[857] | 1 | /*
|
---|
| 2 | *
|
---|
| 3 | * Copyright (c) 2002
|
---|
| 4 | * John Maddock
|
---|
| 5 | *
|
---|
| 6 | * Use, modification and distribution are subject to the
|
---|
| 7 | * Boost Software License, Version 1.0. (See accompanying file
|
---|
| 8 | * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
|
---|
| 9 | *
|
---|
| 10 | */
|
---|
| 11 |
|
---|
| 12 | /*
|
---|
| 13 | * LOCATION: see http://www.boost.org for most recent version.
|
---|
| 14 | * FILE perl_matcher_common.cpp
|
---|
| 15 | * VERSION see <boost/version.hpp>
|
---|
| 16 | * DESCRIPTION: Definitions of perl_matcher member functions that are
|
---|
| 17 | * specific to the recursive implementation.
|
---|
| 18 | */
|
---|
| 19 |
|
---|
| 20 | #ifndef BOOST_REGEX_V4_PERL_MATCHER_RECURSIVE_HPP
|
---|
| 21 | #define BOOST_REGEX_V4_PERL_MATCHER_RECURSIVE_HPP
|
---|
| 22 |
|
---|
| 23 | #ifdef BOOST_HAS_ABI_HEADERS
|
---|
| 24 | # include BOOST_ABI_PREFIX
|
---|
| 25 | #endif
|
---|
| 26 |
|
---|
| 27 | namespace boost{
|
---|
| 28 | namespace re_detail{
|
---|
| 29 |
|
---|
| 30 | template <class BidiIterator>
|
---|
| 31 | class backup_subex
|
---|
| 32 | {
|
---|
| 33 | int index;
|
---|
| 34 | sub_match<BidiIterator> sub;
|
---|
| 35 | public:
|
---|
| 36 | template <class A>
|
---|
| 37 | backup_subex(const match_results<BidiIterator, A>& w, int i)
|
---|
| 38 | : index(i), sub(w[i], false) {}
|
---|
| 39 | template <class A>
|
---|
| 40 | void restore(match_results<BidiIterator, A>& w)
|
---|
| 41 | {
|
---|
| 42 | w.set_first(sub.first, index);
|
---|
| 43 | w.set_second(sub.second, index, sub.matched);
|
---|
| 44 | }
|
---|
| 45 | const sub_match<BidiIterator>& get() { return sub; }
|
---|
| 46 | };
|
---|
| 47 |
|
---|
| 48 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 49 | bool perl_matcher<BidiIterator, Allocator, traits>::match_all_states()
|
---|
| 50 | {
|
---|
| 51 | static matcher_proc_type const s_match_vtable[29] =
|
---|
| 52 | {
|
---|
| 53 | (&perl_matcher<BidiIterator, Allocator, traits>::match_startmark),
|
---|
| 54 | &perl_matcher<BidiIterator, Allocator, traits>::match_endmark,
|
---|
| 55 | &perl_matcher<BidiIterator, Allocator, traits>::match_literal,
|
---|
| 56 | &perl_matcher<BidiIterator, Allocator, traits>::match_start_line,
|
---|
| 57 | &perl_matcher<BidiIterator, Allocator, traits>::match_end_line,
|
---|
| 58 | &perl_matcher<BidiIterator, Allocator, traits>::match_wild,
|
---|
| 59 | &perl_matcher<BidiIterator, Allocator, traits>::match_match,
|
---|
| 60 | &perl_matcher<BidiIterator, Allocator, traits>::match_word_boundary,
|
---|
| 61 | &perl_matcher<BidiIterator, Allocator, traits>::match_within_word,
|
---|
| 62 | &perl_matcher<BidiIterator, Allocator, traits>::match_word_start,
|
---|
| 63 | &perl_matcher<BidiIterator, Allocator, traits>::match_word_end,
|
---|
| 64 | &perl_matcher<BidiIterator, Allocator, traits>::match_buffer_start,
|
---|
| 65 | &perl_matcher<BidiIterator, Allocator, traits>::match_buffer_end,
|
---|
| 66 | &perl_matcher<BidiIterator, Allocator, traits>::match_backref,
|
---|
| 67 | &perl_matcher<BidiIterator, Allocator, traits>::match_long_set,
|
---|
| 68 | &perl_matcher<BidiIterator, Allocator, traits>::match_set,
|
---|
| 69 | &perl_matcher<BidiIterator, Allocator, traits>::match_jump,
|
---|
| 70 | &perl_matcher<BidiIterator, Allocator, traits>::match_alt,
|
---|
| 71 | &perl_matcher<BidiIterator, Allocator, traits>::match_rep,
|
---|
| 72 | &perl_matcher<BidiIterator, Allocator, traits>::match_combining,
|
---|
| 73 | &perl_matcher<BidiIterator, Allocator, traits>::match_soft_buffer_end,
|
---|
| 74 | &perl_matcher<BidiIterator, Allocator, traits>::match_restart_continue,
|
---|
| 75 | (::boost::is_random_access_iterator<BidiIterator>::value ? &perl_matcher<BidiIterator, Allocator, traits>::match_dot_repeat_fast : &perl_matcher<BidiIterator, Allocator, traits>::match_dot_repeat_slow),
|
---|
| 76 | &perl_matcher<BidiIterator, Allocator, traits>::match_char_repeat,
|
---|
