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 <cutl/details/boost/version.hpp>
16   *   DESCRIPTION: Definitions of perl_matcher member functions that are
17   *                common to both the recursive and non-recursive versions.
18   */
19 
20 #ifndef BOOST_REGEX_V4_PERL_MATCHER_COMMON_HPP
21 #define BOOST_REGEX_V4_PERL_MATCHER_COMMON_HPP
22 
23 #ifdef BOOST_MSVC
24 #pragma warning(push)
25 #pragma warning(disable: 4103)
26 #endif
27 #ifdef BOOST_HAS_ABI_HEADERS
28 #  include BOOST_ABI_PREFIX
29 #endif
30 #ifdef BOOST_MSVC
31 #pragma warning(pop)
32 #endif
33 
34 #ifdef __BORLANDC__
35 #  pragma option push -w-8008 -w-8066
36 #endif
37 #ifdef BOOST_MSVC
38 #  pragma warning(push)
39 #  pragma warning(disable: 4800)
40 #endif
41 
42 namespace cutl_details_boost{
43 namespace re_detail{
44 
45 template <class BidiIterator, class Allocator, class traits>
construct_init(const basic_regex<char_type,traits> & e,match_flag_type f)46 void perl_matcher<BidiIterator, Allocator, traits>::construct_init(const basic_regex<char_type, traits>& e, match_flag_type f)
47 {
48    typedef typename regex_iterator_traits<BidiIterator>::iterator_category category;
49    typedef typename basic_regex<char_type, traits>::flag_type expression_flag_type;
50 
51    if(e.empty())
52    {
53       // precondition failure: e is not a valid regex.
54       std::invalid_argument ex("Invalid regular expression object");
55       cutl_details_boost::throw_exception(ex);
56    }
57    pstate = 0;
58    m_match_flags = f;
59    estimate_max_state_count(static_cast<category*>(0));
60    expression_flag_type re_f = re.flags();
61    icase = re_f & regex_constants::icase;
62    if(!(m_match_flags & (match_perl|match_posix)))
63    {
64       if((re_f & (regbase::main_option_type|regbase::no_perl_ex)) == 0)
65          m_match_flags |= match_perl;
66       else if((re_f & (regbase::main_option_type|regbase::emacs_ex)) == (regbase::basic_syntax_group|regbase::emacs_ex))
67          m_match_flags |= match_perl;
68       else if((re_f & (regbase::main_option_type|regbase::literal)) == (regbase::literal))
69          m_match_flags |= match_perl;
70       else
71          m_match_flags |= match_posix;
72    }
73    if(m_match_flags & match_posix)
74    {
75       m_temp_match.reset(new match_results<BidiIterator, Allocator>());
76       m_presult = m_temp_match.get();
77    }
78    else
79       m_presult = &m_result;
80 #ifdef BOOST_REGEX_NON_RECURSIVE
81    m_stack_base = 0;
82    m_backup_state = 0;
83 #endif
84    // find the value to use for matching word boundaries:
85    m_word_mask = re.get_data().m_word_mask;
86    // find bitmask to use for matching '.':
87    match_any_mask = static_cast<unsigned char>((f & match_not_dot_newline) ? re_detail::test_not_newline : re_detail::test_newline);
88 }
89 
90 template <class BidiIterator, class Allocator, class traits>
estimate_max_state_count(std::random_access_iterator_tag *)91 void perl_matcher<BidiIterator, Allocator, traits>::estimate_max_state_count(std::random_access_iterator_tag*)
92 {
93    //
94    // How many states should we allow our machine to visit before giving up?
95    // This is a heuristic: it takes the greater of O(N^2) and O(NS^2)
96    // where N is the length of the string, and S is the number of states
97    // in the machine.  It's tempting to up this to O(N^2S) or even O(N^2S^2)
98    // but these take unreasonably amounts of time to bale out in pathological
99    // cases.
