3 * Copyright (c) 1998-2002
6 * Permission to use, copy, modify, distribute and sell this software
7 * and its documentation for any purpose is hereby granted without fee,
8 * provided that the above copyright notice appear in all copies and
9 * that both that copyright notice and this permission notice appear
10 * in supporting documentation. Dr John Maddock makes no representations
11 * about the suitability of this software for any purpose.
12 * It is provided "as is" without express or implied warranty.
17 * LOCATION: see http://www.boost.org for most recent version.
19 * VERSION: see <boost/version.hpp>
20 * DESCRIPTION: Implements high level class boost::RexEx
24 #define BOOST_REGEX_SOURCE
26 #include <boost/cregex.hpp>
27 #include <boost/regex.hpp>
28 #if !defined(BOOST_NO_STD_STRING)
31 #include <boost/regex/detail/fileiter.hpp>
37 #if __BORLANDC__ < 0x530
39 // we need to instantiate the vector classes we use
40 // since declaring a reference to type doesn't seem to
42 std::vector<std::size_t> inst1;
43 std::vector<std::string> inst2;
49 template <class iterator>
50 std::string to_string(iterator i, iterator j)
62 inline std::string to_string(const char* i, const char* j)
64 return std::string(i, j);
81 #ifndef BOOST_REGEX_NO_FILEITER
82 match_results<mapfile::iterator, regex::allocator_type> fm;
87 #ifndef BOOST_REGEX_NO_FILEITER
88 mapfile::iterator fbase;
90 std::map<int, std::string, std::less<int> > strings;
91 std::map<int, std::ptrdiff_t, std::less<int> > positions;
94 RegExData() : e(), m(),
95 #ifndef BOOST_REGEX_NO_FILEITER
98 t(type_copy), pbase(0), line(0),
99 #ifndef BOOST_REGEX_NO_FILEITER
102 strings(), positions() {}
105 void RegExData::update()
108 strings.erase(strings.begin(), strings.end());
109 positions.erase(positions.begin(), positions.end());
112 for(unsigned int i = 0; i < m.size(); ++i)
114 if(m[i].matched) strings[i] = std::string(m[i].first, m[i].second);
115 positions[i] = m[i].matched ? m[i].first - pbase : -1;
119 #ifndef BOOST_REGEX_NO_FILEITER
122 for(unsigned int i = 0; i < fm.size(); ++i)
124 if(fm[i].matched) strings[i] = to_string(fm[i].first, fm[i].second);
125 positions[i] = fm[i].matched ? fm[i].first - fbase : -1;
133 void RegExData::clean()
136 #ifndef BOOST_REGEX_NO_FILEITER
137 fbase = mapfile::iterator();
138 fm = match_results<mapfile::iterator, regex::allocator_type>();
147 pdata = new re_detail::RegExData();
150 RegEx::RegEx(const RegEx& o)
153 pdata = new re_detail::RegExData(*(o.pdata));
162 RegEx::RegEx(const char* c, bool icase)
165 pdata = new re_detail::RegExData();
166 SetExpression(c, icase);
169 RegEx::RegEx(const std::string& s, bool icase)
172 pdata = new re_detail::RegExData();
173 SetExpression(s.c_str(), icase);
176 RegEx& RegEx::operator=(const RegEx& o)
183 RegEx& RegEx::operator=(const char* p)
186 SetExpression(p, false);
190 unsigned int RegEx::SetExpression(const char* p, bool icase)
193 boost::uint_fast32_t f = icase ? regbase::normal | regbase::use_except | regbase::icase : regbase::normal | regbase::use_except;
194 return pdata->e.set_expression(p, f);
197 unsigned int RegEx::error_code()const
199 return pdata->e.error_code();
203 std::string RegEx::Expression()const
206 return pdata->e.expression();
210 // now matching operators:
212 bool RegEx::Match(const char* p, unsigned int flags)
215 pdata->t = re_detail::RegExData::type_pc;
220 if(regex_match(p, end, pdata->m, pdata->e, flags))
228 bool RegEx::Search(const char* p, unsigned int flags)
231 pdata->t = re_detail::RegExData::type_pc;
236 if(regex_search(p, end, pdata->m, pdata->e, flags))
248 pred1(GrepCallback c, RegEx* i) : cb(c), pe(i) {}
249 bool operator()(const cmatch& m)
256 unsigned int RegEx::Grep(GrepCallback cb, const char* p, unsigned int flags)
259 pdata->t = re_detail::RegExData::type_pc;