| 77 | &perl_matcher<BidiIterator, Allocator, traits>::match_set_repeat,
|
---|
| 78 | &perl_matcher<BidiIterator, Allocator, traits>::match_long_set_repeat,
|
---|
| 79 | &perl_matcher<BidiIterator, Allocator, traits>::match_backstep,
|
---|
| 80 | &perl_matcher<BidiIterator, Allocator, traits>::match_assert_backref,
|
---|
| 81 | &perl_matcher<BidiIterator, Allocator, traits>::match_toggle_case,
|
---|
| 82 | };
|
---|
| 83 |
|
---|
| 84 | if(state_count > max_state_count)
|
---|
| 85 | raise_error(traits_inst, regex_constants::error_space);
|
---|
| 86 | while(pstate)
|
---|
| 87 | {
|
---|
| 88 | matcher_proc_type proc = s_match_vtable[pstate->type];
|
---|
| 89 | ++state_count;
|
---|
| 90 | if(!(this->*proc)())
|
---|
| 91 | {
|
---|
| 92 | if((m_match_flags & match_partial) && (position == last) && (position != search_base))
|
---|
| 93 | m_has_partial_match = true;
|
---|
| 94 | return 0;
|
---|
| 95 | }
|
---|
| 96 | }
|
---|
| 97 | return true;
|
---|
| 98 | }
|
---|
| 99 |
|
---|
| 100 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 101 | bool perl_matcher<BidiIterator, Allocator, traits>::match_startmark()
|
---|
| 102 | {
|
---|
| 103 | int index = static_cast<const re_brace*>(pstate)->index;
|
---|
| 104 | bool r = true;
|
---|
| 105 | switch(index)
|
---|
| 106 | {
|
---|
| 107 | case 0:
|
---|
| 108 | pstate = pstate->next.p;
|
---|
| 109 | break;
|
---|
| 110 | case -1:
|
---|
| 111 | case -2:
|
---|
| 112 | {
|
---|
| 113 | // forward lookahead assert:
|
---|
| 114 | BidiIterator old_position(position);
|
---|
| 115 | const re_syntax_base* next_pstate = static_cast<const re_jump*>(pstate->next.p)->alt.p->next.p;
|
---|
| 116 | pstate = pstate->next.p->next.p;
|
---|
| 117 | r = match_all_states();
|
---|
| 118 | pstate = next_pstate;
|
---|
| 119 | position = old_position;
|
---|
| 120 | if((r && (index != -1)) || (!r && (index != -2)))
|
---|
| 121 | r = false;
|
---|
| 122 | else
|
---|
| 123 | r = true;
|
---|
| 124 | break;
|
---|
| 125 | }
|
---|
| 126 | case -3:
|
---|
| 127 | {
|
---|
| 128 | // independent sub-expression:
|
---|
| 129 | bool old_independent = m_independent;
|
---|
| 130 | m_independent = true;
|
---|
| 131 | const re_syntax_base* next_pstate = static_cast<const re_jump*>(pstate->next.p)->alt.p->next.p;
|
---|
| 132 | pstate = pstate->next.p->next.p;
|
---|
| 133 | r = match_all_states();
|
---|
| 134 | pstate = next_pstate;
|
---|
| 135 | m_independent = old_independent;
|
---|
| 136 | #ifdef BOOST_REGEX_MATCH_EXTRA
|
---|
| 137 | if(r && (m_match_flags & match_extra))
|
---|
| 138 | {
|
---|
| 139 | //
|
---|
| 140 | // our captures have been stored in *m_presult
|
---|
| 141 | // we need to unpack them, and insert them
|
---|
| 142 | // back in the right order when we unwind the stack:
|
---|
| 143 | //
|
---|
| 144 | unsigned i;
|
---|
| 145 | match_results<BidiIterator, Allocator> tm(*m_presult);
|
---|
| 146 | for(i = 0; i < tm.size(); ++i)
|
---|
| 147 | (*m_presult)[i].get_captures().clear();
|
---|
| 148 | // match everything else:
|
---|
| 149 | r = match_all_states();
|
---|
| 150 | // now place the stored captures back:
|
---|
| 151 | for(i = 0; i < tm.size(); ++i)
|
---|
| 152 | {
|
---|
| 153 | typedef typename sub_match<BidiIterator>::capture_sequence_type seq;
|
---|
| 154 | seq& s1 = (*m_presult)[i].get_captures();
|
---|
| 155 | const seq& s2 = tm[i].captures();
|
---|
| 156 | s1.insert(
|
---|
| 157 | s1.end(),
|
---|
| 158 | s2.begin(),
|
---|
| 159 | s2.end());
|
---|
| 160 | }
|
---|
| 161 | }
|
---|
| 162 | #endif
|
---|
| 163 | break;
|
---|
| 164 | }
|
---|
| 165 | case -4:
|
---|
| 166 | {
|
---|
| 167 | // conditional expression:
|
---|
| 168 | const re_alt* alt = static_cast<const re_alt*>(pstate->next.p);
|
---|
| 169 | BOOST_ASSERT(alt->type == syntax_element_alt);
|
---|
| 170 | pstate = alt->next.p;
|
---|
| 171 | if(pstate->type == syntax_element_assert_backref)
|
---|
| 172 | {
|
---|
| 173 | if(!match_assert_backref())
|
---|
| 174 | pstate = alt->alt.p;
|
---|
| 175 | break;
|
---|
| 176 | }
|
---|