100    //
101    // Calculate NS^2 first:
102    //
103    static const std::ptrdiff_t k = 100000;
104    std::ptrdiff_t dist = cutl_details_boost::re_detail::distance(base, last);
105    if(dist == 0)
106       dist = 1;
107    std::ptrdiff_t states = re.size();
108    if(states == 0)
109       states = 1;
110    states *= states;
111    if((std::numeric_limits<std::ptrdiff_t>::max)() / dist < states)
112    {
113       max_state_count = (std::min)((std::ptrdiff_t)BOOST_REGEX_MAX_STATE_COUNT, (std::numeric_limits<std::ptrdiff_t>::max)() - 2);
114       return;
115    }
116    states *= dist;
117    if((std::numeric_limits<std::ptrdiff_t>::max)() - k < states)
118    {
119       max_state_count = (std::min)((std::ptrdiff_t)BOOST_REGEX_MAX_STATE_COUNT, (std::numeric_limits<std::ptrdiff_t>::max)() - 2);
120       return;
121    }
122    states += k;
123 
124    max_state_count = states;
125 
126    //
127    // Now calculate N^2:
128    //
129    states = dist;
130    if((std::numeric_limits<std::ptrdiff_t>::max)() / dist < states)
131    {
132       max_state_count = (std::min)((std::ptrdiff_t)BOOST_REGEX_MAX_STATE_COUNT, (std::numeric_limits<std::ptrdiff_t>::max)() - 2);
133       return;
134    }
135    states *= dist;
136    if((std::numeric_limits<std::ptrdiff_t>::max)() - k < states)
137    {
138       max_state_count = (std::min)((std::ptrdiff_t)BOOST_REGEX_MAX_STATE_COUNT, (std::numeric_limits<std::ptrdiff_t>::max)() - 2);
139       return;
140    }
141    states += k;
142    //
143    // N^2 can be a very large number indeed, to prevent things getting out
144    // of control, cap the max states:
145    //
146    if(states > BOOST_REGEX_MAX_STATE_COUNT)
147       states = BOOST_REGEX_MAX_STATE_COUNT;
148    //
149    // If (the possibly capped) N^2 is larger than our first estimate,
150    // use this instead:
151    //
152    if(states > max_state_count)
153       max_state_count = states;
154 }
155 
156 template <class BidiIterator, class Allocator, class traits>
estimate_max_state_count(void *)157 inline void perl_matcher<BidiIterator, Allocator, traits>::estimate_max_state_count(void*)
158 {
159    // we don't know how long the sequence is:
160    max_state_count = BOOST_REGEX_MAX_STATE_COUNT;
161 }
162 
163 #ifdef BOOST_REGEX_HAS_MS_STACK_GUARD
164 template <class BidiIterator, class Allocator, class traits>
protected_call(protected_proc_type proc)165 inline bool perl_matcher<BidiIterator, Allocator, traits>::protected_call(
166    protected_proc_type proc)
167 {
168    ::cutl_details_boost::re_detail::concrete_protected_call
169       <perl_matcher<BidiIterator, Allocator, traits> >
170       obj(this, proc);
171    return obj.execute();
172 
173 }
174 #endif
175 
176 template <class BidiIterator, class Allocator, class traits>
match()177 inline bool perl_matcher<BidiIterator, Allocator, traits>::match()
178 {
179 #ifdef BOOST_REGEX_HAS_MS_STACK_GUARD
180    return protected_call(&perl_matcher<BidiIterator, Allocator, traits>::match_imp);
181 #else
182    return match_imp();
183 #endif
184 }
185 
186 template <class BidiIterator, class Allocator, class traits>
match_imp()187 bool perl_matcher<BidiIterator, Allocator, traits>::match_imp()
188 {
189    // initialise our stack if we are non-recursive:
190 #ifdef BOOST_REGEX_NON_RECURSIVE
191    save_state_init init(&m_stack_base, &m_backup_state);
192    used_block_count = BOOST_REGEX_MAX_BLOCKS;
193 #if !defined(BOOST_NO_EXCEPTIONS)
194    try{
195 #endif
196 #endif
197 
198    // reset our state machine:
199    position = base;
200    search_base = base;
201    state_count = 0;
202    m_match_flags |= regex_constants::match_all;
203    m_presult->set_size((m_match_flags & match_nosubs) ? 1 : re.mark_count(), search_base, last);
204    m_presult->set_base(base);
205    m_presult->set_named_subs(this->re.get_named_subs());
206    if(m_match_flags & match_posix)
207       m_result = *m_presult;
208    verify_options(re.flags(), m_match_flags);
209    if(0 == match_prefix())
210       return false;
211    return (m_result[0].second == last) && (m_result[0].first == base);
212 
213 #if defined(BOOST_REGEX_NON_RECURSIVE) && !defined(BOOST_NO_EXCEPTIONS)
214    }
215    catch(...)
