jumps(newline, LABEL()); | set_jumps(newline, LABEL()); |
2613 | } | } |
2614 | ||
2615 | JUMPHERE(end); | OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); |
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | ||
2616 | } | } |
2617 | ||
2618 | start = JUMP(SLJIT_JUMP); | start = JUMP(SLJIT_JUMP); |
# | Line 2034 if (newlinecheck) | Line 2624 if (newlinecheck) |
2624 | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2625 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2626 | 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); |
2627 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2628 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2629 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
2630 | #endif | #endif |
2631 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2632 | nl = JUMP(SLJIT_JUMP); | nl = JUMP(SLJIT_JUMP); |
# | Line 2057 if (newlinecheck) | Line 2647 if (newlinecheck) |
2647 | 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); |
2648 | ||
2649 | 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)); |
2650 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #if defined SUPPORT_UTF && !defined COMPILE_PCRE32 |
2651 | #if defined COMPILE_PCRE8 | |
2652 | if (common->utf) | if (common->utf) |
2653 | { | { |
2654 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
2655 | 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_sw)PRIV(utf8_table4) - 0xc0); |
2656 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2657 | JUMPHERE(singlechar); | JUMPHERE(singlechar); |
2658 | } | } |
2659 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
2660 | if (common->utf) | if (common->utf) |
2661 | { | { |
2662 | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
2663 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
2664 | 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); |
2665 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2666 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
2667 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2668 | JUMPHERE(singlechar); | JUMPHERE(singlechar); |
2669 | } | } |
2670 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
2671 | #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ | |
2672 | JUMPHERE(start); | JUMPHERE(start); |
2673 | ||
2674 | if (newlinecheck) | if (newlinecheck) |
# | Line 2089 if (newlinecheck) | Line 2680 if (newlinecheck) |
2680 | return mainloop; | return mainloop; |
2681 | } | } |
2682 | ||
2683 | #define MAX_N_CHARS 3 | |
2684 | ||
2685 | static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline) | |
2686 | { | |
2687 | DEFINE_COMPILER; | |
2688 | struct sljit_label *start; | |
2689 | struct sljit_jump *quit; | |
2690 | pcre_uint32 chars[MAX_N_CHARS * 2]; | |
2691 | pcre_uchar *cc = common->start + 1 + LINK_SIZE; | |
2692 | int location = 0; | |
2693 | pcre_int32 len, c, bit, caseless; | |
2694 | int must_stop; | |
2695 | ||
2696 | /* We do not support alternatives now. */ | |
2697 | if (*(common->start + GET(common->start, 1)) == OP_ALT) | |
2698 | return FALSE; | |
2699 | ||
2700 | while (TRUE) | |
2701 | { | |
2702 | caseless = 0; | |
2703 | must_stop = 1; | |
2704 | switch(*cc) | |
2705 | { | |
2706 | case OP_CHAR: | |
2707 | must_stop = 0; | |
2708 | cc++; | |
2709 | break; | |
2710 | ||
2711 | case OP_CHARI: | |
2712 | caseless = 1; | |
2713 | must_stop = 0; | |
2714 | cc++; | |
2715 | break; | |
2716 | ||
2717 | case OP_SOD: | |
2718 | case OP_SOM: | |
2719 | case OP_SET_SOM: | |
2720 | case OP_NOT_WORD_BOUNDARY: | |
2721 | case OP_WORD_BOUNDARY: | |
2722 | case OP_EODN: | |
2723 | case OP_EOD: | |
2724 | case OP_CIRC: | |
2725 | case OP_CIRCM: | |
2726 | case OP_DOLL: | |
2727 | case OP_DOLLM: | |
2728 | /* Zero width assertions. */ | |
2729 | cc++; | |
2730 | continue; | |
2731 | ||
2732 | case OP_PLUS: | |
2733 | case OP_MINPLUS: | |
2734 | case OP_POSPLUS: | |
2735 | cc++; | |
2736 | break; | |
2737 | ||
2738 | case OP_EXACT: | |
2739 | cc += 1 + IMM2_SIZE; | |
2740 | break; | |
2741 | ||
2742 | case OP_PLUSI: | |
2743 | case OP_MINPLUSI: | |
2744 | case OP_POSPLUSI: | |
2745 | caseless = 1; | |
2746 | cc++; | |
2747 | break; | |
2748 | ||
2749 | case OP_EXACTI: | |
2750 | caseless = 1; | |
2751 | cc += 1 + IMM2_SIZE; | |
2752 | break; | |
2753 | ||
2754 | default: | |
2755 | must_stop = 2; | |
2756 | break; | |
2757 | } | |
2758 | ||
2759 | if (must_stop == 2) | |
2760 | break; | |
2761 | ||
2762 | len = 1; | |
2763 | #ifdef SUPPORT_UTF | |
2764 | if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]); | |
2765 | #endif | |
2766 | ||
2767 | if (caseless && char_has_othercase(common, cc)) | |
2768 | { | |
2769 | caseless = char_get_othercase_bit(common, cc); | |
2770 | if (caseless == 0) | |
2771 | return FALSE; | |
2772 | #ifdef COMPILE_PCRE8 | |
2773 | caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8)); | |
2774 | #else | |
2775 | if ((caseless & 0x100) != 0) | |
2776 | caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9)); | |
2777 | else | |
2778 | caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9)); | |
2779 | #endif | |
2780 | } | |
2781 | else | |
2782 | caseless = 0; | |
2783 | ||
2784 | while (len > 0 && location < MAX_N_CHARS * 2) | |
2785 | { | |
2786 | c = *cc; | |
2787 | bit = 0; | |
2788 | if (len == (caseless & 0xff)) | |
2789 | { | |
2790 | bit = caseless >> 8; | |
2791 | c |= bit; | |
2792 | } | |
2793 | ||
2794 | chars[location] = c; | |
2795 | chars[location + 1] = bit; | |
2796 | ||
2797 | len--; | |
2798 | location += 2; | |
2799 | cc++; | |
2800 | } | |
2801 | ||
2802 | if (location >= MAX_N_CHARS * 2 || must_stop != 0) | |
2803 | break; | |
2804 | } | |
2805 | ||
2806 | /* At least two characters are required. */ | |
2807 | if (location < 2 * 2) | |
2808 | return FALSE; | |
2809 | ||
2810 | if (firstline) | |
2811 | { | |
2812 | SLJIT_ASSERT(common->first_line_end != 0); | |
2813 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | |
2814 | OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); | |
2815 | } | |
2816 | else | |
2817 | OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); | |
2818 | ||
2819 | start = LABEL(); | |
2820 | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | |
2821 | ||
2822 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); | |
2823 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2824 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2825 | if (chars[1] != 0) | |
2826 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]); | |
2827 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start); | |
2828 | if (location > 2 * 2) | |
2829 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); | |
2830 | if (chars[3] != 0) | |
2831 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]); | |
2832 | CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start); | |
2833 | if (location > 2 * 2) | |
2834 | { | |
2835 | if (chars[5] != 0) | |
2836 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]); | |
2837 | CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start); | |
2838 | } | |
2839 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); | |