| 177 | else
|
---|
| 178 | {
|
---|
| 179 | // zero width assertion, have to match this recursively:
|
---|
| 180 | BOOST_ASSERT(pstate->type == syntax_element_startmark);
|
---|
| 181 | bool negated = static_cast<const re_brace*>(pstate)->index == -2;
|
---|
| 182 | BidiIterator saved_position = position;
|
---|
| 183 | const re_syntax_base* next_pstate = static_cast<const re_jump*>(pstate->next.p)->alt.p->next.p;
|
---|
| 184 | pstate = pstate->next.p->next.p;
|
---|
| 185 | bool r = match_all_states();
|
---|
| 186 | position = saved_position;
|
---|
| 187 | if(negated)
|
---|
| 188 | r = !r;
|
---|
| 189 | if(r)
|
---|
| 190 | pstate = next_pstate;
|
---|
| 191 | else
|
---|
| 192 | pstate = alt->alt.p;
|
---|
| 193 | break;
|
---|
| 194 | }
|
---|
| 195 | }
|
---|
| 196 | default:
|
---|
| 197 | {
|
---|
| 198 | BOOST_ASSERT(index > 0);
|
---|
| 199 | if((m_match_flags & match_nosubs) == 0)
|
---|
| 200 | {
|
---|
| 201 | backup_subex<BidiIterator> sub(*m_presult, index);
|
---|
| 202 | m_presult->set_first(position, index);
|
---|
| 203 | pstate = pstate->next.p;
|
---|
| 204 | r = match_all_states();
|
---|
| 205 | if(r == false)
|
---|
| 206 | sub.restore(*m_presult);
|
---|
| 207 | #ifdef BOOST_REGEX_MATCH_EXTRA
|
---|
| 208 | //
|
---|
| 209 | // we have a match, push the capture information onto the stack:
|
---|
| 210 | //
|
---|
| 211 | else if(sub.get().matched && (match_extra & m_match_flags))
|
---|
| 212 | ((*m_presult)[index]).get_captures().push_back(sub.get());
|
---|
| 213 | #endif
|
---|
| 214 | }
|
---|
| 215 | else
|
---|
| 216 | {
|
---|
| 217 | pstate = pstate->next.p;
|
---|
| 218 | }
|
---|
| 219 | break;
|
---|
| 220 | }
|
---|
| 221 | }
|
---|
| 222 | return r;
|
---|
| 223 | }
|
---|
| 224 |
|
---|
| 225 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 226 | bool perl_matcher<BidiIterator, Allocator, traits>::match_alt()
|
---|
| 227 | {
|
---|
| 228 | bool take_first, take_second;
|
---|
| 229 | const re_alt* jmp = static_cast<const re_alt*>(pstate);
|
---|
| 230 |
|
---|
| 231 | // find out which of these two alternatives we need to take:
|
---|
| 232 | if(position == last)
|
---|
| 233 | {
|
---|
| 234 | take_first = jmp->can_be_null & mask_take;
|
---|
| 235 | take_second = jmp->can_be_null & mask_skip;
|
---|
| 236 | }
|
---|
| 237 | else
|
---|
| 238 | {
|
---|
| 239 | take_first = can_start(*position, jmp->_map, (unsigned char)mask_take);
|
---|
| 240 | take_second = can_start(*position, jmp->_map, (unsigned char)mask_skip);
|
---|
| 241 | }
|
---|
| 242 |
|
---|
| 243 | if(take_first)
|
---|
| 244 | {
|
---|
| 245 | // we can take the first alternative,
|
---|
| 246 | // see if we need to push next alternative:
|
---|
| 247 | if(take_second)
|
---|
| 248 | {
|
---|
| 249 | BidiIterator oldposition(position);
|
---|
| 250 | const re_syntax_base* old_pstate = jmp->alt.p;
|
---|
| 251 | pstate = pstate->next.p;
|
---|
| 252 | if(!match_all_states())
|
---|
| 253 | {
|
---|
| 254 | pstate = old_pstate;
|
---|
| 255 | position = oldposition;
|
---|
| 256 | }
|
---|
| 257 | return true;
|
---|
| 258 | }
|
---|
| 259 | pstate = pstate->next.p;
|
---|
| 260 | return true;
|
---|
| 261 | }
|
---|
| 262 | if(take_second)
|
---|
| 263 | {
|
---|
| 264 | pstate = jmp->alt.p;
|
---|
| 265 | return true;
|
---|
| 266 | }
|
---|
| 267 | return false; // neither option is possible
|
---|
| 268 | }
|
---|
| 269 |
|
---|
| 270 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 271 | bool perl_matcher<BidiIterator, Allocator, traits>::match_rep()
|
---|
| 272 | {
|
---|
| 273 | #ifdef BOOST_MSVC
|
---|
| 274 | #pragma warning(push)
|
---|
| 275 | #pragma warning(disable:4127 4244)
|
---|
| 276 | #endif
|
---|
| 277 | const re_repeat* rep = static_cast<const re_repeat*>(pstate);
|
---|
| 278 | if(next_count->get_id() != rep->id)
|
---|
| 279 | {
|
---|
| 280 | // we're moving to a different repeat from the last
|
---|
| 281 | // one, so set up a counter object and recurse:
|
---|
| 282 | repeater_count<BidiIterator> r(rep->id, &next_count, position);