216    {
217       // unwind all pushed states, apart from anything else this
218       // ensures that all the states are correctly destructed
219       // not just the memory freed.
220       while(unwind(true)){}
221       throw;
222    }
223 #endif
224 }
225 
226 template <class BidiIterator, class Allocator, class traits>
find()227 inline bool perl_matcher<BidiIterator, Allocator, traits>::find()
228 {
229 #ifdef BOOST_REGEX_HAS_MS_STACK_GUARD
230    return protected_call(&perl_matcher<BidiIterator, Allocator, traits>::find_imp);
231 #else
232    return find_imp();
233 #endif
234 }
235 
236 template <class BidiIterator, class Allocator, class traits>
find_imp()237 bool perl_matcher<BidiIterator, Allocator, traits>::find_imp()
238 {
239    static matcher_proc_type const s_find_vtable[7] =
240    {
241       &perl_matcher<BidiIterator, Allocator, traits>::find_restart_any,
242       &perl_matcher<BidiIterator, Allocator, traits>::find_restart_word,
243       &perl_matcher<BidiIterator, Allocator, traits>::find_restart_line,
244       &perl_matcher<BidiIterator, Allocator, traits>::find_restart_buf,
245       &perl_matcher<BidiIterator, Allocator, traits>::match_prefix,
246       &perl_matcher<BidiIterator, Allocator, traits>::find_restart_lit,
247       &perl_matcher<BidiIterator, Allocator, traits>::find_restart_lit,
248    };
249 
250    // initialise our stack if we are non-recursive:
251 #ifdef BOOST_REGEX_NON_RECURSIVE
252    save_state_init init(&m_stack_base, &m_backup_state);
253    used_block_count = BOOST_REGEX_MAX_BLOCKS;
254 #if !defined(BOOST_NO_EXCEPTIONS)
255    try{
256 #endif
257 #endif
258 
259    state_count = 0;
260    if((m_match_flags & regex_constants::match_init) == 0)
261    {
262       // reset our state machine:
263       search_base = position = base;
264       pstate = re.get_first_state();
265       m_presult->set_size((m_match_flags & match_nosubs) ? 1 : re.mark_count(), base, last);
266       m_presult->set_base(base);
267       m_presult->set_named_subs(this->re.get_named_subs());
268       m_match_flags |= regex_constants::match_init;
269    }
270    else
271    {
272       // start again:
273       search_base = position = m_result[0].second;
274       // If last match was null and match_not_null was not set then increment
275       // our start position, otherwise we go into an infinite loop:
276       if(((m_match_flags & match_not_null) == 0) && (m_result.length() == 0))
277       {
278          if(position == last)
279             return false;
280          else
281             ++position;
282       }
283       // reset $` start:
284       m_presult->set_size((m_match_flags & match_nosubs) ? 1 : re.mark_count(), search_base, last);
285       //if((base != search_base) && (base == backstop))
286       //   m_match_flags |= match_prev_avail;
287    }
288    if(m_match_flags & match_posix)
289    {
290       m_result.set_size(re.mark_count(), base, last);
291       m_result.set_base(base);
292    }
293 
294    verify_options(re.flags(), m_match_flags);
295    // find out what kind of expression we have:
296    unsigned type = (m_match_flags & match_continuous) ?
297       static_cast<unsigned int>(regbase::restart_continue)
298          : static_cast<unsigned int>(re.get_restart_type());
299 
300    // call the appropriate search routine:
301    matcher_proc_type proc = s_find_vtable[type];
302    return (this->*proc)();
303 
304 #if defined(BOOST_REGEX_NON_RECURSIVE) && !defined(BOOST_NO_EXCEPTIONS)
305    }
306    catch(...)
307    {
308       // unwind all pushed states, apart from anything else this
309       // ensures that all the states are correctly destructed
310       // not just the memory freed.
