[279] | 1 | /**********************************************************************
|
---|
| 2 | regexec.c - Onigmo (Oniguruma-mod) (regular expression library)
|
---|
| 3 | **********************************************************************/
|
---|
| 4 | /*-
|
---|
| 5 | * Copyright (c) 2002-2008 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
|
---|
| 6 | * Copyright (c) 2011-2014 K.Takata <kentkt AT csc DOT jp>
|
---|
| 7 | * All rights reserved.
|
---|
| 8 | *
|
---|
| 9 | * Redistribution and use in source and binary forms, with or without
|
---|
| 10 | * modification, are permitted provided that the following conditions
|
---|
| 11 | * are met:
|
---|
| 12 | * 1. Redistributions of source code must retain the above copyright
|
---|
| 13 | * notice, this list of conditions and the following disclaimer.
|
---|
| 14 | * 2. Redistributions in binary form must reproduce the above copyright
|
---|
| 15 | * notice, this list of conditions and the following disclaimer in the
|
---|
| 16 | * documentation and/or other materials provided with the distribution.
|
---|
| 17 | *
|
---|
| 18 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
---|
| 19 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
---|
| 20 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
---|
| 21 | * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
---|
| 22 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
---|
| 23 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
---|
| 24 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
---|
| 25 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
---|
| 26 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
---|
| 27 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
---|
| 28 | * SUCH DAMAGE.
|
---|
| 29 | */
|
---|
| 30 |
|
---|
| 31 | #include "regint.h"
|
---|
| 32 |
|
---|
| 33 | #define USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
| 34 |
|
---|
| 35 | #ifndef USE_DIRECT_THREADED_VM
|
---|
| 36 | # ifdef __GNUC__
|
---|
| 37 | # define USE_DIRECT_THREADED_VM 1
|
---|
| 38 | # else
|
---|
| 39 | # define USE_DIRECT_THREADED_VM 0
|
---|
| 40 | # endif
|
---|
| 41 | #endif
|
---|
| 42 |
|
---|
| 43 | #ifdef USE_CRNL_AS_LINE_TERMINATOR
|
---|
| 44 | #define ONIGENC_IS_MBC_CRNL(enc,p,end) \
|
---|
| 45 | (ONIGENC_MBC_TO_CODE(enc,p,end) == 13 && \
|
---|
| 46 | ONIGENC_MBC_TO_CODE(enc,(p+enclen(enc,p)),end) == 10)
|
---|
| 47 | #define ONIGENC_IS_MBC_NEWLINE_EX(enc,p,start,end,option,check_prev) \
|
---|
| 48 | is_mbc_newline_ex((enc),(p),(start),(end),(option),(check_prev))
|
---|
| 49 | static int
|
---|
| 50 | is_mbc_newline_ex(OnigEncoding enc, const UChar *p, const UChar *start,
|
---|
| 51 | const UChar *end, OnigOptionType option, int check_prev)
|
---|
| 52 | {
|
---|
| 53 | if (IS_NEWLINE_CRLF(option)) {
|
---|
| 54 | if (ONIGENC_MBC_TO_CODE(enc, p, end) == 0x0a) {
|
---|
| 55 | if (check_prev) {
|
---|
| 56 | const UChar *prev = onigenc_get_prev_char_head(enc, start, p);
|
---|
| 57 | if ((prev != NULL) && ONIGENC_MBC_TO_CODE(enc, prev, end) == 0x0d)
|
---|
| 58 | return 0;
|
---|
| 59 | else
|
---|
| 60 | return 1;
|
---|
| 61 | }
|
---|
| 62 | else
|
---|
| 63 | return 1;
|
---|
| 64 | }
|
---|
| 65 | else {
|
---|
| 66 | const UChar *pnext = p + enclen(enc, p);
|
---|
| 67 | if (pnext < end &&
|
---|
| 68 | ONIGENC_MBC_TO_CODE(enc, p, end) == 0x0d &&
|
---|
| 69 | ONIGENC_MBC_TO_CODE(enc, pnext, end) == 0x0a)
|
---|
| 70 | return 1;
|
---|
| 71 | if (ONIGENC_IS_MBC_NEWLINE(enc, p, end))
|
---|
| 72 | return 1;
|
---|
| 73 | return 0;
|
---|
| 74 | }
|
---|
| 75 | }
|
---|
| 76 | else {
|
---|
| 77 | return ONIGENC_IS_MBC_NEWLINE(enc, p, end);
|
---|
| 78 | }
|
---|
| 79 | }
|
---|
| 80 | #else /* USE_CRNL_AS_LINE_TERMINATOR */
|
---|
| 81 | #define ONIGENC_IS_MBC_NEWLINE_EX(enc,p,start,end,option,check_prev) \
|
---|
| 82 | ONIGENC_IS_MBC_NEWLINE((enc), (p), (end))
|
---|
| 83 | #endif /* USE_CRNL_AS_LINE_TERMINATOR */
|
---|
| 84 |
|
---|
| 85 | #ifdef USE_CAPTURE_HISTORY
|
---|
| 86 | static void history_tree_free(OnigCaptureTreeNode* node);
|
---|
| 87 |
|
---|
| 88 | static void
|
---|
| 89 | history_tree_clear(OnigCaptureTreeNode* node)
|
---|
| 90 | {
|
---|
| 91 | int i;
|
---|
| 92 |
|
---|
| 93 | if (IS_NOT_NULL(node)) {
|
---|
| 94 | for (i = 0; i < node->num_childs; i++) {
|
---|
| 95 | if (IS_NOT_NULL(node->childs[i])) {
|
---|
| 96 | history_tree_free(node->childs[i]);
|
---|
| 97 | }
|
---|
| 98 | }
|
---|
| 99 | for (i = 0; i < node->allocated; i++) {
|
---|
| 100 | node->childs[i] = (OnigCaptureTreeNode* )0;
|
---|
| 101 | }
|
---|
| 102 | node->num_childs = 0;
|
---|
| 103 | node->beg = ONIG_REGION_NOTPOS;
|
---|
| 104 | node->end = ONIG_REGION_NOTPOS;
|
---|
| 105 | node->group = -1;
|
---|
| 106 | xfree(node->childs);
|
---|
| 107 | node->childs = (OnigCaptureTreeNode** )0;
|
---|
| 108 | }
|
---|
| 109 | }
|
---|
| 110 |
|
---|
| 111 | static void
|
---|
| 112 | history_tree_free(OnigCaptureTreeNode* node)
|
---|
| 113 | {
|
---|
| 114 | history_tree_clear(node);
|
---|
| 115 | xfree(node);
|
---|
| 116 | }
|
---|
| 117 |
|
---|
| 118 | static void
|
---|
| 119 | history_root_free(OnigRegion* r)
|
---|
| 120 | {
|
---|
| 121 | if (IS_NOT_NULL(r->history_root)) {
|
---|
| 122 | history_tree_free(r->history_root);
|
---|
| 123 | r->history_root = (OnigCaptureTreeNode* )0;
|
---|
| 124 | }
|
---|
| 125 | }
|
---|
| 126 |
|
---|
| 127 | static OnigCaptureTreeNode*
|
---|
| 128 | history_node_new(void)
|
---|
| 129 | {
|
---|
| 130 | OnigCaptureTreeNode* node;
|
---|
| 131 |
|
---|
| 132 | node = (OnigCaptureTreeNode* )xmalloc(sizeof(OnigCaptureTreeNode));
|
---|
| 133 | CHECK_NULL_RETURN(node);
|
---|
| 134 | node->childs = (OnigCaptureTreeNode** )0;
|
---|
| 135 | node->allocated = 0;
|
---|
| 136 | node->num_childs = 0;
|
---|
| 137 | node->group = -1;
|
---|
| 138 | node->beg = ONIG_REGION_NOTPOS;
|
---|
| 139 | node->end = ONIG_REGION_NOTPOS;
|
---|
| 140 |
|
---|
| 141 | return node;
|
---|
| 142 | }
|
---|
| 143 |
|
---|
| 144 | static int
|
---|
| 145 | history_tree_add_child(OnigCaptureTreeNode* parent, OnigCaptureTreeNode* child)
|
---|
| 146 | {
|
---|
| 147 | #define HISTORY_TREE_INIT_ALLOC_SIZE 8
|
---|
| 148 |
|
---|
| 149 | if (parent->num_childs >= parent->allocated) {
|
---|
| 150 | int n, i;
|
---|
| 151 |
|
---|
| 152 | if (IS_NULL(parent->childs)) {
|
---|
| 153 | n = HISTORY_TREE_INIT_ALLOC_SIZE;
|
---|
| 154 | parent->childs =
|
---|
| 155 | (OnigCaptureTreeNode** )xmalloc(sizeof(OnigCaptureTreeNode*) * n);
|
---|
| 156 | CHECK_NULL_RETURN_MEMERR(parent->childs);
|
---|
| 157 | }
|
---|
| 158 | else {
|
---|
| 159 | OnigCaptureTreeNode** tmp;
|
---|
| 160 | n = parent->allocated * 2;
|
---|
| 161 | tmp =
|
---|
| 162 | (OnigCaptureTreeNode** )xrealloc(parent->childs,
|
---|
| 163 | sizeof(OnigCaptureTreeNode*) * n);
|
---|
| 164 | if (tmp == 0) {
|
---|
| 165 | history_tree_clear(parent);
|
---|
| 166 | return ONIGERR_MEMORY;
|
---|
| 167 | }
|
---|
| 168 | parent->childs = tmp;
|
---|
| 169 | }
|
---|
| 170 | for (i = parent->allocated; i < n; i++) {
|
---|
| 171 | parent->childs[i] = (OnigCaptureTreeNode* )0;
|
---|
| 172 | }
|
---|
| 173 | parent->allocated = n;
|
---|
| 174 | }
|
---|
| 175 |
|
---|
| 176 | parent->childs[parent->num_childs] = child;
|
---|
| 177 | parent->num_childs++;
|
---|
| 178 | return 0;
|
---|
| 179 | }
|
---|
| 180 |
|
---|
| 181 | static OnigCaptureTreeNode*
|
---|
| 182 | history_tree_clone(OnigCaptureTreeNode* node)
|
---|
| 183 | {
|
---|
| 184 | int i, r;
|
---|
| 185 | OnigCaptureTreeNode *clone, *child;
|
---|
| 186 |
|
---|
| 187 | clone = history_node_new();
|
---|
| 188 | CHECK_NULL_RETURN(clone);
|
---|
| 189 |
|
---|
| 190 | clone->beg = node->beg;
|
---|
| 191 | clone->end = node->end;
|
---|
| 192 | for (i = 0; i < node->num_childs; i++) {
|
---|
| 193 | child = history_tree_clone(node->childs[i]);
|
---|
| 194 | if (IS_NULL(child)) {
|
---|
| 195 | history_tree_free(clone);
|
---|
| 196 | return (OnigCaptureTreeNode* )0;
|
---|
| 197 | }
|
---|
| 198 | r = history_tree_add_child(clone, child);
|
---|
| 199 | if (r != 0) {
|
---|
| 200 | history_tree_free(child);
|
---|
| 201 | history_tree_free(clone);
|
---|
| 202 | return (OnigCaptureTreeNode* )0;
|
---|
| 203 | }
|
---|
| 204 | }
|
---|
| 205 |
|
---|
| 206 | return clone;
|
---|
| 207 | }
|
---|
| 208 |
|
---|
| 209 | extern OnigCaptureTreeNode*
|
---|
| 210 | onig_get_capture_tree(OnigRegion* region)
|
---|
| 211 | {
|
---|
| 212 | return region->history_root;
|
---|
| 213 | }
|
---|
| 214 | #endif /* USE_CAPTURE_HISTORY */
|
---|
| 215 |
|
---|
| 216 | extern void
|
---|
| 217 | onig_region_clear(OnigRegion* region)
|
---|
| 218 | {
|
---|
| 219 | int i;
|
---|
| 220 |
|
---|
| 221 | for (i = 0; i < region->num_regs; i++) {
|
---|
| 222 | region->beg[i] = region->end[i] = ONIG_REGION_NOTPOS;
|
---|
| 223 | }
|
---|
| 224 | #ifdef USE_CAPTURE_HISTORY
|
---|
| 225 | history_root_free(region);
|
---|
| 226 | #endif
|
---|
| 227 | }
|
---|
| 228 |
|
---|
| 229 | extern int
|
---|
| 230 | onig_region_resize(OnigRegion* region, int n)
|
---|
| 231 | {
|
---|
| 232 | region->num_regs = n;
|
---|
| 233 |
|
---|
| 234 | if (n < ONIG_NREGION)
|
---|
| 235 | n = ONIG_NREGION;
|
---|
| 236 |
|
---|
| 237 | if (region->allocated == 0) {
|
---|
| 238 | region->beg = (OnigPosition* )xmalloc(n * sizeof(OnigPosition));
|
---|
| 239 | if (region->beg == 0)
|
---|
| 240 | return ONIGERR_MEMORY;
|
---|
| 241 |
|
---|
| 242 | region->end = (OnigPosition* )xmalloc(n * sizeof(OnigPosition));
|
---|
| 243 | if (region->end == 0) {
|
---|
| 244 | xfree(region->beg);
|
---|
| 245 | return ONIGERR_MEMORY;
|
---|
| 246 | }
|
---|
| 247 |
|
---|
| 248 | region->allocated = n;
|
---|
| 249 | }
|
---|
| 250 | else if (region->allocated < n) {
|
---|
| 251 | OnigPosition *tmp;
|
---|
| 252 |
|
---|
| 253 | region->allocated = 0;
|
---|
| 254 | tmp = (OnigPosition* )xrealloc(region->beg, n * sizeof(OnigPosition));
|
---|
| 255 | if (tmp == 0) {
|
---|
| 256 | xfree(region->beg);
|
---|
| 257 | xfree(region->end);
|
---|
| 258 | return ONIGERR_MEMORY;
|
---|
| 259 | }
|
---|
| 260 | region->beg = tmp;
|
---|
| 261 | tmp = (OnigPosition* )xrealloc(region->end, n * sizeof(OnigPosition));
|
---|
| 262 | if (tmp == 0) {
|
---|
| 263 | xfree(region->beg);
|
---|
| 264 | xfree(region->end);
|
---|
| 265 | return ONIGERR_MEMORY;
|
---|
| 266 | }
|
---|
| 267 | region->end = tmp;
|
---|
| 268 |
|
---|
| 269 | region->allocated = n;
|
---|
| 270 | }
|
---|
| 271 |
|
---|
| 272 | return 0;
|
---|
| 273 | }
|
---|
| 274 |
|
---|
| 275 | static int
|
---|
| 276 | onig_region_resize_clear(OnigRegion* region, int n)
|
---|
| 277 | {
|
---|
| 278 | int r;
|
---|
| 279 |
|
---|
| 280 | r = onig_region_resize(region, n);
|
---|
| 281 | if (r != 0) return r;
|
---|
| 282 | onig_region_clear(region);
|
---|
| 283 | return 0;
|
---|
| 284 | }
|
---|
| 285 |
|
---|
| 286 | extern int
|
---|
| 287 | onig_region_set(OnigRegion* region, int at, int beg, int end)
|
---|
| 288 | {
|
---|
| 289 | if (at < 0) return ONIGERR_INVALID_ARGUMENT;
|
---|
| 290 |
|
---|
| 291 | if (at >= region->allocated) {
|
---|
| 292 | int r = onig_region_resize(region, at + 1);
|
---|
| 293 | if (r < 0) return r;
|
---|
| 294 | }
|
---|
| 295 |
|
---|
| 296 | region->beg[at] = beg;
|
---|
| 297 | region->end[at] = end;
|
---|
| 298 | return 0;
|
---|
| 299 | }
|
---|
| 300 |
|
---|
| 301 | extern void
|
---|
| 302 | onig_region_init(OnigRegion* region)
|
---|
| 303 | {
|
---|
| 304 | region->num_regs = 0;
|
---|
| 305 | region->allocated = 0;
|
---|
| 306 | region->beg = (OnigPosition* )0;
|
---|
| 307 | region->end = (OnigPosition* )0;
|
---|
| 308 | region->history_root = (OnigCaptureTreeNode* )0;
|
---|
| 309 | }
|
---|
| 310 |
|
---|
| 311 | extern OnigRegion*
|
---|
| 312 | onig_region_new(void)
|
---|
| 313 | {
|
---|
| 314 | OnigRegion* r;
|
---|
| 315 |
|
---|
| 316 | r = (OnigRegion* )xmalloc(sizeof(OnigRegion));
|
---|
| 317 | if (r)
|
---|
| 318 | onig_region_init(r);
|
---|
| 319 | return r;
|
---|
| 320 | }
|
---|
| 321 |
|
---|
| 322 | extern void
|
---|
| 323 | onig_region_free(OnigRegion* r, int free_self)
|
---|
| 324 | {
|
---|
| 325 | if (r) {
|
---|
| 326 | if (r->allocated > 0) {
|
---|
| 327 | if (r->beg) xfree(r->beg);
|
---|
| 328 | if (r->end) xfree(r->end);
|
---|
| 329 | r->allocated = 0;
|
---|
| 330 | }
|
---|
| 331 | #ifdef USE_CAPTURE_HISTORY
|
---|
| 332 | history_root_free(r);
|
---|
| 333 | #endif
|
---|
| 334 | if (free_self) xfree(r);
|
---|
| 335 | }
|
---|
| 336 | }
|
---|
| 337 |
|
---|
| 338 | extern void
|
---|
| 339 | onig_region_copy(OnigRegion* to, OnigRegion* from)
|
---|
| 340 | {
|
---|
| 341 | #define RREGC_SIZE (sizeof(int) * from->num_regs)
|
---|
| 342 | int i, r;
|
---|
| 343 |
|
---|
| 344 | if (to == from) return;
|
---|
| 345 |
|
---|
| 346 | r = onig_region_resize(to, from->num_regs);
|
---|
| 347 | if (r) return;
|
---|
| 348 |
|
---|
| 349 | for (i = 0; i < from->num_regs; i++) {
|
---|
| 350 | to->beg[i] = from->beg[i];
|
---|
| 351 | to->end[i] = from->end[i];
|
---|
| 352 | }
|
---|
| 353 | to->num_regs = from->num_regs;
|
---|
| 354 |
|
---|
| 355 | #ifdef USE_CAPTURE_HISTORY
|
---|
| 356 | history_root_free(to);
|
---|
| 357 |
|
---|
| 358 | if (IS_NOT_NULL(from->history_root)) {
|
---|
| 359 | to->history_root = history_tree_clone(from->history_root);
|
---|
| 360 | }
|
---|
| 361 | #endif
|
---|
| 362 | }
|
---|
| 363 |
|
---|
| 364 |
|
---|
| 365 | /** stack **/
|
---|
| 366 | #define INVALID_STACK_INDEX -1
|
---|
| 367 |
|
---|
| 368 | /* stack type */
|
---|
| 369 | /* used by normal-POP */
|
---|
| 370 | #define STK_ALT 0x0001
|
---|
| 371 | #define STK_LOOK_BEHIND_NOT 0x0002
|
---|
| 372 | #define STK_POS_NOT 0x0003
|
---|
| 373 | /* handled by normal-POP */
|
---|
| 374 | #define STK_MEM_START 0x0100
|
---|
| 375 | #define STK_MEM_END 0x8200
|
---|
| 376 | #define STK_REPEAT_INC 0x0300
|
---|
| 377 | #define STK_STATE_CHECK_MARK 0x1000
|
---|
| 378 | /* avoided by normal-POP */
|
---|
| 379 | #define STK_NULL_CHECK_START 0x3000
|
---|
| 380 | #define STK_NULL_CHECK_END 0x5000 /* for recursive call */
|
---|
| 381 | #define STK_MEM_END_MARK 0x8400
|
---|
| 382 | #define STK_POS 0x0500 /* used when POP-POS */
|
---|
| 383 | #define STK_STOP_BT 0x0600 /* mark for "(?>...)" */
|
---|
| 384 | #define STK_REPEAT 0x0700
|
---|
| 385 | #define STK_CALL_FRAME 0x0800
|
---|
| 386 | #define STK_RETURN 0x0900
|
---|
| 387 | #define STK_VOID 0x0a00 /* for fill a blank */
|
---|
| 388 |
|
---|
| 389 | /* stack type check mask */
|
---|
| 390 | #define STK_MASK_POP_USED 0x00ff
|
---|
| 391 | #define STK_MASK_TO_VOID_TARGET 0x10ff
|
---|
| 392 | #define STK_MASK_MEM_END_OR_MARK 0x8000 /* MEM_END or MEM_END_MARK */
|
---|
| 393 |
|
---|
| 394 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
| 395 | #define MATCH_ARG_INIT(msa, arg_option, arg_region, arg_start, arg_gpos) do {\
|
---|
| 396 | (msa).stack_p = (void* )0;\
|
---|
| 397 | (msa).options = (arg_option);\
|
---|
| 398 | (msa).region = (arg_region);\
|
---|
| 399 | (msa).start = (arg_start);\
|
---|
| 400 | (msa).gpos = (arg_gpos);\
|
---|
| 401 | (msa).best_len = ONIG_MISMATCH;\
|
---|
| 402 | } while(0)
|
---|
| 403 | #else
|
---|
| 404 | #define MATCH_ARG_INIT(msa, arg_option, arg_region, arg_start, arg_gpos) do {\
|
---|
| 405 | (msa).stack_p = (void* )0;\
|
---|
| 406 | (msa).options = (arg_option);\
|
---|
| 407 | (msa).region = (arg_region);\
|
---|
| 408 | (msa).start = (arg_start);\
|
---|
| 409 | (msa).gpos = (arg_gpos);\
|
---|
| 410 | } while(0)
|
---|
| 411 | #endif
|
---|
| 412 |
|
---|
| 413 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 414 |
|
---|
| 415 | #define STATE_CHECK_BUFF_MALLOC_THRESHOLD_SIZE 16
|
---|
| 416 |
|
---|
| 417 | #define STATE_CHECK_BUFF_INIT(msa, str_len, offset, state_num) do { \
|
---|
| 418 | if ((state_num) > 0 && str_len >= STATE_CHECK_STRING_THRESHOLD_LEN) {\
|
---|
| 419 | unsigned int size = (unsigned int )(((str_len) + 1) * (state_num) + 7) >> 3;\
|
---|
| 420 | offset = ((offset) * (state_num)) >> 3;\
|
---|
| 421 | if (size > 0 && offset < size && size < STATE_CHECK_BUFF_MAX_SIZE) {\
|
---|
| 422 | if (size >= STATE_CHECK_BUFF_MALLOC_THRESHOLD_SIZE) {\
|
---|
| 423 | (msa).state_check_buff = (void* )xmalloc(size);\
|
---|
| 424 | CHECK_NULL_RETURN_MEMERR((msa).state_check_buff);\
|
---|
| 425 | }\
|
---|
| 426 | else \
|
---|
| 427 | (msa).state_check_buff = (void* )xalloca(size);\
|
---|
| 428 | xmemset(((char* )((msa).state_check_buff)+(offset)), 0, \
|
---|
| 429 | (size_t )(size - (offset))); \
|
---|
| 430 | (msa).state_check_buff_size = size;\
|
---|
| 431 | }\
|
---|
| 432 | else {\
|
---|
| 433 | (msa).state_check_buff = (void* )0;\
|
---|
| 434 | (msa).state_check_buff_size = 0;\
|
---|
| 435 | }\
|
---|
| 436 | }\
|
---|
| 437 | else {\
|
---|
| 438 | (msa).state_check_buff = (void* )0;\
|
---|
| 439 | (msa).state_check_buff_size = 0;\
|
---|
| 440 | }\
|
---|
| 441 | } while(0)
|
---|
| 442 |
|
---|
| 443 | #define MATCH_ARG_FREE(msa) do {\
|
---|
| 444 | if ((msa).stack_p) xfree((msa).stack_p);\
|
---|
| 445 | if ((msa).state_check_buff_size >= STATE_CHECK_BUFF_MALLOC_THRESHOLD_SIZE) { \
|
---|
| 446 | if ((msa).state_check_buff) xfree((msa).state_check_buff);\
|
---|
| 447 | }\
|
---|
| 448 | } while(0)
|
---|
| 449 | #else /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
| 450 | #define STATE_CHECK_BUFF_INIT(msa, str_len, offset, state_num)
|
---|
| 451 | #define MATCH_ARG_FREE(msa) if ((msa).stack_p) xfree((msa).stack_p)
|
---|
| 452 | #endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
| 453 |
|
---|
| 454 |
|
---|
| 455 |
|
---|
| 456 | #define MAX_PTR_NUM 100
|
---|
| 457 |
|
---|
| 458 | #define STACK_INIT(alloc_addr, heap_addr, ptr_num, stack_num) do {\
|
---|
| 459 | if (ptr_num > MAX_PTR_NUM) {\
|
---|
| 460 | alloc_addr = (char* )xmalloc(sizeof(OnigStackIndex) * (ptr_num));\
|
---|
| 461 | heap_addr = alloc_addr;\
|
---|
| 462 | if (msa->stack_p) {\
|
---|
| 463 | stk_alloc = (OnigStackType* )(msa->stack_p);\
|
---|
| 464 | stk_base = stk_alloc;\
|
---|
| 465 | stk = stk_base;\
|
---|
| 466 | stk_end = stk_base + msa->stack_n;\
|
---|
| 467 | } else {\
|
---|
| 468 | stk_alloc = (OnigStackType* )xmalloc(sizeof(OnigStackType) * (stack_num));\
|
---|
| 469 | stk_base = stk_alloc;\
|
---|
| 470 | stk = stk_base;\
|
---|
| 471 | stk_end = stk_base + (stack_num);\
|
---|
| 472 | }\
|
---|
| 473 | } else if (msa->stack_p) {\
|
---|
| 474 | alloc_addr = (char* )xmalloc(sizeof(OnigStackIndex) * (ptr_num));\
|
---|
| 475 | heap_addr = NULL;\
|
---|
| 476 | stk_alloc = (OnigStackType* )(msa->stack_p);\
|
---|
| 477 | stk_base = stk_alloc;\
|
---|
| 478 | stk = stk_base;\
|
---|
| 479 | stk_end = stk_base + msa->stack_n;\
|
---|
| 480 | }\
|
---|
| 481 | else {\
|
---|
| 482 | alloc_addr = (char* )xmalloc(sizeof(OnigStackIndex) * (ptr_num)\
|
---|
| 483 | + sizeof(OnigStackType) * (stack_num));\
|
---|
| 484 | heap_addr = NULL;\
|
---|
| 485 | stk_alloc = (OnigStackType* )(alloc_addr + sizeof(OnigStackIndex) * (ptr_num));\
|
---|
| 486 | stk_base = stk_alloc;\
|
---|
| 487 | stk = stk_base;\
|
---|
| 488 | stk_end = stk_base + (stack_num);\
|
---|
| 489 | }\
|
---|
| 490 | } while(0)
|
---|
| 491 |
|
---|
| 492 | #define STACK_SAVE do{\
|
---|
| 493 | if (stk_base != stk_alloc) {\
|
---|
| 494 | msa->stack_p = stk_base;\
|
---|
| 495 | msa->stack_n = stk_end - stk_base; /* TODO: check overflow */\
|
---|
| 496 | };\
|
---|
| 497 | } while(0)
|
---|
| 498 |
|
---|
| 499 | static unsigned int MatchStackLimitSize = DEFAULT_MATCH_STACK_LIMIT_SIZE;
|
---|
| 500 |
|
---|
| 501 | extern unsigned int
|
---|
| 502 | onig_get_match_stack_limit_size(void)
|
---|
| 503 | {
|
---|
| 504 | return MatchStackLimitSize;
|
---|
| 505 | }
|
---|
| 506 |
|
---|
| 507 | extern int
|
---|
| 508 | onig_set_match_stack_limit_size(unsigned int size)
|
---|
| 509 | {
|
---|
| 510 | MatchStackLimitSize = size;
|
---|
| 511 | return 0;
|
---|
| 512 | }
|
---|
| 513 |
|
---|
| 514 | static int
|
---|
| 515 | stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end,
|
---|
| 516 | OnigStackType** arg_stk, OnigStackType* stk_alloc, OnigMatchArg* msa)
|
---|
| 517 | {
|
---|
| 518 | size_t n;
|
---|
| 519 | OnigStackType *x, *stk_base, *stk_end, *stk;
|
---|
| 520 |
|
---|
| 521 | stk_base = *arg_stk_base;
|
---|
| 522 | stk_end = *arg_stk_end;
|
---|
| 523 | stk = *arg_stk;
|
---|
| 524 |
|
---|
| 525 | n = stk_end - stk_base;
|
---|
| 526 | if (stk_base == stk_alloc && IS_NULL(msa->stack_p)) {
|
---|
| 527 | x = (OnigStackType* )xmalloc(sizeof(OnigStackType) * n * 2);
|
---|
| 528 | if (IS_NULL(x)) {
|
---|
| 529 | STACK_SAVE;
|
---|
| 530 | return ONIGERR_MEMORY;
|
---|
| 531 | }
|
---|
| 532 | xmemcpy(x, stk_base, n * sizeof(OnigStackType));
|
---|
| 533 | n *= 2;
|
---|
| 534 | }
|
---|
| 535 | else {
|
---|
| 536 | unsigned int limit_size = MatchStackLimitSize;
|
---|
| 537 | n *= 2;
|
---|
| 538 | if (limit_size != 0 && n > limit_size) {
|
---|
| 539 | if ((unsigned int )(stk_end - stk_base) == limit_size)
|
---|
| 540 | return ONIGERR_MATCH_STACK_LIMIT_OVER;
|
---|
| 541 | else
|
---|
| 542 | n = limit_size;
|
---|
| 543 | }
|
---|