|
---|
| 283 | return match_rep();
|
---|
| 284 | }
|
---|
| 285 | //
|
---|
| 286 | // If we've had at least one repeat already, and the last one
|
---|
| 287 | // matched the NULL string then set the repeat count to
|
---|
| 288 | // maximum:
|
---|
| 289 | //
|
---|
| 290 | next_count->check_null_repeat(position, rep->max);
|
---|
| 291 |
|
---|
| 292 | // find out which of these two alternatives we need to take:
|
---|
| 293 | bool take_first, take_second;
|
---|
| 294 | if(position == last)
|
---|
| 295 | {
|
---|
| 296 | take_first = rep->can_be_null & mask_take;
|
---|
| 297 | take_second = rep->can_be_null & mask_skip;
|
---|
| 298 | }
|
---|
| 299 | else
|
---|
| 300 | {
|
---|
| 301 | take_first = can_start(*position, rep->_map, (unsigned char)mask_take);
|
---|
| 302 | take_second = can_start(*position, rep->_map, (unsigned char)mask_skip);
|
---|
| 303 | }
|
---|
| 304 |
|
---|
| 305 | if(next_count->get_count() < rep->min)
|
---|
| 306 | {
|
---|
| 307 | // we must take the repeat:
|
---|
| 308 | if(take_first)
|
---|
| 309 | {
|
---|
| 310 | // increase the counter:
|
---|
| 311 | ++(*next_count);
|
---|
| 312 | pstate = rep->next.p;
|
---|
| 313 | return match_all_states();
|
---|
| 314 | }
|
---|
| 315 | return false;
|
---|
| 316 | }
|
---|
| 317 | bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);
|
---|
| 318 | if(greedy)
|
---|
| 319 | {
|
---|
| 320 | // try and take the repeat if we can:
|
---|
| 321 | if((next_count->get_count() < rep->max) && take_first)
|
---|
| 322 | {
|
---|
| 323 | // store position in case we fail:
|
---|
| 324 | BidiIterator pos = position;
|
---|
| 325 | // increase the counter:
|
---|
| 326 | ++(*next_count);
|
---|
| 327 | pstate = rep->next.p;
|
---|
| 328 | if(match_all_states())
|
---|
| 329 | return true;
|
---|
| 330 | // failed repeat, reset posistion and fall through for alternative:
|
---|
| 331 | position = pos;
|
---|
| 332 | }
|
---|
| 333 | if(take_second)
|
---|
| 334 | {
|
---|
| 335 | pstate = rep->alt.p;
|
---|
| 336 | return true;
|
---|
| 337 | }
|
---|
| 338 | return false; // can't take anything, fail...
|
---|
| 339 | }
|
---|
| 340 | else // non-greedy
|
---|
| 341 | {
|
---|
| 342 | // try and skip the repeat if we can:
|
---|
| 343 | if(take_second)
|
---|
| 344 | {
|
---|
| 345 | // store position in case we fail:
|
---|
| 346 | BidiIterator pos = position;
|
---|
| 347 | pstate = rep->alt.p;
|
---|
| 348 | if(match_all_states())
|
---|
| 349 | return true;
|
---|
| 350 | // failed alternative, reset posistion and fall through for repeat:
|
---|
| 351 | position = pos;
|
---|
| 352 | }
|
---|
| 353 | if((next_count->get_count() < rep->max) && take_first)
|
---|
| 354 | {
|
---|
| 355 | // increase the counter:
|
---|
| 356 | ++(*next_count);
|
---|
| 357 | pstate = rep->next.p;
|
---|
| 358 | return match_all_states();
|
---|
| 359 | }
|
---|
| 360 | }
|
---|
| 361 | return false;
|
---|
| 362 | #ifdef BOOST_MSVC
|
---|
| 363 | #pragma warning(pop)
|
---|
| 364 | #endif
|
---|
| 365 | }
|
---|
| 366 |
|
---|
| 367 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 368 | bool perl_matcher<BidiIterator, Allocator, traits>::match_dot_repeat_slow()
|
---|
| 369 | {
|
---|
| 370 | #ifdef BOOST_MSVC
|
---|
| 371 | #pragma warning(push)
|
---|
| 372 | #pragma warning(disable:4127)
|
---|
| 373 | #endif
|
---|
| 374 | unsigned count = 0;
|
---|
| 375 | const re_repeat* rep = static_cast<const re_repeat*>(pstate);
|
---|
| 376 | re_syntax_base* psingle = rep->next.p;
|
---|
| 377 | // match compulsary repeats first:
|
---|
| 378 | while(count < rep->min)
|
---|
| 379 | {
|
---|
| 380 | pstate = psingle;
|
---|
| 381 | if(!match_wild())
|
---|
| 382 | return false;
|
---|
| 383 | ++count;
|
---|
| 384 | }
|
---|
| 385 | bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);
|
---|
| 386 | if(greedy)
|
---|
| 387 | {
|
---|
| 388 | // normal repeat:
|
---|
| 389 | while(count < rep->max)
|
---|
| 390 | {
|
---|
| 391 | pstate = psingle;
|
---|