311       while(unwind(true)){}
312       throw;
313    }
314 #endif
315 }
316 
317 template <class BidiIterator, class Allocator, class traits>
match_prefix()318 bool perl_matcher<BidiIterator, Allocator, traits>::match_prefix()
319 {
320    m_has_partial_match = false;
321    m_has_found_match = false;
322    pstate = re.get_first_state();
323    m_presult->set_first(position);
324    restart = position;
325    match_all_states();
326    if(!m_has_found_match && m_has_partial_match && (m_match_flags & match_partial))
327    {
328       m_has_found_match = true;
329       m_presult->set_second(last, 0, false);
330       position = last;
331       if((m_match_flags & match_posix) == match_posix)
332       {
333          m_result.maybe_assign(*m_presult);
334       }
335    }
336 #ifdef BOOST_REGEX_MATCH_EXTRA
337    if(m_has_found_match && (match_extra & m_match_flags))
338    {
339       //
340       // we have a match, reverse the capture information:
341       //
342       for(unsigned i = 0; i < m_presult->size(); ++i)
343       {
344          typename sub_match<BidiIterator>::capture_sequence_type & seq = ((*m_presult)[i]).get_captures();
345          std::reverse(seq.begin(), seq.end());
346       }
347    }
348 #endif
349    if(!m_has_found_match)
350       position = restart; // reset search postion
351    return m_has_found_match;
352 }
353 
354 template <class BidiIterator, class Allocator, class traits>
match_literal()355 bool perl_matcher<BidiIterator, Allocator, traits>::match_literal()
356 {
357    unsigned int len = static_cast<const re_literal*>(pstate)->length;
358    const char_type* what = reinterpret_cast<const char_type*>(static_cast<const re_literal*>(pstate) + 1);
359    //
360    // compare string with what we stored in
361    // our records:
362    for(unsigned int i = 0; i < len; ++i, ++position)
363    {
364       if((position == last) || (traits_inst.translate(*position, icase) != what[i]))
365          return false;
366    }
367    pstate = pstate->next.p;
368    return true;
369 }
370 
371 template <class BidiIterator, class Allocator, class traits>
match_start_line()372 bool perl_matcher<BidiIterator, Allocator, traits>::match_start_line()
373 {
374    if(position == backstop)
375    {
376       if((m_match_flags & match_prev_avail) == 0)
377       {
378          if((m_match_flags & match_not_bol) == 0)
379          {
380             pstate = pstate->next.p;
381             return true;
382          }
383          return false;
384       }
385    }
386    else if(m_match_flags & match_single_line)
387       return false;
388 
389    // check the previous value character:
390    BidiIterator t(position);
391    --t;
392    if(position != last)
393    {
394       if(is_separator(*t) && !((*t == static_cast<char_type>('\r')) && (*position == static_cast<char_type>('\n'))) )
395       {
396          pstate = pstate->next.p;
397          return true;
398       }
399    }
400    else if(is_separator(*t))
401    {
402       pstate = pstate->next.p;
403       return true;
404    }
405    return false;
406 }
407 
408 template <class BidiIterator, class Allocator, class traits>
match_end_line()409 bool perl_matcher<BidiIterator, Allocator, traits>::match_end_line()
410 {
411    if(position != last)
412    {
413       if(m_match_flags & match_single_line)
414          return false;
415       // we're not yet at the end so *first is always valid:
416       if(is_separator(*position))
417       {
418          if((position != backstop) || (m_match_flags & match_prev_avail))
419          {
420             // check that we're not in the middle of \r\n sequence
421             BidiIterator t(position);
422             --t;
423             if((*t == static_cast<char_type>('\r')) && (*position == static_cast<char_type>('\n')))
424             {
425                return false;
426             }
427          }
428          pstate = pstate->next.p;
429          return true;
430       }
431    }
432    else if((m_match_flags & match_not_eol) == 0)
433    {
434       pstate = pstate->next.p;
435       return true;
436    }
437    return false;
438 }
439 
440 template <class BidiIterator, class Allocator, class traits>
match_wild()441 bool perl_matcher<BidiIterator, Allocator, traits>::match_wild()
442 {
443    if(position == last)
444       return false;
445    if(is_separator(*position) && ((match_any_mask & static_cast<const re_dot*>(pstate)->mask) == 0))
446       return false;
447    if((*position == char_type(0)) && (m_match_flags & match_not_dot_null))
448       return false;
449    pstate = pstate->next.p;
450    ++position;
451    return true;
452 }
453 