| 544 | x = (OnigStackType* )xrealloc(stk_base, sizeof(OnigStackType) * n);
|
---|
| 545 | if (IS_NULL(x)) {
|
---|
| 546 | STACK_SAVE;
|
---|
| 547 | return ONIGERR_MEMORY;
|
---|
| 548 | }
|
---|
| 549 | }
|
---|
| 550 | *arg_stk = x + (stk - stk_base);
|
---|
| 551 | *arg_stk_base = x;
|
---|
| 552 | *arg_stk_end = x + n;
|
---|
| 553 | return 0;
|
---|
| 554 | }
|
---|
| 555 |
|
---|
| 556 | #define STACK_ENSURE(n) do {\
|
---|
| 557 | if (stk_end - stk < (n)) {\
|
---|
| 558 | int r = stack_double(&stk_base, &stk_end, &stk, stk_alloc, msa);\
|
---|
| 559 | if (r != 0) {\
|
---|
| 560 | STACK_SAVE;\
|
---|
| 561 | if (xmalloc_base) xfree(xmalloc_base);\
|
---|
| 562 | return r;\
|
---|
| 563 | }\
|
---|
| 564 | }\
|
---|
| 565 | } while(0)
|
---|
| 566 |
|
---|
| 567 | #define STACK_AT(index) (stk_base + (index))
|
---|
| 568 | #define GET_STACK_INDEX(stk) ((stk) - stk_base)
|
---|
| 569 |
|
---|
| 570 | #define STACK_PUSH_TYPE(stack_type) do {\
|
---|
| 571 | STACK_ENSURE(1);\
|
---|
| 572 | stk->type = (stack_type);\
|
---|
| 573 | STACK_INC;\
|
---|
| 574 | } while(0)
|
---|
| 575 |
|
---|
| 576 | #define IS_TO_VOID_TARGET(stk) (((stk)->type & STK_MASK_TO_VOID_TARGET) != 0)
|
---|
| 577 |
|
---|
| 578 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 579 | #define STATE_CHECK_POS(s,snum) \
|
---|
| 580 | (((s) - str) * num_comb_exp_check + ((snum) - 1))
|
---|
| 581 | #define STATE_CHECK_VAL(v,snum) do {\
|
---|
| 582 | if (state_check_buff != NULL) {\
|
---|
| 583 | int x = STATE_CHECK_POS(s,snum);\
|
---|
| 584 | (v) = state_check_buff[x/8] & (1<<(x%8));\
|
---|
| 585 | }\
|
---|
| 586 | else (v) = 0;\
|
---|
| 587 | } while(0)
|
---|
| 588 |
|
---|
| 589 |
|
---|
| 590 | #define ELSE_IF_STATE_CHECK_MARK(stk) \
|
---|
| 591 | else if ((stk)->type == STK_STATE_CHECK_MARK) { \
|
---|
| 592 | int x = STATE_CHECK_POS(stk->u.state.pstr, stk->u.state.state_check);\
|
---|
| 593 | state_check_buff[x/8] |= (1<<(x%8)); \
|
---|
| 594 | }
|
---|
| 595 |
|
---|
| 596 | #define STACK_PUSH(stack_type,pat,s,sprev,keep) do {\
|
---|
| 597 | STACK_ENSURE(1);\
|
---|
| 598 | stk->type = (stack_type);\
|
---|
| 599 | stk->u.state.pcode = (pat);\
|
---|
| 600 | stk->u.state.pstr = (s);\
|
---|
| 601 | stk->u.state.pstr_prev = (sprev);\
|
---|
| 602 | stk->u.state.state_check = 0;\
|
---|
| 603 | stk->u.state.pkeep = (keep);\
|
---|
| 604 | STACK_INC;\
|
---|
| 605 | } while(0)
|
---|
| 606 |
|
---|
| 607 | #define STACK_PUSH_ENSURED(stack_type,pat) do {\
|
---|
| 608 | stk->type = (stack_type);\
|
---|
| 609 | stk->u.state.pcode = (pat);\
|
---|
| 610 | stk->u.state.state_check = 0;\
|
---|
| 611 | STACK_INC;\
|
---|
| 612 | } while(0)
|
---|
| 613 |
|
---|
| 614 | #define STACK_PUSH_ALT_WITH_STATE_CHECK(pat,s,sprev,snum,keep) do {\
|
---|
| 615 | STACK_ENSURE(1);\
|
---|
| 616 | stk->type = STK_ALT;\
|
---|
| 617 | stk->u.state.pcode = (pat);\
|
---|
| 618 | stk->u.state.pstr = (s);\
|
---|
| 619 | stk->u.state.pstr_prev = (sprev);\
|
---|
| 620 | stk->u.state.state_check = ((state_check_buff != NULL) ? (snum) : 0);\
|
---|
| 621 | stk->u.state.pkeep = (keep);\
|
---|
| 622 | STACK_INC;\
|
---|
| 623 | } while(0)
|
---|
| 624 |
|
---|
| 625 | #define STACK_PUSH_STATE_CHECK(s,snum) do {\
|
---|
| 626 | if (state_check_buff != NULL) {\
|
---|
| 627 | STACK_ENSURE(1);\
|
---|
| 628 | stk->type = STK_STATE_CHECK_MARK;\
|
---|
| 629 | stk->u.state.pstr = (s);\
|
---|
| 630 | stk->u.state.state_check = (snum);\
|
---|
| 631 | STACK_INC;\
|
---|
| 632 | }\
|
---|
| 633 | } while(0)
|
---|
| 634 |
|
---|
| 635 | #else /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
| 636 |
|
---|
| 637 | #define ELSE_IF_STATE_CHECK_MARK(stk)
|
---|
| 638 |
|
---|
| 639 | #define STACK_PUSH(stack_type,pat,s,sprev,keep) do {\
|
---|
| 640 | STACK_ENSURE(1);\
|
---|
| 641 | stk->type = (stack_type);\
|
---|
| 642 | stk->u.state.pcode = (pat);\
|
---|
| 643 | stk->u.state.pstr = (s);\
|
---|
| 644 | stk->u.state.pstr_prev = (sprev);\
|
---|
| 645 | stk->u.state.pkeep = (keep);\
|
---|
| 646 | STACK_INC;\
|
---|
| 647 | } while(0)
|
---|
| 648 |
|
---|
| 649 | #define STACK_PUSH_ENSURED(stack_type,pat) do {\
|
---|
| 650 | stk->type = (stack_type);\
|
---|
| 651 | stk->u.state.pcode = (pat);\
|
---|
| 652 | STACK_INC;\
|
---|
| 653 | } while(0)
|
---|
| 654 | #endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
| 655 |
|
---|
| 656 | #define STACK_PUSH_ALT(pat,s,sprev,keep) STACK_PUSH(STK_ALT,pat,s,sprev,keep)
|
---|
| 657 | #define STACK_PUSH_POS(s,sprev,keep) STACK_PUSH(STK_POS,NULL_UCHARP,s,sprev,keep)
|
---|
| 658 | #define STACK_PUSH_POS_NOT(pat,s,sprev,keep) STACK_PUSH(STK_POS_NOT,pat,s,sprev,keep)
|
---|
| 659 | #define STACK_PUSH_STOP_BT STACK_PUSH_TYPE(STK_STOP_BT)
|
---|
| 660 | #define STACK_PUSH_LOOK_BEHIND_NOT(pat,s,sprev,keep) \
|
---|
| 661 | STACK_PUSH(STK_LOOK_BEHIND_NOT,pat,s,sprev,keep)
|
---|
| 662 |
|
---|
| 663 | #define STACK_PUSH_REPEAT(id, pat) do {\
|
---|
| 664 | STACK_ENSURE(1);\
|
---|
| 665 | stk->type = STK_REPEAT;\
|
---|
| 666 | stk->u.repeat.num = (id);\
|
---|
| 667 | stk->u.repeat.pcode = (pat);\
|
---|
| 668 | stk->u.repeat.count = 0;\
|
---|
| 669 | STACK_INC;\
|
---|
| 670 | } while(0)
|
---|
| 671 |
|
---|
| 672 | #define STACK_PUSH_REPEAT_INC(sindex) do {\
|
---|
| 673 | STACK_ENSURE(1);\
|
---|
| 674 | stk->type = STK_REPEAT_INC;\
|
---|
| 675 | stk->u.repeat_inc.si = (sindex);\
|
---|
| 676 | STACK_INC;\
|
---|
| 677 | } while(0)
|
---|
| 678 |
|
---|
| 679 | #define STACK_PUSH_MEM_START(mnum, s) do {\
|
---|
| 680 | STACK_ENSURE(1);\
|
---|
| 681 | stk->type = STK_MEM_START;\
|
---|
| 682 | stk->u.mem.num = (mnum);\
|
---|
| 683 | stk->u.mem.pstr = (s);\
|
---|
| 684 | stk->u.mem.start = mem_start_stk[mnum];\
|
---|
| 685 | stk->u.mem.end = mem_end_stk[mnum];\
|
---|
| 686 | mem_start_stk[mnum] = GET_STACK_INDEX(stk);\
|
---|
| 687 | mem_end_stk[mnum] = INVALID_STACK_INDEX;\
|
---|
| 688 | STACK_INC;\
|
---|
| 689 | } while(0)
|
---|
| 690 |
|
---|
| 691 | #define STACK_PUSH_MEM_END(mnum, s) do {\
|
---|
| 692 | STACK_ENSURE(1);\
|
---|
| 693 | stk->type = STK_MEM_END;\
|
---|
| 694 | stk->u.mem.num = (mnum);\
|
---|
| 695 | stk->u.mem.pstr = (s);\
|
---|
| 696 | stk->u.mem.start = mem_start_stk[mnum];\
|
---|
| 697 | stk->u.mem.end = mem_end_stk[mnum];\
|
---|
| 698 | mem_end_stk[mnum] = GET_STACK_INDEX(stk);\
|
---|
| 699 | STACK_INC;\
|
---|
| 700 | } while(0)
|
---|
| 701 |
|
---|
| 702 | #define STACK_PUSH_MEM_END_MARK(mnum) do {\
|
---|
| 703 | STACK_ENSURE(1);\
|
---|
| 704 | stk->type = STK_MEM_END_MARK;\
|
---|
| 705 | stk->u.mem.num = (mnum);\
|
---|
| 706 | STACK_INC;\
|
---|
| 707 | } while(0)
|
---|
| 708 |
|
---|
| 709 | #define STACK_GET_MEM_START(mnum, k) do {\
|
---|
| 710 | int level = 0;\
|
---|
| 711 | k = stk;\
|
---|
| 712 | while (k > stk_base) {\
|
---|
| 713 | k--;\
|
---|
| 714 | if ((k->type & STK_MASK_MEM_END_OR_MARK) != 0 \
|
---|
| 715 | && k->u.mem.num == (mnum)) {\
|
---|
| 716 | level++;\
|
---|
| 717 | }\
|
---|
| 718 | else if (k->type == STK_MEM_START && k->u.mem.num == (mnum)) {\
|
---|
| 719 | if (level == 0) break;\
|
---|
| 720 | level--;\
|
---|
| 721 | }\
|
---|
| 722 | }\
|
---|
| 723 | } while(0)
|
---|
| 724 |
|
---|
| 725 | #define STACK_GET_MEM_RANGE(k, mnum, start, end) do {\
|
---|
| 726 | int level = 0;\
|
---|
| 727 | while (k < stk) {\
|
---|
| 728 | if (k->type == STK_MEM_START && k->u.mem.num == (mnum)) {\
|
---|
| 729 | if (level == 0) (start) = k->u.mem.pstr;\
|
---|
| 730 | level++;\
|
---|
| 731 | }\
|
---|
| 732 | else if (k->type == STK_MEM_END && k->u.mem.num == (mnum)) {\
|
---|
| 733 | level--;\
|
---|
| 734 | if (level == 0) {\
|
---|
| 735 | (end) = k->u.mem.pstr;\
|
---|
| 736 | break;\
|
---|
| 737 | }\
|
---|
| 738 | }\
|
---|
| 739 | k++;\
|
---|
| 740 | }\
|
---|
| 741 | } while(0)
|
---|
| 742 |
|
---|
| 743 | #define STACK_PUSH_NULL_CHECK_START(cnum, s) do {\
|
---|
| 744 | STACK_ENSURE(1);\
|
---|
| 745 | stk->type = STK_NULL_CHECK_START;\
|
---|
| 746 | stk->u.null_check.num = (cnum);\
|
---|
| 747 | stk->u.null_check.pstr = (s);\
|
---|
| 748 | STACK_INC;\
|
---|
| 749 | } while(0)
|
---|
| 750 |
|
---|
| 751 | #define STACK_PUSH_NULL_CHECK_END(cnum) do {\
|
---|
| 752 | STACK_ENSURE(1);\
|
---|
| 753 | stk->type = STK_NULL_CHECK_END;\
|
---|
| 754 | stk->u.null_check.num = (cnum);\
|
---|
| 755 | STACK_INC;\
|
---|
| 756 | } while(0)
|
---|
| 757 |
|
---|
| 758 | #define STACK_PUSH_CALL_FRAME(pat) do {\
|
---|
| 759 | STACK_ENSURE(1);\
|
---|
| 760 | stk->type = STK_CALL_FRAME;\
|
---|
| 761 | stk->u.call_frame.ret_addr = (pat);\
|
---|
| 762 | STACK_INC;\
|
---|
| 763 | } while(0)
|
---|
| 764 |
|
---|
| 765 | #define STACK_PUSH_RETURN do {\
|
---|
| 766 | STACK_ENSURE(1);\
|
---|
| 767 | stk->type = STK_RETURN;\
|
---|
| 768 | STACK_INC;\
|
---|
| 769 | } while(0)
|
---|
| 770 |
|
---|
| 771 |
|
---|
| 772 | #ifdef ONIG_DEBUG
|
---|
| 773 | #define STACK_BASE_CHECK(p, at) \
|
---|
| 774 | if ((p) < stk_base) {\
|
---|
| 775 | fprintf(stderr, "at %s\n", at);\
|
---|
| 776 | goto stack_error;\
|
---|
| 777 | }
|
---|
| 778 | #else
|
---|
| 779 | #define STACK_BASE_CHECK(p, at)
|
---|
| 780 | #endif
|
---|
| 781 |
|
---|
| 782 | #define STACK_POP_ONE do {\
|
---|
| 783 | stk--;\
|
---|
| 784 | STACK_BASE_CHECK(stk, "STACK_POP_ONE"); \
|
---|
| 785 | } while(0)
|
---|
| 786 |
|
---|
| 787 | #define STACK_POP do {\
|
---|
| 788 | switch (pop_level) {\
|
---|
| 789 | case STACK_POP_LEVEL_FREE:\
|
---|
| 790 | while (1) {\
|
---|
| 791 | stk--;\
|
---|
| 792 | STACK_BASE_CHECK(stk, "STACK_POP"); \
|
---|
| 793 | if ((stk->type & STK_MASK_POP_USED) != 0) break;\
|
---|
| 794 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
| 795 | }\
|
---|
| 796 | break;\
|
---|
| 797 | case STACK_POP_LEVEL_MEM_START:\
|
---|
| 798 | while (1) {\
|
---|
| 799 | stk--;\
|
---|
| 800 | STACK_BASE_CHECK(stk, "STACK_POP 2"); \
|
---|
| 801 | if ((stk->type & STK_MASK_POP_USED) != 0) break;\
|
---|
| 802 | else if (stk->type == STK_MEM_START) {\
|
---|
| 803 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
| 804 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
| 805 | }\
|
---|
| 806 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
| 807 | }\
|
---|
| 808 | break;\
|
---|
| 809 | default:\
|
---|
| 810 | while (1) {\
|
---|
| 811 | stk--;\
|
---|
| 812 | STACK_BASE_CHECK(stk, "STACK_POP 3"); \
|
---|
| 813 | if ((stk->type & STK_MASK_POP_USED) != 0) break;\
|
---|
| 814 | else if (stk->type == STK_MEM_START) {\
|
---|
| 815 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
| 816 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
| 817 | }\
|
---|
| 818 | else if (stk->type == STK_REPEAT_INC) {\
|
---|
| 819 | STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
---|
| 820 | }\
|
---|
| 821 | else if (stk->type == STK_MEM_END) {\
|
---|
| 822 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
| 823 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
| 824 | }\
|
---|
| 825 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
| 826 | }\
|
---|
| 827 | break;\
|
---|
| 828 | }\
|
---|
| 829 | } while(0)
|
---|
| 830 |
|
---|
| 831 | #define STACK_POP_TIL_POS_NOT do {\
|
---|
| 832 | while (1) {\
|
---|
| 833 | stk--;\
|
---|
| 834 | STACK_BASE_CHECK(stk, "STACK_POP_TIL_POS_NOT"); \
|
---|
| 835 | if (stk->type == STK_POS_NOT) break;\
|
---|
| 836 | else if (stk->type == STK_MEM_START) {\
|
---|
| 837 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
| 838 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
| 839 | }\
|
---|
| 840 | else if (stk->type == STK_REPEAT_INC) {\
|
---|
| 841 | STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
---|
| 842 | }\
|
---|
| 843 | else if (stk->type == STK_MEM_END) {\
|
---|
| 844 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
| 845 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
| 846 | }\
|
---|
| 847 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
| 848 | }\
|
---|
| 849 | } while(0)
|
---|
| 850 |
|
---|
| 851 | #define STACK_POP_TIL_LOOK_BEHIND_NOT do {\
|
---|
| 852 | while (1) {\
|
---|
| 853 | stk--;\
|
---|
| 854 | STACK_BASE_CHECK(stk, "STACK_POP_TIL_LOOK_BEHIND_NOT"); \
|
---|
| 855 | if (stk->type == STK_LOOK_BEHIND_NOT) break;\
|
---|
| 856 | else if (stk->type == STK_MEM_START) {\
|
---|
| 857 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
| 858 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
| 859 | }\
|
---|
| 860 | else if (stk->type == STK_REPEAT_INC) {\
|
---|
| 861 | STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
---|
| 862 | }\
|
---|
| 863 | else if (stk->type == STK_MEM_END) {\
|
---|
| 864 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
| 865 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
| 866 | }\
|
---|
| 867 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
| 868 | }\
|
---|
| 869 | } while(0)
|
---|
| 870 |
|
---|
| 871 | #define STACK_POS_END(k) do {\
|
---|
| 872 | k = stk;\
|
---|
| 873 | while (1) {\
|
---|
| 874 | k--;\
|
---|
| 875 | STACK_BASE_CHECK(k, "STACK_POS_END"); \
|
---|
| 876 | if (IS_TO_VOID_TARGET(k)) {\
|
---|
| 877 | k->type = STK_VOID;\
|
---|
| 878 | }\
|
---|
| 879 | else if (k->type == STK_POS) {\
|
---|
| 880 | k->type = STK_VOID;\
|
---|
| 881 | break;\
|
---|
| 882 | }\
|
---|
| 883 | }\
|
---|
| 884 | } while(0)
|
---|
| 885 |
|
---|
| 886 | #define STACK_STOP_BT_END do {\
|
---|
| 887 | OnigStackType *k = stk;\
|
---|
| 888 | while (1) {\
|
---|
| 889 | k--;\
|
---|
| 890 | STACK_BASE_CHECK(k, "STACK_STOP_BT_END"); \
|
---|
| 891 | if (IS_TO_VOID_TARGET(k)) {\
|
---|
| 892 | k->type = STK_VOID;\
|
---|
| 893 | }\
|
---|
| 894 | else if (k->type == STK_STOP_BT) {\
|
---|
| 895 | k->type = STK_VOID;\
|
---|
| 896 | break;\
|
---|
| 897 | }\
|
---|
| 898 | }\
|
---|
| 899 | } while(0)
|
---|
| 900 |
|
---|
| 901 | #define STACK_NULL_CHECK(isnull,id,s) do {\
|
---|
| 902 | OnigStackType* k = stk;\
|
---|
| 903 | while (1) {\
|
---|
| 904 | k--;\
|
---|
| 905 | STACK_BASE_CHECK(k, "STACK_NULL_CHECK"); \
|
---|
| 906 | if (k->type == STK_NULL_CHECK_START) {\
|
---|
| 907 | if (k->u.null_check.num == (id)) {\
|
---|
| 908 | (isnull) = (k->u.null_check.pstr == (s));\
|
---|
| 909 | break;\
|
---|
| 910 | }\
|
---|
| 911 | }\
|
---|
| 912 | }\
|
---|
| 913 | } while(0)
|
---|
| 914 |
|
---|
| 915 | #define STACK_NULL_CHECK_REC(isnull,id,s) do {\
|
---|
| 916 | int level = 0;\
|
---|
| 917 | OnigStackType* k = stk;\
|
---|
| 918 | while (1) {\
|
---|
| 919 | k--;\
|
---|
| 920 | STACK_BASE_CHECK(k, "STACK_NULL_CHECK_REC"); \
|
---|
| 921 | if (k->type == STK_NULL_CHECK_START) {\
|
---|
| 922 | if (k->u.null_check.num == (id)) {\
|
---|
| 923 | if (level == 0) {\
|
---|
| 924 | (isnull) = (k->u.null_check.pstr == (s));\
|
---|
| 925 | break;\
|
---|
| 926 | }\
|
---|
| 927 | else level--;\
|
---|
| 928 | }\
|
---|
| 929 | }\
|
---|
| 930 | else if (k->type == STK_NULL_CHECK_END) {\
|
---|
| 931 | level++;\
|
---|
| 932 | }\
|
---|
| 933 | }\
|
---|
| 934 | } while(0)
|
---|
| 935 |
|
---|
| 936 | #define STACK_NULL_CHECK_MEMST(isnull,id,s,reg) do {\
|
---|
| 937 | OnigStackType* k = stk;\
|
---|
| 938 | while (1) {\
|
---|
| 939 | k--;\
|
---|
| 940 | STACK_BASE_CHECK(k, "STACK_NULL_CHECK_MEMST"); \
|
---|
| 941 | if (k->type == STK_NULL_CHECK_START) {\
|
---|
| 942 | if (k->u.null_check.num == (id)) {\
|
---|
| 943 | if (k->u.null_check.pstr != (s)) {\
|
---|
| 944 | (isnull) = 0;\
|
---|
| 945 | break;\
|
---|
| 946 | }\
|
---|
| 947 | else {\
|
---|
| 948 | UChar* endp;\
|
---|
| 949 | (isnull) = 1;\
|
---|
| 950 | while (k < stk) {\
|
---|
| 951 | if (k->type == STK_MEM_START) {\
|
---|
| 952 | if (k->u.mem.end == INVALID_STACK_INDEX) {\
|
---|
| 953 | (isnull) = 0; break;\
|
---|
| 954 | }\
|
---|
| 955 | if (BIT_STATUS_AT(reg->bt_mem_end, k->u.mem.num))\
|
---|
| 956 | endp = STACK_AT(k->u.mem.end)->u.mem.pstr;\
|
---|
| 957 | else\
|
---|
| 958 | endp = (UChar* )k->u.mem.end;\
|
---|
| 959 | if (STACK_AT(k->u.mem.start)->u.mem.pstr != endp) {\
|
---|
| 960 | (isnull) = 0; break;\
|
---|
| 961 | }\
|
---|
| 962 | else if (endp != s) {\
|
---|
| 963 | (isnull) = -1; /* empty, but position changed */ \
|
---|
| 964 | }\
|
---|
| 965 | }\
|
---|
| 966 | k++;\
|
---|
| 967 | }\
|
---|
| 968 | break;\
|
---|
| 969 | }\
|
---|
| 970 | }\
|
---|
| 971 | }\
|
---|
| 972 | }\
|
---|
| 973 | } while(0)
|
---|
| 974 |
|
---|
| 975 | #define STACK_NULL_CHECK_MEMST_REC(isnull,id,s,reg) do {\
|
---|
| 976 | int level = 0;\
|
---|
| 977 | OnigStackType* k = stk;\
|
---|
| 978 | while (1) {\
|
---|
| 979 | k--;\
|
---|
| 980 | STACK_BASE_CHECK(k, "STACK_NULL_CHECK_MEMST_REC"); \
|
---|
| 981 | if (k->type == STK_NULL_CHECK_START) {\
|
---|
| 982 | if (k->u.null_check.num == (id)) {\
|
---|
| 983 | if (level == 0) {\
|
---|
| 984 | if (k->u.null_check.pstr != (s)) {\
|
---|
| 985 | (isnull) = 0;\
|
---|
| 986 | break;\
|
---|
| 987 | }\
|
---|
| 988 | else {\
|
---|
| 989 | UChar* endp;\
|
---|
| 990 | (isnull) = 1;\
|
---|
| 991 | while (k < stk) {\
|
---|
| 992 | if (k->type == STK_MEM_START) {\
|
---|
| 993 | if (k->u.mem.end == INVALID_STACK_INDEX) {\
|
---|
| 994 | (isnull) = 0; break;\
|
---|
| 995 | }\
|
---|
| 996 | if (BIT_STATUS_AT(reg->bt_mem_end, k->u.mem.num))\
|
---|
| 997 | endp = STACK_AT(k->u.mem.end)->u.mem.pstr;\
|
---|
| 998 | else\
|
---|
| 999 | endp = (UChar* )k->u.mem.end;\
|
---|
| 1000 | if (STACK_AT(k->u.mem.start)->u.mem.pstr != endp) {\
|
---|
| 1001 | (isnull) = 0; break;\
|
---|
| 1002 | }\
|
---|
| 1003 | else if (endp != s) {\
|
---|
| 1004 | (isnull) = -1; /* empty, but position changed */ \
|
---|
| 1005 | }\
|
---|
| 1006 | }\
|
---|
| 1007 | k++;\
|
---|
| 1008 | }\
|
---|
| 1009 | break;\
|
---|
| 1010 | }\
|
---|
| 1011 | }\
|
---|
| 1012 | else {\
|
---|
| 1013 | level--;\
|
---|
| 1014 | }\
|
---|
| 1015 | }\
|
---|
| 1016 | }\
|
---|
| 1017 | else if (k->type == STK_NULL_CHECK_END) {\
|
---|
| 1018 | if (k->u.null_check.num == (id)) level++;\
|
---|
| 1019 | }\
|
---|
| 1020 | }\
|
---|
| 1021 | } while(0)
|
---|
| 1022 |
|
---|
| 1023 | #define STACK_GET_REPEAT(id, k) do {\
|
---|
| 1024 | int level = 0;\
|
---|
| 1025 | k = stk;\
|
---|
| 1026 | while (1) {\
|
---|
| 1027 | k--;\
|
---|
| 1028 | STACK_BASE_CHECK(k, "STACK_GET_REPEAT"); \
|
---|
| 1029 | if (k->type == STK_REPEAT) {\
|
---|
| 1030 | if (level == 0) {\
|
---|
| 1031 | if (k->u.repeat.num == (id)) {\
|
---|
| 1032 | break;\
|
---|
| 1033 | }\
|
---|
| 1034 | }\
|
---|
| 1035 | }\
|
---|
| 1036 | else if (k->type == STK_CALL_FRAME) level--;\
|
---|
| 1037 | else if (k->type == STK_RETURN) level++;\
|
---|
| 1038 | }\
|
---|
| 1039 | } while(0)
|
---|
| 1040 |
|
---|
| 1041 | #define STACK_RETURN(addr) do {\
|
---|
| 1042 | int level = 0;\
|
---|
| 1043 | OnigStackType* k = stk;\
|
---|
| 1044 | while (1) {\
|
---|
| 1045 | k--;\
|
---|
| 1046 | STACK_BASE_CHECK(k, "STACK_RETURN"); \
|
---|
| 1047 | if (k->type == STK_CALL_FRAME) {\
|
---|
| 1048 | if (level == 0) {\
|
---|
| 1049 | (addr) = k->u.call_frame.ret_addr;\
|
---|
| 1050 | break;\
|
---|
| 1051 | }\
|
---|
| 1052 | else level--;\
|
---|
| 1053 | }\
|
---|
| 1054 | else if (k->type == STK_RETURN)\
|
---|
| 1055 | level++;\
|
---|
| 1056 | }\
|
---|
| 1057 | } while(0)
|
---|
| 1058 |
|
---|
| 1059 |
|
---|
| 1060 | #define STRING_CMP(s1,s2,len) do {\
|
---|
| 1061 | while (len-- > 0) {\
|
---|
| 1062 | if (*s1++ != *s2++) goto fail;\
|
---|
| 1063 | }\
|
---|
| 1064 | } while(0)
|
---|
| 1065 |
|
---|
| 1066 | #define STRING_CMP_IC(case_fold_flag,s1,ps2,len) do {\
|
---|
| 1067 | if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len) == 0) \
|
---|
| 1068 | goto fail; \
|
---|
| 1069 | } while(0)
|
---|
| 1070 |
|
---|
| 1071 | static int string_cmp_ic(OnigEncoding enc, int case_fold_flag,
|
---|
| 1072 | UChar* s1, UChar** ps2, OnigDistance mblen)
|
---|
| 1073 | {
|
---|
| 1074 | UChar buf1[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
| 1075 | UChar buf2[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
| 1076 | UChar *p1, *p2, *end1, *s2, *end2;
|
---|
| 1077 | int len1, len2;
|
---|
| 1078 |
|
---|
| 1079 | s2 = *ps2;
|
---|
| 1080 | end1 = s1 + mblen;
|
---|
| 1081 | end2 = s2 + mblen;
|
---|
| 1082 | while (s1 < end1) {
|
---|
| 1083 | len1 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s1, end1, buf1);
|
---|
| 1084 | len2 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s2, end2, buf2);
|
---|
| 1085 | if (len1 != len2) return 0;
|
---|
| 1086 | p1 = buf1;
|
---|
| 1087 | p2 = buf2;
|
---|
| 1088 | while (len1-- > 0) {
|
---|
| 1089 | if (*p1 != *p2) return 0;
|
---|
| 1090 | p1++;
|
---|
| 1091 | p2++;
|
---|
| 1092 | }
|
---|
| 1093 | }
|
---|
| 1094 |
|
---|
| 1095 | *ps2 = s2;
|
---|
| 1096 | return 1;
|
---|
| 1097 | }
|
---|
| 1098 |
|
---|
| 1099 | #define STRING_CMP_VALUE(s1,s2,len,is_fail) do {\
|
---|
| 1100 | is_fail = 0;\
|
---|
| 1101 | while (len-- > 0) {\
|
---|
| 1102 | if (*s1++ != *s2++) {\
|
---|
| 1103 | is_fail = 1; break;\
|
---|
| 1104 | }\
|
---|
| 1105 | }\
|
---|
| 1106 | } while(0)
|
---|
| 1107 |
|
---|
| 1108 | #define STRING_CMP_VALUE_IC(case_fold_flag,s1,ps2,len,is_fail) do {\
|
---|
| 1109 | if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len) == 0) \
|
---|
| 1110 | is_fail = 1; \
|
---|