| 392 | if(!match_wild())
|
---|
| 393 | break;
|
---|
| 394 | ++count;
|
---|
| 395 | }
|
---|
| 396 | if((rep->leading) && (count < rep->max))
|
---|
| 397 | restart = position;
|
---|
| 398 | pstate = rep;
|
---|
| 399 | return backtrack_till_match(count - rep->min);
|
---|
| 400 | }
|
---|
| 401 | else
|
---|
| 402 | {
|
---|
| 403 | // non-greedy, keep trying till we get a match:
|
---|
| 404 | BidiIterator save_pos;
|
---|
| 405 | do
|
---|
| 406 | {
|
---|
| 407 | if((rep->leading) && (rep->max == UINT_MAX))
|
---|
| 408 | restart = position;
|
---|
| 409 | pstate = rep->alt.p;
|
---|
| 410 | save_pos = position;
|
---|
| 411 | ++state_count;
|
---|
| 412 | if(match_all_states())
|
---|
| 413 | return true;
|
---|
| 414 | if(count >= rep->max)
|
---|
| 415 | return false;
|
---|
| 416 | ++count;
|
---|
| 417 | pstate = psingle;
|
---|
| 418 | position = save_pos;
|
---|
| 419 | if(!match_wild())
|
---|
| 420 | return false;
|
---|
| 421 | }while(true);
|
---|
| 422 | }
|
---|
| 423 | #ifdef BOOST_MSVC
|
---|
| 424 | #pragma warning(pop)
|
---|
| 425 | #endif
|
---|
| 426 | }
|
---|
| 427 |
|
---|
| 428 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 429 | bool perl_matcher<BidiIterator, Allocator, traits>::match_dot_repeat_fast()
|
---|
| 430 | {
|
---|
| 431 | #ifdef BOOST_MSVC
|
---|
| 432 | #pragma warning(push)
|
---|
| 433 | #pragma warning(disable:4127)
|
---|
| 434 | #endif
|
---|
| 435 | if(m_match_flags & match_not_dot_null)
|
---|
| 436 | return match_dot_repeat_slow();
|
---|
| 437 | if((static_cast<const re_dot*>(pstate->next.p)->mask & match_any_mask) == 0)
|
---|
| 438 | return match_dot_repeat_slow();
|
---|
| 439 | //
|
---|
| 440 | // start by working out how much we can skip:
|
---|
| 441 | //
|
---|
| 442 | const re_repeat* rep = static_cast<const re_repeat*>(pstate);
|
---|
| 443 | #ifdef BOOST_MSVC
|
---|
| 444 | #pragma warning(push)
|
---|
| 445 | #pragma warning(disable:4267)
|
---|
| 446 | #endif
|
---|
| 447 | bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);
|
---|
| 448 | std::size_t count = (std::min)(static_cast<std::size_t>(::boost::re_detail::distance(position, last)), static_cast<std::size_t>(greedy ? rep->max : rep->min));
|
---|
| 449 | if(rep->min > count)
|
---|
| 450 | {
|
---|
| 451 | position = last;
|
---|
| 452 | return false; // not enough text left to match
|
---|
| 453 | }
|
---|
| 454 | std::advance(position, count);
|
---|
| 455 | #ifdef BOOST_MSVC
|
---|
| 456 | #pragma warning(pop)
|
---|
| 457 | #endif
|
---|
| 458 | if((rep->leading) && (count < rep->max) && greedy)
|
---|
| 459 | restart = position;
|
---|
| 460 | if(greedy)
|
---|
| 461 | return backtrack_till_match(count - rep->min);
|
---|
| 462 |
|
---|
| 463 | // non-greedy, keep trying till we get a match:
|
---|
| 464 | BidiIterator save_pos;
|
---|
| 465 | do
|
---|
| 466 | {
|
---|
| 467 | while((position != last) && (count < rep->max) && !can_start(*position, rep->_map, mask_skip))
|
---|
| 468 | {
|
---|
| 469 | ++position;
|
---|
| 470 | ++count;
|
---|
| 471 | }
|
---|
| 472 | if((rep->leading) && (rep->max == UINT_MAX))
|
---|
| 473 | restart = position;
|
---|
| 474 | pstate = rep->alt.p;
|
---|
| 475 | save_pos = position;
|
---|
| 476 | ++state_count;
|
---|
| 477 | if(match_all_states())
|
---|
| 478 | return true;
|
---|
| 479 | if(count >= rep->max)
|
---|
| 480 | return false;
|
---|
| 481 | if(position == last)
|
---|
| 482 | return false;
|
---|
| 483 | position = ++save_pos;
|
---|
| 484 | ++count;
|
---|
| 485 | }while(true);
|
---|
| 486 | #ifdef BOOST_MSVC
|
---|
| 487 | #pragma warning(pop)
|
---|
| 488 | #endif
|
---|
| 489 | }
|
---|
| 490 |
|
---|
| 491 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 492 | bool perl_matcher<BidiIterator, Allocator, traits>::match_char_repeat()
|
---|
| 493 | {
|
---|
| 494 | #ifdef BOOST_MSVC
|
---|
| 495 | #pragma warning(push)
|
---|
| 496 | #pragma warning(disable:4127)
|
---|
| 497 | #endif
|
---|
| 498 | #ifdef __BORLANDC__
|
---|
| 499 | #pragma option push -w-8008 -w-8066 -w-8004