264 unsigned int result = regex_grep(re_detail::pred1(cb, this), p, end, pdata->e, flags);
272 std::vector<std::string>& v;
274 pred2(std::vector<std::string>& o, RegEx* e) : v(o), pe(e) {}
275 bool operator()(const cmatch& m)
278 v.push_back(std::string(m[0].first, m[0].second));
282 pred2& operator=(const pred2&);
286 unsigned int RegEx::Grep(std::vector<std::string>& v, const char* p, unsigned int flags)
289 pdata->t = re_detail::RegExData::type_pc;
294 unsigned int result = regex_grep(re_detail::pred2(v, this), p, end, pdata->e, flags);
302 std::vector<std::size_t>& v;
305 pred3(std::vector<std::size_t>& o, const char* pb, RegEx* p) : v(o), base(pb), pe(p) {}
306 bool operator()(const cmatch& m)
309 v.push_back(static_cast<std::size_t>(m[0].first - base));
313 pred3& operator=(const pred3&);
316 unsigned int RegEx::Grep(std::vector<std::size_t>& v, const char* p, unsigned int flags)
319 pdata->t = re_detail::RegExData::type_pc;
324 unsigned int result = regex_grep(re_detail::pred3(v, p, this), p, end, pdata->e, flags);
329 #ifndef BOOST_REGEX_NO_FILEITER
337 pred4(GrepFileCallback c, RegEx* i, const char* f) : cb(c), pe(i), file(f), ok(true) {}
338 bool operator()(const match_results<mapfile::iterator, regex::allocator_type>& m)
340 pe->pdata->t = RegExData::type_pf;
349 void BuildFileList(std::list<std::string>* pl, const char* files, bool recurse)
352 file_iterator start(files);
356 // go through sub directories:
358 std::strcpy(buf, start.root());
361 std::strcpy(buf, ".");
362 std::strcat(buf, directory_iterator::separator());
363 std::strcat(buf, "*");
367 std::strcat(buf, directory_iterator::separator());
368 std::strcat(buf, "*");
370 directory_iterator dstart(buf);
371 directory_iterator dend;
373 // now get the file mask bit of "files":
374 const char* ptr = files;
376 while((ptr != files) && (*ptr != *directory_iterator::separator()) && (*ptr != '/'))--ptr;
377 if(ptr != files) ++ptr;
379 while(dstart != dend)
381 std::sprintf(buf, "%s%s%s", dstart.path(), directory_iterator::separator(), ptr);
382 BuildFileList(pl, buf, recurse);
388 pl->push_back(*start);
394 unsigned int RegEx::GrepFiles(GrepFileCallback cb, const char* files, bool recurse, unsigned int flags)
397 unsigned int result = 0;
398 std::list<std::string> file_list;
399 BuildFileList(&file_list, files, recurse);
400 std::list<std::string>::iterator start, end;
401 start = file_list.begin();
402 end = file_list.end();
406 mapfile map((*start).c_str());
407 pdata->t = re_detail::RegExData::type_pf;
408 pdata->fbase = map.begin();
409 re_detail::pred4 pred(cb, this, (*start).c_str());
410 int r = regex_grep(pred, map.begin(), map.end(), pdata->e, flags);
422 unsigned int RegEx::FindFiles(FindFilesCallback cb, const char* files, bool recurse, unsigned int flags)
425 unsigned int result = 0;
426 std::list<std::string> file_list;
427 BuildFileList(&file_list, files, recurse);
428 std::list<std::string>::iterator start, end;
429 start = file_list.begin();
430 end = file_list.end();
434 mapfile map((*start).c_str());
435 pdata->t = re_detail::RegExData::type_pf;
436 pdata->fbase = map.begin();
438 if(regex_search(map.begin(), map.end(), pdata->fm, pdata->e, flags))
441 if(false == cb((*start).c_str()))
453 std::string RegEx::Merge(const std::string& in, const std::string& fmt,
454 bool copy, unsigned int flags)
457 re_detail::string_out_iterator<std::string> i(result);
458 if(!copy) flags |= format_no_copy;
459 regex_merge(i, in.begin(), in.end(), pdata->e, fmt.c_str(), flags);
463 std::string RegEx::Merge(const char* in, const char* fmt,
464 bool copy, unsigned int flags)
467 if(!copy) flags |= format_no_copy;
468 re_detail::string_out_iterator<std::string> i(result);
469 regex_merge(i, in, in + std::strlen(in), pdata->e, fmt, flags);
473 std::size_t RegEx::Split(std::vector<std::string>& v,