2840 | ||
2841 | JUMPHERE(quit); | |
2842 | ||
2843 | if (firstline) | |
2844 | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); | |
2845 | else | |
2846 | OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); | |
2847 | return TRUE; | |
2848 | } | |
2849 | ||
2850 | #undef MAX_N_CHARS | |
2851 | ||
2852 | 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) |
2853 | { | { |
2854 | DEFINE_COMPILER; | DEFINE_COMPILER; |
2855 | struct sljit_label *start; | struct sljit_label *start; |
2856 | struct sljit_jump *leave; | struct sljit_jump *quit; |
2857 | struct sljit_jump *found; | struct sljit_jump *found; |
2858 | pcre_uchar oc, bit; | pcre_uchar oc, bit; |
2859 | ||
2860 | if (firstline) | if (firstline) |
2861 | { | { |
2862 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2863 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | |
2864 | 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); |
2865 | } | } |
2866 | ||
2867 | start = LABEL(); | start = LABEL(); |
2868 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2869 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2870 | ||
2871 | oc = first_char; | oc = first_char; |
# | Line 2121 if (first_char == oc) | Line 2882 if (first_char == oc) |
2882 | else | else |
2883 | { | { |
2884 | bit = first_char ^ oc; | bit = first_char ^ oc; |
2885 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
2886 | { | { |
2887 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); |
2888 | 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 2129 else | Line 2890 else |
2890 | else | else |
2891 | { | { |
2892 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, first_char); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, first_char); |
2893 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2894 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); |
2895 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
2896 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_C_NOT_ZERO); |
2897 | } | } |
2898 | } | } |
2899 | ||
2900 | 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 | ||
2901 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
2902 | JUMPHERE(found); | JUMPHERE(found); |
2903 | JUMPHERE(leave); | JUMPHERE(quit); |
2904 | ||
2905 | if (firstline) | if (firstline) |
2906 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
2907 | } | } |
2908 | ||
2909 | 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 2912 DEFINE_COMPILER; |
2912 | struct sljit_label *loop; | struct sljit_label *loop; |
2913 | struct sljit_jump *lastchar; | struct sljit_jump *lastchar; |
2914 | struct sljit_jump *firstchar; | struct sljit_jump *firstchar; |
2915 | struct sljit_jump *leave; | struct sljit_jump *quit; |
2916 | struct sljit_jump *foundcr = NULL; | struct sljit_jump *foundcr = NULL; |
2917 | struct sljit_jump *notfoundnl; | struct sljit_jump *notfoundnl; |
2918 | jump_list *newline = NULL; | jump_list *newline = NULL; |
2919 | ||
2920 | if (firstline) | if (firstline) |
2921 | { | { |
2922 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
2923 | OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); | |
2924 | 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); |
2925 | } | } |
2926 | ||
# | Line 2191 if (common->nltype == NLTYPE_FIXED && co | Line 2934 if (common->nltype == NLTYPE_FIXED && co |
2934 | ||
2935 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); |
2936 | 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); |
2937 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER_EQUAL); |
2938 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2939 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); |
2940 | #endif | #endif |
2941 | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); | OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); |
2942 | ||
2943 | loop = LABEL(); | loop = LABEL(); |
2944 | 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)); |
2945 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2946 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); |
2947 | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); | OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); |
2948 | 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); |
2949 | 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); |
2950 | ||
2951 | JUMPHERE(leave); | JUMPHERE(quit); |
2952 | JUMPHERE(firstchar); | JUMPHERE(firstchar); |
2953 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
2954 | ||
# | Line 2229 set_jumps(newline, loop); | Line 2972 set_jumps(newline, loop); |
2972 | ||
2973 | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) | if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) |
2974 | { | { |
2975 | leave = JUMP(SLJIT_JUMP); | quit = JUMP(SLJIT_JUMP); |
2976 | JUMPHERE(foundcr); | JUMPHERE(foundcr); |
2977 | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
2978 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
2979 | 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); |
2980 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
2981 | #ifdef COMPILE_PCRE16 | #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
2982 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); |
2983 | #endif | #endif |
2984 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
2985 | JUMPHERE(notfoundnl); | JUMPHERE(notfoundnl); |
2986 | JUMPHERE(leave); | JUMPHERE(quit); |
2987 | } | } |
2988 | JUMPHERE(lastchar); | JUMPHERE(lastchar); |
2989 | JUMPHERE(firstchar); | JUMPHERE(firstchar); |
2990 | ||
2991 | if (firstline) | if (firstline) |
2992 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); |
2993 | } | } |
2994 | ||
2995 | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks); | |
2996 | ||
2997 | 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) |
2998 | { | { |
2999 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3000 | struct sljit_label *start; | struct sljit_label *start; |
3001 | struct sljit_jump *leave; | struct sljit_jump *quit; |
3002 | struct sljit_jump *found; | struct sljit_jump *found = NULL; |
3003 | jump_list *matches = NULL; | |
3004 | pcre_uint8 inverted_start_bits[32]; | |
3005 | int i; | |
3006 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3007 | struct sljit_jump *jump; | struct sljit_jump *jump; |
3008 | #endif | #endif |
3009 | ||
3010 | for (i = 0; i < 32; ++i) | |
3011 | inverted_start_bits[i] = ~(((pcre_uint8*)start_bits)[i]); | |
3012 | ||
3013 | if (firstline) | if (firstline) |
3014 | { | { |
3015 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); | SLJIT_ASSERT(common->first_line_end != 0); |
3016 | OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); | |
3017 | 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); |
3018 | } | } |
3019 | ||
3020 | start = LABEL(); | start = LABEL(); |
3021 | leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); | quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); |
3022 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
3023 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
3024 | if (common->utf) | if (common->utf) |