| 1111 | else \
|
---|
| 1112 | is_fail = 0; \
|
---|
| 1113 | } while(0)
|
---|
| 1114 |
|
---|
| 1115 |
|
---|
| 1116 | #define IS_EMPTY_STR (str == end)
|
---|
| 1117 | #define ON_STR_BEGIN(s) ((s) == str)
|
---|
| 1118 | #define ON_STR_END(s) ((s) == end)
|
---|
| 1119 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
| 1120 | #define DATA_ENSURE_CHECK1 (s < right_range)
|
---|
| 1121 | #define DATA_ENSURE_CHECK(n) (s + (n) <= right_range)
|
---|
| 1122 | #define DATA_ENSURE(n) if (s + (n) > right_range) goto fail
|
---|
| 1123 | #else
|
---|
| 1124 | #define DATA_ENSURE_CHECK1 (s < end)
|
---|
| 1125 | #define DATA_ENSURE_CHECK(n) (s + (n) <= end)
|
---|
| 1126 | #define DATA_ENSURE(n) if (s + (n) > end) goto fail
|
---|
| 1127 | #endif /* USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE */
|
---|
| 1128 |
|
---|
| 1129 |
|
---|
| 1130 | #ifdef USE_CAPTURE_HISTORY
|
---|
| 1131 | static int
|
---|
| 1132 | make_capture_history_tree(OnigCaptureTreeNode* node, OnigStackType** kp,
|
---|
| 1133 | OnigStackType* stk_top, UChar* str, regex_t* reg)
|
---|
| 1134 | {
|
---|
| 1135 | int n, r;
|
---|
| 1136 | OnigCaptureTreeNode* child;
|
---|
| 1137 | OnigStackType* k = *kp;
|
---|
| 1138 |
|
---|
| 1139 | while (k < stk_top) {
|
---|
| 1140 | if (k->type == STK_MEM_START) {
|
---|
| 1141 | n = k->u.mem.num;
|
---|
| 1142 | if (n <= ONIG_MAX_CAPTURE_HISTORY_GROUP &&
|
---|
| 1143 | BIT_STATUS_AT(reg->capture_history, n) != 0) {
|
---|
| 1144 | child = history_node_new();
|
---|
| 1145 | CHECK_NULL_RETURN_MEMERR(child);
|
---|
| 1146 | child->group = n;
|
---|
| 1147 | child->beg = k->u.mem.pstr - str;
|
---|
| 1148 | r = history_tree_add_child(node, child);
|
---|
| 1149 | if (r != 0) {
|
---|
| 1150 | history_tree_free(child);
|
---|
| 1151 | return r;
|
---|
| 1152 | }
|
---|
| 1153 | *kp = (k + 1);
|
---|
| 1154 | r = make_capture_history_tree(child, kp, stk_top, str, reg);
|
---|
| 1155 | if (r != 0) return r;
|
---|
| 1156 |
|
---|
| 1157 | k = *kp;
|
---|
| 1158 | child->end = k->u.mem.pstr - str;
|
---|
| 1159 | }
|
---|
| 1160 | }
|
---|
| 1161 | else if (k->type == STK_MEM_END) {
|
---|
| 1162 | if (k->u.mem.num == node->group) {
|
---|
| 1163 | node->end = k->u.mem.pstr - str;
|
---|
| 1164 | *kp = k;
|
---|
| 1165 | return 0;
|
---|
| 1166 | }
|
---|
| 1167 | }
|
---|
| 1168 | k++;
|
---|
| 1169 | }
|
---|
| 1170 |
|
---|
| 1171 | return 1; /* 1: root node ending. */
|
---|
| 1172 | }
|
---|
| 1173 | #endif /* USE_CAPTURE_HISTORY */
|
---|
| 1174 |
|
---|
| 1175 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
| 1176 | static int mem_is_in_memp(int mem, int num, UChar* memp)
|
---|
| 1177 | {
|
---|
| 1178 | int i;
|
---|
| 1179 | MemNumType m;
|
---|
| 1180 |
|
---|
| 1181 | for (i = 0; i < num; i++) {
|
---|
| 1182 | GET_MEMNUM_INC(m, memp);
|
---|
| 1183 | if (mem == (int )m) return 1;
|
---|
| 1184 | }
|
---|
| 1185 | return 0;
|
---|
| 1186 | }
|
---|
| 1187 |
|
---|
| 1188 | static int backref_match_at_nested_level(regex_t* reg
|
---|
| 1189 | , OnigStackType* top, OnigStackType* stk_base
|
---|
| 1190 | , int ignore_case, int case_fold_flag
|
---|
| 1191 | , int nest, int mem_num, UChar* memp, UChar** s, const UChar* send)
|
---|
| 1192 | {
|
---|
| 1193 | UChar *ss, *p, *pstart, *pend = NULL_UCHARP;
|
---|
| 1194 | int level;
|
---|
| 1195 | OnigStackType* k;
|
---|
| 1196 |
|
---|
| 1197 | level = 0;
|
---|
| 1198 | k = top;
|
---|
| 1199 | k--;
|
---|
| 1200 | while (k >= stk_base) {
|
---|
| 1201 | if (k->type == STK_CALL_FRAME) {
|
---|
| 1202 | level--;
|
---|
| 1203 | }
|
---|
| 1204 | else if (k->type == STK_RETURN) {
|
---|
| 1205 | level++;
|
---|
| 1206 | }
|
---|
| 1207 | else if (level == nest) {
|
---|
| 1208 | if (k->type == STK_MEM_START) {
|
---|
| 1209 | if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) {
|
---|
| 1210 | pstart = k->u.mem.pstr;
|
---|
| 1211 | if (pend != NULL_UCHARP) {
|
---|
| 1212 | if (pend - pstart > send - *s) return 0; /* or goto next_mem; */
|
---|
| 1213 | p = pstart;
|
---|
| 1214 | ss = *s;
|
---|
| 1215 |
|
---|
| 1216 | if (ignore_case != 0) {
|
---|
| 1217 | if (string_cmp_ic(reg->enc, case_fold_flag,
|
---|
| 1218 | pstart, &ss, pend - pstart) == 0)
|
---|
| 1219 | return 0; /* or goto next_mem; */
|
---|
| 1220 | }
|
---|
| 1221 | else {
|
---|
| 1222 | while (p < pend) {
|
---|
| 1223 | if (*p++ != *ss++) return 0; /* or goto next_mem; */
|
---|
| 1224 | }
|
---|
| 1225 | }
|
---|
| 1226 |
|
---|
| 1227 | *s = ss;
|
---|
| 1228 | return 1;
|
---|
| 1229 | }
|
---|
| 1230 | }
|
---|
| 1231 | }
|
---|
| 1232 | else if (k->type == STK_MEM_END) {
|
---|
| 1233 | if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) {
|
---|
| 1234 | pend = k->u.mem.pstr;
|
---|
| 1235 | }
|
---|
| 1236 | }
|
---|
| 1237 | }
|
---|
| 1238 | k--;
|
---|
| 1239 | }
|
---|
| 1240 |
|
---|
| 1241 | return 0;
|
---|
| 1242 | }
|
---|
| 1243 | #endif /* USE_BACKREF_WITH_LEVEL */
|
---|
| 1244 |
|
---|
| 1245 |
|
---|
| 1246 | #ifdef ONIG_DEBUG_STATISTICS
|
---|
| 1247 |
|
---|
| 1248 | #define USE_TIMEOFDAY
|
---|
| 1249 |
|
---|
| 1250 | #ifdef USE_TIMEOFDAY
|
---|
| 1251 | #ifdef HAVE_SYS_TIME_H
|
---|
| 1252 | #include <sys/time.h>
|
---|
| 1253 | #endif
|
---|
| 1254 | #ifdef HAVE_UNISTD_H
|
---|
| 1255 | #include <unistd.h>
|
---|
| 1256 | #endif
|
---|
| 1257 | static struct timeval ts, te;
|
---|
| 1258 | #define GETTIME(t) gettimeofday(&(t), (struct timezone* )0)
|
---|
| 1259 | #define TIMEDIFF(te,ts) (((te).tv_usec - (ts).tv_usec) + \
|
---|
| 1260 | (((te).tv_sec - (ts).tv_sec)*1000000))
|
---|
| 1261 | #else /* USE_TIMEOFDAY */
|
---|
| 1262 | #ifdef HAVE_SYS_TIMES_H
|
---|
| 1263 | #include <sys/times.h>
|
---|
| 1264 | #endif
|
---|
| 1265 | static struct tms ts, te;
|
---|
| 1266 | #define GETTIME(t) times(&(t))
|
---|
| 1267 | #define TIMEDIFF(te,ts) ((te).tms_utime - (ts).tms_utime)
|
---|
| 1268 | #endif /* USE_TIMEOFDAY */
|
---|
| 1269 |
|
---|
| 1270 | static int OpCounter[256];
|
---|
| 1271 | static int OpPrevCounter[256];
|
---|
| 1272 | static unsigned long OpTime[256];
|
---|
| 1273 | static int OpCurr = OP_FINISH;
|
---|
| 1274 | static int OpPrevTarget = OP_FAIL;
|
---|
| 1275 | static int MaxStackDepth = 0;
|
---|
| 1276 |
|
---|
| 1277 | #define MOP_IN(opcode) do {\
|
---|
| 1278 | if (opcode == OpPrevTarget) OpPrevCounter[OpCurr]++;\
|
---|
| 1279 | OpCurr = opcode;\
|
---|
| 1280 | OpCounter[opcode]++;\
|
---|
| 1281 | GETTIME(ts);\
|
---|
| 1282 | } while(0)
|
---|
| 1283 |
|
---|
| 1284 | #define MOP_OUT do {\
|
---|
| 1285 | GETTIME(te);\
|
---|
| 1286 | OpTime[OpCurr] += TIMEDIFF(te, ts);\
|
---|
| 1287 | } while(0)
|
---|
| 1288 |
|
---|
| 1289 | extern void
|
---|
| 1290 | onig_statistics_init(void)
|
---|
| 1291 | {
|
---|
| 1292 | int i;
|
---|
| 1293 | for (i = 0; i < 256; i++) {
|
---|
| 1294 | OpCounter[i] = OpPrevCounter[i] = 0; OpTime[i] = 0;
|
---|
| 1295 | }
|
---|
| 1296 | MaxStackDepth = 0;
|
---|
| 1297 | }
|
---|
| 1298 |
|
---|
| 1299 | extern void
|
---|
| 1300 | onig_print_statistics(FILE* f)
|
---|
| 1301 | {
|
---|
| 1302 | int i;
|
---|
| 1303 | fprintf(f, " count prev time\n");
|
---|
| 1304 | for (i = 0; OnigOpInfo[i].opcode >= 0; i++) {
|
---|
| 1305 | fprintf(f, "%8d: %8d: %10lu: %s\n",
|
---|
| 1306 | OpCounter[i], OpPrevCounter[i], OpTime[i], OnigOpInfo[i].name);
|
---|
| 1307 | }
|
---|
| 1308 | fprintf(f, "\nmax stack depth: %d\n", MaxStackDepth);
|
---|
| 1309 | }
|
---|
| 1310 |
|
---|
| 1311 | #define STACK_INC do {\
|
---|
| 1312 | stk++;\
|
---|
| 1313 | if (stk - stk_base > MaxStackDepth) \
|
---|
| 1314 | MaxStackDepth = stk - stk_base;\
|
---|
| 1315 | } while(0)
|
---|
| 1316 |
|
---|
| 1317 | #else /* ONIG_DEBUG_STATISTICS */
|
---|
| 1318 | #define STACK_INC stk++
|
---|
| 1319 |
|
---|
| 1320 | #define MOP_IN(opcode)
|
---|
| 1321 | #define MOP_OUT
|
---|
| 1322 | #endif /* ONIG_DEBUG_STATISTICS */
|
---|
| 1323 |
|
---|
| 1324 |
|
---|
| 1325 |
|
---|
| 1326 | /* matching region of POSIX API */
|
---|
| 1327 | typedef int regoff_t;
|
---|
| 1328 |
|
---|
| 1329 | typedef struct {
|
---|
| 1330 | regoff_t rm_so;
|
---|
| 1331 | regoff_t rm_eo;
|
---|
| 1332 | } posix_regmatch_t;
|
---|
| 1333 |
|
---|
| 1334 | /* match data(str - end) from position (sstart). */
|
---|
| 1335 | /* if sstart == str then set sprev to NULL. */
|
---|
| 1336 | static OnigPosition
|
---|
| 1337 | match_at_(regex_t* reg, const UChar* str, const UChar* end,
|
---|
| 1338 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
| 1339 | const UChar* right_range,
|
---|
| 1340 | #endif
|
---|
| 1341 | const UChar* sstart, UChar* sprev, OnigMatchArg* msa, char **palloca_base)
|
---|
| 1342 | {
|
---|
| 1343 | static const UChar FinishCode[] = { OP_FINISH };
|
---|
| 1344 |
|
---|
| 1345 | int i, num_mem, pop_level;
|
---|
| 1346 | ptrdiff_t n, best_len;
|
---|
| 1347 | LengthType tlen, tlen2;
|
---|
| 1348 | MemNumType mem;
|
---|
| 1349 | RelAddrType addr;
|
---|
| 1350 | OnigOptionType option = reg->options;
|
---|
| 1351 | OnigEncoding encode = reg->enc;
|
---|
| 1352 | OnigCaseFoldType case_fold_flag = reg->case_fold_flag;
|
---|
| 1353 | UChar *s, *q, *sbegin;
|
---|
| 1354 | UChar *p = reg->p;
|
---|
| 1355 | UChar *pkeep;
|
---|
| 1356 | char *xmalloc_base = NULL;
|
---|
| 1357 | OnigStackType *stk_alloc, *stk_base, *stk, *stk_end;
|
---|
| 1358 | OnigStackType *stkp; /* used as any purpose. */
|
---|
| 1359 | OnigStackIndex si;
|
---|
| 1360 | OnigStackIndex *repeat_stk;
|
---|
| 1361 | OnigStackIndex *mem_start_stk, *mem_end_stk;
|
---|
| 1362 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 1363 | int scv;
|
---|
| 1364 | unsigned char* state_check_buff = msa->state_check_buff;
|
---|
| 1365 | int num_comb_exp_check = reg->num_comb_exp_check;
|
---|
| 1366 | #endif
|
---|
| 1367 |
|
---|
| 1368 | #if USE_DIRECT_THREADED_VM
|
---|
| 1369 | #define VM_LOOP JUMP;
|
---|
| 1370 | #define VM_LOOP_END
|
---|
| 1371 | #define CASE(x) L_##x: sbegin = s; OPCODE_EXEC_HOOK;
|
---|
| 1372 | #define DEFAULT L_DEFAULT:
|
---|
| 1373 | #define NEXT sprev = sbegin; JUMP
|
---|
| 1374 | #define JUMP goto *oplabels[*p++]
|
---|
| 1375 |
|
---|
| 1376 | static const void *oplabels[] = {
|
---|
| 1377 | &&L_OP_FINISH, /* matching process terminator (no more alternative) */
|
---|
| 1378 | &&L_OP_END, /* pattern code terminator (success end) */
|
---|
| 1379 |
|
---|
| 1380 | &&L_OP_EXACT1, /* single byte, N = 1 */
|
---|
| 1381 | &&L_OP_EXACT2, /* single byte, N = 2 */
|
---|
| 1382 | &&L_OP_EXACT3, /* single byte, N = 3 */
|
---|
| 1383 | &&L_OP_EXACT4, /* single byte, N = 4 */
|
---|
| 1384 | &&L_OP_EXACT5, /* single byte, N = 5 */
|
---|
| 1385 | &&L_OP_EXACTN, /* single byte */
|
---|
| 1386 | &&L_OP_EXACTMB2N1, /* mb-length = 2 N = 1 */
|
---|
| 1387 | &&L_OP_EXACTMB2N2, /* mb-length = 2 N = 2 */
|
---|
| 1388 | &&L_OP_EXACTMB2N3, /* mb-length = 2 N = 3 */
|
---|
| 1389 | &&L_OP_EXACTMB2N, /* mb-length = 2 */
|
---|
| 1390 | &&L_OP_EXACTMB3N, /* mb-length = 3 */
|
---|
| 1391 | &&L_OP_EXACTMBN, /* other length */
|
---|
| 1392 |
|
---|
| 1393 | &&L_OP_EXACT1_IC, /* single byte, N = 1, ignore case */
|
---|
| 1394 | &&L_OP_EXACTN_IC, /* single byte, ignore case */
|
---|
| 1395 |
|
---|
| 1396 | &&L_OP_CCLASS,
|
---|
| 1397 | &&L_OP_CCLASS_MB,
|
---|
| 1398 | &&L_OP_CCLASS_MIX,
|
---|
| 1399 | &&L_OP_CCLASS_NOT,
|
---|
| 1400 | &&L_OP_CCLASS_MB_NOT,
|
---|
| 1401 | &&L_OP_CCLASS_MIX_NOT,
|
---|
| 1402 | &&L_OP_CCLASS_NODE, /* pointer to CClassNode node */
|
---|
| 1403 |
|
---|
| 1404 | &&L_OP_ANYCHAR, /* "." */
|
---|
| 1405 | &&L_OP_ANYCHAR_ML, /* "." multi-line */
|
---|
| 1406 | &&L_OP_ANYCHAR_STAR, /* ".*" */
|
---|
| 1407 | &&L_OP_ANYCHAR_ML_STAR, /* ".*" multi-line */
|
---|
| 1408 | &&L_OP_ANYCHAR_STAR_PEEK_NEXT,
|
---|
| 1409 | &&L_OP_ANYCHAR_ML_STAR_PEEK_NEXT,
|
---|
| 1410 |
|
---|
| 1411 | &&L_OP_WORD,
|
---|
| 1412 | &&L_OP_NOT_WORD,
|
---|
| 1413 | &&L_OP_WORD_BOUND,
|
---|
| 1414 | &&L_OP_NOT_WORD_BOUND,
|
---|
| 1415 | #ifdef USE_WORD_BEGIN_END
|
---|
| 1416 | &&L_OP_WORD_BEGIN,
|
---|
| 1417 | &&L_OP_WORD_END,
|
---|
| 1418 | #else
|
---|
| 1419 | &&L_DEFAULT,
|
---|
| 1420 | &&L_DEFAULT,
|
---|
| 1421 | #endif
|
---|
| 1422 | &&L_OP_ASCII_WORD,
|
---|
| 1423 | &&L_OP_NOT_ASCII_WORD,
|
---|
| 1424 | &&L_OP_ASCII_WORD_BOUND,
|
---|
| 1425 | &&L_OP_NOT_ASCII_WORD_BOUND,
|
---|
| 1426 | #ifdef USE_WORD_BEGIN_END
|
---|
| 1427 | &&L_OP_ASCII_WORD_BEGIN,
|
---|
| 1428 | &&L_OP_ASCII_WORD_END,
|
---|
| 1429 | #else
|
---|
| 1430 | &&L_DEFAULT,
|
---|
| 1431 | &&L_DEFAULT,
|
---|
| 1432 | #endif
|
---|
| 1433 |
|
---|
| 1434 | &&L_OP_BEGIN_BUF,
|
---|
| 1435 | &&L_OP_END_BUF,
|
---|
| 1436 | &&L_OP_BEGIN_LINE,
|
---|
| 1437 | &&L_OP_END_LINE,
|
---|
| 1438 | &&L_OP_SEMI_END_BUF,
|
---|
| 1439 | &&L_OP_BEGIN_POSITION,
|
---|
| 1440 | &&L_OP_BEGIN_POS_OR_LINE, /* used for implicit anchor optimization */
|
---|
| 1441 |
|
---|
| 1442 | &&L_OP_BACKREF1,
|
---|
| 1443 | &&L_OP_BACKREF2,
|
---|
| 1444 | &&L_OP_BACKREFN,
|
---|
| 1445 | &&L_OP_BACKREFN_IC,
|
---|
| 1446 | &&L_OP_BACKREF_MULTI,
|
---|
| 1447 | &&L_OP_BACKREF_MULTI_IC,
|
---|
| 1448 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
| 1449 | &&L_OP_BACKREF_WITH_LEVEL, /* \k<xxx+n>, \k<xxx-n> */
|
---|
| 1450 | #else
|
---|
| 1451 | &&L_DEFAULT,
|
---|
| 1452 | #endif
|
---|
| 1453 | &&L_OP_MEMORY_START,
|
---|
| 1454 | &&L_OP_MEMORY_START_PUSH, /* push back-tracker to stack */
|
---|
| 1455 | &&L_OP_MEMORY_END_PUSH, /* push back-tracker to stack */
|
---|
| 1456 | #ifdef USE_SUBEXP_CALL
|
---|
| 1457 | &&L_OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */
|
---|
| 1458 | #else
|
---|
| 1459 | &&L_DEFAULT,
|
---|
| 1460 | #endif
|
---|
| 1461 | &&L_OP_MEMORY_END,
|
---|
| 1462 | #ifdef USE_SUBEXP_CALL
|
---|
| 1463 | &&L_OP_MEMORY_END_REC, /* push marker to stack */
|
---|
| 1464 | #else
|
---|
| 1465 | &&L_DEFAULT,
|
---|
| 1466 | #endif
|
---|
| 1467 |
|
---|
| 1468 | &&L_OP_KEEP,
|
---|
| 1469 |
|
---|
| 1470 | &&L_OP_FAIL, /* pop stack and move */
|
---|
| 1471 | &&L_OP_JUMP,
|
---|
| 1472 | &&L_OP_PUSH,
|
---|
| 1473 | &&L_OP_POP,
|
---|
| 1474 | &&L_OP_PUSH_OR_JUMP_EXACT1, /* if match exact then push, else jump. */
|
---|
| 1475 | &&L_OP_PUSH_IF_PEEK_NEXT, /* if match exact then push, else none. */
|
---|
| 1476 | &&L_OP_REPEAT, /* {n,m} */
|
---|
| 1477 | &&L_OP_REPEAT_NG, /* {n,m}? (non greedy) */
|
---|
| 1478 | &&L_OP_REPEAT_INC,
|
---|
| 1479 | &&L_OP_REPEAT_INC_NG, /* non greedy */
|
---|
| 1480 | &&L_OP_REPEAT_INC_SG, /* search and get in stack */
|
---|
| 1481 | &&L_OP_REPEAT_INC_NG_SG, /* search and get in stack (non greedy) */
|
---|
| 1482 | &&L_OP_NULL_CHECK_START, /* null loop checker start */
|
---|
| 1483 | &&L_OP_NULL_CHECK_END, /* null loop checker end */
|
---|
| 1484 | #ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT
|
---|
| 1485 | &&L_OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */
|
---|
| 1486 | #else
|
---|
| 1487 | &&L_DEFAULT,
|
---|
| 1488 | #endif
|
---|
| 1489 | #ifdef USE_SUBEXP_CALL
|
---|
| 1490 | &&L_OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */
|
---|
| 1491 | #else
|
---|
| 1492 | &&L_DEFAULT,
|
---|
| 1493 | #endif
|
---|
| 1494 |
|
---|
| 1495 | &&L_OP_PUSH_POS, /* (?=...) start */
|
---|
| 1496 | &&L_OP_POP_POS, /* (?=...) end */
|
---|
| 1497 | &&L_OP_PUSH_POS_NOT, /* (?!...) start */
|
---|
| 1498 | &&L_OP_FAIL_POS, /* (?!...) end */
|
---|
| 1499 | &&L_OP_PUSH_STOP_BT, /* (?>...) start */
|
---|
| 1500 | &&L_OP_POP_STOP_BT, /* (?>...) end */
|
---|
| 1501 | &&L_OP_LOOK_BEHIND, /* (?<=...) start (no needs end opcode) */
|
---|
| 1502 | &&L_OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */
|
---|
| 1503 | &&L_OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end */
|
---|
| 1504 |
|
---|
| 1505 | #ifdef USE_SUBEXP_CALL
|
---|
| 1506 | &&L_OP_CALL, /* \g<name> */
|
---|
| 1507 | &&L_OP_RETURN,
|
---|
| 1508 | #else
|
---|
| 1509 | &&L_DEFAULT,
|
---|
| 1510 | &&L_DEFAULT,
|
---|
| 1511 | #endif
|
---|
| 1512 | &&L_OP_CONDITION,
|
---|
| 1513 |
|
---|
| 1514 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 1515 | &&L_OP_STATE_CHECK_PUSH, /* combination explosion check and push */
|
---|
| 1516 | &&L_OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump */
|
---|
| 1517 | &&L_OP_STATE_CHECK, /* check only */
|
---|
| 1518 | #else
|
---|
| 1519 | &&L_DEFAULT,
|
---|
| 1520 | &&L_DEFAULT,
|
---|
| 1521 | &&L_DEFAULT,
|
---|
| 1522 | #endif
|
---|
| 1523 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 1524 | &&L_OP_STATE_CHECK_ANYCHAR_STAR,
|
---|
| 1525 | &&L_OP_STATE_CHECK_ANYCHAR_ML_STAR,
|
---|
| 1526 | #else
|
---|
| 1527 | &&L_DEFAULT,
|
---|
| 1528 | &&L_DEFAULT,
|
---|
| 1529 | #endif
|
---|
| 1530 | /* no need: IS_DYNAMIC_OPTION() == 0 */
|
---|
| 1531 | #if 0 /* no need: IS_DYNAMIC_OPTION() == 0 */
|
---|
| 1532 | &&L_OP_SET_OPTION_PUSH, /* set option and push recover option */
|
---|
| 1533 | &&L_OP_SET_OPTION /* set option */
|
---|
| 1534 | #else
|
---|
| 1535 | &&L_DEFAULT,
|
---|
| 1536 | &&L_DEFAULT
|
---|
| 1537 | #endif
|
---|
| 1538 | };
|
---|
| 1539 | #else
|
---|
| 1540 |
|
---|
| 1541 | #define VM_LOOP \
|
---|
| 1542 | while (1) { \
|
---|
| 1543 | OPCODE_EXEC_HOOK; \
|
---|
| 1544 | sbegin = s; \
|
---|
| 1545 | switch (*p++) {
|
---|
| 1546 | #define VM_LOOP_END } sprev = sbegin; }
|
---|
| 1547 | #define CASE(x) case x:
|
---|
| 1548 | #define DEFAULT default:
|
---|
| 1549 | #define NEXT break
|
---|
| 1550 | #define JUMP continue; break
|
---|
| 1551 | #endif
|
---|
| 1552 |
|
---|
| 1553 |
|
---|
| 1554 | #ifdef USE_SUBEXP_CALL
|
---|
| 1555 | /* Stack #0 is used to store the pattern itself and used for (?R), \g<0>, etc. */
|
---|
| 1556 | n = reg->num_repeat + (reg->num_mem + 1) * 2;
|
---|
| 1557 |
|
---|
| 1558 | STACK_INIT(*palloca_base, xmalloc_base, n, INIT_MATCH_STACK_SIZE);
|
---|
| 1559 | pop_level = reg->stack_pop_level;
|
---|
| 1560 | num_mem = reg->num_mem;
|
---|
| 1561 | repeat_stk = (OnigStackIndex* )*palloca_base;
|
---|
| 1562 |
|
---|
| 1563 | mem_start_stk = (OnigStackIndex* )(repeat_stk + reg->num_repeat);
|
---|
| 1564 | mem_end_stk = mem_start_stk + (num_mem + 1);
|
---|
| 1565 | #else /* USE_SUBEXP_CALL */
|
---|
| 1566 | /* Stack #0 not is used. */
|
---|
| 1567 | n = reg->num_repeat + reg->num_mem * 2;
|
---|
| 1568 |
|
---|
| 1569 | STACK_INIT(*palloca_base, xmalloc_base, n, INIT_MATCH_STACK_SIZE);
|
---|
| 1570 | pop_level = reg->stack_pop_level;
|
---|
| 1571 | num_mem = reg->num_mem;
|
---|
| 1572 | repeat_stk = (OnigStackIndex* )*palloca_base;
|
---|
| 1573 |
|
---|
| 1574 | mem_start_stk = (OnigStackIndex* )(repeat_stk + reg->num_repeat);
|
---|
| 1575 | mem_end_stk = mem_start_stk + num_mem;
|
---|
| 1576 | mem_start_stk--; /* for index start from 1,
|
---|
| 1577 | mem_start_stk[1]..mem_start_stk[num_mem] */
|
---|
| 1578 | mem_end_stk--; /* for index start from 1,
|
---|
| 1579 | mem_end_stk[1]..mem_end_stk[num_mem] */
|
---|
| 1580 | #endif /* USE_SUBEXP_CALL */
|
---|
| 1581 | {
|
---|
| 1582 | OnigStackIndex *pp = mem_start_stk;
|
---|
| 1583 | for (; pp < (repeat_stk + n); pp+=2) {
|
---|
| 1584 | pp[0] = INVALID_STACK_INDEX;
|
---|
| 1585 | pp[1] = INVALID_STACK_INDEX;
|
---|
| 1586 | }
|
---|
| 1587 | }
|
---|
| 1588 |
|
---|
| 1589 | #ifdef ONIG_DEBUG_MATCH
|
---|
| 1590 | fprintf(stderr, "match_at: str: %"PRIdPTR" (%p), end: %"PRIdPTR" (%p), start: %"PRIdPTR" (%p), sprev: %"PRIdPTR" (%p)\n",
|
---|
| 1591 | (intptr_t )str, str, (intptr_t )end, end, (intptr_t )sstart, sstart, (intptr_t )sprev, sprev);
|
---|
| 1592 | fprintf(stderr, "size: %d, start offset: %d\n",
|
---|
| 1593 | (int )(end - str), (int )(sstart - str));
|
---|
| 1594 | #endif
|
---|
| 1595 |
|
---|
| 1596 | STACK_PUSH_ENSURED(STK_ALT, (UChar* )FinishCode); /* bottom stack */
|
---|
| 1597 | best_len = ONIG_MISMATCH;
|
---|
| 1598 | s = (UChar* )sstart;
|
---|
| 1599 | pkeep = (UChar* )sstart;
|
---|
| 1600 |
|
---|
| 1601 |
|
---|
| 1602 | #ifdef ONIG_DEBUG_MATCH
|
---|
| 1603 | #define OPCODE_EXEC_HOOK \
|
---|
| 1604 | if (s) { \
|
---|
| 1605 | UChar *op, *q, *bp, buf[50]; \
|
---|
| 1606 | int len; \
|
---|
| 1607 | op = p - 1; \
|
---|
| 1608 | fprintf(stderr, "%4"PRIdPTR"> \"", (*op == OP_FINISH) ? (ptrdiff_t )-1 : s - str); \
|
---|
| 1609 | bp = buf; \
|
---|
| 1610 | q = s; \
|
---|
| 1611 | if (*op != OP_FINISH) { /* s may not be a valid pointer if OP_FINISH. */ \
|
---|
| 1612 | for (i = 0; i < 7 && q < end; i++) { \
|
---|
| 1613 | len = enclen(encode, q); \
|
---|
| 1614 | while (len-- > 0) *bp++ = *q++; \
|
---|
| 1615 | } \
|
---|
| 1616 | } \
|
---|
| 1617 | if (q < end) { xmemcpy(bp, "...\"", 4); bp += 4; } \
|
---|
| 1618 | else { xmemcpy(bp, "\"", 1); bp += 1; } \
|
---|
| 1619 | *bp = 0; \
|
---|
| 1620 | fputs((char* )buf, stderr); \
|
---|
| 1621 | for (i = 0; i < 20 - (bp - buf); i++) fputc(' ', stderr); \
|
---|
| 1622 | fprintf(stderr, "%4"PRIdPTR":", (op == FinishCode) ? (ptrdiff_t )-1 : op - reg->p); \
|
---|
| 1623 | onig_print_compiled_byte_code(stderr, op, NULL, encode); \
|
---|
| 1624 | fprintf(stderr, "\n"); \
|
---|
| 1625 | }
|
---|
| 1626 | #else
|
---|
| 1627 | #define OPCODE_EXEC_HOOK ((void) 0)
|
---|
| 1628 | #endif
|
---|
| 1629 |
|
---|
| 1630 |
|
---|
| 1631 | VM_LOOP {
|
---|