454 template <class BidiIterator, class Allocator, class traits>
match_word_boundary()455 bool perl_matcher<BidiIterator, Allocator, traits>::match_word_boundary()
456 {
457    bool b; // indcates whether next character is a word character
458    if(position != last)
459    {
460       // prev and this character must be opposites:
461    #if defined(BOOST_REGEX_USE_C_LOCALE) && defined(__GNUC__) && (__GNUC__ == 2) && (__GNUC_MINOR__ < 95)
462       b = traits::isctype(*position, m_word_mask);
463    #else
464       b = traits_inst.isctype(*position, m_word_mask);
465    #endif
466    }
467    else
468    {
469       b = (m_match_flags & match_not_eow) ? true : false;
470    }
471    if((position == backstop) && ((m_match_flags & match_prev_avail) == 0))
472    {
473       if(m_match_flags & match_not_bow)
474          b ^= true;
475       else
476          b ^= false;
477    }
478    else
479    {
480       --position;
481       b ^= traits_inst.isctype(*position, m_word_mask);
482       ++position;
483    }
484    if(b)
485    {
486       pstate = pstate->next.p;
487       return true;
488    }
489    return false; // no match if we get to here...
490 }
491 
492 template <class BidiIterator, class Allocator, class traits>
match_within_word()493 bool perl_matcher<BidiIterator, Allocator, traits>::match_within_word()
494 {
495    if(position == last)
496       return false;
497    // both prev and this character must be m_word_mask:
498    bool prev = traits_inst.isctype(*position, m_word_mask);
499    {
500       bool b;
501       if((position == backstop) && ((m_match_flags & match_prev_avail) == 0))
502          return false;
503       else
504       {
505          --position;
506          b = traits_inst.isctype(*position, m_word_mask);
507          ++position;
508       }
509       if(b == prev)
510       {
511          pstate = pstate->next.p;
512          return true;
513       }
514    }
515    return false;
516 }
517 
518 template <class BidiIterator, class Allocator, class traits>
match_word_start()519 bool perl_matcher<BidiIterator, Allocator, traits>::match_word_start()
520 {
521    if(position == last)
522       return false; // can't be starting a word if we're already at the end of input
523    if(!traits_inst.isctype(*position, m_word_mask))
524       return false; // next character isn't a word character
525    if((position == backstop) && ((m_match_flags & match_prev_avail) == 0))
526    {
527       if(m_match_flags & match_not_bow)
528          return false; // no previous input
529    }
530    else
531    {
532       // otherwise inside buffer:
533       BidiIterator t(position);
534       --t;
535       if(traits_inst.isctype(*t, m_word_mask))
536          return false; // previous character not non-word
537    }
538    // OK we have a match:
539    pstate = pstate->next.p;
540    return true;
541 }
542 
543 template <class BidiIterator, class Allocator, class traits>
match_word_end()544 bool perl_matcher<BidiIterator, Allocator, traits>::match_word_end()
545 {
546    if((position == backstop) && ((m_match_flags & match_prev_avail) == 0))
547       return false;  // start of buffer can't be end of word
548    BidiIterator t(position);
549    --t;
550    if(traits_inst.isctype(*t, m_word_mask) == false)
551       return false;  // previous character wasn't a word character
552 
553    if(position == last)
554    {
555       if(m_match_flags & match_not_eow)
556          return false; // end of buffer but not end of word
557    }
558    else
559    {
560       // otherwise inside buffer:
561       if(traits_inst.isctype(*position, m_word_mask))
562          return false; // next character is a word character
563    }
564    pstate = pstate->next.p;
565    return true;      // if we fall through to here then we've succeeded
566 }
567 
568 template <class BidiIterator, class Allocator, class traits>
match_buffer_start()569 bool perl_matcher<BidiIterator, Allocator, traits>::match_buffer_start()
570 {
571    if((position != backstop) || (m_match_flags & match_not_bob))
572       return false;
573    // OK match:
574    pstate = pstate->next.p;
575    return true;
576 }
577 
578 template <class BidiIterator, class Allocator, class traits>
match_buffer_end()579 bool perl_matcher<BidiIterator, Allocator, traits>::match_buffer_end()
580 {
581    if((position != last) || (m_match_flags & match_not_eob))
582       return false;
583    // OK match:
584    pstate = pstate->next.p;
585    return true;
586 }
587 
588 template <class BidiIterator, class Allocator, class traits>
match_backref()589 bool perl_matcher<BidiIterator, Allocator, traits>::match_backref()
590 {
591    //
592    // Compare with what we previously matched.