3025 | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); | OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); |
3026 | #endif | #endif |
3027 | ||
3028 | if (!check_class_ranges(common, inverted_start_bits, (inverted_start_bits[31] & 0x80) != 0, &matches)) | |
3029 | { | |
3030 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3031 | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); |
3032 | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); | OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); |
3033 | JUMPHERE(jump); | JUMPHERE(jump); |
3034 | #endif | #endif |
3035 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
3036 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
3037 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits); |
3038 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); |
3039 | 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); |
3040 | found = JUMP(SLJIT_C_NOT_ZERO); | found = JUMP(SLJIT_C_NOT_ZERO); |
3041 | } | |
3042 | ||
3043 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
3044 | if (common->utf) | if (common->utf) |
3045 | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); | OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); |
3046 | #endif | #endif |
3047 | 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)); |
3048 | #if defined SUPPORT_UTF && defined COMPILE_PCRE8 | #ifdef SUPPORT_UTF |
3049 | #if defined COMPILE_PCRE8 | |
3050 | if (common->utf) | if (common->utf) |
3051 | { | { |
3052 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); |
3053 | 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_sw)PRIV(utf8_table4) - 0xc0); |
3054 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3055 | } | } |
3056 | #endif | #elif defined COMPILE_PCRE16 |
#if defined SUPPORT_UTF && defined COMPILE_PCRE16 | ||
3057 | if (common->utf) | if (common->utf) |
3058 | { | { |
3059 | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); | CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); |
3060 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
3061 | 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); |
3062 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3063 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
3064 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
3065 | } | } |
3066 | #endif | #endif /* COMPILE_PCRE[8|16] */ |
3067 | #endif /* SUPPORT_UTF */ | |
3068 | JUMPTO(SLJIT_JUMP, start); | JUMPTO(SLJIT_JUMP, start); |
3069 | JUMPHERE(found); | if (found != NULL) |
3070 | JUMPHERE(leave); | JUMPHERE(found); |
3071 | if (matches != NULL) | |
3072 | set_jumps(matches, LABEL()); | |
3073 | JUMPHERE(quit); | |
3074 | ||
3075 | if (firstline) | if (firstline) |
3076 | OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); | OP1(SLJIT_MOV, STR_END, 0, RETURN_ADDR, 0); |
3077 | } | } |
3078 | ||
3079 | 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 3085 struct sljit_jump *alreadyfound; |
3085 | struct sljit_jump *found; | struct sljit_jump *found; |
3086 | struct sljit_jump *foundoc = NULL; | struct sljit_jump *foundoc = NULL; |
3087 | struct sljit_jump *notfound; | struct sljit_jump *notfound; |
3088 | pcre_uchar oc, bit; | pcre_uint32 oc, bit; |
3089 | ||
3090 | SLJIT_ASSERT(common->req_char_ptr != 0); | SLJIT_ASSERT(common->req_char_ptr != 0); |
3091 | 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 3116 if (req_char == oc) |
3116 | else | else |
3117 | { | { |
3118 | bit = req_char ^ oc; | bit = req_char ^ oc; |
3119 | if (ispowerof2(bit)) | if (is_powerof2(bit)) |
3120 | { | { |
3121 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); |
3122 | 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 2392 GET_LOCAL_BASE(TMP3, 0, 0); | Line 3152 GET_LOCAL_BASE(TMP3, 0, 0); |
3152 | /* Drop frames until we reach STACK_TOP. */ | /* Drop frames until we reach STACK_TOP. */ |
3153 | mainloop = LABEL(); | mainloop = LABEL(); |
3154 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); |
3155 | jump = CMP(SLJIT_C_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); | OP2(SLJIT_SUB | SLJIT_SET_S, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0); |
3156 | jump = JUMP(SLJIT_C_SIG_LESS_EQUAL); | |
3157 | ||
3158 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); |
3159 | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); |
3160 | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_w), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_w)); | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_sw), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_sw)); |
3161 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 3 * sizeof(sljit_w)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 3 * sizeof(sljit_sw)); |
3162 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
3163 | ||
3164 | JUMPHERE(jump); | JUMPHERE(jump); |
3165 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); | jump = JUMP(SLJIT_C_SIG_LESS); |
3166 | /* End of dropping frames. */ | /* End of dropping frames. */ |
3167 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3168 | ||
3169 | JUMPHERE(jump); | JUMPHERE(jump); |
3170 | jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setstrbegin); | OP1(SLJIT_NEG, TMP2, 0, TMP2, 0); |
3171 | /* Set string begin. */ | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); |
3172 | OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); |
3173 | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); | OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_sw)); |
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); | ||
JUMPTO(SLJIT_JUMP, mainloop); | ||
JUMPHERE(jump); | ||
if (common->mark_ptr != 0) | ||
{ | ||
jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setmark); | ||
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); | ||
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); | ||
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP2, 0); | ||
JUMPTO(SLJIT_JUMP, mainloop); | ||
JUMPHERE(jump); | ||
} | ||
/* Unknown command. */ | ||
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); | ||
3174 | JUMPTO(SLJIT_JUMP, mainloop); | JUMPTO(SLJIT_JUMP, mainloop); |
3175 | } | } |
3176 | ||
# | Line 2458 if (common->use_ucp) | Line 3203 if (common->use_ucp) |
3203 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
3204 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
3205 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); |
3206 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3207 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); |
3208 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); |
3209 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3210 | JUMPHERE(jump); | JUMPHERE(jump); |
3211 | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); | OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); |
3212 | } | } |
# | Line 2502 if (common->use_ucp) | Line 3247 if (common->use_ucp) |
3247 | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); |
3248 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); |
3249 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); |