| 1632 | CASE(OP_END) MOP_IN(OP_END);
|
---|
| 1633 | n = s - sstart;
|
---|
| 1634 | if (n > best_len) {
|
---|
| 1635 | OnigRegion* region;
|
---|
| 1636 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
| 1637 | if (IS_FIND_LONGEST(option)) {
|
---|
| 1638 | if (n > msa->best_len) {
|
---|
| 1639 | msa->best_len = n;
|
---|
| 1640 | msa->best_s = (UChar* )sstart;
|
---|
| 1641 | }
|
---|
| 1642 | else
|
---|
| 1643 | goto end_best_len;
|
---|
| 1644 | }
|
---|
| 1645 | #endif
|
---|
| 1646 | best_len = n;
|
---|
| 1647 | region = msa->region;
|
---|
| 1648 | if (region) {
|
---|
| 1649 | #ifdef USE_POSIX_API_REGION_OPTION
|
---|
| 1650 | if (IS_POSIX_REGION(msa->options)) {
|
---|
| 1651 | posix_regmatch_t* rmt = (posix_regmatch_t* )region;
|
---|
| 1652 |
|
---|
| 1653 | rmt[0].rm_so = (regoff_t )(((pkeep > s) ? s : pkeep) - str);
|
---|
| 1654 | rmt[0].rm_eo = (regoff_t )(s - str);
|
---|
| 1655 | for (i = 1; i <= num_mem; i++) {
|
---|
| 1656 | if (mem_end_stk[i] != INVALID_STACK_INDEX) {
|
---|
| 1657 | if (BIT_STATUS_AT(reg->bt_mem_start, i))
|
---|
| 1658 | rmt[i].rm_so = (regoff_t )(STACK_AT(mem_start_stk[i])->u.mem.pstr - str);
|
---|
| 1659 | else
|
---|
| 1660 | rmt[i].rm_so = (regoff_t )((UChar* )((void* )(mem_start_stk[i])) - str);
|
---|
| 1661 |
|
---|
| 1662 | rmt[i].rm_eo = (regoff_t )((BIT_STATUS_AT(reg->bt_mem_end, i)
|
---|
| 1663 | ? STACK_AT(mem_end_stk[i])->u.mem.pstr
|
---|
| 1664 | : (UChar* )((void* )mem_end_stk[i])) - str);
|
---|
| 1665 | }
|
---|
| 1666 | else {
|
---|
| 1667 | rmt[i].rm_so = rmt[i].rm_eo = ONIG_REGION_NOTPOS;
|
---|
| 1668 | }
|
---|
| 1669 | }
|
---|
| 1670 | }
|
---|
| 1671 | else {
|
---|
| 1672 | #endif /* USE_POSIX_API_REGION_OPTION */
|
---|
| 1673 | region->beg[0] = ((pkeep > s) ? s : pkeep) - str;
|
---|
| 1674 | region->end[0] = s - str;
|
---|
| 1675 | for (i = 1; i <= num_mem; i++) {
|
---|
| 1676 | if (mem_end_stk[i] != INVALID_STACK_INDEX) {
|
---|
| 1677 | if (BIT_STATUS_AT(reg->bt_mem_start, i))
|
---|
| 1678 | region->beg[i] = STACK_AT(mem_start_stk[i])->u.mem.pstr - str;
|
---|
| 1679 | else
|
---|
| 1680 | region->beg[i] = (UChar* )((void* )mem_start_stk[i]) - str;
|
---|
| 1681 |
|
---|
| 1682 | region->end[i] = (BIT_STATUS_AT(reg->bt_mem_end, i)
|
---|
| 1683 | ? STACK_AT(mem_end_stk[i])->u.mem.pstr
|
---|
| 1684 | : (UChar* )((void* )mem_end_stk[i])) - str;
|
---|
| 1685 | }
|
---|
| 1686 | else {
|
---|
| 1687 | region->beg[i] = region->end[i] = ONIG_REGION_NOTPOS;
|
---|
| 1688 | }
|
---|
| 1689 | }
|
---|
| 1690 |
|
---|
| 1691 | #ifdef USE_CAPTURE_HISTORY
|
---|
| 1692 | if (reg->capture_history != 0) {
|
---|
| 1693 | int r;
|
---|
| 1694 | OnigCaptureTreeNode* node;
|
---|
| 1695 |
|
---|
| 1696 | if (IS_NULL(region->history_root)) {
|
---|
| 1697 | region->history_root = node = history_node_new();
|
---|
| 1698 | CHECK_NULL_RETURN_MEMERR(node);
|
---|
| 1699 | }
|
---|
| 1700 | else {
|
---|
| 1701 | node = region->history_root;
|
---|
| 1702 | history_tree_clear(node);
|
---|
| 1703 | }
|
---|
| 1704 |
|
---|
| 1705 | node->group = 0;
|
---|
| 1706 | node->beg = ((pkeep > s) ? s : pkeep) - str;
|
---|
| 1707 | node->end = s - str;
|
---|
| 1708 |
|
---|
| 1709 | stkp = stk_base;
|
---|
| 1710 | r = make_capture_history_tree(region->history_root, &stkp,
|
---|
| 1711 | stk, (UChar* )str, reg);
|
---|
| 1712 | if (r < 0) {
|
---|
| 1713 | best_len = r; /* error code */
|
---|
| 1714 | goto finish;
|
---|
| 1715 | }
|
---|
| 1716 | }
|
---|
| 1717 | #endif /* USE_CAPTURE_HISTORY */
|
---|
| 1718 | #ifdef USE_POSIX_API_REGION_OPTION
|
---|
| 1719 | } /* else IS_POSIX_REGION() */
|
---|
| 1720 | #endif
|
---|
| 1721 | } /* if (region) */
|
---|
| 1722 | } /* n > best_len */
|
---|
| 1723 |
|
---|
| 1724 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
| 1725 | end_best_len:
|
---|
| 1726 | #endif
|
---|
| 1727 | MOP_OUT;
|
---|
| 1728 |
|
---|
| 1729 | if (IS_FIND_CONDITION(option)) {
|
---|
| 1730 | if (IS_FIND_NOT_EMPTY(option) && s == sstart) {
|
---|
| 1731 | best_len = ONIG_MISMATCH;
|
---|
| 1732 | goto fail; /* for retry */
|
---|
| 1733 | }
|
---|
| 1734 | if (IS_FIND_LONGEST(option) && DATA_ENSURE_CHECK1) {
|
---|
| 1735 | goto fail; /* for retry */
|
---|
| 1736 | }
|
---|
| 1737 | }
|
---|
| 1738 |
|
---|
| 1739 | /* default behavior: return first-matching result. */
|
---|
| 1740 | goto finish;
|
---|
| 1741 | NEXT;
|
---|
| 1742 |
|
---|
| 1743 | CASE(OP_EXACT1) MOP_IN(OP_EXACT1);
|
---|
| 1744 | #if 0
|
---|
| 1745 | DATA_ENSURE(1);
|
---|
| 1746 | if (*p != *s) goto fail;
|
---|
| 1747 | p++; s++;
|
---|
| 1748 | #endif
|
---|
| 1749 | if (*p != *s++) goto fail;
|
---|
| 1750 | DATA_ENSURE(0);
|
---|
| 1751 | p++;
|
---|
| 1752 | MOP_OUT;
|
---|
| 1753 | NEXT;
|
---|
| 1754 |
|
---|
| 1755 | CASE(OP_EXACT1_IC) MOP_IN(OP_EXACT1_IC);
|
---|
| 1756 | {
|
---|
| 1757 | int len;
|
---|
| 1758 | UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
| 1759 |
|
---|
| 1760 | DATA_ENSURE(1);
|
---|
| 1761 | len = ONIGENC_MBC_CASE_FOLD(encode,
|
---|
| 1762 | /* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */
|
---|
| 1763 | case_fold_flag,
|
---|
| 1764 | &s, end, lowbuf);
|
---|
| 1765 | DATA_ENSURE(0);
|
---|
| 1766 | q = lowbuf;
|
---|
| 1767 | while (len-- > 0) {
|
---|
| 1768 | if (*p != *q) {
|
---|
| 1769 | goto fail;
|
---|
| 1770 | }
|
---|
| 1771 | p++; q++;
|
---|
| 1772 | }
|
---|
| 1773 | }
|
---|
| 1774 | MOP_OUT;
|
---|
| 1775 | NEXT;
|
---|
| 1776 |
|
---|
| 1777 | CASE(OP_EXACT2) MOP_IN(OP_EXACT2);
|
---|
| 1778 | DATA_ENSURE(2);
|
---|
| 1779 | if (*p != *s) goto fail;
|
---|
| 1780 | p++; s++;
|
---|
| 1781 | if (*p != *s) goto fail;
|
---|
| 1782 | sprev = s;
|
---|
| 1783 | p++; s++;
|
---|
| 1784 | MOP_OUT;
|
---|
| 1785 | JUMP;
|
---|
| 1786 |
|
---|
| 1787 | CASE(OP_EXACT3) MOP_IN(OP_EXACT3);
|
---|
| 1788 | DATA_ENSURE(3);
|
---|
| 1789 | if (*p != *s) goto fail;
|
---|
| 1790 | p++; s++;
|
---|
| 1791 | if (*p != *s) goto fail;
|
---|
| 1792 | p++; s++;
|
---|
| 1793 | if (*p != *s) goto fail;
|
---|
| 1794 | sprev = s;
|
---|
| 1795 | p++; s++;
|
---|
| 1796 | MOP_OUT;
|
---|
| 1797 | JUMP;
|
---|
| 1798 |
|
---|
| 1799 | CASE(OP_EXACT4) MOP_IN(OP_EXACT4);
|
---|
| 1800 | DATA_ENSURE(4);
|
---|
| 1801 | if (*p != *s) goto fail;
|
---|
| 1802 | p++; s++;
|
---|
| 1803 | if (*p != *s) goto fail;
|
---|
| 1804 | p++; s++;
|
---|
| 1805 | if (*p != *s) goto fail;
|
---|
| 1806 | p++; s++;
|
---|
| 1807 | if (*p != *s) goto fail;
|
---|
| 1808 | sprev = s;
|
---|
| 1809 | p++; s++;
|
---|
| 1810 | MOP_OUT;
|
---|
| 1811 | JUMP;
|
---|
| 1812 |
|
---|
| 1813 | CASE(OP_EXACT5) MOP_IN(OP_EXACT5);
|
---|
| 1814 | DATA_ENSURE(5);
|
---|
| 1815 | if (*p != *s) goto fail;
|
---|
| 1816 | p++; s++;
|
---|
| 1817 | if (*p != *s) goto fail;
|
---|
| 1818 | p++; s++;
|
---|
| 1819 | if (*p != *s) goto fail;
|
---|
| 1820 | p++; s++;
|
---|
| 1821 | if (*p != *s) goto fail;
|
---|
| 1822 | p++; s++;
|
---|
| 1823 | if (*p != *s) goto fail;
|
---|
| 1824 | sprev = s;
|
---|
| 1825 | p++; s++;
|
---|
| 1826 | MOP_OUT;
|
---|
| 1827 | JUMP;
|
---|
| 1828 |
|
---|
| 1829 | CASE(OP_EXACTN) MOP_IN(OP_EXACTN);
|
---|
| 1830 | GET_LENGTH_INC(tlen, p);
|
---|
| 1831 | DATA_ENSURE(tlen);
|
---|
| 1832 | while (tlen-- > 0) {
|
---|
| 1833 | if (*p++ != *s++) goto fail;
|
---|
| 1834 | }
|
---|
| 1835 | sprev = s - 1;
|
---|
| 1836 | MOP_OUT;
|
---|
| 1837 | JUMP;
|
---|
| 1838 |
|
---|
| 1839 | CASE(OP_EXACTN_IC) MOP_IN(OP_EXACTN_IC);
|
---|
| 1840 | {
|
---|
| 1841 | int len;
|
---|
| 1842 | UChar *q, *endp, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
| 1843 |
|
---|
| 1844 | GET_LENGTH_INC(tlen, p);
|
---|
| 1845 | endp = p + tlen;
|
---|
| 1846 |
|
---|
| 1847 | while (p < endp) {
|
---|
| 1848 | sprev = s;
|
---|
| 1849 | DATA_ENSURE(1);
|
---|
| 1850 | len = ONIGENC_MBC_CASE_FOLD(encode,
|
---|
| 1851 | /* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */
|
---|
| 1852 | case_fold_flag,
|
---|
| 1853 | &s, end, lowbuf);
|
---|
| 1854 | DATA_ENSURE(0);
|
---|
| 1855 | q = lowbuf;
|
---|
| 1856 | while (len-- > 0) {
|
---|
| 1857 | if (*p != *q) goto fail;
|
---|
| 1858 | p++; q++;
|
---|
| 1859 | }
|
---|
| 1860 | }
|
---|
| 1861 | }
|
---|
| 1862 |
|
---|
| 1863 | MOP_OUT;
|
---|
| 1864 | JUMP;
|
---|
| 1865 |
|
---|
| 1866 | CASE(OP_EXACTMB2N1) MOP_IN(OP_EXACTMB2N1);
|
---|
| 1867 | DATA_ENSURE(2);
|
---|
| 1868 | if (*p != *s) goto fail;
|
---|
| 1869 | p++; s++;
|
---|
| 1870 | if (*p != *s) goto fail;
|
---|
| 1871 | p++; s++;
|
---|
| 1872 | MOP_OUT;
|
---|
| 1873 | NEXT;
|
---|
| 1874 |
|
---|
| 1875 | CASE(OP_EXACTMB2N2) MOP_IN(OP_EXACTMB2N2);
|
---|
| 1876 | DATA_ENSURE(4);
|
---|
| 1877 | if (*p != *s) goto fail;
|
---|
| 1878 | p++; s++;
|
---|
| 1879 | if (*p != *s) goto fail;
|
---|
| 1880 | p++; s++;
|
---|
| 1881 | sprev = s;
|
---|
| 1882 | if (*p != *s) goto fail;
|
---|
| 1883 | p++; s++;
|
---|
| 1884 | if (*p != *s) goto fail;
|
---|
| 1885 | p++; s++;
|
---|
| 1886 | MOP_OUT;
|
---|
| 1887 | JUMP;
|
---|
| 1888 |
|
---|
| 1889 | CASE(OP_EXACTMB2N3) MOP_IN(OP_EXACTMB2N3);
|
---|
| 1890 | DATA_ENSURE(6);
|
---|
| 1891 | if (*p != *s) goto fail;
|
---|
| 1892 | p++; s++;
|
---|
| 1893 | if (*p != *s) goto fail;
|
---|
| 1894 | p++; s++;
|
---|
| 1895 | if (*p != *s) goto fail;
|
---|
| 1896 | p++; s++;
|
---|
| 1897 | if (*p != *s) goto fail;
|
---|
| 1898 | p++; s++;
|
---|
| 1899 | sprev = s;
|
---|
| 1900 | if (*p != *s) goto fail;
|
---|
| 1901 | p++; s++;
|
---|
| 1902 | if (*p != *s) goto fail;
|
---|
| 1903 | p++; s++;
|
---|
| 1904 | MOP_OUT;
|
---|
| 1905 | JUMP;
|
---|
| 1906 |
|
---|
| 1907 | CASE(OP_EXACTMB2N) MOP_IN(OP_EXACTMB2N);
|
---|
| 1908 | GET_LENGTH_INC(tlen, p);
|
---|
| 1909 | DATA_ENSURE(tlen * 2);
|
---|
| 1910 | while (tlen-- > 0) {
|
---|
| 1911 | if (*p != *s) goto fail;
|
---|
| 1912 | p++; s++;
|
---|
| 1913 | if (*p != *s) goto fail;
|
---|
| 1914 | p++; s++;
|
---|
| 1915 | }
|
---|
| 1916 | sprev = s - 2;
|
---|
| 1917 | MOP_OUT;
|
---|
| 1918 | JUMP;
|
---|
| 1919 |
|
---|
| 1920 | CASE(OP_EXACTMB3N) MOP_IN(OP_EXACTMB3N);
|
---|
| 1921 | GET_LENGTH_INC(tlen, p);
|
---|
| 1922 | DATA_ENSURE(tlen * 3);
|
---|
| 1923 | while (tlen-- > 0) {
|
---|
| 1924 | if (*p != *s) goto fail;
|
---|
| 1925 | p++; s++;
|
---|
| 1926 | if (*p != *s) goto fail;
|
---|
| 1927 | p++; s++;
|
---|
| 1928 | if (*p != *s) goto fail;
|
---|
| 1929 | p++; s++;
|
---|
| 1930 | }
|
---|
| 1931 | sprev = s - 3;
|
---|
| 1932 | MOP_OUT;
|
---|
| 1933 | JUMP;
|
---|
| 1934 |
|
---|
| 1935 | CASE(OP_EXACTMBN) MOP_IN(OP_EXACTMBN);
|
---|
| 1936 | GET_LENGTH_INC(tlen, p); /* mb-len */
|
---|
| 1937 | GET_LENGTH_INC(tlen2, p); /* string len */
|
---|
| 1938 | tlen2 *= tlen;
|
---|
| 1939 | DATA_ENSURE(tlen2);
|
---|
| 1940 | while (tlen2-- > 0) {
|
---|
| 1941 | if (*p != *s) goto fail;
|
---|
| 1942 | p++; s++;
|
---|
| 1943 | }
|
---|
| 1944 | sprev = s - tlen;
|
---|
| 1945 | MOP_OUT;
|
---|
| 1946 | JUMP;
|
---|
| 1947 |
|
---|
| 1948 | CASE(OP_CCLASS) MOP_IN(OP_CCLASS);
|
---|
| 1949 | DATA_ENSURE(1);
|
---|
| 1950 | if (BITSET_AT(((BitSetRef )p), *s) == 0) goto fail;
|
---|
| 1951 | p += SIZE_BITSET;
|
---|
| 1952 | s += enclen(encode, s); /* OP_CCLASS can match mb-code. \D, \S */
|
---|
| 1953 | MOP_OUT;
|
---|
| 1954 | NEXT;
|
---|
| 1955 |
|
---|
| 1956 | CASE(OP_CCLASS_MB) MOP_IN(OP_CCLASS_MB);
|
---|
| 1957 | if (! ONIGENC_IS_MBC_HEAD(encode, s)) goto fail;
|
---|
| 1958 |
|
---|
| 1959 | cclass_mb:
|
---|
| 1960 | GET_LENGTH_INC(tlen, p);
|
---|
| 1961 | {
|
---|
| 1962 | OnigCodePoint code;
|
---|
| 1963 | UChar *ss;
|
---|
| 1964 | int mb_len;
|
---|
| 1965 |
|
---|
| 1966 | DATA_ENSURE(1);
|
---|
| 1967 | mb_len = enclen(encode, s);
|
---|
| 1968 | DATA_ENSURE(mb_len);
|
---|
| 1969 | ss = s;
|
---|
| 1970 | s += mb_len;
|
---|
| 1971 | code = ONIGENC_MBC_TO_CODE(encode, ss, s);
|
---|
| 1972 |
|
---|
| 1973 | #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
---|
| 1974 | if (! onig_is_in_code_range(p, code)) goto fail;
|
---|
| 1975 | #else
|
---|
| 1976 | q = p;
|
---|
| 1977 | ALIGNMENT_RIGHT(q);
|
---|
| 1978 | if (! onig_is_in_code_range(q, code)) goto fail;
|
---|
| 1979 | #endif
|
---|
| 1980 | }
|
---|
| 1981 | p += tlen;
|
---|
| 1982 | MOP_OUT;
|
---|
| 1983 | NEXT;
|
---|
| 1984 |
|
---|
| 1985 | CASE(OP_CCLASS_MIX) MOP_IN(OP_CCLASS_MIX);
|
---|
| 1986 | DATA_ENSURE(1);
|
---|
| 1987 | if (ONIGENC_IS_MBC_HEAD(encode, s)) {
|
---|
| 1988 | p += SIZE_BITSET;
|
---|
| 1989 | goto cclass_mb;
|
---|
| 1990 | }
|
---|
| 1991 | else {
|
---|
| 1992 | if (BITSET_AT(((BitSetRef )p), *s) == 0)
|
---|
| 1993 | goto fail;
|
---|
| 1994 |
|
---|
| 1995 | p += SIZE_BITSET;
|
---|
| 1996 | GET_LENGTH_INC(tlen, p);
|
---|
| 1997 | p += tlen;
|
---|
| 1998 | s++;
|
---|
| 1999 | }
|
---|
| 2000 | MOP_OUT;
|
---|
| 2001 | NEXT;
|
---|
| 2002 |
|
---|
| 2003 | CASE(OP_CCLASS_NOT) MOP_IN(OP_CCLASS_NOT);
|
---|
| 2004 | DATA_ENSURE(1);
|
---|
| 2005 | if (BITSET_AT(((BitSetRef )p), *s) != 0) goto fail;
|
---|
| 2006 | p += SIZE_BITSET;
|
---|
| 2007 | s += enclen(encode, s);
|
---|
| 2008 | MOP_OUT;
|
---|
| 2009 | NEXT;
|
---|
| 2010 |
|
---|
| 2011 | CASE(OP_CCLASS_MB_NOT) MOP_IN(OP_CCLASS_MB_NOT);
|
---|
| 2012 | DATA_ENSURE(1);
|
---|
| 2013 | if (! ONIGENC_IS_MBC_HEAD(encode, s)) {
|
---|
| 2014 | s++;
|
---|
| 2015 | GET_LENGTH_INC(tlen, p);
|
---|
| 2016 | p += tlen;
|
---|
| 2017 | goto cc_mb_not_success;
|
---|
| 2018 | }
|
---|
| 2019 |
|
---|
| 2020 | cclass_mb_not:
|
---|
| 2021 | GET_LENGTH_INC(tlen, p);
|
---|
| 2022 | {
|
---|
| 2023 | OnigCodePoint code;
|
---|
| 2024 | UChar *ss;
|
---|
| 2025 | int mb_len = enclen(encode, s);
|
---|
| 2026 |
|
---|
| 2027 | if (! DATA_ENSURE_CHECK(mb_len)) {
|
---|
| 2028 | DATA_ENSURE(1);
|
---|
| 2029 | s = (UChar* )end;
|
---|
| 2030 | p += tlen;
|
---|
| 2031 | goto cc_mb_not_success;
|
---|
| 2032 | }
|
---|
| 2033 |
|
---|
| 2034 | ss = s;
|
---|
| 2035 | s += mb_len;
|
---|
| 2036 | code = ONIGENC_MBC_TO_CODE(encode, ss, s);
|
---|
| 2037 |
|
---|
| 2038 | #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
---|
| 2039 | if (onig_is_in_code_range(p, code)) goto fail;
|
---|
| 2040 | #else
|
---|
| 2041 | q = p;
|
---|
| 2042 | ALIGNMENT_RIGHT(q);
|
---|
| 2043 | if (onig_is_in_code_range(q, code)) goto fail;
|
---|
| 2044 | #endif
|
---|
| 2045 | }
|
---|
| 2046 | p += tlen;
|
---|
| 2047 |
|
---|
| 2048 | cc_mb_not_success:
|
---|
| 2049 | MOP_OUT;
|
---|
| 2050 | NEXT;
|
---|
| 2051 |
|
---|
| 2052 | CASE(OP_CCLASS_MIX_NOT) MOP_IN(OP_CCLASS_MIX_NOT);
|
---|
| 2053 | DATA_ENSURE(1);
|
---|
| 2054 | if (ONIGENC_IS_MBC_HEAD(encode, s)) {
|
---|
| 2055 | p += SIZE_BITSET;
|
---|
| 2056 | goto cclass_mb_not;
|
---|
| 2057 | }
|
---|
| 2058 | else {
|
---|
| 2059 | if (BITSET_AT(((BitSetRef )p), *s) != 0)
|
---|
| 2060 | goto fail;
|
---|
| 2061 |
|
---|
| 2062 | p += SIZE_BITSET;
|
---|
| 2063 | GET_LENGTH_INC(tlen, p);
|
---|
| 2064 | p += tlen;
|
---|
| 2065 | s++;
|
---|
| 2066 | }
|
---|
| 2067 | MOP_OUT;
|
---|
| 2068 | NEXT;
|
---|
| 2069 |
|
---|
| 2070 | CASE(OP_CCLASS_NODE) MOP_IN(OP_CCLASS_NODE);
|
---|
| 2071 | {
|
---|
| 2072 | OnigCodePoint code;
|
---|
| 2073 | void *node;
|
---|
| 2074 | int mb_len;
|
---|
| 2075 | UChar *ss;
|
---|
| 2076 |
|
---|
| 2077 | DATA_ENSURE(1);
|
---|
| 2078 | GET_POINTER_INC(node, p);
|
---|
| 2079 | mb_len = enclen(encode, s);
|
---|
| 2080 | ss = s;
|
---|
| 2081 | s += mb_len;
|
---|
| 2082 | DATA_ENSURE(0);
|
---|
| 2083 | code = ONIGENC_MBC_TO_CODE(encode, ss, s);
|
---|
| 2084 | if (onig_is_code_in_cc_len(mb_len, code, node) == 0) goto fail;
|
---|
| 2085 | }
|
---|
| 2086 | MOP_OUT;
|
---|
| 2087 | NEXT;
|
---|
| 2088 |
|
---|
| 2089 | CASE(OP_ANYCHAR) MOP_IN(OP_ANYCHAR);
|
---|
| 2090 | DATA_ENSURE(1);
|
---|
| 2091 | n = enclen(encode, s);
|
---|
| 2092 | DATA_ENSURE(n);
|
---|
| 2093 | if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
|
---|
| 2094 | s += n;
|
---|
| 2095 | MOP_OUT;
|
---|
| 2096 | NEXT;
|
---|
| 2097 |
|
---|
| 2098 | CASE(OP_ANYCHAR_ML) MOP_IN(OP_ANYCHAR_ML);
|
---|
| 2099 | DATA_ENSURE(1);
|
---|
| 2100 | n = enclen(encode, s);
|
---|
| 2101 | DATA_ENSURE(n);
|
---|
| 2102 | s += n;
|
---|
| 2103 | MOP_OUT;
|
---|
| 2104 | NEXT;
|
---|
| 2105 |
|
---|
| 2106 | CASE(OP_ANYCHAR_STAR) MOP_IN(OP_ANYCHAR_STAR);
|
---|
| 2107 | while (DATA_ENSURE_CHECK1) {
|
---|
| 2108 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
| 2109 | n = enclen(encode, s);
|
---|
| 2110 | DATA_ENSURE(n);
|
---|
| 2111 | if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
|
---|
| 2112 | sprev = s;
|
---|
| 2113 | s += n;
|
---|
| 2114 | }
|
---|
| 2115 | MOP_OUT;
|
---|
| 2116 | NEXT;
|
---|
| 2117 |
|
---|
| 2118 | CASE(OP_ANYCHAR_ML_STAR) MOP_IN(OP_ANYCHAR_ML_STAR);
|
---|
| 2119 | while (DATA_ENSURE_CHECK1) {
|
---|
| 2120 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
| 2121 | n = enclen(encode, s);
|
---|
| 2122 | if (n > 1) {
|
---|
| 2123 | DATA_ENSURE(n);
|
---|
| 2124 | sprev = s;
|
---|
| 2125 | s += n;
|
---|
| 2126 | }
|
---|
| 2127 | else {
|
---|
| 2128 | sprev = s;
|
---|
| 2129 | s++;
|
---|
| 2130 | }
|
---|
| 2131 | }
|
---|
| 2132 | MOP_OUT;
|
---|
| 2133 | NEXT;
|
---|
| 2134 |
|
---|
| 2135 | CASE(OP_ANYCHAR_STAR_PEEK_NEXT) MOP_IN(OP_ANYCHAR_STAR_PEEK_NEXT);
|
---|
| 2136 | while (DATA_ENSURE_CHECK1) {
|
---|
| 2137 | if (*p == *s) {
|
---|
| 2138 | STACK_PUSH_ALT(p + 1, s, sprev, pkeep);
|
---|
| 2139 | }
|
---|
| 2140 | n = enclen(encode, s);
|
---|
| 2141 | DATA_ENSURE(n);
|
---|
| 2142 | if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
|
---|
| 2143 | sprev = s;
|
---|
| 2144 | s += n;
|
---|
| 2145 | }
|
---|
| 2146 | p++;
|
---|
| 2147 | MOP_OUT;
|
---|
| 2148 | NEXT;
|
---|
| 2149 |
|
---|
| 2150 | CASE(OP_ANYCHAR_ML_STAR_PEEK_NEXT)MOP_IN(OP_ANYCHAR_ML_STAR_PEEK_NEXT);
|
---|
| 2151 | while (DATA_ENSURE_CHECK1) {
|
---|
| 2152 | if (*p == *s) {
|
---|
| 2153 | STACK_PUSH_ALT(p + 1, s, sprev, pkeep);
|
---|
| 2154 | }
|
---|
| 2155 | n = enclen(encode, s);
|
---|
| 2156 | if (n > 1) {
|
---|
| 2157 | DATA_ENSURE(n);
|
---|
| 2158 | sprev = s;
|
---|
| 2159 | s += n;
|
---|
| 2160 | }
|
---|
| 2161 | else {
|
---|
| 2162 | sprev = s;
|
---|
| 2163 | s++;
|
---|
| 2164 | }
|
---|
| 2165 | }
|
---|
| 2166 | p++;
|
---|
| 2167 | MOP_OUT;
|
---|
| 2168 | NEXT;
|
---|
| 2169 |
|
---|
| 2170 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 2171 | CASE(OP_STATE_CHECK_ANYCHAR_STAR) MOP_IN(OP_STATE_CHECK_ANYCHAR_STAR);
|
---|
| 2172 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
| 2173 | while (DATA_ENSURE_CHECK1) {
|
---|
| 2174 | STATE_CHECK_VAL(scv, mem);
|
---|
| 2175 | if (scv) goto fail;
|
---|
| 2176 |
|
---|
| 2177 | STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep);
|
---|
| 2178 | n = enclen(encode, s);
|
---|
| 2179 | DATA_ENSURE(n);
|
---|
| 2180 | if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
|
---|
| 2181 | sprev = s;
|
---|
| 2182 | s += n;
|
---|
| 2183 | }
|
---|
| 2184 | MOP_OUT;
|
---|
| 2185 | NEXT;
|
---|
| 2186 |
|
---|
| 2187 | CASE(OP_STATE_CHECK_ANYCHAR_ML_STAR)
|
---|
| 2188 | MOP_IN(OP_STATE_CHECK_ANYCHAR_ML_STAR);
|
---|
| 2189 |
|
---|
| 2190 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
| 2191 | while (DATA_ENSURE_CHECK1) {
|
---|
| 2192 | STATE_CHECK_VAL(scv, mem);
|
---|
| 2193 | if (scv) goto fail;
|
---|
| 2194 |
|
---|
| 2195 | STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep);
|
---|
| 2196 | n = enclen(encode, s);
|
---|
| 2197 | if (n > 1) {
|
---|
| 2198 | DATA_ENSURE(n);
|
---|
| 2199 | sprev = s;
|
---|
| 2200 | s += n;
|
---|
| 2201 | }
|
---|
| 2202 | else {
|
---|
| 2203 | sprev = s;
|
---|
| 2204 | s++;
|
---|
| 2205 | }
|
---|
| 2206 | }
|
---|
| 2207 | MOP_OUT;
|
---|
| 2208 | NEXT;
|
---|
| 2209 | #endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
| 2210 |
|
---|
| 2211 | CASE(OP_WORD) MOP_IN(OP_WORD);
|
---|
| 2212 | DATA_ENSURE(1);
|
---|
| 2213 | if (! ONIGENC_IS_MBC_WORD(encode, s, end))
|
---|
| 2214 | goto fail;
|
---|
| 2215 |
|
---|
| 2216 | s += enclen(encode, s);
|
---|
| 2217 | MOP_OUT;
|
---|
| 2218 | NEXT;
|
---|
| 2219 |
|
---|
| 2220 | CASE(OP_ASCII_WORD) MOP_IN(OP_ASCII_WORD);
|
---|
| 2221 | DATA_ENSURE(1);
|
---|
| 2222 | if (! ONIGENC_IS_MBC_ASCII_WORD(encode, s, end))
|
---|
| 2223 | goto fail;
|
---|
| 2224 |
|
---|
| 2225 | s += enclen(encode, s);
|
---|
| 2226 | MOP_OUT;
|
---|
| 2227 | NEXT;
|
---|
| 2228 |
|
---|
| 2229 | CASE(OP_NOT_WORD) MOP_IN(OP_NOT_WORD);
|
---|
| 2230 | DATA_ENSURE(1);
|
---|
| 2231 | if (ONIGENC_IS_MBC_WORD(encode, s, end))
|
---|
| 2232 | goto fail;
|
---|
| 2233 |
|
---|
| 2234 | s += enclen(encode, s);
|
---|
| 2235 | MOP_OUT;
|
---|
| 2236 | NEXT;
|
---|
| 2237 |
|
---|
| 2238 | CASE(OP_NOT_ASCII_WORD) MOP_IN(OP_NOT_ASCII_WORD);
|
---|
| 2239 | DATA_ENSURE(1);
|
---|
| 2240 | if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end))
|
---|
| 2241 | goto fail;
|
---|
| 2242 |
|
---|
| 2243 | s += enclen(encode, s);
|
---|
| 2244 | MOP_OUT;
|
---|
| 2245 | NEXT;
|
---|
| 2246 |
|
---|
| 2247 | CASE(OP_WORD_BOUND) MOP_IN(OP_WORD_BOUND);
|
---|
| 2248 | if (ON_STR_BEGIN(s)) {
|
---|
| 2249 | DATA_ENSURE(1);
|
---|
| 2250 | if (! ONIGENC_IS_MBC_WORD(encode, s, end))
|
---|
| 2251 | goto fail;
|
---|
| 2252 | }
|
---|