|
---|
| 500 | #endif
|
---|
| 501 | const re_repeat* rep = static_cast<const re_repeat*>(pstate);
|
---|
| 502 | BOOST_ASSERT(1 == static_cast<const re_literal*>(rep->next.p)->length);
|
---|
| 503 | const char_type what = *reinterpret_cast<const char_type*>(static_cast<const re_literal*>(rep->next.p) + 1);
|
---|
| 504 | unsigned count = 0;
|
---|
| 505 | //
|
---|
| 506 | // start by working out how much we can skip:
|
---|
| 507 | //
|
---|
| 508 | bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);
|
---|
| 509 | std::size_t desired = greedy ? rep->max : rep->min;
|
---|
| 510 | if(::boost::is_random_access_iterator<BidiIterator>::value)
|
---|
| 511 | {
|
---|
| 512 | BidiIterator end = position;
|
---|
| 513 | std::advance(end, (std::min)((std::size_t)::boost::re_detail::distance(position, last), desired));
|
---|
| 514 | BidiIterator origin(position);
|
---|
| 515 | while((position != end) && (traits_inst.translate(*position, icase) == what))
|
---|
| 516 | {
|
---|
| 517 | ++position;
|
---|
| 518 | }
|
---|
| 519 | count = (unsigned)::boost::re_detail::distance(origin, position);
|
---|
| 520 | }
|
---|
| 521 | else
|
---|
| 522 | {
|
---|
| 523 | while((count < desired) && (position != last) && (traits_inst.translate(*position, icase) == what))
|
---|
| 524 | {
|
---|
| 525 | ++position;
|
---|
| 526 | ++count;
|
---|
| 527 | }
|
---|
| 528 | }
|
---|
| 529 | if((rep->leading) && (count < rep->max) && greedy)
|
---|
| 530 | restart = position;
|
---|
| 531 | if(count < rep->min)
|
---|
| 532 | return false;
|
---|
| 533 |
|
---|
| 534 | if(greedy)
|
---|
| 535 | return backtrack_till_match(count - rep->min);
|
---|
| 536 |
|
---|
| 537 | // non-greedy, keep trying till we get a match:
|
---|
| 538 | BidiIterator save_pos;
|
---|
| 539 | do
|
---|
| 540 | {
|
---|
| 541 | while((position != last) && (count < rep->max) && !can_start(*position, rep->_map, mask_skip))
|
---|
| 542 | {
|
---|
| 543 | if((traits_inst.translate(*position, icase) == what))
|
---|
| 544 | {
|
---|
| 545 | ++position;
|
---|
| 546 | ++count;
|
---|
| 547 | }
|
---|
| 548 | else
|
---|
| 549 | return false; // counldn't repeat even though it was the only option
|
---|
| 550 | }
|
---|
| 551 | if((rep->leading) && (rep->max == UINT_MAX))
|
---|
| 552 | restart = position;
|
---|
| 553 | pstate = rep->alt.p;
|
---|
| 554 | save_pos = position;
|
---|
| 555 | ++state_count;
|
---|
| 556 | if(match_all_states())
|
---|
| 557 | return true;
|
---|
| 558 | if(count >= rep->max)
|
---|
| 559 | return false;
|
---|
| 560 | if(position == last)
|
---|
| 561 | return false;
|
---|
| 562 | position = save_pos;
|
---|
| 563 | if(traits_inst.translate(*position, icase) == what)
|
---|
| 564 | {
|
---|
| 565 | ++position;
|
---|
| 566 | ++count;
|
---|
| 567 | }
|
---|
| 568 | else
|
---|
| 569 | {
|
---|
| 570 | return false;
|
---|
| 571 | }
|
---|
| 572 | }while(true);
|
---|
| 573 | #ifdef __BORLANDC__
|
---|
| 574 | #pragma option pop
|
---|
| 575 | #endif
|
---|
| 576 | #ifdef BOOST_MSVC
|
---|
| 577 | #pragma warning(pop)
|
---|
| 578 | #endif
|
---|
| 579 | }
|
---|
| 580 |
|
---|
| 581 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 582 | bool perl_matcher<BidiIterator, Allocator, traits>::match_set_repeat()
|
---|
| 583 | {
|
---|
| 584 | #ifdef BOOST_MSVC
|
---|
| 585 | #pragma warning(push)
|
---|
| 586 | #pragma warning(disable:4127)
|
---|
| 587 | #endif
|
---|
| 588 | #ifdef __BORLANDC__
|
---|
| 589 | #pragma option push -w-8008 -w-8066 -w-8004
|
---|
| 590 | #endif
|
---|
| 591 | const re_repeat* rep = static_cast<const re_repeat*>(pstate);
|
---|
| 592 | const unsigned char* map = static_cast<const re_set*>(rep->next.p)->_map;
|
---|
| 593 | unsigned count = 0;
|
---|
| 594 | //
|
---|
| 595 | // start by working out how much we can skip:
|
---|
| 596 | //
|
---|
| 597 | bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);
|
---|
| 598 | std::size_t desired = greedy ? rep->max : rep->min;
|
---|
| 599 | if(::boost::is_random_access_iterator<BidiIterator>::value)