3250 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3251 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); |
3252 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); |
3253 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3254 | JUMPHERE(jump); | JUMPHERE(jump); |
3255 | } | } |
3256 | else | else |
# | Line 2537 OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE | Line 3282 OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE |
3282 | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); | sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); |
3283 | } | } |
3284 | ||
3285 | /* | |
3286 | range format: | |
3287 | ||
3288 | ranges[0] = length of the range (max MAX_RANGE_SIZE, -1 means invalid range). | |
3289 | ranges[1] = first bit (0 or 1) | |
3290 | ranges[2-length] = position of the bit change (when the current bit is not equal to the previous) | |
3291 | */ | |
3292 | ||
3293 | static BOOL check_ranges(compiler_common *common, int *ranges, jump_list **backtracks, BOOL readch) | |
3294 | { | |
3295 | DEFINE_COMPILER; | |
3296 | struct sljit_jump *jump; | |
3297 | ||
3298 | if (ranges[0] < 0) | |
3299 | return FALSE; | |
3300 | ||
3301 | switch(ranges[0]) | |
3302 | { | |
3303 | case 1: | |
3304 | if (readch) | |
3305 | read_char(common); | |
3306 | add_jump(compiler, backtracks, CMP(ranges[1] == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | |
3307 | return TRUE; | |
3308 | ||
3309 | case 2: | |
3310 | if (readch) | |
3311 | read_char(common); | |
3312 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]); | |
3313 | add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2])); | |
3314 | return TRUE; | |
3315 | ||
3316 | case 4: | |
3317 | if (ranges[2] + 1 == ranges[3] && ranges[4] + 1 == ranges[5]) | |
3318 | { | |
3319 | if (readch) | |
3320 | read_char(common); | |
3321 | if (ranges[1] != 0) | |
3322 | { | |
3323 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); | |
3324 | add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | |
3325 | } | |
3326 | else | |
3327 | { | |
3328 | jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]); | |
3329 | add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); | |
3330 | JUMPHERE(jump); | |
3331 | } | |
3332 | return TRUE; | |
3333 | } | |
3334 | if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && is_powerof2(ranges[4] - ranges[2])) | |
3335 | { | |
3336 | if (readch) | |
3337 | read_char(common); | |
3338 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4] - ranges[2]); | |
3339 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4]); | |
3340 | add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[5] - ranges[4])); | |
3341 | return TRUE; | |
3342 | } | |
3343 | return FALSE; | |
3344 | ||
3345 | default: | |
3346 | return FALSE; | |
3347 | } | |
3348 | } | |
3349 | ||
3350 | static void get_ctype_ranges(compiler_common *common, int flag, int *ranges) | |
3351 | { | |
3352 | int i, bit, length; | |
3353 | const pcre_uint8 *ctypes = (const pcre_uint8*)common->ctypes; | |
3354 | ||
3355 | bit = ctypes[0] & flag; | |
3356 | ranges[0] = -1; | |
3357 | ranges[1] = bit != 0 ? 1 : 0; | |
3358 | length = 0; | |
3359 | ||
3360 | for (i = 1; i < 256; i++) | |
3361 | if ((ctypes[i] & flag) != bit) | |
3362 | { | |
3363 | if (length >= MAX_RANGE_SIZE) | |
3364 | return; | |
3365 | ranges[2 + length] = i; | |
3366 | length++; | |
3367 | bit ^= flag; | |
3368 | } | |
3369 | ||
3370 | if (bit != 0) | |
3371 | { | |
3372 | if (length >= MAX_RANGE_SIZE) | |
3373 | return; | |
3374 | ranges[2 + length] = 256; | |
3375 | length++; | |
3376 | } | |
3377 | ranges[0] = length; | |
3378 | } | |
3379 | ||
3380 | static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks) | |
3381 | { | |
3382 | int ranges[2 + MAX_RANGE_SIZE]; | |
3383 | pcre_uint8 bit, cbit, all; | |
3384 | int i, byte, length = 0; | |
3385 | ||
3386 | bit = bits[0] & 0x1; | |
3387 | ranges[1] = bit; | |
3388 | /* Can be 0 or 255. */ | |
3389 | all = -bit; | |
3390 | ||
3391 | for (i = 0; i < 256; ) | |
3392 | { | |
3393 | byte = i >> 3; | |
3394 | if ((i & 0x7) == 0 && bits[byte] == all) | |
3395 | i += 8; | |
3396 | else | |
3397 | { | |
3398 | cbit = (bits[byte] >> (i & 0x7)) & 0x1; | |
3399 | if (cbit != bit) | |
3400 | { | |
3401 | if (length >= MAX_RANGE_SIZE) | |
3402 | return FALSE; | |
3403 | ranges[2 + length] = i; | |
3404 | length++; | |
3405 | bit = cbit; | |
3406 | all = -cbit; | |
3407 | } | |
3408 | i++; | |
3409 | } | |
3410 | } | |
3411 | ||
3412 | if (((bit == 0) && nclass) || ((bit == 1) && !nclass)) | |
3413 | { | |
3414 | if (length >= MAX_RANGE_SIZE) | |
3415 | return FALSE; | |
3416 | ranges[2 + length] = 256; | |
3417 | length++; | |
3418 | } | |
3419 | ranges[0] = length; | |
3420 | ||
3421 | return check_ranges(common, ranges, backtracks, FALSE); | |
3422 | } | |
3423 | ||
3424 | static void check_anynewline(compiler_common *common) | static void check_anynewline(compiler_common *common) |
3425 | { | { |
3426 | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ | /* Check whether TMP1 contains a newline character. TMP2 destroyed. */ |
# | Line 2546 sljit_emit_fast_enter(compiler, RETURN_A | Line 3430 sljit_emit_fast_enter(compiler, RETURN_A |
3430 | ||
3431 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); |
3432 | 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); |
3433 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3434 | 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); |
3435 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3436 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3437 | if (common->utf) | if (common->utf) |
3438 | { | { |
3439 | #endif | #endif |
3440 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3441 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); |
3442 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); |
3443 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3444 | } | } |
3445 | #endif | #endif |
3446 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3447 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3448 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3449 | } | } |
3450 | ||
# | Line 2572 DEFINE_COMPILER; | Line 3456 DEFINE_COMPILER; |
3456 | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); | sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); |
3457 | ||
3458 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x09); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x09); |
3459 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
3460 | 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); |
3461 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3462 | 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); |
3463 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3464 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3465 | if (common->utf) | if (common->utf) |
3466 | { | { |
3467 | #endif | #endif |
3468 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3469 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x1680); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x1680); |