| 2253 | else if (ON_STR_END(s)) {
|
---|
| 2254 | if (! ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
---|
| 2255 | goto fail;
|
---|
| 2256 | }
|
---|
| 2257 | else {
|
---|
| 2258 | if (ONIGENC_IS_MBC_WORD(encode, s, end)
|
---|
| 2259 | == ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
---|
| 2260 | goto fail;
|
---|
| 2261 | }
|
---|
| 2262 | MOP_OUT;
|
---|
| 2263 | JUMP;
|
---|
| 2264 |
|
---|
| 2265 | CASE(OP_ASCII_WORD_BOUND) MOP_IN(OP_ASCII_WORD_BOUND);
|
---|
| 2266 | if (ON_STR_BEGIN(s)) {
|
---|
| 2267 | DATA_ENSURE(1);
|
---|
| 2268 | if (! ONIGENC_IS_MBC_ASCII_WORD(encode, s, end))
|
---|
| 2269 | goto fail;
|
---|
| 2270 | }
|
---|
| 2271 | else if (ON_STR_END(s)) {
|
---|
| 2272 | if (! ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end))
|
---|
| 2273 | goto fail;
|
---|
| 2274 | }
|
---|
| 2275 | else {
|
---|
| 2276 | if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)
|
---|
| 2277 | == ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end))
|
---|
| 2278 | goto fail;
|
---|
| 2279 | }
|
---|
| 2280 | MOP_OUT;
|
---|
| 2281 | JUMP;
|
---|
| 2282 |
|
---|
| 2283 | CASE(OP_NOT_WORD_BOUND) MOP_IN(OP_NOT_WORD_BOUND);
|
---|
| 2284 | if (ON_STR_BEGIN(s)) {
|
---|
| 2285 | if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_WORD(encode, s, end))
|
---|
| 2286 | goto fail;
|
---|
| 2287 | }
|
---|
| 2288 | else if (ON_STR_END(s)) {
|
---|
| 2289 | if (ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
---|
| 2290 | goto fail;
|
---|
| 2291 | }
|
---|
| 2292 | else {
|
---|
| 2293 | if (ONIGENC_IS_MBC_WORD(encode, s, end)
|
---|
| 2294 | != ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
---|
| 2295 | goto fail;
|
---|
| 2296 | }
|
---|
| 2297 | MOP_OUT;
|
---|
| 2298 | JUMP;
|
---|
| 2299 |
|
---|
| 2300 | CASE(OP_NOT_ASCII_WORD_BOUND) MOP_IN(OP_NOT_ASCII_WORD_BOUND);
|
---|
| 2301 | if (ON_STR_BEGIN(s)) {
|
---|
| 2302 | if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_ASCII_WORD(encode, s, end))
|
---|
| 2303 | goto fail;
|
---|
| 2304 | }
|
---|
| 2305 | else if (ON_STR_END(s)) {
|
---|
| 2306 | if (ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end))
|
---|
| 2307 | goto fail;
|
---|
| 2308 | }
|
---|
| 2309 | else {
|
---|
| 2310 | if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)
|
---|
| 2311 | != ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end))
|
---|
| 2312 | goto fail;
|
---|
| 2313 | }
|
---|
| 2314 | MOP_OUT;
|
---|
| 2315 | JUMP;
|
---|
| 2316 |
|
---|
| 2317 | #ifdef USE_WORD_BEGIN_END
|
---|
| 2318 | CASE(OP_WORD_BEGIN) MOP_IN(OP_WORD_BEGIN);
|
---|
| 2319 | if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_WORD(encode, s, end)) {
|
---|
| 2320 | if (ON_STR_BEGIN(s) || !ONIGENC_IS_MBC_WORD(encode, sprev, end)) {
|
---|
| 2321 | MOP_OUT;
|
---|
| 2322 | JUMP;
|
---|
| 2323 | }
|
---|
| 2324 | }
|
---|
| 2325 | goto fail;
|
---|
| 2326 | NEXT;
|
---|
| 2327 |
|
---|
| 2328 | CASE(OP_ASCII_WORD_BEGIN) MOP_IN(OP_ASCII_WORD_BEGIN);
|
---|
| 2329 | if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) {
|
---|
| 2330 | if (ON_STR_BEGIN(s) || !ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) {
|
---|
| 2331 | MOP_OUT;
|
---|
| 2332 | JUMP;
|
---|
| 2333 | }
|
---|
| 2334 | }
|
---|
| 2335 | goto fail;
|
---|
| 2336 | NEXT;
|
---|
| 2337 |
|
---|
| 2338 | CASE(OP_WORD_END) MOP_IN(OP_WORD_END);
|
---|
| 2339 | if (!ON_STR_BEGIN(s) && ONIGENC_IS_MBC_WORD(encode, sprev, end)) {
|
---|
| 2340 | if (ON_STR_END(s) || !ONIGENC_IS_MBC_WORD(encode, s, end)) {
|
---|
| 2341 | MOP_OUT;
|
---|
| 2342 | JUMP;
|
---|
| 2343 | }
|
---|
| 2344 | }
|
---|
| 2345 | goto fail;
|
---|
| 2346 | NEXT;
|
---|
| 2347 |
|
---|
| 2348 | CASE(OP_ASCII_WORD_END) MOP_IN(OP_ASCII_WORD_END);
|
---|
| 2349 | if (!ON_STR_BEGIN(s) && ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) {
|
---|
| 2350 | if (ON_STR_END(s) || !ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) {
|
---|
| 2351 | MOP_OUT;
|
---|
| 2352 | JUMP;
|
---|
| 2353 | }
|
---|
| 2354 | }
|
---|
| 2355 | goto fail;
|
---|
| 2356 | NEXT;
|
---|
| 2357 | #endif
|
---|
| 2358 |
|
---|
| 2359 | CASE(OP_BEGIN_BUF) MOP_IN(OP_BEGIN_BUF);
|
---|
| 2360 | if (! ON_STR_BEGIN(s)) goto fail;
|
---|
| 2361 | if (IS_NOTBOS(msa->options)) goto fail;
|
---|
| 2362 |
|
---|
| 2363 | MOP_OUT;
|
---|
| 2364 | JUMP;
|
---|
| 2365 |
|
---|
| 2366 | CASE(OP_END_BUF) MOP_IN(OP_END_BUF);
|
---|
| 2367 | if (! ON_STR_END(s)) goto fail;
|
---|
| 2368 | if (IS_NOTEOS(msa->options)) goto fail;
|
---|
| 2369 |
|
---|
| 2370 | MOP_OUT;
|
---|
| 2371 | JUMP;
|
---|
| 2372 |
|
---|
| 2373 | CASE(OP_BEGIN_LINE) MOP_IN(OP_BEGIN_LINE);
|
---|
| 2374 | op_begin_line:
|
---|
| 2375 | if (ON_STR_BEGIN(s)) {
|
---|
| 2376 | if (IS_NOTBOL(msa->options)) goto fail;
|
---|
| 2377 | MOP_OUT;
|
---|
| 2378 | JUMP;
|
---|
| 2379 | }
|
---|
| 2380 | else if (ONIGENC_IS_MBC_NEWLINE(encode, sprev, end)
|
---|
| 2381 | #ifdef USE_CRNL_AS_LINE_TERMINATOR
|
---|
| 2382 | && !(IS_NEWLINE_CRLF(option)
|
---|
| 2383 | && ONIGENC_IS_MBC_CRNL(encode, sprev, end))
|
---|
| 2384 | #endif
|
---|
| 2385 | && !ON_STR_END(s)) {
|
---|
| 2386 | MOP_OUT;
|
---|
| 2387 | JUMP;
|
---|
| 2388 | }
|
---|
| 2389 | goto fail;
|
---|
| 2390 | NEXT;
|
---|
| 2391 |
|
---|
| 2392 | CASE(OP_END_LINE) MOP_IN(OP_END_LINE);
|
---|
| 2393 | if (ON_STR_END(s)) {
|
---|
| 2394 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
| 2395 | if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE_EX(encode, sprev, str, end, option, 1)) {
|
---|
| 2396 | #endif
|
---|
| 2397 | if (IS_NOTEOL(msa->options)) goto fail;
|
---|
| 2398 | MOP_OUT;
|
---|
| 2399 | JUMP;
|
---|
| 2400 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
| 2401 | }
|
---|
| 2402 | #endif
|
---|
| 2403 | }
|
---|
| 2404 | else if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 1)) {
|
---|
| 2405 | MOP_OUT;
|
---|
| 2406 | JUMP;
|
---|
| 2407 | }
|
---|
| 2408 | goto fail;
|
---|
| 2409 | NEXT;
|
---|
| 2410 |
|
---|
| 2411 | CASE(OP_SEMI_END_BUF) MOP_IN(OP_SEMI_END_BUF);
|
---|
| 2412 | if (ON_STR_END(s)) {
|
---|
| 2413 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
| 2414 | if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE_EX(encode, sprev, str, end, option, 1)) {
|
---|
| 2415 | #endif
|
---|
| 2416 | if (IS_NOTEOL(msa->options)) goto fail;
|
---|
| 2417 | MOP_OUT;
|
---|
| 2418 | JUMP;
|
---|
| 2419 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
| 2420 | }
|
---|
| 2421 | #endif
|
---|
| 2422 | }
|
---|
| 2423 | else if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 1)) {
|
---|
| 2424 | UChar* ss = s + enclen(encode, s);
|
---|
| 2425 | if (ON_STR_END(ss)) {
|
---|
| 2426 | MOP_OUT;
|
---|
| 2427 | JUMP;
|
---|
| 2428 | }
|
---|
| 2429 | #ifdef USE_CRNL_AS_LINE_TERMINATOR
|
---|
| 2430 | else if (IS_NEWLINE_CRLF(option)
|
---|
| 2431 | && ONIGENC_IS_MBC_CRNL(encode, s, end)) {
|
---|
| 2432 | ss += enclen(encode, ss);
|
---|
| 2433 | if (ON_STR_END(ss)) {
|
---|
| 2434 | MOP_OUT;
|
---|
| 2435 | JUMP;
|
---|
| 2436 | }
|
---|
| 2437 | }
|
---|
| 2438 | #endif
|
---|
| 2439 | }
|
---|
| 2440 | goto fail;
|
---|
| 2441 | NEXT;
|
---|
| 2442 |
|
---|
| 2443 | CASE(OP_BEGIN_POSITION) MOP_IN(OP_BEGIN_POSITION);
|
---|
| 2444 | if (s != msa->gpos)
|
---|
| 2445 | goto fail;
|
---|
| 2446 |
|
---|
| 2447 | MOP_OUT;
|
---|
| 2448 | JUMP;
|
---|
| 2449 |
|
---|
| 2450 | CASE(OP_BEGIN_POS_OR_LINE) MOP_IN(OP_BEGIN_POS_OR_LINE);
|
---|
| 2451 | if (s != msa->gpos)
|
---|
| 2452 | goto op_begin_line;
|
---|
| 2453 |
|
---|
| 2454 | MOP_OUT;
|
---|
| 2455 | JUMP;
|
---|
| 2456 |
|
---|
| 2457 | CASE(OP_MEMORY_START_PUSH) MOP_IN(OP_MEMORY_START_PUSH);
|
---|
| 2458 | GET_MEMNUM_INC(mem, p);
|
---|
| 2459 | STACK_PUSH_MEM_START(mem, s);
|
---|
| 2460 | MOP_OUT;
|
---|
| 2461 | JUMP;
|
---|
| 2462 |
|
---|
| 2463 | CASE(OP_MEMORY_START) MOP_IN(OP_MEMORY_START);
|
---|
| 2464 | GET_MEMNUM_INC(mem, p);
|
---|
| 2465 | mem_start_stk[mem] = (OnigStackIndex )((void* )s);
|
---|
| 2466 | MOP_OUT;
|
---|
| 2467 | JUMP;
|
---|
| 2468 |
|
---|
| 2469 | CASE(OP_MEMORY_END_PUSH) MOP_IN(OP_MEMORY_END_PUSH);
|
---|
| 2470 | GET_MEMNUM_INC(mem, p);
|
---|
| 2471 | STACK_PUSH_MEM_END(mem, s);
|
---|
| 2472 | MOP_OUT;
|
---|
| 2473 | JUMP;
|
---|
| 2474 |
|
---|
| 2475 | CASE(OP_MEMORY_END) MOP_IN(OP_MEMORY_END);
|
---|
| 2476 | GET_MEMNUM_INC(mem, p);
|
---|
| 2477 | mem_end_stk[mem] = (OnigStackIndex )((void* )s);
|
---|
| 2478 | MOP_OUT;
|
---|
| 2479 | JUMP;
|
---|
| 2480 |
|
---|
| 2481 | CASE(OP_KEEP) MOP_IN(OP_KEEP);
|
---|
| 2482 | pkeep = s;
|
---|
| 2483 | MOP_OUT;
|
---|
| 2484 | JUMP;
|
---|
| 2485 |
|
---|
| 2486 | #ifdef USE_SUBEXP_CALL
|
---|
| 2487 | CASE(OP_MEMORY_END_PUSH_REC) MOP_IN(OP_MEMORY_END_PUSH_REC);
|
---|
| 2488 | GET_MEMNUM_INC(mem, p);
|
---|
| 2489 | STACK_GET_MEM_START(mem, stkp); /* should be before push mem-end. */
|
---|
| 2490 | STACK_PUSH_MEM_END(mem, s);
|
---|
| 2491 | mem_start_stk[mem] = GET_STACK_INDEX(stkp);
|
---|
| 2492 | MOP_OUT;
|
---|
| 2493 | JUMP;
|
---|
| 2494 |
|
---|
| 2495 | CASE(OP_MEMORY_END_REC) MOP_IN(OP_MEMORY_END_REC);
|
---|
| 2496 | GET_MEMNUM_INC(mem, p);
|
---|
| 2497 | mem_end_stk[mem] = (OnigStackIndex )((void* )s);
|
---|
| 2498 | STACK_GET_MEM_START(mem, stkp);
|
---|
| 2499 |
|
---|
| 2500 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
| 2501 | mem_start_stk[mem] = GET_STACK_INDEX(stkp);
|
---|
| 2502 | else
|
---|
| 2503 | mem_start_stk[mem] = (OnigStackIndex )((void* )stkp->u.mem.pstr);
|
---|
| 2504 |
|
---|
| 2505 | STACK_PUSH_MEM_END_MARK(mem);
|
---|
| 2506 | MOP_OUT;
|
---|
| 2507 | JUMP;
|
---|
| 2508 | #endif
|
---|
| 2509 |
|
---|
| 2510 | CASE(OP_BACKREF1) MOP_IN(OP_BACKREF1);
|
---|
| 2511 | mem = 1;
|
---|
| 2512 | goto backref;
|
---|
| 2513 | NEXT;
|
---|
| 2514 |
|
---|
| 2515 | CASE(OP_BACKREF2) MOP_IN(OP_BACKREF2);
|
---|
| 2516 | mem = 2;
|
---|
| 2517 | goto backref;
|
---|
| 2518 | NEXT;
|
---|
| 2519 |
|
---|
| 2520 | CASE(OP_BACKREFN) MOP_IN(OP_BACKREFN);
|
---|
| 2521 | GET_MEMNUM_INC(mem, p);
|
---|
| 2522 | backref:
|
---|
| 2523 | {
|
---|
| 2524 | int len;
|
---|
| 2525 | UChar *pstart, *pend;
|
---|
| 2526 |
|
---|
| 2527 | /* if you want to remove following line,
|
---|
| 2528 | you should check in parse and compile time. */
|
---|
| 2529 | if (mem > num_mem) goto fail;
|
---|
| 2530 | if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
---|
| 2531 | if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
---|
| 2532 |
|
---|
| 2533 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
| 2534 | pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
---|
| 2535 | else
|
---|
| 2536 | pstart = (UChar* )((void* )mem_start_stk[mem]);
|
---|
| 2537 |
|
---|
| 2538 | pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
---|
| 2539 | ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
---|
| 2540 | : (UChar* )((void* )mem_end_stk[mem]));
|
---|
| 2541 | n = pend - pstart;
|
---|
| 2542 | DATA_ENSURE(n);
|
---|
| 2543 | sprev = s;
|
---|
| 2544 | STRING_CMP(pstart, s, n);
|
---|
| 2545 | while (sprev + (len = enclen(encode, sprev)) < s)
|
---|
| 2546 | sprev += len;
|
---|
| 2547 |
|
---|
| 2548 | MOP_OUT;
|
---|
| 2549 | JUMP;
|
---|
| 2550 | }
|
---|
| 2551 |
|
---|
| 2552 | CASE(OP_BACKREFN_IC) MOP_IN(OP_BACKREFN_IC);
|
---|
| 2553 | GET_MEMNUM_INC(mem, p);
|
---|
| 2554 | {
|
---|
| 2555 | int len;
|
---|
| 2556 | UChar *pstart, *pend;
|
---|
| 2557 |
|
---|
| 2558 | /* if you want to remove following line,
|
---|
| 2559 | you should check in parse and compile time. */
|
---|
| 2560 | if (mem > num_mem) goto fail;
|
---|
| 2561 | if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
---|
| 2562 | if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
---|
| 2563 |
|
---|
| 2564 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
| 2565 | pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
---|
| 2566 | else
|
---|
| 2567 | pstart = (UChar* )((void* )mem_start_stk[mem]);
|
---|
| 2568 |
|
---|
| 2569 | pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
---|
| 2570 | ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
---|
| 2571 | : (UChar* )((void* )mem_end_stk[mem]));
|
---|
| 2572 | n = pend - pstart;
|
---|
| 2573 | DATA_ENSURE(n);
|
---|
| 2574 | sprev = s;
|
---|
| 2575 | STRING_CMP_IC(case_fold_flag, pstart, &s, n);
|
---|
| 2576 | while (sprev + (len = enclen(encode, sprev)) < s)
|
---|
| 2577 | sprev += len;
|
---|
| 2578 |
|
---|
| 2579 | MOP_OUT;
|
---|
| 2580 | JUMP;
|
---|
| 2581 | }
|
---|
| 2582 | NEXT;
|
---|
| 2583 |
|
---|
| 2584 | CASE(OP_BACKREF_MULTI) MOP_IN(OP_BACKREF_MULTI);
|
---|
| 2585 | {
|
---|
| 2586 | int len, is_fail;
|
---|
| 2587 | UChar *pstart, *pend, *swork;
|
---|
| 2588 |
|
---|
| 2589 | GET_LENGTH_INC(tlen, p);
|
---|
| 2590 | for (i = 0; i < tlen; i++) {
|
---|
| 2591 | GET_MEMNUM_INC(mem, p);
|
---|
| 2592 |
|
---|
| 2593 | if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue;
|
---|
| 2594 | if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue;
|
---|
| 2595 |
|
---|
| 2596 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
| 2597 | pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
---|
| 2598 | else
|
---|
| 2599 | pstart = (UChar* )((void* )mem_start_stk[mem]);
|
---|
| 2600 |
|
---|
| 2601 | pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
---|
| 2602 | ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
---|
| 2603 | : (UChar* )((void* )mem_end_stk[mem]));
|
---|
| 2604 | n = pend - pstart;
|
---|
| 2605 | DATA_ENSURE(n);
|
---|
| 2606 | sprev = s;
|
---|
| 2607 | swork = s;
|
---|
| 2608 | STRING_CMP_VALUE(pstart, swork, n, is_fail);
|
---|
| 2609 | if (is_fail) continue;
|
---|
| 2610 | s = swork;
|
---|
| 2611 | while (sprev + (len = enclen(encode, sprev)) < s)
|
---|
| 2612 | sprev += len;
|
---|
| 2613 |
|
---|
| 2614 | p += (SIZE_MEMNUM * (tlen - i - 1));
|
---|
| 2615 | break; /* success */
|
---|
| 2616 | }
|
---|
| 2617 | if (i == tlen) goto fail;
|
---|
| 2618 | MOP_OUT;
|
---|
| 2619 | JUMP;
|
---|
| 2620 | }
|
---|
| 2621 | NEXT;
|
---|
| 2622 |
|
---|
| 2623 | CASE(OP_BACKREF_MULTI_IC) MOP_IN(OP_BACKREF_MULTI_IC);
|
---|
| 2624 | {
|
---|
| 2625 | int len, is_fail;
|
---|
| 2626 | UChar *pstart, *pend, *swork;
|
---|
| 2627 |
|
---|
| 2628 | GET_LENGTH_INC(tlen, p);
|
---|
| 2629 | for (i = 0; i < tlen; i++) {
|
---|
| 2630 | GET_MEMNUM_INC(mem, p);
|
---|
| 2631 |
|
---|
| 2632 | if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue;
|
---|
| 2633 | if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue;
|
---|
| 2634 |
|
---|
| 2635 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
| 2636 | pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
---|
| 2637 | else
|
---|
| 2638 | pstart = (UChar* )((void* )mem_start_stk[mem]);
|
---|
| 2639 |
|
---|
| 2640 | pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
---|
| 2641 | ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
---|
| 2642 | : (UChar* )((void* )mem_end_stk[mem]));
|
---|
| 2643 | n = pend - pstart;
|
---|
| 2644 | DATA_ENSURE(n);
|
---|
| 2645 | sprev = s;
|
---|
| 2646 | swork = s;
|
---|
| 2647 | STRING_CMP_VALUE_IC(case_fold_flag, pstart, &swork, n, is_fail);
|
---|
| 2648 | if (is_fail) continue;
|
---|
| 2649 | s = swork;
|
---|
| 2650 | while (sprev + (len = enclen(encode, sprev)) < s)
|
---|
| 2651 | sprev += len;
|
---|
| 2652 |
|
---|
| 2653 | p += (SIZE_MEMNUM * (tlen - i - 1));
|
---|
| 2654 | break; /* success */
|
---|
| 2655 | }
|
---|
| 2656 | if (i == tlen) goto fail;
|
---|
| 2657 | MOP_OUT;
|
---|
| 2658 | JUMP;
|
---|
| 2659 | }
|
---|
| 2660 |
|
---|
| 2661 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
| 2662 | CASE(OP_BACKREF_WITH_LEVEL)
|
---|
| 2663 | {
|
---|
| 2664 | int len;
|
---|
| 2665 | OnigOptionType ic;
|
---|
| 2666 | LengthType level;
|
---|
| 2667 |
|
---|
| 2668 | GET_OPTION_INC(ic, p);
|
---|
| 2669 | GET_LENGTH_INC(level, p);
|
---|
| 2670 | GET_LENGTH_INC(tlen, p);
|
---|
| 2671 |
|
---|
| 2672 | sprev = s;
|
---|
| 2673 | if (backref_match_at_nested_level(reg, stk, stk_base, ic
|
---|
| 2674 | , case_fold_flag, (int )level, (int )tlen, p, &s, end)) {
|
---|
| 2675 | while (sprev + (len = enclen(encode, sprev)) < s)
|
---|
| 2676 | sprev += len;
|
---|
| 2677 |
|
---|
| 2678 | p += (SIZE_MEMNUM * tlen);
|
---|
| 2679 | }
|
---|
| 2680 | else
|
---|
| 2681 | goto fail;
|
---|
| 2682 |
|
---|
| 2683 | MOP_OUT;
|
---|
| 2684 | JUMP;
|
---|
| 2685 | }
|
---|
| 2686 |
|
---|
| 2687 | #endif
|
---|
| 2688 |
|
---|
| 2689 | #if 0 /* no need: IS_DYNAMIC_OPTION() == 0 */
|
---|
| 2690 | CASE(OP_SET_OPTION_PUSH) MOP_IN(OP_SET_OPTION_PUSH);
|
---|
| 2691 | GET_OPTION_INC(option, p);
|
---|
| 2692 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
| 2693 | p += SIZE_OP_SET_OPTION + SIZE_OP_FAIL;
|
---|
| 2694 | MOP_OUT;
|
---|
| 2695 | JUMP;
|
---|
| 2696 |
|
---|
| 2697 | CASE(OP_SET_OPTION) MOP_IN(OP_SET_OPTION);
|
---|
| 2698 | GET_OPTION_INC(option, p);
|
---|
| 2699 | MOP_OUT;
|
---|
| 2700 | JUMP;
|
---|
| 2701 | #endif
|
---|
| 2702 |
|
---|
| 2703 | CASE(OP_NULL_CHECK_START) MOP_IN(OP_NULL_CHECK_START);
|
---|
| 2704 | GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
---|
| 2705 | STACK_PUSH_NULL_CHECK_START(mem, s);
|
---|
| 2706 | MOP_OUT;
|
---|
| 2707 | JUMP;
|
---|
| 2708 |
|
---|
| 2709 | CASE(OP_NULL_CHECK_END) MOP_IN(OP_NULL_CHECK_END);
|
---|
| 2710 | {
|
---|
| 2711 | int isnull;
|
---|
| 2712 |
|
---|
| 2713 | GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
---|
| 2714 | STACK_NULL_CHECK(isnull, mem, s);
|
---|
| 2715 | if (isnull) {
|
---|
| 2716 | #ifdef ONIG_DEBUG_MATCH
|
---|
| 2717 | fprintf(stderr, "NULL_CHECK_END: skip id:%d, s:%"PRIdPTR" (%p)\n",
|
---|
| 2718 | (int )mem, (intptr_t )s, s);
|
---|
| 2719 | #endif
|
---|
| 2720 | null_check_found:
|
---|
| 2721 | /* empty loop founded, skip next instruction */
|
---|
| 2722 | switch (*p++) {
|
---|
| 2723 | case OP_JUMP:
|
---|
| 2724 | case OP_PUSH:
|
---|
| 2725 | p += SIZE_RELADDR;
|
---|
| 2726 | break;
|
---|
| 2727 | case OP_REPEAT_INC:
|
---|
| 2728 | case OP_REPEAT_INC_NG:
|
---|
| 2729 | case OP_REPEAT_INC_SG:
|
---|
| 2730 | case OP_REPEAT_INC_NG_SG:
|
---|
| 2731 | p += SIZE_MEMNUM;
|
---|
| 2732 | break;
|
---|
| 2733 | default:
|
---|
| 2734 | goto unexpected_bytecode_error;
|
---|
| 2735 | break;
|
---|
| 2736 | }
|
---|
| 2737 | }
|
---|
| 2738 | }
|
---|
| 2739 | MOP_OUT;
|
---|
| 2740 | JUMP;
|
---|
| 2741 |
|
---|
| 2742 | #ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT
|
---|
| 2743 | CASE(OP_NULL_CHECK_END_MEMST) MOP_IN(OP_NULL_CHECK_END_MEMST);
|
---|
| 2744 | {
|
---|
| 2745 | int isnull;
|
---|
| 2746 |
|
---|
| 2747 | GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
---|
| 2748 | STACK_NULL_CHECK_MEMST(isnull, mem, s, reg);
|
---|
| 2749 | if (isnull) {
|
---|
| 2750 | #ifdef ONIG_DEBUG_MATCH
|
---|
| 2751 | fprintf(stderr, "NULL_CHECK_END_MEMST: skip id:%d, s:%"PRIdPTR" (%p)\n",
|
---|
| 2752 | (int )mem, (intptr_t )s, s);
|
---|
| 2753 | #endif
|
---|
| 2754 | if (isnull == -1) goto fail;
|
---|
| 2755 | goto null_check_found;
|
---|
| 2756 | }
|
---|
| 2757 | }
|
---|
| 2758 | MOP_OUT;
|
---|
| 2759 | JUMP;
|
---|
| 2760 | #endif
|
---|
| 2761 |
|
---|
| 2762 | #ifdef USE_SUBEXP_CALL
|
---|
| 2763 | CASE(OP_NULL_CHECK_END_MEMST_PUSH)
|
---|
| 2764 | MOP_IN(OP_NULL_CHECK_END_MEMST_PUSH);
|
---|
| 2765 | {
|
---|
| 2766 | int isnull;
|
---|
| 2767 |
|
---|
| 2768 | GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
---|
| 2769 | #ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT
|
---|
| 2770 | STACK_NULL_CHECK_MEMST_REC(isnull, mem, s, reg);
|
---|
| 2771 | #else
|
---|
| 2772 | STACK_NULL_CHECK_REC(isnull, mem, s);
|
---|
| 2773 | #endif
|
---|
| 2774 | if (isnull) {
|
---|
| 2775 | #ifdef ONIG_DEBUG_MATCH
|
---|
| 2776 | fprintf(stderr, "NULL_CHECK_END_MEMST_PUSH: skip id:%d, s:%"PRIdPTR" (%p)\n",
|
---|
| 2777 | (int )mem, (intptr_t )s, s);
|
---|
| 2778 | #endif
|
---|
| 2779 | if (isnull == -1) goto fail;
|
---|
| 2780 | goto null_check_found;
|
---|
| 2781 | }
|
---|
| 2782 | else {
|
---|
| 2783 | STACK_PUSH_NULL_CHECK_END(mem);
|
---|
| 2784 | }
|
---|
| 2785 | }
|
---|
| 2786 | MOP_OUT;
|
---|
| 2787 | JUMP;
|
---|
| 2788 | #endif
|
---|
| 2789 |
|
---|
| 2790 | CASE(OP_JUMP) MOP_IN(OP_JUMP);
|
---|
| 2791 | GET_RELADDR_INC(addr, p);
|
---|
| 2792 | p += addr;
|
---|
| 2793 | MOP_OUT;
|
---|
| 2794 | CHECK_INTERRUPT_IN_MATCH_AT;
|
---|
| 2795 | JUMP;
|
---|
| 2796 |
|
---|
| 2797 | CASE(OP_PUSH) MOP_IN(OP_PUSH);
|
---|
| 2798 | GET_RELADDR_INC(addr, p);
|
---|
| 2799 | STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
|
---|
| 2800 | MOP_OUT;
|
---|
| 2801 | JUMP;
|
---|
| 2802 |
|
---|
| 2803 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 2804 | CASE(OP_STATE_CHECK_PUSH) MOP_IN(OP_STATE_CHECK_PUSH);
|
---|
| 2805 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
| 2806 | STATE_CHECK_VAL(scv, mem);
|
---|
| 2807 | if (scv) goto fail;
|
---|
| 2808 |
|
---|
| 2809 | GET_RELADDR_INC(addr, p);
|
---|
| 2810 | STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem, pkeep);
|
---|
| 2811 | MOP_OUT;
|
---|
| 2812 | JUMP;
|
---|
| 2813 |
|
---|
| 2814 | CASE(OP_STATE_CHECK_PUSH_OR_JUMP) MOP_IN(OP_STATE_CHECK_PUSH_OR_JUMP);
|
---|