|
---|
| 600 | {
|
---|
| 601 | BidiIterator end = position;
|
---|
| 602 | std::advance(end, (std::min)((std::size_t)::boost::re_detail::distance(position, last), desired));
|
---|
| 603 | BidiIterator origin(position);
|
---|
| 604 | while((position != end) && map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])
|
---|
| 605 | {
|
---|
| 606 | ++position;
|
---|
| 607 | }
|
---|
| 608 | count = (unsigned)::boost::re_detail::distance(origin, position);
|
---|
| 609 | }
|
---|
| 610 | else
|
---|
| 611 | {
|
---|
| 612 | while((count < desired) && (position != last) && map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])
|
---|
| 613 | {
|
---|
| 614 | ++position;
|
---|
| 615 | ++count;
|
---|
| 616 | }
|
---|
| 617 | }
|
---|
| 618 | if((rep->leading) && (count < rep->max) && greedy)
|
---|
| 619 | restart = position;
|
---|
| 620 | if(count < rep->min)
|
---|
| 621 | return false;
|
---|
| 622 |
|
---|
| 623 | if(greedy)
|
---|
| 624 | return backtrack_till_match(count - rep->min);
|
---|
| 625 |
|
---|
| 626 | // non-greedy, keep trying till we get a match:
|
---|
| 627 | BidiIterator save_pos;
|
---|
| 628 | do
|
---|
| 629 | {
|
---|
| 630 | while((position != last) && (count < rep->max) && !can_start(*position, rep->_map, mask_skip))
|
---|
| 631 | {
|
---|
| 632 | if(map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])
|
---|
| 633 | {
|
---|
| 634 | ++position;
|
---|
| 635 | ++count;
|
---|
| 636 | }
|
---|
| 637 | else
|
---|
| 638 | return false; // counldn't repeat even though it was the only option
|
---|
| 639 | }
|
---|
| 640 | if((rep->leading) && (rep->max == UINT_MAX))
|
---|
| 641 | restart = position;
|
---|
| 642 | pstate = rep->alt.p;
|
---|
| 643 | save_pos = position;
|
---|
| 644 | ++state_count;
|
---|
| 645 | if(match_all_states())
|
---|
| 646 | return true;
|
---|
| 647 | if(count >= rep->max)
|
---|
| 648 | return false;
|
---|
| 649 | if(position == last)
|
---|
| 650 | return false;
|
---|
| 651 | position = save_pos;
|
---|
| 652 | if(map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])
|
---|
| 653 | {
|
---|
| 654 | ++position;
|
---|
| 655 | ++count;
|
---|
| 656 | }
|
---|
| 657 | else
|
---|
| 658 | {
|
---|
| 659 | return false;
|
---|
| 660 | }
|
---|
| 661 | }while(true);
|
---|
| 662 | #ifdef __BORLANDC__
|
---|
| 663 | #pragma option pop
|
---|
| 664 | #endif
|
---|
| 665 | #ifdef BOOST_MSVC
|
---|
| 666 | #pragma warning(pop)
|
---|
| 667 | #endif
|
---|
| 668 | }
|
---|
| 669 |
|
---|
| 670 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 671 | bool perl_matcher<BidiIterator, Allocator, traits>::match_long_set_repeat()
|
---|
| 672 | {
|
---|
| 673 | #ifdef BOOST_MSVC
|
---|
| 674 | #pragma warning(push)
|
---|
| 675 | #pragma warning(disable:4127)
|
---|
| 676 | #endif
|
---|
| 677 | #ifdef __BORLANDC__
|
---|
| 678 | #pragma option push -w-8008 -w-8066 -w-8004
|
---|
| 679 | #endif
|
---|
| 680 | typedef typename traits::char_class_type char_class_type;
|
---|
| 681 | const re_repeat* rep = static_cast<const re_repeat*>(pstate);
|
---|
| 682 | const re_set_long<char_class_type>* set = static_cast<const re_set_long<char_class_type>*>(pstate->next.p);
|
---|
| 683 | unsigned count = 0;
|
---|
| 684 | //
|
---|
| 685 | // start by working out how much we can skip:
|
---|
| 686 | //
|
---|
| 687 | bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);
|
---|
| 688 | std::size_t desired = greedy ? rep->max : rep->min;
|
---|
| 689 | if(::boost::is_random_access_iterator<BidiIterator>::value)
|
---|
| 690 | {
|
---|
| 691 | BidiIterator end = position;
|
---|
| 692 | std::advance(end, (std::min)((std::size_t)::boost::re_detail::distance(position, last), desired));
|
---|
| 693 | BidiIterator origin(position);
|
---|
| 694 | while((position != end) && (position != re_is_set_member(position, last, set, re.get_data(), icase)))
|
---|
| 695 | {
|
---|
| 696 | ++position;
|
---|
| 697 | }
|
---|
| 698 | count = (unsigned)::boost::re_detail::distance(origin, position);