3470 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3471 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e); |
3472 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3473 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000); |
3474 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x200A - 0x2000); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x200A - 0x2000); |
3475 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
3476 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x202f - 0x2000); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x202f - 0x2000); |
3477 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3478 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x205f - 0x2000); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x205f - 0x2000); |
3479 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3480 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x3000 - 0x2000); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x3000 - 0x2000); |
3481 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3482 | } | } |
3483 | #endif | #endif |
3484 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3485 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3486 | ||
3487 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3488 | } | } |
# | Line 2612 sljit_emit_fast_enter(compiler, RETURN_A | Line 3496 sljit_emit_fast_enter(compiler, RETURN_A |
3496 | ||
3497 | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); | OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); |
3498 | 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); |
3499 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
3500 | 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); |
3501 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 | #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3502 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3503 | if (common->utf) | if (common->utf) |
3504 | { | { |
3505 | #endif | #endif |
3506 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3507 | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); | OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); |
3508 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); |
3509 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3510 | } | } |
3511 | #endif | #endif |
3512 | #endif /* SUPPORT_UTF || COMPILE_PCRE16 */ | #endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ |
3513 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
3514 | ||
3515 | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); | sljit_emit_fast_return(compiler, RETURN_ADDR, 0); |
3516 | } | } |
# | Line 2712 sljit_emit_fast_return(compiler, RETURN_ | Line 3596 sljit_emit_fast_return(compiler, RETURN_ |
3596 | ||
3597 | #if defined SUPPORT_UTF && defined SUPPORT_UCP | #if defined SUPPORT_UTF && defined SUPPORT_UCP |
3598 | ||
3599 | 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) |
3600 | { | { |
3601 | /* This function would be ineffective to do in JIT level. */ | /* This function would be ineffective to do in JIT level. */ |
3602 | int c1, c2; | pcre_uint32 c1, c2; |
3603 | const pcre_uchar *src2 = args->uchar_ptr; | const pcre_uchar *src2 = args->uchar_ptr; |
3604 | const pcre_uchar *end2 = args->end; | const pcre_uchar *end2 = args->end; |
3605 | const ucd_record *ur; | |
3606 | const pcre_uint32 *pp; | |
3607 | ||
3608 | while (src1 < end1) | while (src1 < end1) |
3609 | { | { |
# | Line 2725 while (src1 < end1) | Line 3611 while (src1 < end1) |
3611 | return (pcre_uchar*)1; | return (pcre_uchar*)1; |
3612 | GETCHARINC(c1, src1); | GETCHARINC(c1, src1); |
3613 | GETCHARINC(c2, src2); | GETCHARINC(c2, src2); |
3614 | if (c1 != c2 && c1 != UCD_OTHERCASE(c2)) return NULL; | ur = GET_UCD(c2); |
3615 | if (c1 != c2 && c1 != c2 + ur->other_case) | |
3616 | { | |
3617 | pp = PRIV(ucd_caseless_sets) + ur->caseset; | |
3618 | for (;;) | |
3619 | { | |
3620 | if (c1 < *pp) return NULL; | |
3621 | if (c1 == *pp++) break; | |
3622 | } | |
3623 | } | |
3624 | } | } |
3625 | return src2; | return src2; |
3626 | } | } |
# | Line 2733 return src2; | Line 3628 return src2; |
3628 | #endif /* SUPPORT_UTF && SUPPORT_UCP */ | #endif /* SUPPORT_UTF && SUPPORT_UCP */ |
3629 | ||
3630 | 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, |
3631 | compare_context* context, jump_list **fallbacks) | compare_context* context, jump_list **backtracks) |
3632 | { | { |
3633 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3634 | unsigned int othercasebit = 0; | unsigned int othercasebit = 0; |
# | Line 2747 if (caseless && char_has_othercase(commo | Line 3642 if (caseless && char_has_othercase(commo |
3642 | othercasebit = char_get_othercase_bit(common, cc); | othercasebit = char_get_othercase_bit(common, cc); |
3643 | SLJIT_ASSERT(othercasebit); | SLJIT_ASSERT(othercasebit); |
3644 | /* Extracting bit difference info. */ | /* Extracting bit difference info. */ |
3645 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3646 | othercasechar = cc + (othercasebit >> 8); | othercasechar = cc + (othercasebit >> 8); |
3647 | othercasebit &= 0xff; | othercasebit &= 0xff; |
3648 | #else | #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 |
3649 | #ifdef COMPILE_PCRE16 | /* Note that this code only handles characters in the BMP. If there |
3650 | ever are characters outside the BMP whose othercase differs in only one | |
3651 | bit from itself (there currently are none), this code will need to be | |
3652 | revised for COMPILE_PCRE32. */ | |
3653 | othercasechar = cc + (othercasebit >> 9); | othercasechar = cc + (othercasebit >> 9); |
3654 | if ((othercasebit & 0x100) != 0) | if ((othercasebit & 0x100) != 0) |
3655 | othercasebit = (othercasebit & 0xff) << 8; | othercasebit = (othercasebit & 0xff) << 8; |
3656 | else | else |
3657 | othercasebit &= 0xff; | othercasebit &= 0xff; |
3658 | #endif | #endif /* COMPILE_PCRE[8|16|32] */ |
#endif | ||
3659 | } | } |
3660 | ||
3661 | if (context->sourcereg == -1) | if (context->sourcereg == -1) |
3662 | { | { |
3663 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3664 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
3665 | if (context->length >= 4) | if (context->length >= 4) |
3666 | 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 3669 if (context->sourcereg == -1) |
3669 | else | else |
3670 | #endif | #endif |
3671 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3672 | #else | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
3673 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED |
3674 | if (context->length >= 4) | if (context->length >= 4) |
3675 | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3676 | else | else |
3677 | #endif | #endif |
3678 | OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3679 | #endif | #elif defined COMPILE_PCRE32 |
3680 | #endif /* COMPILE_PCRE8 */ | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3681 | #endif /* COMPILE_PCRE[8|16|32] */ | |
3682 | context->sourcereg = TMP2; | context->sourcereg = TMP2; |
3683 | } | } |
3684 | ||