| 2815 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
| 2816 | GET_RELADDR_INC(addr, p);
|
---|
| 2817 | STATE_CHECK_VAL(scv, mem);
|
---|
| 2818 | if (scv) {
|
---|
| 2819 | p += addr;
|
---|
| 2820 | }
|
---|
| 2821 | else {
|
---|
| 2822 | STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem, pkeep);
|
---|
| 2823 | }
|
---|
| 2824 | MOP_OUT;
|
---|
| 2825 | JUMP;
|
---|
| 2826 |
|
---|
| 2827 | CASE(OP_STATE_CHECK) MOP_IN(OP_STATE_CHECK);
|
---|
| 2828 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
| 2829 | STATE_CHECK_VAL(scv, mem);
|
---|
| 2830 | if (scv) goto fail;
|
---|
| 2831 |
|
---|
| 2832 | STACK_PUSH_STATE_CHECK(s, mem);
|
---|
| 2833 | MOP_OUT;
|
---|
| 2834 | JUMP;
|
---|
| 2835 | #endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
| 2836 |
|
---|
| 2837 | CASE(OP_POP) MOP_IN(OP_POP);
|
---|
| 2838 | STACK_POP_ONE;
|
---|
| 2839 | MOP_OUT;
|
---|
| 2840 | JUMP;
|
---|
| 2841 |
|
---|
| 2842 | CASE(OP_PUSH_OR_JUMP_EXACT1) MOP_IN(OP_PUSH_OR_JUMP_EXACT1);
|
---|
| 2843 | GET_RELADDR_INC(addr, p);
|
---|
| 2844 | if (*p == *s && DATA_ENSURE_CHECK1) {
|
---|
| 2845 | p++;
|
---|
| 2846 | STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
|
---|
| 2847 | MOP_OUT;
|
---|
| 2848 | JUMP;
|
---|
| 2849 | }
|
---|
| 2850 | p += (addr + 1);
|
---|
| 2851 | MOP_OUT;
|
---|
| 2852 | JUMP;
|
---|
| 2853 |
|
---|
| 2854 | CASE(OP_PUSH_IF_PEEK_NEXT) MOP_IN(OP_PUSH_IF_PEEK_NEXT);
|
---|
| 2855 | GET_RELADDR_INC(addr, p);
|
---|
| 2856 | if (*p == *s) {
|
---|
| 2857 | p++;
|
---|
| 2858 | STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
|
---|
| 2859 | MOP_OUT;
|
---|
| 2860 | JUMP;
|
---|
| 2861 | }
|
---|
| 2862 | p++;
|
---|
| 2863 | MOP_OUT;
|
---|
| 2864 | JUMP;
|
---|
| 2865 |
|
---|
| 2866 | CASE(OP_REPEAT) MOP_IN(OP_REPEAT);
|
---|
| 2867 | {
|
---|
| 2868 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
| 2869 | GET_RELADDR_INC(addr, p);
|
---|
| 2870 |
|
---|
| 2871 | STACK_ENSURE(1);
|
---|
| 2872 | repeat_stk[mem] = GET_STACK_INDEX(stk);
|
---|
| 2873 | STACK_PUSH_REPEAT(mem, p);
|
---|
| 2874 |
|
---|
| 2875 | if (reg->repeat_range[mem].lower == 0) {
|
---|
| 2876 | STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
|
---|
| 2877 | }
|
---|
| 2878 | }
|
---|
| 2879 | MOP_OUT;
|
---|
| 2880 | JUMP;
|
---|
| 2881 |
|
---|
| 2882 | CASE(OP_REPEAT_NG) MOP_IN(OP_REPEAT_NG);
|
---|
| 2883 | {
|
---|
| 2884 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
| 2885 | GET_RELADDR_INC(addr, p);
|
---|
| 2886 |
|
---|
| 2887 | STACK_ENSURE(1);
|
---|
| 2888 | repeat_stk[mem] = GET_STACK_INDEX(stk);
|
---|
| 2889 | STACK_PUSH_REPEAT(mem, p);
|
---|
| 2890 |
|
---|
| 2891 | if (reg->repeat_range[mem].lower == 0) {
|
---|
| 2892 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
| 2893 | p += addr;
|
---|
| 2894 | }
|
---|
| 2895 | }
|
---|
| 2896 | MOP_OUT;
|
---|
| 2897 | JUMP;
|
---|
| 2898 |
|
---|
| 2899 | CASE(OP_REPEAT_INC) MOP_IN(OP_REPEAT_INC);
|
---|
| 2900 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
| 2901 | si = repeat_stk[mem];
|
---|
| 2902 | stkp = STACK_AT(si);
|
---|
| 2903 |
|
---|
| 2904 | repeat_inc:
|
---|
| 2905 | stkp->u.repeat.count++;
|
---|
| 2906 | if (stkp->u.repeat.count >= reg->repeat_range[mem].upper) {
|
---|
| 2907 | /* end of repeat. Nothing to do. */
|
---|
| 2908 | }
|
---|
| 2909 | else if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) {
|
---|
| 2910 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
| 2911 | p = STACK_AT(si)->u.repeat.pcode; /* Don't use stkp after PUSH. */
|
---|
| 2912 | }
|
---|
| 2913 | else {
|
---|
| 2914 | p = stkp->u.repeat.pcode;
|
---|
| 2915 | }
|
---|
| 2916 | STACK_PUSH_REPEAT_INC(si);
|
---|
| 2917 | MOP_OUT;
|
---|
| 2918 | CHECK_INTERRUPT_IN_MATCH_AT;
|
---|
| 2919 | JUMP;
|
---|
| 2920 |
|
---|
| 2921 | CASE(OP_REPEAT_INC_SG) MOP_IN(OP_REPEAT_INC_SG);
|
---|
| 2922 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
| 2923 | STACK_GET_REPEAT(mem, stkp);
|
---|
| 2924 | si = GET_STACK_INDEX(stkp);
|
---|
| 2925 | goto repeat_inc;
|
---|
| 2926 | NEXT;
|
---|
| 2927 |
|
---|
| 2928 | CASE(OP_REPEAT_INC_NG) MOP_IN(OP_REPEAT_INC_NG);
|
---|
| 2929 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
| 2930 | si = repeat_stk[mem];
|
---|
| 2931 | stkp = STACK_AT(si);
|
---|
| 2932 |
|
---|
| 2933 | repeat_inc_ng:
|
---|
| 2934 | stkp->u.repeat.count++;
|
---|
| 2935 | if (stkp->u.repeat.count < reg->repeat_range[mem].upper) {
|
---|
| 2936 | if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) {
|
---|
| 2937 | UChar* pcode = stkp->u.repeat.pcode;
|
---|
| 2938 |
|
---|
| 2939 | STACK_PUSH_REPEAT_INC(si);
|
---|
| 2940 | STACK_PUSH_ALT(pcode, s, sprev, pkeep);
|
---|
| 2941 | }
|
---|
| 2942 | else {
|
---|
| 2943 | p = stkp->u.repeat.pcode;
|
---|
| 2944 | STACK_PUSH_REPEAT_INC(si);
|
---|
| 2945 | }
|
---|
| 2946 | }
|
---|
| 2947 | else if (stkp->u.repeat.count == reg->repeat_range[mem].upper) {
|
---|
| 2948 | STACK_PUSH_REPEAT_INC(si);
|
---|
| 2949 | }
|
---|
| 2950 | MOP_OUT;
|
---|
| 2951 | CHECK_INTERRUPT_IN_MATCH_AT;
|
---|
| 2952 | JUMP;
|
---|
| 2953 |
|
---|
| 2954 | CASE(OP_REPEAT_INC_NG_SG) MOP_IN(OP_REPEAT_INC_NG_SG);
|
---|
| 2955 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
| 2956 | STACK_GET_REPEAT(mem, stkp);
|
---|
| 2957 | si = GET_STACK_INDEX(stkp);
|
---|
| 2958 | goto repeat_inc_ng;
|
---|
| 2959 | NEXT;
|
---|
| 2960 |
|
---|
| 2961 | CASE(OP_PUSH_POS) MOP_IN(OP_PUSH_POS);
|
---|
| 2962 | STACK_PUSH_POS(s, sprev, pkeep);
|
---|
| 2963 | MOP_OUT;
|
---|
| 2964 | JUMP;
|
---|
| 2965 |
|
---|
| 2966 | CASE(OP_POP_POS) MOP_IN(OP_POP_POS);
|
---|
| 2967 | {
|
---|
| 2968 | STACK_POS_END(stkp);
|
---|
| 2969 | s = stkp->u.state.pstr;
|
---|
| 2970 | sprev = stkp->u.state.pstr_prev;
|
---|
| 2971 | }
|
---|
| 2972 | MOP_OUT;
|
---|
| 2973 | JUMP;
|
---|
| 2974 |
|
---|
| 2975 | CASE(OP_PUSH_POS_NOT) MOP_IN(OP_PUSH_POS_NOT);
|
---|
| 2976 | GET_RELADDR_INC(addr, p);
|
---|
| 2977 | STACK_PUSH_POS_NOT(p + addr, s, sprev, pkeep);
|
---|
| 2978 | MOP_OUT;
|
---|
| 2979 | JUMP;
|
---|
| 2980 |
|
---|
| 2981 | CASE(OP_FAIL_POS) MOP_IN(OP_FAIL_POS);
|
---|
| 2982 | STACK_POP_TIL_POS_NOT;
|
---|
| 2983 | goto fail;
|
---|
| 2984 | NEXT;
|
---|
| 2985 |
|
---|
| 2986 | CASE(OP_PUSH_STOP_BT) MOP_IN(OP_PUSH_STOP_BT);
|
---|
| 2987 | STACK_PUSH_STOP_BT;
|
---|
| 2988 | MOP_OUT;
|
---|
| 2989 | JUMP;
|
---|
| 2990 |
|
---|
| 2991 | CASE(OP_POP_STOP_BT) MOP_IN(OP_POP_STOP_BT);
|
---|
| 2992 | STACK_STOP_BT_END;
|
---|
| 2993 | MOP_OUT;
|
---|
| 2994 | JUMP;
|
---|
| 2995 |
|
---|
| 2996 | CASE(OP_LOOK_BEHIND) MOP_IN(OP_LOOK_BEHIND);
|
---|
| 2997 | GET_LENGTH_INC(tlen, p);
|
---|
| 2998 | s = (UChar* )ONIGENC_STEP_BACK(encode, str, s, (int )tlen);
|
---|
| 2999 | if (IS_NULL(s)) goto fail;
|
---|
| 3000 | sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s);
|
---|
| 3001 | MOP_OUT;
|
---|
| 3002 | JUMP;
|
---|
| 3003 |
|
---|
| 3004 | CASE(OP_PUSH_LOOK_BEHIND_NOT) MOP_IN(OP_PUSH_LOOK_BEHIND_NOT);
|
---|
| 3005 | GET_RELADDR_INC(addr, p);
|
---|
| 3006 | GET_LENGTH_INC(tlen, p);
|
---|
| 3007 | q = (UChar* )ONIGENC_STEP_BACK(encode, str, s, (int )tlen);
|
---|
| 3008 | if (IS_NULL(q)) {
|
---|
| 3009 | /* too short case -> success. ex. /(?<!XXX)a/.match("a")
|
---|
| 3010 | If you want to change to fail, replace following line. */
|
---|
| 3011 | p += addr;
|
---|
| 3012 | /* goto fail; */
|
---|
| 3013 | }
|
---|
| 3014 | else {
|
---|
| 3015 | STACK_PUSH_LOOK_BEHIND_NOT(p + addr, s, sprev, pkeep);
|
---|
| 3016 | s = q;
|
---|
| 3017 | sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s);
|
---|
| 3018 | }
|
---|
| 3019 | MOP_OUT;
|
---|
| 3020 | JUMP;
|
---|
| 3021 |
|
---|
| 3022 | CASE(OP_FAIL_LOOK_BEHIND_NOT) MOP_IN(OP_FAIL_LOOK_BEHIND_NOT);
|
---|
| 3023 | STACK_POP_TIL_LOOK_BEHIND_NOT;
|
---|
| 3024 | goto fail;
|
---|
| 3025 | NEXT;
|
---|
| 3026 |
|
---|
| 3027 | #ifdef USE_SUBEXP_CALL
|
---|
| 3028 | CASE(OP_CALL) MOP_IN(OP_CALL);
|
---|
| 3029 | GET_ABSADDR_INC(addr, p);
|
---|
| 3030 | STACK_PUSH_CALL_FRAME(p);
|
---|
| 3031 | p = reg->p + addr;
|
---|
| 3032 | MOP_OUT;
|
---|
| 3033 | JUMP;
|
---|
| 3034 |
|
---|
| 3035 | CASE(OP_RETURN) MOP_IN(OP_RETURN);
|
---|
| 3036 | STACK_RETURN(p);
|
---|
| 3037 | STACK_PUSH_RETURN;
|
---|
| 3038 | MOP_OUT;
|
---|
| 3039 | JUMP;
|
---|
| 3040 | #endif
|
---|
| 3041 |
|
---|
| 3042 | CASE(OP_CONDITION) MOP_IN(OP_CONDITION);
|
---|
| 3043 | GET_MEMNUM_INC(mem, p);
|
---|
| 3044 | GET_RELADDR_INC(addr, p);
|
---|
| 3045 | if ((mem > num_mem) ||
|
---|
| 3046 | (mem_end_stk[mem] == INVALID_STACK_INDEX) ||
|
---|
| 3047 | (mem_start_stk[mem] == INVALID_STACK_INDEX)) {
|
---|
| 3048 | p += addr;
|
---|
| 3049 | }
|
---|
| 3050 | MOP_OUT;
|
---|
| 3051 | JUMP;
|
---|
| 3052 |
|
---|
| 3053 | CASE(OP_FINISH)
|
---|
| 3054 | goto finish;
|
---|
| 3055 | NEXT;
|
---|
| 3056 |
|
---|
| 3057 | CASE(OP_FAIL)
|
---|
| 3058 | if (0) {
|
---|
| 3059 | /* fall */
|
---|
| 3060 | fail:
|
---|
| 3061 | MOP_OUT;
|
---|
| 3062 | }
|
---|
| 3063 | MOP_IN(OP_FAIL);
|
---|
| 3064 | STACK_POP;
|
---|
| 3065 | p = stk->u.state.pcode;
|
---|
| 3066 | s = stk->u.state.pstr;
|
---|
| 3067 | sprev = stk->u.state.pstr_prev;
|
---|
| 3068 | pkeep = stk->u.state.pkeep;
|
---|
| 3069 |
|
---|
| 3070 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 3071 | if (stk->u.state.state_check != 0) {
|
---|
| 3072 | stk->type = STK_STATE_CHECK_MARK;
|
---|
| 3073 | stk++;
|
---|
| 3074 | }
|
---|
| 3075 | #endif
|
---|
| 3076 |
|
---|
| 3077 | MOP_OUT;
|
---|
| 3078 | JUMP;
|
---|
| 3079 |
|
---|
| 3080 | DEFAULT
|
---|
| 3081 | goto bytecode_error;
|
---|
| 3082 | } VM_LOOP_END
|
---|
| 3083 |
|
---|
| 3084 | finish:
|
---|
| 3085 | STACK_SAVE;
|
---|
| 3086 | if (xmalloc_base) xfree(xmalloc_base);
|
---|
| 3087 | return best_len;
|
---|
| 3088 |
|
---|
| 3089 | #ifdef ONIG_DEBUG
|
---|
| 3090 | stack_error:
|
---|
| 3091 | STACK_SAVE;
|
---|
| 3092 | if (xmalloc_base) xfree(xmalloc_base);
|
---|
| 3093 | return ONIGERR_STACK_BUG;
|
---|
| 3094 | #endif
|
---|
| 3095 |
|
---|
| 3096 | bytecode_error:
|
---|
| 3097 | STACK_SAVE;
|
---|
| 3098 | if (xmalloc_base) xfree(xmalloc_base);
|
---|
| 3099 | return ONIGERR_UNDEFINED_BYTECODE;
|
---|
| 3100 |
|
---|
| 3101 | unexpected_bytecode_error:
|
---|
| 3102 | STACK_SAVE;
|
---|
| 3103 | if (xmalloc_base) xfree(xmalloc_base);
|
---|
| 3104 | return ONIGERR_UNEXPECTED_BYTECODE;
|
---|
| 3105 | }
|
---|
| 3106 |
|
---|
| 3107 | static OnigPosition
|
---|
| 3108 | match_at(regex_t* reg, const UChar* str, const UChar* end,
|
---|
| 3109 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
| 3110 | const UChar* right_range,
|
---|
| 3111 | #endif
|
---|
| 3112 | const UChar* sstart, UChar* sprev, OnigMatchArg* msa)
|
---|
| 3113 | {
|
---|
| 3114 | char *alloca_base = NULL;
|
---|
| 3115 | OnigPosition ret = match_at_(reg, str, end,
|
---|
| 3116 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
| 3117 | right_range,
|
---|
| 3118 | #endif
|
---|
| 3119 | sstart, sprev, msa, &alloca_base);
|
---|
| 3120 | if(alloca_base) xfree(alloca_base);
|
---|
| 3121 | return ret;
|
---|
| 3122 | }
|
---|
| 3123 |
|
---|
| 3124 | static UChar*
|
---|
| 3125 | slow_search(OnigEncoding enc, UChar* target, UChar* target_end,
|
---|
| 3126 | const UChar* text, const UChar* text_end, UChar* text_range)
|
---|
| 3127 | {
|
---|
| 3128 | UChar *t, *p, *s, *end;
|
---|
| 3129 |
|
---|
| 3130 | end = (UChar* )text_end;
|
---|
| 3131 | end -= target_end - target - 1;
|
---|
| 3132 | if (end > text_range)
|
---|
| 3133 | end = text_range;
|
---|
| 3134 |
|
---|
| 3135 | s = (UChar* )text;
|
---|
| 3136 |
|
---|
| 3137 | if (enc->max_enc_len == enc->min_enc_len) {
|
---|
| 3138 | int n = enc->max_enc_len;
|
---|
| 3139 |
|
---|
| 3140 | while (s < end) {
|
---|
| 3141 | if (*s == *target) {
|
---|
| 3142 | p = s + 1;
|
---|
| 3143 | t = target + 1;
|
---|
| 3144 | if (target_end == t || memcmp(t, p, target_end - t) == 0)
|
---|
| 3145 | return s;
|
---|
| 3146 | }
|
---|
| 3147 | s += n;
|
---|
| 3148 | }
|
---|
| 3149 | return (UChar* )NULL;
|
---|
| 3150 | }
|
---|
| 3151 | while (s < end) {
|
---|
| 3152 | if (*s == *target) {
|
---|
| 3153 | p = s + 1;
|
---|
| 3154 | t = target + 1;
|
---|
| 3155 | if (target_end == t || memcmp(t, p, target_end - t) == 0)
|
---|
| 3156 | return s;
|
---|
| 3157 | }
|
---|
| 3158 | s += enclen(enc, s);
|
---|
| 3159 | }
|
---|
| 3160 |
|
---|
| 3161 | return (UChar* )NULL;
|
---|
| 3162 | }
|
---|
| 3163 |
|
---|
| 3164 | static int
|
---|
| 3165 | str_lower_case_match(OnigEncoding enc, int case_fold_flag,
|
---|
| 3166 | const UChar* t, const UChar* tend,
|
---|
| 3167 | const UChar* p, const UChar* end)
|
---|
| 3168 | {
|
---|
| 3169 | int lowlen;
|
---|
| 3170 | UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
| 3171 |
|
---|
| 3172 | while (t < tend) {
|
---|
| 3173 | lowlen = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &p, end, lowbuf);
|
---|
| 3174 | q = lowbuf;
|
---|
| 3175 | while (lowlen > 0) {
|
---|
| 3176 | if (*t++ != *q++) return 0;
|
---|
| 3177 | lowlen--;
|
---|
| 3178 | }
|
---|
| 3179 | }
|
---|
| 3180 |
|
---|
| 3181 | return 1;
|
---|
| 3182 | }
|
---|
| 3183 |
|
---|
| 3184 | static UChar*
|
---|
| 3185 | slow_search_ic(OnigEncoding enc, int case_fold_flag,
|
---|
| 3186 | UChar* target, UChar* target_end,
|
---|
| 3187 | const UChar* text, const UChar* text_end, UChar* text_range)
|
---|
| 3188 | {
|
---|
| 3189 | UChar *s, *end;
|
---|
| 3190 |
|
---|
| 3191 | end = (UChar* )text_end;
|
---|
| 3192 | end -= target_end - target - 1;
|
---|
| 3193 | if (end > text_range)
|
---|
| 3194 | end = text_range;
|
---|
| 3195 |
|
---|
| 3196 | s = (UChar* )text;
|
---|
| 3197 |
|
---|
| 3198 | while (s < end) {
|
---|
| 3199 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
| 3200 | s, text_end))
|
---|
| 3201 | return s;
|
---|
| 3202 |
|
---|
| 3203 | s += enclen(enc, s);
|
---|
| 3204 | }
|
---|
| 3205 |
|
---|
| 3206 | return (UChar* )NULL;
|
---|
| 3207 | }
|
---|
| 3208 |
|
---|
| 3209 | static UChar*
|
---|
| 3210 | slow_search_backward(OnigEncoding enc, UChar* target, UChar* target_end,
|
---|
| 3211 | const UChar* text, const UChar* adjust_text,
|
---|
| 3212 | const UChar* text_end, const UChar* text_start)
|
---|
| 3213 | {
|
---|
| 3214 | UChar *t, *p, *s;
|
---|
| 3215 |
|
---|
| 3216 | s = (UChar* )text_end;
|
---|
| 3217 | s -= (target_end - target);
|
---|
| 3218 | if (s > text_start)
|
---|
| 3219 | s = (UChar* )text_start;
|
---|
| 3220 | else
|
---|
| 3221 | s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s);
|
---|
| 3222 |
|
---|
| 3223 | while (s >= text) {
|
---|
| 3224 | if (*s == *target) {
|
---|
| 3225 | p = s + 1;
|
---|
| 3226 | t = target + 1;
|
---|
| 3227 | while (t < target_end) {
|
---|
| 3228 | if (*t != *p++)
|
---|
| 3229 | break;
|
---|
| 3230 | t++;
|
---|
| 3231 | }
|
---|
| 3232 | if (t == target_end)
|
---|
| 3233 | return s;
|
---|
| 3234 | }
|
---|
| 3235 | s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s);
|
---|
| 3236 | }
|
---|
| 3237 |
|
---|
| 3238 | return (UChar* )NULL;
|
---|
| 3239 | }
|
---|
| 3240 |
|
---|
| 3241 | static UChar*
|
---|
| 3242 | slow_search_backward_ic(OnigEncoding enc, int case_fold_flag,
|
---|
| 3243 | UChar* target, UChar* target_end,
|
---|
| 3244 | const UChar* text, const UChar* adjust_text,
|
---|
| 3245 | const UChar* text_end, const UChar* text_start)
|
---|
| 3246 | {
|
---|
| 3247 | UChar *s;
|
---|
| 3248 |
|
---|
| 3249 | s = (UChar* )text_end;
|
---|
| 3250 | s -= (target_end - target);
|
---|
| 3251 | if (s > text_start)
|
---|
| 3252 | s = (UChar* )text_start;
|
---|
| 3253 | else
|
---|
| 3254 | s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s);
|
---|
| 3255 |
|
---|
| 3256 | while (s >= text) {
|
---|
| 3257 | if (str_lower_case_match(enc, case_fold_flag,
|
---|
| 3258 | target, target_end, s, text_end))
|
---|
| 3259 | return s;
|
---|
| 3260 |
|
---|
| 3261 | s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s);
|
---|
| 3262 | }
|
---|
| 3263 |
|
---|
| 3264 | return (UChar* )NULL;
|
---|
| 3265 | }
|
---|
| 3266 |
|
---|
| 3267 | #ifndef USE_SUNDAY_QUICK_SEARCH
|
---|
| 3268 | /* Boyer-Moore-Horspool search applied to a multibyte string */
|
---|
| 3269 | static UChar*
|
---|
| 3270 | bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
| 3271 | const UChar* text, const UChar* text_end,
|
---|
| 3272 | const UChar* text_range)
|
---|
| 3273 | {
|
---|
| 3274 | const UChar *s, *se, *t, *p, *end;
|
---|
| 3275 | const UChar *tail;
|
---|
| 3276 | ptrdiff_t skip, tlen1;
|
---|
| 3277 |
|
---|
| 3278 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3279 | fprintf(stderr, "bm_search_notrev: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n",
|
---|
| 3280 | text, text, text_end, text_end, text_range, text_range);
|
---|
| 3281 | #endif
|
---|
| 3282 |
|
---|
| 3283 | tail = target_end - 1;
|
---|
| 3284 | tlen1 = tail - target;
|
---|
| 3285 | end = text_range;
|
---|
| 3286 | if (end + tlen1 > text_end)
|
---|
| 3287 | end = text_end - tlen1;
|
---|
| 3288 |
|
---|
| 3289 | s = text;
|
---|
| 3290 |
|
---|
| 3291 | if (IS_NULL(reg->int_map)) {
|
---|
| 3292 | while (s < end) {
|
---|
| 3293 | p = se = s + tlen1;
|
---|
| 3294 | t = tail;
|
---|
| 3295 | while (*p == *t) {
|
---|
| 3296 | if (t == target) return (UChar* )s;
|
---|
| 3297 | p--; t--;
|
---|
| 3298 | }
|
---|
| 3299 | skip = reg->map[*se];
|
---|
| 3300 | t = s;
|
---|
| 3301 | do {
|
---|
| 3302 | s += enclen(reg->enc, s);
|
---|
| 3303 | } while ((s - t) < skip && s < end);
|
---|
| 3304 | }
|
---|
| 3305 | }
|
---|
| 3306 | else {
|
---|
| 3307 | while (s < end) {
|
---|
| 3308 | p = se = s + tlen1;
|
---|
| 3309 | t = tail;
|
---|
| 3310 | while (*p == *t) {
|
---|
| 3311 | if (t == target) return (UChar* )s;
|
---|
| 3312 | p--; t--;
|
---|
| 3313 | }
|
---|
| 3314 | skip = reg->int_map[*se];
|
---|
| 3315 | t = s;
|
---|
| 3316 | do {
|
---|
| 3317 | s += enclen(reg->enc, s);
|
---|
| 3318 | } while ((s - t) < skip && s < end);
|
---|
| 3319 | }
|
---|
| 3320 | }
|
---|
| 3321 |
|
---|
| 3322 | return (UChar* )NULL;
|
---|
| 3323 | }
|
---|
| 3324 |
|
---|
| 3325 | /* Boyer-Moore-Horspool search */
|
---|
| 3326 | static UChar*
|
---|
| 3327 | bm_search(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
| 3328 | const UChar* text, const UChar* text_end, const UChar* text_range)
|
---|
| 3329 | {
|
---|
| 3330 | const UChar *s, *t, *p, *end;
|
---|
| 3331 | const UChar *tail;
|
---|
| 3332 |
|
---|
| 3333 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3334 | fprintf(stderr, "bm_search: text: %"PRIuPTR", text_end: %"PRIuPTR", text_range: %"PRIuPTR"\n",
|
---|
| 3335 | text, text_end, text_range);
|
---|
| 3336 | #endif
|
---|
| 3337 |
|
---|
| 3338 | end = text_range + (target_end - target) - 1;
|
---|
| 3339 | if (end > text_end)
|
---|
| 3340 | end = text_end;
|
---|
| 3341 |
|
---|
| 3342 | tail = target_end - 1;
|
---|
| 3343 | s = text + (target_end - target) - 1;
|
---|
| 3344 | if (IS_NULL(reg->int_map)) {
|
---|
| 3345 | while (s < end) {
|
---|
| 3346 | p = s;
|
---|
| 3347 | t = tail;
|
---|
| 3348 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3349 | fprintf(stderr, "bm_search_loop: pos: %"PRIdPTR" %s\n",
|
---|
| 3350 | (intptr_t )(s - text), s);
|
---|
| 3351 | #endif
|
---|
| 3352 | while (*p == *t) {
|
---|
| 3353 | if (t == target) return (UChar* )p;
|
---|
| 3354 | p--; t--;
|
---|
| 3355 | }
|
---|
| 3356 | s += reg->map[*s];
|
---|
| 3357 | }
|
---|
| 3358 | }
|
---|
| 3359 | else { /* see int_map[] */
|
---|
| 3360 | while (s < end) {
|
---|
| 3361 | p = s;
|
---|
| 3362 | t = tail;
|
---|
| 3363 | while (*p == *t) {
|
---|
| 3364 | if (t == target) return (UChar* )p;
|
---|
| 3365 | p--; t--;
|
---|
| 3366 | }
|
---|
| 3367 | s += reg->int_map[*s];
|
---|
| 3368 | }
|
---|
| 3369 | }
|
---|
| 3370 | return (UChar* )NULL;
|
---|
| 3371 | }
|
---|
| 3372 |
|
---|
| 3373 | /* Boyer-Moore-Horspool search applied to a multibyte string (ignore case) */
|
---|
| 3374 | static UChar*
|
---|
| 3375 | bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
| 3376 | const UChar* text, const UChar* text_end,
|
---|
| 3377 | const UChar* text_range)
|
---|
| 3378 | {
|
---|
| 3379 | const UChar *s, *se, *t, *end;
|
---|
| 3380 | const UChar *tail;
|
---|
| 3381 | ptrdiff_t skip, tlen1;
|
---|
| 3382 | OnigEncoding enc = reg->enc;
|
---|
| 3383 | int case_fold_flag = reg->case_fold_flag;
|
---|
| 3384 |
|
---|
| 3385 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3386 | fprintf(stderr, "bm_search_notrev_ic: text: %d (%p), text_end: %d (%p), text_range: %d (%p)\n",
|
---|
| 3387 | (int )text, text, (int )text_end, text_end, (int )text_range, text_range);
|
---|
| 3388 | #endif
|
---|
| 3389 |
|
---|
| 3390 | tail = target_end - 1;
|
---|
| 3391 | tlen1 = tail - target;
|
---|
| 3392 | end = text_range;
|
---|
| 3393 | if (end + tlen1 > text_end)
|
---|
| 3394 | end = text_end - tlen1;
|
---|
| 3395 |
|
---|
| 3396 | s = text;
|
---|
| 3397 |
|
---|
| 3398 | if (IS_NULL(reg->int_map)) {
|
---|
| 3399 | while (s < end) {
|
---|
| 3400 | se = s + tlen1;
|
---|
| 3401 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
| 3402 | s, se + 1))
|
---|
| 3403 | return (UChar* )s;