|
---|
| 699 | }
|
---|
| 700 | else
|
---|
| 701 | {
|
---|
| 702 | while((count < desired) && (position != last) && (position != re_is_set_member(position, last, set, re.get_data(), icase)))
|
---|
| 703 | {
|
---|
| 704 | ++position;
|
---|
| 705 | ++count;
|
---|
| 706 | }
|
---|
| 707 | }
|
---|
| 708 | if((rep->leading) && (count < rep->max) && greedy)
|
---|
| 709 | restart = position;
|
---|
| 710 | if(count < rep->min)
|
---|
| 711 | return false;
|
---|
| 712 |
|
---|
| 713 | if(greedy)
|
---|
| 714 | return backtrack_till_match(count - rep->min);
|
---|
| 715 |
|
---|
| 716 | // non-greedy, keep trying till we get a match:
|
---|
| 717 | BidiIterator save_pos;
|
---|
| 718 | do
|
---|
| 719 | {
|
---|
| 720 | while((position != last) && (count < rep->max) && !can_start(*position, rep->_map, mask_skip))
|
---|
| 721 | {
|
---|
| 722 | if(position != re_is_set_member(position, last, set, re.get_data(), icase))
|
---|
| 723 | {
|
---|
| 724 | ++position;
|
---|
| 725 | ++count;
|
---|
| 726 | }
|
---|
| 727 | else
|
---|
| 728 | return false; // counldn't repeat even though it was the only option
|
---|
| 729 | }
|
---|
| 730 | if((rep->leading) && (rep->max == UINT_MAX))
|
---|
| 731 | restart = position;
|
---|
| 732 | pstate = rep->alt.p;
|
---|
| 733 | save_pos = position;
|
---|
| 734 | ++state_count;
|
---|
| 735 | if(match_all_states())
|
---|
| 736 | return true;
|
---|
| 737 | if(count >= rep->max)
|
---|
| 738 | return false;
|
---|
| 739 | if(position == last)
|
---|
| 740 | return false;
|
---|
| 741 | position = save_pos;
|
---|
| 742 | if(position != re_is_set_member(position, last, set, re.get_data(), icase))
|
---|
| 743 | {
|
---|
| 744 | ++position;
|
---|
| 745 | ++count;
|
---|
| 746 | }
|
---|
| 747 | else
|
---|
| 748 | {
|
---|
| 749 | return false;
|
---|
| 750 | }
|
---|
| 751 | }while(true);
|
---|
| 752 | #ifdef __BORLANDC__
|
---|
| 753 | #pragma option pop
|
---|
| 754 | #endif
|
---|
| 755 | #ifdef BOOST_MSVC
|
---|
| 756 | #pragma warning(pop)
|
---|
| 757 | #endif
|
---|
| 758 | }
|
---|
| 759 |
|
---|
| 760 | template <class BidiIterator, class Allocator, class traits>
|
---|
| 761 | bool perl_matcher<BidiIterator, Allocator, traits>::backtrack_till_match(std::size_t count)
|
---|
| 762 | {
|
---|
| 763 | #ifdef BOOST_MSVC
|
---|
| 764 | #pragma warning(push)
|
---|
| 765 | #pragma warning(disable:4127)
|
---|
| 766 | #endif
|
---|
| 767 | if((m_match_flags & match_partial) && (position == last))
|
---|
| 768 | m_has_partial_match = true;
|
---|
| 769 |
|
---|
| 770 | const re_repeat* rep = static_cast<const re_repeat*>(pstate);
|
---|
| 771 | BidiIterator backtrack = position;
|
---|
| 772 | if(position == last)
|
---|
| 773 | {
|
---|
| 774 | if(rep->can_be_null & mask_skip)
|
---|
| 775 | {
|
---|
| 776 | pstate = rep->alt.p;
|
---|
| 777 | if(match_all_states())
|
---|
| 778 | return true;
|
---|
| 779 | }
|
---|
| 780 | if(count)
|
---|
| 781 | {
|
---|
| 782 | position = --backtrack;
|
---|
| 783 | --count;
|
---|
| 784 | }
|
---|
| 785 | else
|
---|
| 786 | return false;
|
---|
| 787 | }
|
---|
| 788 | do
|
---|
| 789 | {
|
---|
| 790 | while(count && !can_start(*position, rep->_map, mask_skip))
|
---|
| 791 | {
|
---|
| 792 | --position;
|
---|
| 793 | --count;
|
---|
| 794 | ++state_count;
|
---|
| 795 | }
|
---|
| 796 | pstate = rep->alt.p;
|
---|
| 797 | backtrack = position;
|
---|
| 798 | if(match_all_states())
|
---|
| 799 | return true;
|
---|
| 800 | if(count == 0)
|
---|
| 801 | return false;
|
---|
| 802 | position = --backtrack;
|
---|
| 803 | ++state_count;
|
---|
| 804 | --count;
|
---|
| 805 | }while(true);
|
---|
| 806 | #ifdef BOOST_MSVC
|
---|
| 807 | #pragma warning(pop)
|
---|
| 808 | #endif
|
---|
| 809 | }
|
---|
| 810 |
|
---|
| 811 | } // namespace re_detail
|
---|
| 812 | } // namespace boost
|
---|
| 813 |
|
---|
| 814 | #ifdef BOOST_HAS_ABI_HEADERS
|
---|
| 815 | # include BOOST_ABI_SUFFIX
|
---|
| 816 | #endif
|
---|
| 817 |
|
---|
| 818 | #endif
|
---|
| 819 |
|
---|