# | Line 2795 do | Line 3692 do |
3692 | #endif | #endif |
3693 | ||
3694 | context->length -= IN_UCHARS(1); | context->length -= IN_UCHARS(1); |
3695 | #if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED | #if (defined SLJIT_UNALIGNED && SLJIT_UNALIGNED) && (defined COMPILE_PCRE8 || defined COMPILE_PCRE16) |
3696 | ||
3697 | /* Unaligned read is supported. */ | /* Unaligned read is supported. */ |
3698 | if (othercasebit != 0 && othercasechar == cc) | if (othercasebit != 0 && othercasechar == cc) |
# | Line 2810 do | Line 3707 do |
3707 | } | } |
3708 | context->ucharptr++; | context->ucharptr++; |
3709 | ||
3710 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
3711 | 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)) |
3712 | #else | #else |
3713 | if (context->ucharptr >= 2 || context->length == 0) | if (context->ucharptr >= 2 || context->length == 0) |
# | Line 2818 do | Line 3715 do |
3715 | { | { |
3716 | if (context->length >= 4) | if (context->length >= 4) |
3717 | 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); |
#ifdef COMPILE_PCRE8 | ||
3718 | else if (context->length >= 2) | else if (context->length >= 2) |
3719 | 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); |
3720 | #if defined COMPILE_PCRE8 | |
3721 | else if (context->length >= 1) | else if (context->length >= 1) |
3722 | 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); |
3723 | #else | #endif /* COMPILE_PCRE8 */ |
else if (context->length >= 2) | ||
OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | ||
#endif | ||
3724 | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; |
3725 | ||
3726 | switch(context->ucharptr) | switch(context->ucharptr) |
# | Line 2834 do | Line 3728 do |
3728 | case 4 / sizeof(pcre_uchar): | case 4 / sizeof(pcre_uchar): |
3729 | if (context->oc.asint != 0) | if (context->oc.asint != 0) |
3730 | 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); |
3731 | 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)); |
3732 | break; | break; |
3733 | ||
3734 | case 2 / sizeof(pcre_uchar): | case 2 / sizeof(pcre_uchar): |
3735 | if (context->oc.asushort != 0) | if (context->oc.asushort != 0) |
3736 | 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); |
3737 | 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)); |
3738 | break; | break; |
3739 | ||
3740 | #ifdef COMPILE_PCRE8 | #ifdef COMPILE_PCRE8 |
3741 | case 1: | case 1: |
3742 | if (context->oc.asbyte != 0) | if (context->oc.asbyte != 0) |
3743 | 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); |
3744 | 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)); |
3745 | break; | break; |
3746 | #endif | #endif |
3747 | ||
# | Line 2860 do | Line 3754 do |
3754 | ||
3755 | #else | #else |
3756 | ||
3757 | /* Unaligned read is unsupported. */ | /* Unaligned read is unsupported or in 32 bit mode. */ |
3758 | #ifdef COMPILE_PCRE8 | if (context->length >= 1) |
3759 | if (context->length > 0) | OP1(MOV_UCHAR, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); |
3760 | OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | |
#else | ||
if (context->length > 0) | ||
OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); | ||
#endif | ||
3761 | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; | context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; |
3762 | ||
3763 | if (othercasebit != 0 && othercasechar == cc) | if (othercasebit != 0 && othercasechar == cc) |
3764 | { | { |
3765 | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); | OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); |
3766 | 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)); |
3767 | } | } |
3768 | else | else |
3769 | 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)); |
3770 | ||
3771 | #endif | #endif |
3772 | ||
# | Line 2912 return cc; | Line 3802 return cc; |
3802 | } \ | } \ |
3803 | charoffset = (value); | charoffset = (value); |
3804 | ||
3805 | 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) |
3806 | { | { |
3807 | DEFINE_COMPILER; | DEFINE_COMPILER; |
3808 | jump_list *found = NULL; | jump_list *found = NULL; |
3809 | jump_list **list = (*cc & XCL_NOT) == 0 ? &found : fallbacks; | jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks; |
3810 | unsigned int c; | pcre_int32 c, charoffset; |
3811 | int compares; | const pcre_uint32 *other_cases; |
3812 | struct sljit_jump *jump = NULL; | struct sljit_jump *jump = NULL; |
3813 | pcre_uchar *ccbegin; | pcre_uchar *ccbegin; |
3814 | int compares, invertcmp, numberofcmps; | |
3815 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
3816 | BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; | BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; |
3817 | BOOL charsaved = FALSE; | BOOL charsaved = FALSE; |
3818 | int typereg = TMP1, scriptreg = TMP1; | int typereg = TMP1, scriptreg = TMP1; |
3819 | unsigned int typeoffset; | pcre_int32 typeoffset; |
3820 | #endif | #endif |
int invertcmp, numberofcmps; | ||
unsigned int charoffset; | ||
3821 | ||
3822 | /* Although SUPPORT_UTF must be defined, we are not necessary in utf mode. */ | /* Although SUPPORT_UTF must be defined, we are |
3823 | fallback_at_str_end(common, fallbacks); | not necessary in utf mode even in 8 bit mode. */ |
3824 | detect_partial_match(common, backtracks); | |
3825 | read_char(common); | read_char(common); |
3826 | ||
3827 | if ((*cc++ & XCL_MAP) != 0) | if ((*cc++ & XCL_MAP) != 0) |
# | Line 2944 if ((*cc++ & XCL_MAP) != 0) | Line 3834 if ((*cc++ & XCL_MAP) != 0) |
3834 | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); | jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); |
3835 | #endif | #endif |
3836 | ||
3837 | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); | if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, list)) |
3838 | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); | { |
3839 | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); | OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); |
3840 | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); | OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); |
3841 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc); |
3842 | add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); | OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); |
3843 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); | |
3844 | add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); | |
3845 | } | |
3846 | ||
3847 | #ifndef COMPILE_PCRE8 | #ifndef COMPILE_PCRE8 |
3848 | JUMPHERE(jump); | JUMPHERE(jump); |
# | Line 3022 while (*cc != XCL_END) | Line 3915 while (*cc != XCL_END) |
3915 | needschar = TRUE; | needschar = TRUE; |
3916 | break; | break; |
3917 | ||
3918 | case PT_CLIST: | |
3919 | needschar = TRUE; | |
3920 | break; | |
3921 | ||
3922 | default: | default: |
3923 | SLJIT_ASSERT_STOP(); | SLJIT_ASSERT_STOP(); |
3924 | break; | break; |
# | Line 3058 if (needstype || needsscript) | Line 3955 if (needstype || needsscript) |
3955 | { | { |
3956 | if (scriptreg == TMP1) | if (scriptreg == TMP1) |
3957 | { | { |