478 return regex_split(std::back_inserter(v), s, pdata->e, flags, max_count);
484 // now operators for returning what matched in more detail:
486 std::size_t RegEx::Position(int i)const
491 case re_detail::RegExData::type_pc:
492 return pdata->m[i].matched ? pdata->m[i].first - pdata->pbase : RegEx::npos;
493 #ifndef BOOST_REGEX_NO_FILEITER
494 case re_detail::RegExData::type_pf:
495 return pdata->fm[i].matched ? pdata->fm[i].first - pdata->fbase : RegEx::npos;
497 case re_detail::RegExData::type_copy:
499 std::map<int, std::ptrdiff_t, std::less<int> >::iterator pos = pdata->positions.find(i);
500 if(pos == pdata->positions.end())
502 return (*pos).second;
508 unsigned int RegEx::Line()const
513 case re_detail::RegExData::type_pc:
514 return pdata->m[0].matched ? pdata->m.line() : RegEx::npos;
515 #ifndef BOOST_REGEX_NO_FILEITER
516 case re_detail::RegExData::type_pf:
517 return pdata->fm[0].matched ? pdata->fm.line() : RegEx::npos;
519 case re_detail::RegExData::type_copy:
527 unsigned int RegEx::Marks()const
530 return pdata->e.mark_count();
534 std::size_t RegEx::Length(int i)const
539 case re_detail::RegExData::type_pc:
540 return pdata->m[i].matched ? pdata->m[i].second - pdata->m[i].first : RegEx::npos;
541 #ifndef BOOST_REGEX_NO_FILEITER
542 case re_detail::RegExData::type_pf:
543 return pdata->fm[i].matched ? pdata->fm[i].second - pdata->fm[i].first : RegEx::npos;
545 case re_detail::RegExData::type_copy:
547 std::map<int, std::string, std::less<int> >::iterator pos = pdata->strings.find(i);
548 if(pos == pdata->strings.end())
550 return (*pos).second.size();
556 bool RegEx::Matched(int i)const
561 case re_detail::RegExData::type_pc:
562 return pdata->m[i].matched;
563 #ifndef BOOST_REGEX_NO_FILEITER
564 case re_detail::RegExData::type_pf:
565 return pdata->fm[i].matched;
567 case re_detail::RegExData::type_copy:
569 std::map<int, std::string, std::less<int> >::iterator pos = pdata->strings.find(i);
570 if(pos == pdata->strings.end())
579 std::string RegEx::What(int i)const
585 case re_detail::RegExData::type_pc:
586 if(pdata->m[i].matched)
587 result.assign(pdata->m[i].first, pdata->m[i].second);
589 case re_detail::RegExData::type_pf:
590 if(pdata->m[i].matched)
591 result.assign(to_string(pdata->m[i].first, pdata->m[i].second));
593 case re_detail::RegExData::type_copy:
595 std::map<int, std::string, std::less<int> >::iterator pos = pdata->strings.find(i);
596 if(pos != pdata->strings.end())
597 result = (*pos).second;
604 const unsigned int RegEx::npos = ~0u;
609 #if defined(__BORLANDC__) && (__BORLANDC__ >= 0x550) && (__BORLANDC__ <= 0x551) && !defined(_RWSTD_COMPILE_INSTANTIATE)
611 // this is an ugly hack to work around an ugly problem:
612 // by default this file will produce unresolved externals during
613 // linking unless _RWSTD_COMPILE_INSTANTIATE is defined (Borland bug).
614 // However if _RWSTD_COMPILE_INSTANTIATE is defined then we get separate
615 // copies of basic_string's static data in the RTL and this DLL, this messes
616 // with basic_string's memory management and results in run-time crashes,
617 // Oh sweet joy of Catch 22....
620 template<> template<>
622 basic_string<char>::replace<const char*>(char* f1, char* f2, const char* i1, const char* i2)
624 unsigned insert_pos = f1 - begin();
625 unsigned remove_len = f2 - f1;
626 unsigned insert_len = i2 - i1;
627 unsigned org_size = size();
628 if(insert_len > remove_len)
630 append(insert_len-remove_len, ' ');
631 std::copy_backward(begin() + insert_pos + remove_len, begin() + org_size, end());
632 std::copy(i1, i2, begin() + insert_pos);
636 std::copy(begin() + insert_pos + remove_len, begin() + org_size, begin() + insert_pos + insert_len);
637 std::copy(i1, i2, begin() + insert_pos);
638 erase(size() + insert_len - remove_len);