593    // Note that this succeeds if the backref did not partisipate
594    // in the match, this is in line with ECMAScript, but not Perl
595    // or PCRE.
596    //
597    int index = static_cast<const re_brace*>(pstate)->index;
598    if(index >= 10000)
599    {
600       named_subexpressions::range_type r = re.get_data().equal_range(index);
601       BOOST_ASSERT(r.first != r.second);
602       do
603       {
604          index = r.first->index;
605          ++r.first;
606       }while((r.first != r.second) && ((*m_presult)[index].matched != true));
607    }
608 
609    if((m_match_flags & match_perl) && !(*m_presult)[index].matched)
610       return false;
611 
612    BidiIterator i = (*m_presult)[index].first;
613    BidiIterator j = (*m_presult)[index].second;
614    while(i != j)
615    {
616       if((position == last) || (traits_inst.translate(*position, icase) != traits_inst.translate(*i, icase)))
617          return false;
618       ++i;
619       ++position;
620    }
621    pstate = pstate->next.p;
622    return true;
623 }
624 
625 template <class BidiIterator, class Allocator, class traits>
match_long_set()626 bool perl_matcher<BidiIterator, Allocator, traits>::match_long_set()
627 {
628    typedef typename traits::char_class_type char_class_type;
629    // let the traits class do the work:
630    if(position == last)
631       return false;
632    BidiIterator t = re_is_set_member(position, last, static_cast<const re_set_long<char_class_type>*>(pstate), re.get_data(), icase);
633    if(t != position)
634    {
635       pstate = pstate->next.p;
636       position = t;
637       return true;
638    }
639    return false;
640 }
641 
642 template <class BidiIterator, class Allocator, class traits>
match_set()643 bool perl_matcher<BidiIterator, Allocator, traits>::match_set()
644 {
645    if(position == last)
646       return false;
647    if(static_cast<const re_set*>(pstate)->_map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])
648    {
649       pstate = pstate->next.p;
650       ++position;
651       return true;
652    }
653    return false;
654 }
655 
656 template <class BidiIterator, class Allocator, class traits>
match_jump()657 bool perl_matcher<BidiIterator, Allocator, traits>::match_jump()
658 {
659    pstate = static_cast<const re_jump*>(pstate)->alt.p;
660    return true;
661 }
662 
663 template <class BidiIterator, class Allocator, class traits>
match_combining()664 bool perl_matcher<BidiIterator, Allocator, traits>::match_combining()
665 {
666    if(position == last)
667       return false;
668    if(is_combining(traits_inst.translate(*position, icase)))
669       return false;
670    ++position;
671    while((position != last) && is_combining(traits_inst.translate(*position, icase)))
672       ++position;
673    pstate = pstate->next.p;
674    return true;
675 }
676 
677 template <class BidiIterator, class Allocator, class traits>
match_soft_buffer_end()678 bool perl_matcher<BidiIterator, Allocator, traits>::match_soft_buffer_end()
679 {
680    if(m_match_flags & match_not_eob)
681       return false;
682    BidiIterator p(position);
683    while((p != last) && is_separator(traits_inst.translate(*p, icase)))++p;
684    if(p != last)
685       return false;
686    pstate = pstate->next.p;
687    return true;
688 }
689 
690 template <class BidiIterator, class Allocator, class traits>
match_restart_continue()691 bool perl_matcher<BidiIterator, Allocator, traits>::match_restart_continue()
692 {
693    if(position == search_base)
694    {
695       pstate = pstate->next.p;
696       return true;
697    }
698    return false;
699 }
700 
701 template <class BidiIterator, class Allocator, class traits>
match_backstep()702 bool perl_matcher<BidiIterator, Allocator, traits>::match_backstep()
703 {
704 #ifdef BOOST_MSVC
705 #pragma warning(push)
706 #pragma warning(disable:4127)
707 #endif
708    if( ::cutl_details_boost::is_random_access_iterator<BidiIterator>::value)
709    {
710       std::ptrdiff_t maxlen = ::cutl_details_boost::re_detail::distance(backstop, position);
711       if(maxlen < static_cast<const re_brace*>(pstate)->index)
712          return false;
713       std::advance(position, -static_cast<const re_brace*>(pstate)->index);
714    }
715    else