|
---|
| 3404 | skip = reg->map[*se];
|
---|
| 3405 | t = s;
|
---|
| 3406 | do {
|
---|
| 3407 | s += enclen(reg->enc, s);
|
---|
| 3408 | } while ((s - t) < skip && s < end);
|
---|
| 3409 | }
|
---|
| 3410 | }
|
---|
| 3411 | else {
|
---|
| 3412 | while (s < end) {
|
---|
| 3413 | se = s + tlen1;
|
---|
| 3414 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
| 3415 | s, se + 1))
|
---|
| 3416 | return (UChar* )s;
|
---|
| 3417 | skip = reg->int_map[*se];
|
---|
| 3418 | t = s;
|
---|
| 3419 | do {
|
---|
| 3420 | s += enclen(reg->enc, s);
|
---|
| 3421 | } while ((s - t) < skip && s < end);
|
---|
| 3422 | }
|
---|
| 3423 | }
|
---|
| 3424 |
|
---|
| 3425 | return (UChar* )NULL;
|
---|
| 3426 | }
|
---|
| 3427 |
|
---|
| 3428 | /* Boyer-Moore-Horspool search (ignore case) */
|
---|
| 3429 | static UChar*
|
---|
| 3430 | bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
| 3431 | const UChar* text, const UChar* text_end, const UChar* text_range)
|
---|
| 3432 | {
|
---|
| 3433 | const UChar *s, *p, *end;
|
---|
| 3434 | const UChar *tail;
|
---|
| 3435 | OnigEncoding enc = reg->enc;
|
---|
| 3436 | int case_fold_flag = reg->case_fold_flag;
|
---|
| 3437 |
|
---|
| 3438 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3439 | fprintf(stderr, "bm_search_ic: text: %d (%p), text_end: %d (%p), text_range: %d (%p)\n",
|
---|
| 3440 | (int )text, text, (int )text_end, text_end, (int )text_range, text_range);
|
---|
| 3441 | #endif
|
---|
| 3442 |
|
---|
| 3443 | end = text_range + (target_end - target) - 1;
|
---|
| 3444 | if (end > text_end)
|
---|
| 3445 | end = text_end;
|
---|
| 3446 |
|
---|
| 3447 | tail = target_end - 1;
|
---|
| 3448 | s = text + (target_end - target) - 1;
|
---|
| 3449 | if (IS_NULL(reg->int_map)) {
|
---|
| 3450 | while (s < end) {
|
---|
| 3451 | p = s - (target_end - target) + 1;
|
---|
| 3452 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
| 3453 | p, s + 1))
|
---|
| 3454 | return (UChar* )p;
|
---|
| 3455 | s += reg->map[*s];
|
---|
| 3456 | }
|
---|
| 3457 | }
|
---|
| 3458 | else { /* see int_map[] */
|
---|
| 3459 | while (s < end) {
|
---|
| 3460 | p = s - (target_end - target) + 1;
|
---|
| 3461 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
| 3462 | p, s + 1))
|
---|
| 3463 | return (UChar* )p;
|
---|
| 3464 | s += reg->int_map[*s];
|
---|
| 3465 | }
|
---|
| 3466 | }
|
---|
| 3467 | return (UChar* )NULL;
|
---|
| 3468 | }
|
---|
| 3469 |
|
---|
| 3470 | #else /* USE_SUNDAY_QUICK_SEARCH */
|
---|
| 3471 |
|
---|
| 3472 | /* Sunday's quick search applied to a multibyte string */
|
---|
| 3473 | static UChar*
|
---|
| 3474 | bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
| 3475 | const UChar* text, const UChar* text_end,
|
---|
| 3476 | const UChar* text_range)
|
---|
| 3477 | {
|
---|
| 3478 | const UChar *s, *se, *t, *p, *end;
|
---|
| 3479 | const UChar *tail;
|
---|
| 3480 | ptrdiff_t skip, tlen1;
|
---|
| 3481 | OnigEncoding enc = reg->enc;
|
---|
| 3482 | int (*mbc_enc_len)(const OnigUChar* p) = enc->mbc_enc_len;
|
---|
| 3483 |
|
---|
| 3484 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3485 | fprintf(stderr, "bm_search_notrev: text: %"PRIdPTR" (%p), text_end: %"PRIdPTR" (%p), text_range: %"PRIdPTR" (%p)\n",
|
---|
| 3486 | (intptr_t )text, text, (intptr_t )text_end, text_end, (intptr_t )text_range, text_range);
|
---|
| 3487 | #endif
|
---|
| 3488 |
|
---|
| 3489 | tail = target_end - 1;
|
---|
| 3490 | tlen1 = tail - target;
|
---|
| 3491 | end = text_range;
|
---|
| 3492 | if (end + tlen1 > text_end)
|
---|
| 3493 | end = text_end - tlen1;
|
---|
| 3494 |
|
---|
| 3495 | s = text;
|
---|
| 3496 |
|
---|
| 3497 | if (IS_NULL(reg->int_map)) {
|
---|
| 3498 | while (s < end) {
|
---|
| 3499 | p = se = s + tlen1;
|
---|
| 3500 | t = tail;
|
---|
| 3501 | while (*p == *t) {
|
---|
| 3502 | if (t == target) return (UChar* )s;
|
---|
| 3503 | p--; t--;
|
---|
| 3504 | }
|
---|
| 3505 | if (s + 1 >= end) break;
|
---|
| 3506 | skip = reg->map[se[1]];
|
---|
| 3507 | t = s;
|
---|
| 3508 | do {
|
---|
| 3509 | s += mbc_enc_len(s);
|
---|
| 3510 | } while ((s - t) < skip && s < end);
|
---|
| 3511 | }
|
---|
| 3512 | }
|
---|
| 3513 | else {
|
---|
| 3514 | while (s < end) {
|
---|
| 3515 | p = se = s + tlen1;
|
---|
| 3516 | t = tail;
|
---|
| 3517 | while (*p == *t) {
|
---|
| 3518 | if (t == target) return (UChar* )s;
|
---|
| 3519 | p--; t--;
|
---|
| 3520 | }
|
---|
| 3521 | if (s + 1 >= end) break;
|
---|
| 3522 | skip = reg->int_map[se[1]];
|
---|
| 3523 | t = s;
|
---|
| 3524 | do {
|
---|
| 3525 | s += mbc_enc_len(s);
|
---|
| 3526 | } while ((s - t) < skip && s < end);
|
---|
| 3527 | }
|
---|
| 3528 | }
|
---|
| 3529 |
|
---|
| 3530 | return (UChar* )NULL;
|
---|
| 3531 | }
|
---|
| 3532 |
|
---|
| 3533 | /* Sunday's quick search */
|
---|
| 3534 | static UChar*
|
---|
| 3535 | bm_search(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
| 3536 | const UChar* text, const UChar* text_end, const UChar* text_range)
|
---|
| 3537 | {
|
---|
| 3538 | const UChar *s, *t, *p, *end;
|
---|
| 3539 | const UChar *tail;
|
---|
| 3540 | ptrdiff_t tlen1;
|
---|
| 3541 |
|
---|
| 3542 | tail = target_end - 1;
|
---|
| 3543 | tlen1 = tail - target;
|
---|
| 3544 | end = text_range + tlen1;
|
---|
| 3545 | if (end > text_end)
|
---|
| 3546 | end = text_end;
|
---|
| 3547 |
|
---|
| 3548 | s = text + tlen1;
|
---|
| 3549 | if (IS_NULL(reg->int_map)) {
|
---|
| 3550 | while (s < end) {
|
---|
| 3551 | p = s;
|
---|
| 3552 | t = tail;
|
---|
| 3553 | while (*p == *t) {
|
---|
| 3554 | if (t == target) return (UChar* )p;
|
---|
| 3555 | p--; t--;
|
---|
| 3556 | }
|
---|
| 3557 | if (s + 1 >= end) break;
|
---|
| 3558 | s += reg->map[s[1]];
|
---|
| 3559 | }
|
---|
| 3560 | }
|
---|
| 3561 | else { /* see int_map[] */
|
---|
| 3562 | while (s < end) {
|
---|
| 3563 | p = s;
|
---|
| 3564 | t = tail;
|
---|
| 3565 | while (*p == *t) {
|
---|
| 3566 | if (t == target) return (UChar* )p;
|
---|
| 3567 | p--; t--;
|
---|
| 3568 | }
|
---|
| 3569 | if (s + 1 >= end) break;
|
---|
| 3570 | s += reg->int_map[s[1]];
|
---|
| 3571 | }
|
---|
| 3572 | }
|
---|
| 3573 | return (UChar* )NULL;
|
---|
| 3574 | }
|
---|
| 3575 |
|
---|
| 3576 | /* Sunday's quick search applied to a multibyte string (ignore case) */
|
---|
| 3577 | static UChar*
|
---|
| 3578 | bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
| 3579 | const UChar* text, const UChar* text_end,
|
---|
| 3580 | const UChar* text_range)
|
---|
| 3581 | {
|
---|
| 3582 | const UChar *s, *se, *t, *end;
|
---|
| 3583 | const UChar *tail;
|
---|
| 3584 | ptrdiff_t skip, tlen1;
|
---|
| 3585 | OnigEncoding enc = reg->enc;
|
---|
| 3586 | int (*mbc_enc_len)(const OnigUChar* p) = enc->mbc_enc_len;
|
---|
| 3587 | int case_fold_flag = reg->case_fold_flag;
|
---|
| 3588 |
|
---|
| 3589 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3590 | fprintf(stderr, "bm_search_notrev_ic: text: %"PRIdPTR" (%p), text_end: %"PRIdPTR" (%p), text_range: %"PRIdPTR" (%p)\n",
|
---|
| 3591 | (intptr_t )text, text, (intptr_t )text_end, text_end, (intptr_t )text_range, text_range);
|
---|
| 3592 | #endif
|
---|
| 3593 |
|
---|
| 3594 | tail = target_end - 1;
|
---|
| 3595 | tlen1 = tail - target;
|
---|
| 3596 | end = text_range;
|
---|
| 3597 | if (end + tlen1 > text_end)
|
---|
| 3598 | end = text_end - tlen1;
|
---|
| 3599 |
|
---|
| 3600 | s = text;
|
---|
| 3601 |
|
---|
| 3602 | if (IS_NULL(reg->int_map)) {
|
---|
| 3603 | while (s < end) {
|
---|
| 3604 | se = s + tlen1;
|
---|
| 3605 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
| 3606 | s, se + 1))
|
---|
| 3607 | return (UChar* )s;
|
---|
| 3608 | if (s + 1 >= end) break;
|
---|
| 3609 | skip = reg->map[se[1]];
|
---|
| 3610 | t = s;
|
---|
| 3611 | do {
|
---|
| 3612 | s += mbc_enc_len(s);
|
---|
| 3613 | } while ((s - t) < skip && s < end);
|
---|
| 3614 | }
|
---|
| 3615 | }
|
---|
| 3616 | else {
|
---|
| 3617 | while (s < end) {
|
---|
| 3618 | se = s + tlen1;
|
---|
| 3619 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
| 3620 | s, se + 1))
|
---|
| 3621 | return (UChar* )s;
|
---|
| 3622 | if (s + 1 >= end) break;
|
---|
| 3623 | skip = reg->int_map[se[1]];
|
---|
| 3624 | t = s;
|
---|
| 3625 | do {
|
---|
| 3626 | s += mbc_enc_len(s);
|
---|
| 3627 | } while ((s - t) < skip && s < end);
|
---|
| 3628 | }
|
---|
| 3629 | }
|
---|
| 3630 |
|
---|
| 3631 | return (UChar* )NULL;
|
---|
| 3632 | }
|
---|
| 3633 |
|
---|
| 3634 | /* Sunday's quick search (ignore case) */
|
---|
| 3635 | static UChar*
|
---|
| 3636 | bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
| 3637 | const UChar* text, const UChar* text_end, const UChar* text_range)
|
---|
| 3638 | {
|
---|
| 3639 | const UChar *s, *p, *end;
|
---|
| 3640 | const UChar *tail;
|
---|
| 3641 | ptrdiff_t tlen1;
|
---|
| 3642 | OnigEncoding enc = reg->enc;
|
---|
| 3643 | int case_fold_flag = reg->case_fold_flag;
|
---|
| 3644 |
|
---|
| 3645 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3646 | fprintf(stderr, "bm_search_ic: text: %"PRIdPTR" (%p), text_end: %"PRIdPTR" (%p), text_range: %"PRIdPTR" (%p)\n",
|
---|
| 3647 | (intptr_t )text, text, (intptr_t )text_end, text_end, (intptr_t )text_range, text_range);
|
---|
| 3648 | #endif
|
---|
| 3649 |
|
---|
| 3650 | tail = target_end - 1;
|
---|
| 3651 | tlen1 = tail - target;
|
---|
| 3652 | end = text_range + tlen1;
|
---|
| 3653 | if (end > text_end)
|
---|
| 3654 | end = text_end;
|
---|
| 3655 |
|
---|
| 3656 | s = text + tlen1;
|
---|
| 3657 | if (IS_NULL(reg->int_map)) {
|
---|
| 3658 | while (s < end) {
|
---|
| 3659 | p = s - tlen1;
|
---|
| 3660 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
| 3661 | p, s + 1))
|
---|
| 3662 | return (UChar* )p;
|
---|
| 3663 | if (s + 1 >= end) break;
|
---|
| 3664 | s += reg->map[s[1]];
|
---|
| 3665 | }
|
---|
| 3666 | }
|
---|
| 3667 | else { /* see int_map[] */
|
---|
| 3668 | while (s < end) {
|
---|
| 3669 | p = s - tlen1;
|
---|
| 3670 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
| 3671 | p, s + 1))
|
---|
| 3672 | return (UChar* )p;
|
---|
| 3673 | if (s + 1 >= end) break;
|
---|
| 3674 | s += reg->int_map[s[1]];
|
---|
| 3675 | }
|
---|
| 3676 | }
|
---|
| 3677 | return (UChar* )NULL;
|
---|
| 3678 | }
|
---|
| 3679 | #endif /* USE_SUNDAY_QUICK_SEARCH */
|
---|
| 3680 |
|
---|
| 3681 | static int
|
---|
| 3682 | set_bm_backward_skip(UChar* s, UChar* end, OnigEncoding enc ARG_UNUSED,
|
---|
| 3683 | int** skip)
|
---|
| 3684 | {
|
---|
| 3685 | int i, len;
|
---|
| 3686 |
|
---|
| 3687 | if (IS_NULL(*skip)) {
|
---|
| 3688 | *skip = (int* )xmalloc(sizeof(int) * ONIG_CHAR_TABLE_SIZE);
|
---|
| 3689 | if (IS_NULL(*skip)) return ONIGERR_MEMORY;
|
---|
| 3690 | }
|
---|
| 3691 |
|
---|
| 3692 | len = (int )(end - s);
|
---|
| 3693 | for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++)
|
---|
| 3694 | (*skip)[i] = len;
|
---|
| 3695 |
|
---|
| 3696 | for (i = len - 1; i > 0; i--)
|
---|
| 3697 | (*skip)[s[i]] = i;
|
---|
| 3698 |
|
---|
| 3699 | return 0;
|
---|
| 3700 | }
|
---|
| 3701 |
|
---|
| 3702 | static UChar*
|
---|
| 3703 | bm_search_backward(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
| 3704 | const UChar* text, const UChar* adjust_text,
|
---|
| 3705 | const UChar* text_end, const UChar* text_start)
|
---|
| 3706 | {
|
---|
| 3707 | const UChar *s, *t, *p;
|
---|
| 3708 |
|
---|
| 3709 | s = text_end - (target_end - target);
|
---|
| 3710 | if (text_start < s)
|
---|
| 3711 | s = text_start;
|
---|
| 3712 | else
|
---|
| 3713 | s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s);
|
---|
| 3714 |
|
---|
| 3715 | while (s >= text) {
|
---|
| 3716 | p = s;
|
---|
| 3717 | t = target;
|
---|
| 3718 | while (t < target_end && *p == *t) {
|
---|
| 3719 | p++; t++;
|
---|
| 3720 | }
|
---|
| 3721 | if (t == target_end)
|
---|
| 3722 | return (UChar* )s;
|
---|
| 3723 |
|
---|
| 3724 | s -= reg->int_map_backward[*s];
|
---|
| 3725 | s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s);
|
---|
| 3726 | }
|
---|
| 3727 |
|
---|
| 3728 | return (UChar* )NULL;
|
---|
| 3729 | }
|
---|
| 3730 |
|
---|
| 3731 | static UChar*
|
---|
| 3732 | map_search(OnigEncoding enc, UChar map[],
|
---|
| 3733 | const UChar* text, const UChar* text_range)
|
---|
| 3734 | {
|
---|
| 3735 | const UChar *s = text;
|
---|
| 3736 |
|
---|
| 3737 | while (s < text_range) {
|
---|
| 3738 | if (map[*s]) return (UChar* )s;
|
---|
| 3739 |
|
---|
| 3740 | s += enclen(enc, s);
|
---|
| 3741 | }
|
---|
| 3742 | return (UChar* )NULL;
|
---|
| 3743 | }
|
---|
| 3744 |
|
---|
| 3745 | static UChar*
|
---|
| 3746 | map_search_backward(OnigEncoding enc, UChar map[],
|
---|
| 3747 | const UChar* text, const UChar* adjust_text,
|
---|
| 3748 | const UChar* text_start)
|
---|
| 3749 | {
|
---|
| 3750 | const UChar *s = text_start;
|
---|
| 3751 |
|
---|
| 3752 | while (s >= text) {
|
---|
| 3753 | if (map[*s]) return (UChar* )s;
|
---|
| 3754 |
|
---|
| 3755 | s = onigenc_get_prev_char_head(enc, adjust_text, s);
|
---|
| 3756 | }
|
---|
| 3757 | return (UChar* )NULL;
|
---|
| 3758 | }
|
---|
| 3759 |
|
---|
| 3760 | extern OnigPosition
|
---|
| 3761 | onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at, OnigRegion* region,
|
---|
| 3762 | OnigOptionType option)
|
---|
| 3763 | {
|
---|
| 3764 | ptrdiff_t r;
|
---|
| 3765 | UChar *prev;
|
---|
| 3766 | OnigMatchArg msa;
|
---|
| 3767 |
|
---|
| 3768 | #if defined(USE_RECOMPILE_API) && defined(USE_MULTI_THREAD_SYSTEM)
|
---|
| 3769 | start:
|
---|
| 3770 | THREAD_ATOMIC_START;
|
---|
| 3771 | if (ONIG_STATE(reg) >= ONIG_STATE_NORMAL) {
|
---|
| 3772 | ONIG_STATE_INC(reg);
|
---|
| 3773 | if (IS_NOT_NULL(reg->chain) && ONIG_STATE(reg) == ONIG_STATE_NORMAL) {
|
---|
| 3774 | onig_chain_reduce(reg);
|
---|
| 3775 | ONIG_STATE_INC(reg);
|
---|
| 3776 | }
|
---|
| 3777 | }
|
---|
| 3778 | else {
|
---|
| 3779 | int n;
|
---|
| 3780 |
|
---|
| 3781 | THREAD_ATOMIC_END;
|
---|
| 3782 | n = 0;
|
---|
| 3783 | while (ONIG_STATE(reg) < ONIG_STATE_NORMAL) {
|
---|
| 3784 | if (++n > THREAD_PASS_LIMIT_COUNT)
|
---|
| 3785 | return ONIGERR_OVER_THREAD_PASS_LIMIT_COUNT;
|
---|
| 3786 | THREAD_PASS;
|
---|
| 3787 | }
|
---|
| 3788 | goto start;
|
---|
| 3789 | }
|
---|
| 3790 | THREAD_ATOMIC_END;
|
---|
| 3791 | #endif /* USE_RECOMPILE_API && USE_MULTI_THREAD_SYSTEM */
|
---|
| 3792 |
|
---|
| 3793 | MATCH_ARG_INIT(msa, option, region, at, at);
|
---|
| 3794 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 3795 | {
|
---|
| 3796 | int offset = at - str;
|
---|
| 3797 | STATE_CHECK_BUFF_INIT(msa, end - str, offset, reg->num_comb_exp_check);
|
---|
| 3798 | }
|
---|
| 3799 | #endif
|
---|
| 3800 |
|
---|
| 3801 | if (region
|
---|
| 3802 | #ifdef USE_POSIX_API_REGION_OPTION
|
---|
| 3803 | && !IS_POSIX_REGION(option)
|
---|
| 3804 | #endif
|
---|
| 3805 | ) {
|
---|
| 3806 | r = onig_region_resize_clear(region, reg->num_mem + 1);
|
---|
| 3807 | }
|
---|
| 3808 | else
|
---|
| 3809 | r = 0;
|
---|
| 3810 |
|
---|
| 3811 | if (r == 0) {
|
---|
| 3812 | prev = (UChar* )onigenc_get_prev_char_head(reg->enc, str, at);
|
---|
| 3813 | r = match_at(reg, str, end,
|
---|
| 3814 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
| 3815 | end,
|
---|
| 3816 | #endif
|
---|
| 3817 | at, prev, &msa);
|
---|
| 3818 | }
|
---|
| 3819 |
|
---|
| 3820 | MATCH_ARG_FREE(msa);
|
---|
| 3821 | ONIG_STATE_DEC_THREAD(reg);
|
---|
| 3822 | return r;
|
---|
| 3823 | }
|
---|
| 3824 |
|
---|
| 3825 | static int
|
---|
| 3826 | forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
|
---|
| 3827 | UChar* range, UChar** low, UChar** high, UChar** low_prev)
|
---|
| 3828 | {
|
---|
| 3829 | UChar *p, *pprev = (UChar* )NULL;
|
---|
| 3830 |
|
---|
| 3831 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3832 | fprintf(stderr, "forward_search_range: str: %"PRIuPTR" (%p), end: %"PRIuPTR" (%p), s: %"PRIuPTR" (%p), range: %"PRIuPTR" (%p)\n",
|
---|
| 3833 | (intptr_t )str, str, (intptr_t )end, end, (intptr_t )s, s, (intptr_t )range, range);
|
---|
| 3834 | #endif
|
---|
| 3835 |
|
---|
| 3836 | p = s;
|
---|
| 3837 | if (reg->dmin > 0) {
|
---|
| 3838 | if (ONIGENC_IS_SINGLEBYTE(reg->enc)) {
|
---|
| 3839 | p += reg->dmin;
|
---|
| 3840 | }
|
---|
| 3841 | else {
|
---|
| 3842 | UChar *q = p + reg->dmin;
|
---|
| 3843 | while (p < q) p += enclen(reg->enc, p);
|
---|
| 3844 | }
|
---|
| 3845 | }
|
---|
| 3846 |
|
---|
| 3847 | retry:
|
---|
| 3848 | switch (reg->optimize) {
|
---|
| 3849 | case ONIG_OPTIMIZE_EXACT:
|
---|
| 3850 | p = slow_search(reg->enc, reg->exact, reg->exact_end, p, end, range);
|
---|
| 3851 | break;
|
---|
| 3852 | case ONIG_OPTIMIZE_EXACT_IC:
|
---|
| 3853 | p = slow_search_ic(reg->enc, reg->case_fold_flag,
|
---|
| 3854 | reg->exact, reg->exact_end, p, end, range);
|
---|
| 3855 | break;
|
---|
| 3856 |
|
---|
| 3857 | case ONIG_OPTIMIZE_EXACT_BM:
|
---|
| 3858 | p = bm_search(reg, reg->exact, reg->exact_end, p, end, range);
|
---|
| 3859 | break;
|
---|
| 3860 |
|
---|
| 3861 | case ONIG_OPTIMIZE_EXACT_BM_NOT_REV:
|
---|
| 3862 | p = bm_search_notrev(reg, reg->exact, reg->exact_end, p, end, range);
|
---|
| 3863 | break;
|
---|
| 3864 |
|
---|
| 3865 | case ONIG_OPTIMIZE_EXACT_BM_IC:
|
---|
| 3866 | p = bm_search_ic(reg, reg->exact, reg->exact_end, p, end, range);
|
---|
| 3867 | break;
|
---|
| 3868 |
|
---|
| 3869 | case ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC:
|
---|
| 3870 | p = bm_search_notrev_ic(reg, reg->exact, reg->exact_end, p, end, range);
|
---|
| 3871 | break;
|
---|
| 3872 |
|
---|
| 3873 | case ONIG_OPTIMIZE_MAP:
|
---|
| 3874 | p = map_search(reg->enc, reg->map, p, range);
|
---|
| 3875 | break;
|
---|
| 3876 | }
|
---|
| 3877 |
|
---|
| 3878 | if (p && p < range) {
|
---|
| 3879 | if (p - reg->dmin < s) {
|
---|
| 3880 | retry_gate:
|
---|
| 3881 | pprev = p;
|
---|
| 3882 | p += enclen(reg->enc, p);
|
---|
| 3883 | goto retry;
|
---|
| 3884 | }
|
---|
| 3885 |
|
---|
| 3886 | if (reg->sub_anchor) {
|
---|
| 3887 | UChar* prev;
|
---|
| 3888 |
|
---|
| 3889 | switch (reg->sub_anchor) {
|
---|
| 3890 | case ANCHOR_BEGIN_LINE:
|
---|
| 3891 | if (!ON_STR_BEGIN(p)) {
|
---|
| 3892 | prev = onigenc_get_prev_char_head(reg->enc,
|
---|
| 3893 | (pprev ? pprev : str), p);
|
---|
| 3894 | if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0))
|
---|
| 3895 | goto retry_gate;
|
---|
| 3896 | }
|
---|
| 3897 | break;
|
---|
| 3898 |
|
---|
| 3899 | case ANCHOR_END_LINE:
|
---|
| 3900 | if (ON_STR_END(p)) {
|
---|
| 3901 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
| 3902 | prev = (UChar* )onigenc_get_prev_char_head(reg->enc,
|
---|
| 3903 | (pprev ? pprev : str), p);
|
---|
| 3904 | if (prev && ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 1))
|
---|
| 3905 | goto retry_gate;
|
---|
| 3906 | #endif
|
---|
| 3907 | }
|
---|
| 3908 | else if (! ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, p, str, end, reg->options, 1))
|
---|
| 3909 | goto retry_gate;
|
---|
| 3910 | break;
|
---|
| 3911 | }
|
---|
| 3912 | }
|
---|
| 3913 |
|
---|
| 3914 | if (reg->dmax == 0) {
|
---|
| 3915 | *low = p;
|
---|
| 3916 | if (low_prev) {
|
---|
| 3917 | if (*low > s)
|
---|
| 3918 | *low_prev = onigenc_get_prev_char_head(reg->enc, s, p);
|
---|
| 3919 | else
|
---|
| 3920 | *low_prev = onigenc_get_prev_char_head(reg->enc,
|
---|
| 3921 | (pprev ? pprev : str), p);
|
---|
| 3922 | }
|
---|
| 3923 | }
|
---|
| 3924 | else {
|
---|
| 3925 | if (reg->dmax != ONIG_INFINITE_DISTANCE) {
|
---|
| 3926 | *low = p - reg->dmax;
|
---|
| 3927 | if (*low > s) {
|
---|
| 3928 | *low = onigenc_get_right_adjust_char_head_with_prev(reg->enc, s,
|
---|
| 3929 | *low, (const UChar** )low_prev);
|
---|
| 3930 | if (low_prev && IS_NULL(*low_prev))
|
---|
| 3931 | *low_prev = onigenc_get_prev_char_head(reg->enc,
|
---|
| 3932 | (pprev ? pprev : s), *low);
|
---|
| 3933 | }
|
---|
| 3934 | else {
|
---|
| 3935 | if (low_prev)
|
---|
| 3936 | *low_prev = onigenc_get_prev_char_head(reg->enc,
|
---|
| 3937 | (pprev ? pprev : str), *low);
|
---|
| 3938 | }
|
---|
| 3939 | }
|
---|
| 3940 | }
|
---|
| 3941 | /* no needs to adjust *high, *high is used as range check only */
|
---|
| 3942 | *high = p - reg->dmin;
|
---|
| 3943 |
|
---|
| 3944 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 3945 | fprintf(stderr,
|
---|
| 3946 | "forward_search_range success: low: %"PRIdPTR", high: %"PRIdPTR", dmin: %"PRIdPTR", dmax: %"PRIdPTR"\n",
|
---|
| 3947 | *low - str, *high - str, reg->dmin, reg->dmax);
|
---|
| 3948 | #endif
|
---|
| 3949 | return 1; /* success */
|
---|
| 3950 | }
|
---|
| 3951 |
|
---|
| 3952 | return 0; /* fail */
|
---|
| 3953 | }
|
---|
| 3954 |
|
---|
| 3955 | #define BM_BACKWARD_SEARCH_LENGTH_THRESHOLD 100
|
---|
| 3956 |
|
---|
| 3957 | static int
|
---|
| 3958 | backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
|
---|
| 3959 | UChar* s, const UChar* range, UChar* adjrange,
|
---|
| 3960 | UChar** low, UChar** high)
|
---|
| 3961 | {
|
---|
| 3962 | int r;
|
---|
| 3963 | UChar *p;
|
---|
| 3964 |
|
---|
| 3965 | range += reg->dmin;
|
---|
| 3966 | p = s;
|
---|
| 3967 |
|
---|
| 3968 | retry:
|
---|
| 3969 | switch (reg->optimize) {
|
---|
| 3970 | case ONIG_OPTIMIZE_EXACT:
|
---|
| 3971 | exact_method:
|
---|