3958 | OP1(SLJIT_MOV, scriptreg, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); | OP1(SLJIT_MOV, scriptreg, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); |
3959 | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3); | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3); |
3960 | } | } |
3961 | else | else |
3962 | { | { |
3963 | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3); | OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3); |
3964 | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); | OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); |
3965 | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0); | OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0); |
3966 | } | } |
3967 | } | } |
# | Line 3082 typeoffset = 0; | Line 3979 typeoffset = 0; |
3979 | while (*cc != XCL_END) | while (*cc != XCL_END) |
3980 | { | { |
3981 | compares--; | compares--; |
3982 | invertcmp = (compares == 0 && list != fallbacks); | invertcmp = (compares == 0 && list != backtracks); |
3983 | jump = NULL; | jump = NULL; |
3984 | ||
3985 | if (*cc == XCL_SINGLE) | if (*cc == XCL_SINGLE) |
# | Line 3100 while (*cc != XCL_END) | Line 3997 while (*cc != XCL_END) |
3997 | if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) | if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) |
3998 | { | { |
3999 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); |
4000 | COND_VALUE(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_EQUAL); |
4001 | numberofcmps++; | numberofcmps++; |
4002 | } | } |
4003 | else if (numberofcmps > 0) | else if (numberofcmps > 0) |
4004 | { | { |
4005 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); |
4006 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
4007 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
4008 | numberofcmps = 0; | numberofcmps = 0; |
4009 | } | } |
# | Line 3139 while (*cc != XCL_END) | Line 4036 while (*cc != XCL_END) |
4036 | if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) | if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) |
4037 | { | { |
4038 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); |
4039 | COND_VALUE(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL); |
4040 | numberofcmps++; | numberofcmps++; |
4041 | } | } |
4042 | else if (numberofcmps > 0) | else if (numberofcmps > 0) |
4043 | { | { |
4044 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); |
4045 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
4046 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
4047 | numberofcmps = 0; | numberofcmps = 0; |
4048 | } | } |
# | Line 3164 while (*cc != XCL_END) | Line 4061 while (*cc != XCL_END) |
4061 | switch(*cc) | switch(*cc) |
4062 | { | { |
4063 | case PT_ANY: | case PT_ANY: |
4064 | if (list != fallbacks) | if (list != backtracks) |
4065 | { | { |
4066 | 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)) |
4067 | continue; | continue; |
# | Line 3176 while (*cc != XCL_END) | Line 4073 while (*cc != XCL_END) |
4073 | ||
4074 | case PT_LAMP: | case PT_LAMP: |
4075 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - typeoffset); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - typeoffset); |
4076 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
4077 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ll - typeoffset); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ll - typeoffset); |
4078 | COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
4079 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lt - typeoffset); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lt - typeoffset); |
4080 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); |
4081 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
4082 | break; | break; |
4083 | ||
# | Line 3207 while (*cc != XCL_END) | Line 4104 while (*cc != XCL_END) |
4104 | } | } |
4105 | SET_CHAR_OFFSET(9); | SET_CHAR_OFFSET(9); |
4106 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 13 - 9); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 13 - 9); |
4107 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); |
4108 | if (*cc == PT_SPACE) | if (*cc == PT_SPACE) |
4109 | JUMPHERE(jump); | JUMPHERE(jump); |
4110 | ||
4111 | SET_TYPE_OFFSET(ucp_Zl); | SET_TYPE_OFFSET(ucp_Zl); |
4112 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl); |
4113 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
4114 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
4115 | break; | break; |
4116 | ||
4117 | case PT_WORD: | case PT_WORD: |
4118 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE - charoffset); | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE - charoffset); |
4119 | COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
4120 | /* ... fall through */ | /* Fall through. */ |
4121 | ||
4122 | case PT_ALNUM: | case PT_ALNUM: |
4123 | SET_TYPE_OFFSET(ucp_Ll); | SET_TYPE_OFFSET(ucp_Ll); |
4124 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); |
4125 | COND_VALUE((*cc == PT_ALNUM) ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS((*cc == PT_ALNUM) ? SLJIT_MOV : SLJIT_OR, TMP2, 0, (*cc == PT_ALNUM) ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL); |
4126 | SET_TYPE_OFFSET(ucp_Nd); | SET_TYPE_OFFSET(ucp_Nd); |
4127 | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd); | OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd); |
4128 | COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL); | OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); |
4129 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | |
4130 | break; | |
4131 | ||
4132 | case PT_CLIST: | |
4133 | other_cases = PRIV(ucd_caseless_sets) + cc[1]; | |
4134 | ||
4135 | /* At least three characters are required. | |
4136 | Otherwise this case would be handled by the normal code path. */ | |
4137 | SLJIT_ASSERT(other_cases[0] != NOTACHAR && other_cases[1] != NOTACHAR && other_cases[2] != NOTACHAR); | |
4138 | SLJIT_ASSERT(other_cases[0] < other_cases[1] && other_cases[1] < other_cases[2]); | |
4139 | ||
4140 | /* Optimizing character pairs, if their difference is power of 2. */ | |
4141 | if (is_powerof2(other_cases[1] ^ other_cases[0])) | |
4142 | { | |
4143 | if (charoffset == 0) | |
4144 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); | |
4145 | else | |
4146 | { | |
4147 | OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset); | |
4148 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); | |
4149 | } | |
4150 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[1]); | |
4151 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
4152 | other_cases += 2; | |
4153 | } | |
4154 | else if (is_powerof2(other_cases[2] ^ other_cases[1])) | |
4155 | { | |
4156 | if (charoffset == 0) | |
4157 | OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[2] ^ other_cases[1]); | |
4158 | else | |
4159 | { | |
4160 | OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset); | |
4161 | OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); | |
4162 | } | |
4163 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]); | |
4164 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
4165 | ||