716    {
717       int c = static_cast<const re_brace*>(pstate)->index;
718       while(c--)
719       {
720          if(position == backstop)
721             return false;
722          --position;
723       }
724    }
725    pstate = pstate->next.p;
726    return true;
727 #ifdef BOOST_MSVC
728 #pragma warning(pop)
729 #endif
730 }
731 
732 template <class BidiIterator, class Allocator, class traits>
match_assert_backref()733 inline bool perl_matcher<BidiIterator, Allocator, traits>::match_assert_backref()
734 {
735    // return true if marked sub-expression N has been matched:
736    int index = static_cast<const re_brace*>(pstate)->index;
737    bool result = false;
738    if(index == 9999)
739    {
740       // Magic value for a (DEFINE) block:
741       return false;
742    }
743    else if(index > 0)
744    {
745       // Have we matched subexpression "index"?
746       // Check if index is a hash value:
747       if(index >= 10000)
748       {
749          named_subexpressions::range_type r = re.get_data().equal_range(index);
750          while(r.first != r.second)
751          {
752             if((*m_presult)[r.first->index].matched)
753             {
754                result = true;
755                break;
756             }
757             ++r.first;
758          }
759       }
760       else
761       {
762          result = (*m_presult)[index].matched;
763       }
764       pstate = pstate->next.p;
765    }
766    else
767    {
768       // Have we recursed into subexpression "index"?
769       // If index == 0 then check for any recursion at all, otherwise for recursion to -index-1.
770       int idx = -index-1;
771       if(idx >= 10000)
772       {
773          named_subexpressions::range_type r = re.get_data().equal_range(idx);
774          int stack_index = recursion_stack.empty() ? -1 : recursion_stack.back().idx;
775          while(r.first != r.second)
776          {
777             result |= (stack_index == r.first->index);
778             if(result)break;
779             ++r.first;
780          }
781       }
782       else
783       {
784          result = !recursion_stack.empty() && ((recursion_stack.back().idx == idx) || (index == 0));
785       }
786       pstate = pstate->next.p;
787    }
788    return result;
789 }
790 
791 template <class BidiIterator, class Allocator, class traits>
match_toggle_case()792 bool perl_matcher<BidiIterator, Allocator, traits>::match_toggle_case()
793 {
794    // change our case sensitivity:
795    this->icase = static_cast<const re_case*>(pstate)->icase;
796    pstate = pstate->next.p;
797    return true;
798 }
799 
800 
801 template <class BidiIterator, class Allocator, class traits>
find_restart_any()802 bool perl_matcher<BidiIterator, Allocator, traits>::find_restart_any()
803 {
804 #ifdef BOOST_MSVC
805 #pragma warning(push)
806 #pragma warning(disable:4127)
807 #endif
808    const unsigned char* _map = re.get_map();
809    while(true)
810    {
811       // skip everything we can't match:
812       while((position != last) && !can_start(*position, _map, (unsigned char)mask_any) )
813          ++position;
814       if(position == last)
815       {
816          // run out of characters, try a null match if possible:
817          if(re.can_be_null())
818             return match_prefix();
819          break;
820       }
821       // now try and obtain a match:
822       if(match_prefix())
823          return true;
824       if(position == last)
825          return false;
826       ++position;
827    }
828    return false;
829 #ifdef BOOST_MSVC
830 #pragma warning(pop)
831 #endif
832 }
833 
834 template <class BidiIterator, class Allocator, class traits>
find_restart_word()835 bool perl_matcher<BidiIterator, Allocator, traits>::find_restart_word()
836 {
837 #ifdef BOOST_MSVC
838 #pragma warning(push)
839 #pragma warning(disable:4127)
840 #endif
841    // do search optimised for word starts:
842    const unsigned char* _map = re.get_map();
843    if((m_match_flags & match_prev_avail) || (position != base))
844       --position;
845    else if(match_prefix())
846       return true;
847    do
848    {
849       while((position != last) && traits_inst.isctype(*position, m_word_mask))
850          ++position;