| 3972 | p = slow_search_backward(reg->enc, reg->exact, reg->exact_end,
|
---|
| 3973 | range, adjrange, end, p);
|
---|
| 3974 | break;
|
---|
| 3975 |
|
---|
| 3976 | case ONIG_OPTIMIZE_EXACT_IC:
|
---|
| 3977 | case ONIG_OPTIMIZE_EXACT_BM_IC:
|
---|
| 3978 | case ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC:
|
---|
| 3979 | p = slow_search_backward_ic(reg->enc, reg->case_fold_flag,
|
---|
| 3980 | reg->exact, reg->exact_end,
|
---|
| 3981 | range, adjrange, end, p);
|
---|
| 3982 | break;
|
---|
| 3983 |
|
---|
| 3984 | case ONIG_OPTIMIZE_EXACT_BM:
|
---|
| 3985 | case ONIG_OPTIMIZE_EXACT_BM_NOT_REV:
|
---|
| 3986 | if (IS_NULL(reg->int_map_backward)) {
|
---|
| 3987 | if (s - range < BM_BACKWARD_SEARCH_LENGTH_THRESHOLD)
|
---|
| 3988 | goto exact_method;
|
---|
| 3989 |
|
---|
| 3990 | r = set_bm_backward_skip(reg->exact, reg->exact_end, reg->enc,
|
---|
| 3991 | &(reg->int_map_backward));
|
---|
| 3992 | if (r) return r;
|
---|
| 3993 | }
|
---|
| 3994 | p = bm_search_backward(reg, reg->exact, reg->exact_end, range, adjrange,
|
---|
| 3995 | end, p);
|
---|
| 3996 | break;
|
---|
| 3997 |
|
---|
| 3998 | case ONIG_OPTIMIZE_MAP:
|
---|
| 3999 | p = map_search_backward(reg->enc, reg->map, range, adjrange, p);
|
---|
| 4000 | break;
|
---|
| 4001 | }
|
---|
| 4002 |
|
---|
| 4003 | if (p) {
|
---|
| 4004 | if (reg->sub_anchor) {
|
---|
| 4005 | UChar* prev;
|
---|
| 4006 |
|
---|
| 4007 | switch (reg->sub_anchor) {
|
---|
| 4008 | case ANCHOR_BEGIN_LINE:
|
---|
| 4009 | if (!ON_STR_BEGIN(p)) {
|
---|
| 4010 | prev = onigenc_get_prev_char_head(reg->enc, str, p);
|
---|
| 4011 | if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0)) {
|
---|
| 4012 | p = prev;
|
---|
| 4013 | goto retry;
|
---|
| 4014 | }
|
---|
| 4015 | }
|
---|
| 4016 | break;
|
---|
| 4017 |
|
---|
| 4018 | case ANCHOR_END_LINE:
|
---|
| 4019 | if (ON_STR_END(p)) {
|
---|
| 4020 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
| 4021 | prev = onigenc_get_prev_char_head(reg->enc, adjrange, p);
|
---|
| 4022 | if (IS_NULL(prev)) goto fail;
|
---|
| 4023 | if (ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 1)) {
|
---|
| 4024 | p = prev;
|
---|
| 4025 | goto retry;
|
---|
| 4026 | }
|
---|
| 4027 | #endif
|
---|
| 4028 | }
|
---|
| 4029 | else if (! ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, p, str, end, reg->options, 1)) {
|
---|
| 4030 | p = onigenc_get_prev_char_head(reg->enc, adjrange, p);
|
---|
| 4031 | if (IS_NULL(p)) goto fail;
|
---|
| 4032 | goto retry;
|
---|
| 4033 | }
|
---|
| 4034 | break;
|
---|
| 4035 | }
|
---|
| 4036 | }
|
---|
| 4037 |
|
---|
| 4038 | /* no needs to adjust *high, *high is used as range check only */
|
---|
| 4039 | if (reg->dmax != ONIG_INFINITE_DISTANCE) {
|
---|
| 4040 | *low = p - reg->dmax;
|
---|
| 4041 | *high = p - reg->dmin;
|
---|
| 4042 | *high = onigenc_get_right_adjust_char_head(reg->enc, adjrange, *high);
|
---|
| 4043 | }
|
---|
| 4044 |
|
---|
| 4045 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 4046 | fprintf(stderr, "backward_search_range: low: %d, high: %d\n",
|
---|
| 4047 | (int )(*low - str), (int )(*high - str));
|
---|
| 4048 | #endif
|
---|
| 4049 | return 1; /* success */
|
---|
| 4050 | }
|
---|
| 4051 |
|
---|
| 4052 | fail:
|
---|
| 4053 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 4054 | fprintf(stderr, "backward_search_range: fail.\n");
|
---|
| 4055 | #endif
|
---|
| 4056 | return 0; /* fail */
|
---|
| 4057 | }
|
---|
| 4058 |
|
---|
| 4059 |
|
---|
| 4060 | extern OnigPosition
|
---|
| 4061 | onig_search(regex_t* reg, const UChar* str, const UChar* end,
|
---|
| 4062 | const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option)
|
---|
| 4063 | {
|
---|
| 4064 | return onig_search_gpos(reg, str, end, start, start, range, region, option);
|
---|
| 4065 | }
|
---|
| 4066 |
|
---|
| 4067 | extern OnigPosition
|
---|
| 4068 | onig_search_gpos(regex_t* reg, const UChar* str, const UChar* end,
|
---|
| 4069 | const UChar* global_pos,
|
---|
| 4070 | const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option)
|
---|
| 4071 | {
|
---|
| 4072 | ptrdiff_t r;
|
---|
| 4073 | UChar *s, *prev;
|
---|
| 4074 | OnigMatchArg msa;
|
---|
| 4075 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
| 4076 | const UChar *orig_start = start;
|
---|
| 4077 | const UChar *orig_range = range;
|
---|
| 4078 | #endif
|
---|
| 4079 |
|
---|
| 4080 | #if defined(USE_RECOMPILE_API) && defined(USE_MULTI_THREAD_SYSTEM)
|
---|
| 4081 | start:
|
---|
| 4082 | THREAD_ATOMIC_START;
|
---|
| 4083 | if (ONIG_STATE(reg) >= ONIG_STATE_NORMAL) {
|
---|
| 4084 | ONIG_STATE_INC(reg);
|
---|
| 4085 | if (IS_NOT_NULL(reg->chain) && ONIG_STATE(reg) == ONIG_STATE_NORMAL) {
|
---|
| 4086 | onig_chain_reduce(reg);
|
---|
| 4087 | ONIG_STATE_INC(reg);
|
---|
| 4088 | }
|
---|
| 4089 | }
|
---|
| 4090 | else {
|
---|
| 4091 | int n;
|
---|
| 4092 |
|
---|
| 4093 | THREAD_ATOMIC_END;
|
---|
| 4094 | n = 0;
|
---|
| 4095 | while (ONIG_STATE(reg) < ONIG_STATE_NORMAL) {
|
---|
| 4096 | if (++n > THREAD_PASS_LIMIT_COUNT)
|
---|
| 4097 | return ONIGERR_OVER_THREAD_PASS_LIMIT_COUNT;
|
---|
| 4098 | THREAD_PASS;
|
---|
| 4099 | }
|
---|
| 4100 | goto start;
|
---|
| 4101 | }
|
---|
| 4102 | THREAD_ATOMIC_END;
|
---|
| 4103 | #endif /* USE_RECOMPILE_API && USE_MULTI_THREAD_SYSTEM */
|
---|
| 4104 |
|
---|
| 4105 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 4106 | fprintf(stderr,
|
---|
| 4107 | "onig_search (entry point): str: %"PRIuPTR" (%p), end: %"PRIuPTR", start: %"PRIuPTR", range: %"PRIuPTR"\n",
|
---|
| 4108 | (intptr_t )str, str, end - str, start - str, range - str);
|
---|
| 4109 | #endif
|
---|
| 4110 |
|
---|
| 4111 | if (region
|
---|
| 4112 | #ifdef USE_POSIX_API_REGION_OPTION
|
---|
| 4113 | && !IS_POSIX_REGION(option)
|
---|
| 4114 | #endif
|
---|
| 4115 | ) {
|
---|
| 4116 | r = onig_region_resize_clear(region, reg->num_mem + 1);
|
---|
| 4117 | if (r) goto finish_no_msa;
|
---|
| 4118 | }
|
---|
| 4119 |
|
---|
| 4120 | if (start > end || start < str) goto mismatch_no_msa;
|
---|
| 4121 |
|
---|
| 4122 |
|
---|
| 4123 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
| 4124 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
| 4125 | #define MATCH_AND_RETURN_CHECK(upper_range) \
|
---|
| 4126 | r = match_at(reg, str, end, (upper_range), s, prev, &msa); \
|
---|
| 4127 | if (r != ONIG_MISMATCH) {\
|
---|
| 4128 | if (r >= 0) {\
|
---|
| 4129 | if (! IS_FIND_LONGEST(reg->options)) {\
|
---|
| 4130 | goto match;\
|
---|
| 4131 | }\
|
---|
| 4132 | }\
|
---|
| 4133 | else goto finish; /* error */ \
|
---|
| 4134 | }
|
---|
| 4135 | #else
|
---|
| 4136 | #define MATCH_AND_RETURN_CHECK(upper_range) \
|
---|
| 4137 | r = match_at(reg, str, end, (upper_range), s, prev, &msa); \
|
---|
| 4138 | if (r != ONIG_MISMATCH) {\
|
---|
| 4139 | if (r >= 0) {\
|
---|
| 4140 | goto match;\
|
---|
| 4141 | }\
|
---|
| 4142 | else goto finish; /* error */ \
|
---|
| 4143 | }
|
---|
| 4144 | #endif /* USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE */
|
---|
| 4145 | #else
|
---|
| 4146 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
| 4147 | #define MATCH_AND_RETURN_CHECK(none) \
|
---|
| 4148 | r = match_at(reg, str, end, s, prev, &msa);\
|
---|
| 4149 | if (r != ONIG_MISMATCH) {\
|
---|
| 4150 | if (r >= 0) {\
|
---|
| 4151 | if (! IS_FIND_LONGEST(reg->options)) {\
|
---|
| 4152 | goto match;\
|
---|
| 4153 | }\
|
---|
| 4154 | }\
|
---|
| 4155 | else goto finish; /* error */ \
|
---|
| 4156 | }
|
---|
| 4157 | #else
|
---|
| 4158 | #define MATCH_AND_RETURN_CHECK(none) \
|
---|
| 4159 | r = match_at(reg, str, end, s, prev, &msa);\
|
---|
| 4160 | if (r != ONIG_MISMATCH) {\
|
---|
| 4161 | if (r >= 0) {\
|
---|
| 4162 | goto match;\
|
---|
| 4163 | }\
|
---|
| 4164 | else goto finish; /* error */ \
|
---|
| 4165 | }
|
---|
| 4166 | #endif /* USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE */
|
---|
| 4167 | #endif /* USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE */
|
---|
| 4168 |
|
---|
| 4169 |
|
---|
| 4170 | /* anchor optimize: resume search range */
|
---|
| 4171 | if (reg->anchor != 0 && str < end) {
|
---|
| 4172 | UChar *min_semi_end, *max_semi_end;
|
---|
| 4173 |
|
---|
| 4174 | if (reg->anchor & ANCHOR_BEGIN_POSITION) {
|
---|
| 4175 | /* search start-position only */
|
---|
| 4176 | begin_position:
|
---|
| 4177 | if (range > start)
|
---|
| 4178 | range = start + 1;
|
---|
| 4179 | else
|
---|
| 4180 | range = start;
|
---|
| 4181 | }
|
---|
| 4182 | else if (reg->anchor & ANCHOR_BEGIN_BUF) {
|
---|
| 4183 | /* search str-position only */
|
---|
| 4184 | if (range > start) {
|
---|
| 4185 | if (start != str) goto mismatch_no_msa;
|
---|
| 4186 | range = str + 1;
|
---|
| 4187 | }
|
---|
| 4188 | else {
|
---|
| 4189 | if (range <= str) {
|
---|
| 4190 | start = str;
|
---|
| 4191 | range = str;
|
---|
| 4192 | }
|
---|
| 4193 | else
|
---|
| 4194 | goto mismatch_no_msa;
|
---|
| 4195 | }
|
---|
| 4196 | }
|
---|
| 4197 | else if (reg->anchor & ANCHOR_END_BUF) {
|
---|
| 4198 | min_semi_end = max_semi_end = (UChar* )end;
|
---|
| 4199 |
|
---|
| 4200 | end_buf:
|
---|
| 4201 | if ((OnigDistance )(max_semi_end - str) < reg->anchor_dmin)
|
---|
| 4202 | goto mismatch_no_msa;
|
---|
| 4203 |
|
---|
| 4204 | if (range > start) {
|
---|
| 4205 | if ((OnigDistance )(min_semi_end - start) > reg->anchor_dmax) {
|
---|
| 4206 | start = min_semi_end - reg->anchor_dmax;
|
---|
| 4207 | if (start < end)
|
---|
| 4208 | start = onigenc_get_right_adjust_char_head(reg->enc, str, start);
|
---|
| 4209 | }
|
---|
| 4210 | if ((OnigDistance )(max_semi_end - (range - 1)) < reg->anchor_dmin) {
|
---|
| 4211 | range = max_semi_end - reg->anchor_dmin + 1;
|
---|
| 4212 | }
|
---|
| 4213 |
|
---|
| 4214 | if (start > range) goto mismatch_no_msa;
|
---|
| 4215 | /* If start == range, match with empty at end.
|
---|
| 4216 | Backward search is used. */
|
---|
| 4217 | }
|
---|
| 4218 | else {
|
---|
| 4219 | if ((OnigDistance )(min_semi_end - range) > reg->anchor_dmax) {
|
---|
| 4220 | range = min_semi_end - reg->anchor_dmax;
|
---|
| 4221 | }
|
---|
| 4222 | if ((OnigDistance )(max_semi_end - start) < reg->anchor_dmin) {
|
---|
| 4223 | start = max_semi_end - reg->anchor_dmin;
|
---|
| 4224 | start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, start);
|
---|
| 4225 | }
|
---|
| 4226 | if (range > start) goto mismatch_no_msa;
|
---|
| 4227 | }
|
---|
| 4228 | }
|
---|
| 4229 | else if (reg->anchor & ANCHOR_SEMI_END_BUF) {
|
---|
| 4230 | UChar* pre_end = ONIGENC_STEP_BACK(reg->enc, str, end, 1);
|
---|
| 4231 |
|
---|
| 4232 | max_semi_end = (UChar* )end;
|
---|
| 4233 | if (ONIGENC_IS_MBC_NEWLINE(reg->enc, pre_end, end)) {
|
---|
| 4234 | min_semi_end = pre_end;
|
---|
| 4235 |
|
---|
| 4236 | #ifdef USE_CRNL_AS_LINE_TERMINATOR
|
---|
| 4237 | pre_end = ONIGENC_STEP_BACK(reg->enc, str, pre_end, 1);
|
---|
| 4238 | if (IS_NOT_NULL(pre_end) &&
|
---|
| 4239 | IS_NEWLINE_CRLF(reg->options) &&
|
---|
| 4240 | ONIGENC_IS_MBC_CRNL(reg->enc, pre_end, end)) {
|
---|
| 4241 | min_semi_end = pre_end;
|
---|
| 4242 | }
|
---|
| 4243 | #endif
|
---|
| 4244 | if (min_semi_end > str && start <= min_semi_end) {
|
---|
| 4245 | goto end_buf;
|
---|
| 4246 | }
|
---|
| 4247 | }
|
---|
| 4248 | else {
|
---|
| 4249 | min_semi_end = (UChar* )end;
|
---|
| 4250 | goto end_buf;
|
---|
| 4251 | }
|
---|
| 4252 | }
|
---|
| 4253 | else if ((reg->anchor & ANCHOR_ANYCHAR_STAR_ML)) {
|
---|
| 4254 | if (! (reg->anchor & ANCHOR_LOOK_BEHIND)) {
|
---|
| 4255 | goto begin_position;
|
---|
| 4256 | }
|
---|
| 4257 | }
|
---|
| 4258 | }
|
---|
| 4259 | else if (str == end) { /* empty string */
|
---|
| 4260 | static const UChar address_for_empty_string[] = "";
|
---|
| 4261 |
|
---|
| 4262 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 4263 | fprintf(stderr, "onig_search: empty string.\n");
|
---|
| 4264 | #endif
|
---|
| 4265 |
|
---|
| 4266 | if (reg->threshold_len == 0) {
|
---|
| 4267 | start = end = str = address_for_empty_string;
|
---|
| 4268 | s = (UChar* )start;
|
---|
| 4269 | prev = (UChar* )NULL;
|
---|
| 4270 |
|
---|
| 4271 | MATCH_ARG_INIT(msa, option, region, start, start);
|
---|
| 4272 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 4273 | msa.state_check_buff = (void* )0;
|
---|
| 4274 | msa.state_check_buff_size = 0; /* NO NEED, for valgrind */
|
---|
| 4275 | #endif
|
---|
| 4276 | MATCH_AND_RETURN_CHECK(end);
|
---|
| 4277 | goto mismatch;
|
---|
| 4278 | }
|
---|
| 4279 | goto mismatch_no_msa;
|
---|
| 4280 | }
|
---|
| 4281 |
|
---|
| 4282 | #ifdef ONIG_DEBUG_SEARCH
|
---|
| 4283 | fprintf(stderr, "onig_search(apply anchor): end: %d, start: %d, range: %d\n",
|
---|
| 4284 | (int )(end - str), (int )(start - str), (int )(range - str));
|
---|
| 4285 | #endif
|
---|
| 4286 |
|
---|
| 4287 | MATCH_ARG_INIT(msa, option, region, start, global_pos);
|
---|
| 4288 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 4289 | {
|
---|
| 4290 | int offset = (MIN(start, range) - str);
|
---|
| 4291 | STATE_CHECK_BUFF_INIT(msa, end - str, offset, reg->num_comb_exp_check);
|
---|
| 4292 | }
|
---|
| 4293 | #endif
|
---|
| 4294 |
|
---|
| 4295 | s = (UChar* )start;
|
---|
| 4296 | if (range > start) { /* forward search */
|
---|
| 4297 | if (s > str)
|
---|
| 4298 | prev = onigenc_get_prev_char_head(reg->enc, str, s);
|
---|
| 4299 | else
|
---|
| 4300 | prev = (UChar* )NULL;
|
---|
| 4301 |
|
---|
| 4302 | if (reg->optimize != ONIG_OPTIMIZE_NONE) {
|
---|
| 4303 | UChar *sch_range, *low, *high, *low_prev;
|
---|
| 4304 |
|
---|
| 4305 | sch_range = (UChar* )range;
|
---|
| 4306 | if (reg->dmax != 0) {
|
---|
| 4307 | if (reg->dmax == ONIG_INFINITE_DISTANCE)
|
---|
| 4308 | sch_range = (UChar* )end;
|
---|
| 4309 | else {
|
---|
| 4310 | sch_range += reg->dmax;
|
---|
| 4311 | if (sch_range > end) sch_range = (UChar* )end;
|
---|
| 4312 | }
|
---|
| 4313 | }
|
---|
| 4314 |
|
---|
| 4315 | if ((end - start) < reg->threshold_len)
|
---|
| 4316 | goto mismatch;
|
---|
| 4317 |
|
---|
| 4318 | if (reg->dmax != ONIG_INFINITE_DISTANCE) {
|
---|
| 4319 | do {
|
---|
| 4320 | if (! forward_search_range(reg, str, end, s, sch_range,
|
---|
| 4321 | &low, &high, &low_prev)) goto mismatch;
|
---|
| 4322 | if (s < low) {
|
---|
| 4323 | s = low;
|
---|
| 4324 | prev = low_prev;
|
---|
| 4325 | }
|
---|
| 4326 | while (s <= high) {
|
---|
| 4327 | MATCH_AND_RETURN_CHECK(orig_range);
|
---|
| 4328 | prev = s;
|
---|
| 4329 | s += enclen(reg->enc, s);
|
---|
| 4330 | }
|
---|
| 4331 | } while (s < range);
|
---|
| 4332 | goto mismatch;
|
---|
| 4333 | }
|
---|
| 4334 | else { /* check only. */
|
---|
| 4335 | if (! forward_search_range(reg, str, end, s, sch_range,
|
---|
| 4336 | &low, &high, (UChar** )NULL)) goto mismatch;
|
---|
| 4337 |
|
---|
| 4338 | if ((reg->anchor & ANCHOR_ANYCHAR_STAR) != 0) {
|
---|
| 4339 | do {
|
---|
| 4340 | if ((reg->anchor & ANCHOR_BEGIN_POSITION) == 0)
|
---|
| 4341 | msa.gpos = s; /* move \G position */
|
---|
| 4342 | MATCH_AND_RETURN_CHECK(orig_range);
|
---|
| 4343 | prev = s;
|
---|
| 4344 | s += enclen(reg->enc, s);
|
---|
| 4345 |
|
---|
| 4346 | if ((reg->anchor & (ANCHOR_LOOK_BEHIND | ANCHOR_PREC_READ_NOT)) == 0) {
|
---|
| 4347 | while (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0)
|
---|
| 4348 | && s < range) {
|
---|
| 4349 | prev = s;
|
---|
| 4350 | s += enclen(reg->enc, s);
|
---|
| 4351 | }
|
---|
| 4352 | }
|
---|
| 4353 | } while (s < range);
|
---|
| 4354 | goto mismatch;
|
---|
| 4355 | }
|
---|
| 4356 | }
|
---|
| 4357 | }
|
---|
| 4358 |
|
---|
| 4359 | do {
|
---|
| 4360 | MATCH_AND_RETURN_CHECK(orig_range);
|
---|
| 4361 | prev = s;
|
---|
| 4362 | s += enclen(reg->enc, s);
|
---|
| 4363 | } while (s < range);
|
---|
| 4364 |
|
---|
| 4365 | if (s == range) { /* because empty match with /$/. */
|
---|
| 4366 | MATCH_AND_RETURN_CHECK(orig_range);
|
---|
| 4367 | }
|
---|
| 4368 | }
|
---|
| 4369 | else { /* backward search */
|
---|
| 4370 | if (reg->optimize != ONIG_OPTIMIZE_NONE) {
|
---|
| 4371 | UChar *low, *high, *adjrange, *sch_start;
|
---|
| 4372 |
|
---|
| 4373 | if (range < end)
|
---|
| 4374 | adjrange = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, range);
|
---|
| 4375 | else
|
---|
| 4376 | adjrange = (UChar* )end;
|
---|
| 4377 |
|
---|
| 4378 | if (reg->dmax != ONIG_INFINITE_DISTANCE &&
|
---|
| 4379 | (end - range) >= reg->threshold_len) {
|
---|
| 4380 | do {
|
---|
| 4381 | sch_start = s + reg->dmax;
|
---|
| 4382 | if (sch_start > end) sch_start = (UChar* )end;
|
---|
| 4383 | if (backward_search_range(reg, str, end, sch_start, range, adjrange,
|
---|
| 4384 | &low, &high) <= 0)
|
---|
| 4385 | goto mismatch;
|
---|
| 4386 |
|
---|
| 4387 | if (s > high)
|
---|
| 4388 | s = high;
|
---|
| 4389 |
|
---|
| 4390 | while (s >= low) {
|
---|
| 4391 | prev = onigenc_get_prev_char_head(reg->enc, str, s);
|
---|
| 4392 | MATCH_AND_RETURN_CHECK(orig_start);
|
---|
| 4393 | s = prev;
|
---|
| 4394 | }
|
---|
| 4395 | } while (s >= range);
|
---|
| 4396 | goto mismatch;
|
---|
| 4397 | }
|
---|
| 4398 | else { /* check only. */
|
---|
| 4399 | if ((end - range) < reg->threshold_len) goto mismatch;
|
---|
| 4400 |
|
---|
| 4401 | sch_start = s;
|
---|
| 4402 | if (reg->dmax != 0) {
|
---|
| 4403 | if (reg->dmax == ONIG_INFINITE_DISTANCE)
|
---|
| 4404 | sch_start = (UChar* )end;
|
---|
| 4405 | else {
|
---|
| 4406 | sch_start += reg->dmax;
|
---|
| 4407 | if (sch_start > end) sch_start = (UChar* )end;
|
---|
| 4408 | else
|
---|
| 4409 | sch_start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc,
|
---|
| 4410 | start, sch_start);
|
---|
| 4411 | }
|
---|
| 4412 | }
|
---|
| 4413 | if (backward_search_range(reg, str, end, sch_start, range, adjrange,
|
---|
| 4414 | &low, &high) <= 0) goto mismatch;
|
---|
| 4415 | }
|
---|
| 4416 | }
|
---|
| 4417 |
|
---|
| 4418 | do {
|
---|
| 4419 | prev = onigenc_get_prev_char_head(reg->enc, str, s);
|
---|
| 4420 | MATCH_AND_RETURN_CHECK(orig_start);
|
---|
| 4421 | s = prev;
|
---|
| 4422 | } while (s >= range);
|
---|
| 4423 | }
|
---|
| 4424 |
|
---|
| 4425 | mismatch:
|
---|
| 4426 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
| 4427 | if (IS_FIND_LONGEST(reg->options)) {
|
---|
| 4428 | if (msa.best_len >= 0) {
|
---|
| 4429 | s = msa.best_s;
|
---|
| 4430 | goto match;
|
---|
| 4431 | }
|
---|
| 4432 | }
|
---|
| 4433 | #endif
|
---|
| 4434 | r = ONIG_MISMATCH;
|
---|
| 4435 |
|
---|
| 4436 | finish:
|
---|
| 4437 | MATCH_ARG_FREE(msa);
|
---|
| 4438 | ONIG_STATE_DEC_THREAD(reg);
|
---|
| 4439 |
|
---|
| 4440 | /* If result is mismatch and no FIND_NOT_EMPTY option,
|
---|
| 4441 | then the region is not set in match_at(). */
|
---|
| 4442 | if (IS_FIND_NOT_EMPTY(reg->options) && region
|
---|
| 4443 | #ifdef USE_POSIX_API_REGION_OPTION
|
---|
| 4444 | && !IS_POSIX_REGION(option)
|
---|
| 4445 | #endif
|
---|
| 4446 | ) {
|
---|
| 4447 | onig_region_clear(region);
|
---|
| 4448 | }
|
---|
| 4449 |
|
---|
| 4450 | #ifdef ONIG_DEBUG
|
---|
| 4451 | if (r != ONIG_MISMATCH)
|
---|
| 4452 | fprintf(stderr, "onig_search: error %"PRIdPTR"\n", r);
|
---|
| 4453 | #endif
|
---|
| 4454 | return r;
|
---|
| 4455 |
|
---|
| 4456 | mismatch_no_msa:
|
---|
| 4457 | r = ONIG_MISMATCH;
|
---|
| 4458 | finish_no_msa:
|
---|
| 4459 | ONIG_STATE_DEC_THREAD(reg);
|
---|
| 4460 | #ifdef ONIG_DEBUG
|
---|
| 4461 | if (r != ONIG_MISMATCH)
|
---|
| 4462 | fprintf(stderr, "onig_search: error %"PRIdPTR"\n", r);
|
---|
| 4463 | #endif
|
---|
| 4464 | return r;
|
---|
| 4465 |
|
---|
| 4466 | match:
|
---|
| 4467 | ONIG_STATE_DEC_THREAD(reg);
|
---|
| 4468 | MATCH_ARG_FREE(msa);
|
---|
| 4469 | return s - str;
|
---|
| 4470 | }
|
---|
| 4471 |
|
---|
| 4472 | extern OnigEncoding
|
---|
| 4473 | onig_get_encoding(regex_t* reg)
|
---|
| 4474 | {
|
---|
| 4475 | return reg->enc;
|
---|
| 4476 | }
|
---|
| 4477 |
|
---|
| 4478 | extern OnigOptionType
|
---|
| 4479 | onig_get_options(regex_t* reg)
|
---|
| 4480 | {
|
---|
| 4481 | return reg->options;
|
---|
| 4482 | }
|
---|
| 4483 |
|
---|
| 4484 | extern OnigCaseFoldType
|
---|
| 4485 | onig_get_case_fold_flag(regex_t* reg)
|
---|
| 4486 | {
|
---|
| 4487 | return reg->case_fold_flag;
|
---|
| 4488 | }
|
---|
| 4489 |
|
---|
| 4490 | extern OnigSyntaxType*
|
---|
| 4491 | onig_get_syntax(regex_t* reg)
|
---|
| 4492 | {
|
---|
| 4493 | return reg->syntax;
|
---|
| 4494 | }
|
---|
| 4495 |
|
---|
| 4496 | extern int
|
---|
| 4497 | onig_number_of_captures(regex_t* reg)
|
---|
| 4498 | {
|
---|
| 4499 | return reg->num_mem;
|
---|
| 4500 | }
|
---|
| 4501 |
|
---|
| 4502 | extern int
|
---|
| 4503 | onig_number_of_capture_histories(regex_t* reg)
|
---|
| 4504 | {
|
---|
| 4505 | #ifdef USE_CAPTURE_HISTORY
|
---|
| 4506 | int i, n;
|
---|
| 4507 |
|
---|
| 4508 | n = 0;
|
---|
| 4509 | for (i = 0; i <= ONIG_MAX_CAPTURE_HISTORY_GROUP; i++) {
|
---|
| 4510 | if (BIT_STATUS_AT(reg->capture_history, i) != 0)
|
---|
| 4511 | n++;
|
---|
| 4512 | }
|
---|
| 4513 | return n;
|
---|
| 4514 | #else
|
---|
| 4515 | return 0;
|
---|
| 4516 | #endif
|
---|
| 4517 | }
|
---|
| 4518 |
|
---|
| 4519 | extern void
|
---|
| 4520 | onig_copy_encoding(OnigEncoding to, OnigEncoding from)
|
---|
| 4521 | {
|
---|
| 4522 | *to = *from;
|
---|
| 4523 | }
|
---|
| 4524 |
|
---|