4166 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, other_cases[0] - charoffset); | |
4167 | OP_FLAGS(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); | |
4168 | ||
4169 | other_cases += 3; | |
4170 | } | |
4171 | else | |
4172 | { | |
4173 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset); | |
4174 | OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); | |
4175 | } | |
4176 | ||
4177 | while (*other_cases != NOTACHAR) | |
4178 | { | |
4179 | OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset); | |
4180 | OP_FLAGS(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); | |
4181 | } | |
4182 | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); | jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); |
4183 | break; | break; |
4184 | } | } |
# | Line 3237 while (*cc != XCL_END) | Line 4187 while (*cc != XCL_END) |
4187 | #endif | #endif |
4188 | ||
4189 | if (jump != NULL) | if (jump != NULL) |
4190 | add_jump(compiler, compares > 0 ? list : fallbacks, jump); | add_jump(compiler, compares > 0 ? list : backtracks, jump); |
4191 | } | } |
4192 | ||
4193 | if (found != NULL) | if (found != NULL) |
# | Line 3249 if (found != NULL) | Line 4199 if (found != NULL) |
4199 | ||
4200 | #endif | #endif |
4201 | ||
4202 | 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) |
4203 | { | { |
4204 | DEFINE_COMPILER; | DEFINE_COMPILER; |
4205 | int length; | int length; |
# | Line 3268 switch(type) | Line 4218 switch(type) |
4218 | case OP_SOD: | case OP_SOD: |
4219 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
4220 | 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)); |
4221 | 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)); |
4222 | return cc; | return cc; |
4223 | ||
4224 | case OP_SOM: | case OP_SOM: |
4225 | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); | OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); |
4226 | 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)); |
4227 | 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)); |
4228 | return cc; | return cc; |
4229 | ||
4230 | case OP_NOT_WORD_BOUNDARY: | case OP_NOT_WORD_BOUNDARY: |
4231 | case OP_WORD_BOUNDARY: | case OP_WORD_BOUNDARY: |
4232 | add_jump(compiler, &common->wordboundary, JUMP(SLJIT_FAST_CALL)); | add_jump(compiler, &common->wordboundary, JUMP(SLJIT_FAST_CALL)); |
4233 | 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)); |
4234 | return cc; | return cc; |
4235 | ||
4236 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
4237 | case OP_DIGIT: | case OP_DIGIT: |
4238 | fallback_at_str_end(common, fallbacks); | /* Digits are usually 0-9, so it is worth to optimize them. */ |
4239 | read_char8_type(common); | if (common->digits[0] == -2) |
4240 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); | get_ctype_ranges(common, ctype_digit, common->digits); |
4241 | add_jump(compiler, fallbacks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | detect_partial_match(common, backtracks); |
4242 | /* Flip the starting bit in the negative case. */ | |
4243 | if (type == OP_NOT_DIGIT) | |
4244 | common->digits[1] ^= 1; | |
4245 | if (!check_ranges(common, common->digits, backtracks, TRUE)) | |
4246 | { | |
4247 | read_char8_type(common); | |
4248 | OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); | |
4249 | add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); | |
4250 | } | |
4251 | if (type == OP_NOT_DIGIT) | |
4252 | common->digits[1] ^= 1; | |
4253 | return cc; | return cc; |
4254 | ||
4255 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
4256 | case OP_WHITESPACE: | case OP_WHITESPACE: |
4257 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4258 | read_char8_type(common); | read_char8_type(common); |
4259 | 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); |
4260 | 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)); |
4261 | return cc; | return cc; |
4262 | ||
4263 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
4264 | case OP_WORDCHAR: | case OP_WORDCHAR: |
4265 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4266 | read_char8_type(common); | read_char8_type(common); |
4267 | 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); |
4268 | 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)); |
4269 | return cc; | return cc; |
4270 | ||
4271 | case OP_ANY: | case OP_ANY: |
4272 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4273 | read_char(common); | read_char(common); |
4274 | if (common->nltype == NLTYPE_FIXED && common->newline > 255) | if (common->nltype == NLTYPE_FIXED && common->newline > 255) |
4275 | { | { |
# | Line 3319 switch(type) | Line 4280 switch(type) |
4280 | jump[1] = check_str_end(common); | jump[1] = check_str_end(common); |
4281 | ||
4282 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
4283 | 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)); |
4284 | if (jump[1] != NULL) | if (jump[1] != NULL) |
4285 | JUMPHERE(jump[1]); | JUMPHERE(jump[1]); |
4286 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4287 | } | } |
4288 | else | else |
4289 | check_newlinechar(common, common->nltype, fallbacks, TRUE); | check_newlinechar(common, common->nltype, backtracks, TRUE); |
4290 | return cc; | return cc; |
4291 | ||
4292 | case OP_ALLANY: | case OP_ALLANY: |
4293 | fallback_at_str_end(common, fallbacks); | detect_partial_match(common, backtracks); |
4294 | #ifdef SUPPORT_UTF | #ifdef SUPPORT_UTF |
4295 | if (common->utf) | if (common->utf) |
4296 | { | { |
4297 | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); | OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); |
4298 | 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)); |
4299 | #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 |
4300 | #if defined COMPILE_PCRE8 | |
4301 | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); |
4302 | 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_sw)PRIV(utf8_table4) - 0xc0); |
4303 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
4304 | #else /* COMPILE_PCRE8 */ | #elif defined COMPILE_PCRE16 |
#ifdef COMPILE_PCRE16 | ||
4305 | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); | jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); |
4306 | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); | OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); |
4307 | 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); |
4308 | COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); | OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); |
4309 | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); | OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); |
4310 | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); | OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); |
4311 | #endif /* COMPILE_PCRE16 */ | #endif |
#endif /* COMPILE_PCRE8 */ | ||
4312 | JUMPHERE(jump[0]); | JUMPHERE(jump[0]); |
4313 | #endif /* COMPILE_PCRE[8|16] */ | |
4314 | return cc; | return cc; |
4315 | } | } |
4316 | #endif | #endif |
# | Line 3357 switch(type) | Line 4318 switch(type) |
4318 | return cc; | return cc; |
4319 | ||
4320 |