851       while((position != last) && !traits_inst.isctype(*position, m_word_mask))
852          ++position;
853       if(position == last)
854          break;
855 
856       if(can_start(*position, _map, (unsigned char)mask_any) )
857       {
858          if(match_prefix())
859             return true;
860       }
861       if(position == last)
862          break;
863    } while(true);
864    return false;
865 #ifdef BOOST_MSVC
866 #pragma warning(pop)
867 #endif
868 }
869 
870 template <class BidiIterator, class Allocator, class traits>
find_restart_line()871 bool perl_matcher<BidiIterator, Allocator, traits>::find_restart_line()
872 {
873    // do search optimised for line starts:
874    const unsigned char* _map = re.get_map();
875    if(match_prefix())
876       return true;
877    while(position != last)
878    {
879       while((position != last) && !is_separator(*position))
880          ++position;
881       if(position == last)
882          return false;
883       ++position;
884       if(position == last)
885       {
886          if(re.can_be_null() && match_prefix())
887             return true;
888          return false;
889       }
890 
891       if( can_start(*position, _map, (unsigned char)mask_any) )
892       {
893          if(match_prefix())
894             return true;
895       }
896       if(position == last)
897          return false;
898       //++position;
899    }
900    return false;
901 }
902 
903 template <class BidiIterator, class Allocator, class traits>
find_restart_buf()904 bool perl_matcher<BidiIterator, Allocator, traits>::find_restart_buf()
905 {
906    if((position == base) && ((m_match_flags & match_not_bob) == 0))
907       return match_prefix();
908    return false;
909 }
910 
911 template <class BidiIterator, class Allocator, class traits>
find_restart_lit()912 bool perl_matcher<BidiIterator, Allocator, traits>::find_restart_lit()
913 {
914 #if 0
915    if(position == last)
916       return false; // can't possibly match if we're at the end already
917 
918    unsigned type = (m_match_flags & match_continuous) ?
919       static_cast<unsigned int>(regbase::restart_continue)
920          : static_cast<unsigned int>(re.get_restart_type());
921 
922    const kmp_info<char_type>* info = access::get_kmp(re);
923    int len = info->len;
924    const char_type* x = info->pstr;
925    int j = 0;
926    while (position != last)
927    {
928       while((j > -1) && (x[j] != traits_inst.translate(*position, icase)))
929          j = info->kmp_next[j];
930       ++position;
931       ++j;
932       if(j >= len)
933       {
934          if(type == regbase::restart_fixed_lit)
935          {
936             std::advance(position, -j);
937             restart = position;
938             std::advance(restart, len);
939             m_result.set_first(position);
940             m_result.set_second(restart);
941             position = restart;
942             return true;
943          }
944          else
945          {
946             restart = position;
947             std::advance(position, -j);
948             if(match_prefix())
949                return true;
950             else
951             {
952                for(int k = 0; (restart != position) && (k < j); ++k, --restart)
953                      {} // dwa 10/20/2000 - warning suppression for MWCW
954                if(restart != last)
955                   ++restart;
956                position = restart;
957                j = 0;  //we could do better than this...
958             }
959          }
960       }
961    }
962    if((m_match_flags & match_partial) && (position == last) && j)
963    {
964       // we need to check for a partial match:
965       restart = position;
966       std::advance(position, -j);
967       return match_prefix();
968    }
969 #endif
970    return false;
971 }
972 
973 } // namespace re_detail
974 
975 } // namespace cutl_details_boost
976 
977 #ifdef BOOST_MSVC
978 #  pragma warning(pop)
979 #endif
980 
981 #ifdef __BORLANDC__
982 #  pragma option pop
983 #endif
984 #ifdef BOOST_MSVC
985 #pragma warning(push)
986 #pragma warning(disable: 4103)
987 #endif
988 #ifdef BOOST_HAS_ABI_HEADERS
989 #  include BOOST_ABI_SUFFIX
990 #endif
991 #ifdef BOOST_MSVC
992 #pragma warning(pop)
993 #endif
994 
995 #endif
996 
997