cregex.cpp

C++的一个好库。。。现在很流行

   GrepFileCallback cb;
   RegEx* pe;
   const char* file;
   bool ok;
   pred4(GrepFileCallback c, RegEx* i, const char* f) : cb(c), pe(i), file(f), ok(true) {}
   bool operator()(const match_results<mapfile::iterator>& m)
   {
      pe->pdata->t = RegExData::type_pf;
      pe->pdata->fm = m;
      pe->pdata->update();
      ok = cb(file, *pe);
      return ok;
   }
};
}
namespace{
void BuildFileList(std::list<std::string>* pl, const char* files, bool recurse)
{
   file_iterator start(files);
   file_iterator end;
   if(recurse)
   {
      // go through sub directories:
      char buf[MAX_PATH];
      re_detail::overflow_error_if_not_zero(re_detail::strcpy_s(buf, MAX_PATH, start.root()));
      if(*buf == 0)
      {
         re_detail::overflow_error_if_not_zero(re_detail::strcpy_s(buf, MAX_PATH, "."));
         re_detail::overflow_error_if_not_zero(re_detail::strcat_s(buf, MAX_PATH, directory_iterator::separator()));
         re_detail::overflow_error_if_not_zero(re_detail::strcat_s(buf, MAX_PATH, "*"));
      }
      else
      {
         re_detail::overflow_error_if_not_zero(re_detail::strcat_s(buf, MAX_PATH, directory_iterator::separator()));
         re_detail::overflow_error_if_not_zero(re_detail::strcat_s(buf, MAX_PATH, "*"));
      }
      directory_iterator dstart(buf);
      directory_iterator dend;

      // now get the file mask bit of "files":
      const char* ptr = files;
      while(*ptr) ++ptr;
      while((ptr != files) && (*ptr != *directory_iterator::separator()) && (*ptr != '/'))--ptr;
      if(ptr != files) ++ptr;

      while(dstart != dend)
      {
#if BOOST_WORKAROUND(BOOST_MSVC, >= 1400)
         (::sprintf_s)(buf, sizeof(buf), "%s%s%s", dstart.path(), directory_iterator::separator(), ptr);
#else
         (std::sprintf)(buf, "%s%s%s", dstart.path(), directory_iterator::separator(), ptr);
#endif
         BuildFileList(pl, buf, recurse);
         ++dstart;
      }
   }
   while(start != end)
   {
      pl->push_back(*start);
      ++start;
   }
}
}

unsigned int RegEx::GrepFiles(GrepFileCallback cb, const char* files, bool recurse, match_flag_type flags)
{
   unsigned int result = 0;
   std::list<std::string> file_list;
   BuildFileList(&file_list, files, recurse);
   std::list<std::string>::iterator start, end;
   start = file_list.begin();
   end = file_list.end();

   while(start != end)
   {
      mapfile map((*start).c_str());
      pdata->t = re_detail::RegExData::type_pf;
      pdata->fbase = map.begin();
      re_detail::pred4 pred(cb, this, (*start).c_str());
      int r = regex_grep(pred, map.begin(), map.end(), pdata->e, flags);
      result += r;
      ++start;
      pdata->clean();
      if(pred.ok == false)
         return result;
   }

   return result;
}


unsigned int RegEx::FindFiles(FindFilesCallback cb, const char* files, bool recurse, match_flag_type flags)
{
   unsigned int result = 0;
   std::list<std::string> file_list;
   BuildFileList(&file_list, files, recurse);
   std::list<std::string>::iterator start, end;
   start = file_list.begin();
   end = file_list.end();

   while(start != end)
   {
      mapfile map((*start).c_str());
      pdata->t = re_detail::RegExData::type_pf;
      pdata->fbase = map.begin();

      if(regex_search(map.begin(), map.end(), pdata->fm, pdata->e, flags))
      {
         ++result;
         if(false == cb((*start).c_str()))
            return result;
      }
      //pdata->update();
      ++start;
      //pdata->clean();
   }

   return result;
}
#endif

#ifdef BOOST_REGEX_V3
#define regex_replace regex_merge
#endif

std::string RegEx::Merge(const std::string& in, const std::string& fmt,
                    bool copy, match_flag_type flags)
{
   std::string result;
   re_detail::string_out_iterator<std::string> i(result);
   if(!copy) flags |= format_no_copy;
   regex_replace(i, in.begin(), in.end(), pdata->e, fmt.c_str(), flags);
   return result;
}

std::string RegEx::Merge(const char* in, const char* fmt,
                    bool copy, match_flag_type flags)
{
   std::string result;
   if(!copy) flags |= format_no_copy;
   re_detail::string_out_iterator<std::string> i(result);
   regex_replace(i, in, in + std::strlen(in), pdata->e, fmt, flags);
   return result;
}

std::size_t RegEx::Split(std::vector<std::string>& v, 
                      std::string& s,
                      match_flag_type flags,
                      unsigned max_count)
{
   return regex_split(std::back_inserter(v), s, pdata->e, flags, max_count);
}



//
// now operators for returning what matched in more detail:
//
std::size_t RegEx::Position(int i)const
{
   switch(pdata->t)
   {
   case re_detail::RegExData::type_pc:
      return pdata->m[i].matched ? pdata->m[i].first - pdata->pbase : RegEx::npos;
#ifndef BOOST_REGEX_NO_FILEITER
   case re_detail::RegExData::type_pf:
      return pdata->fm[i].matched ? pdata->fm[i].first - pdata->fbase : RegEx::npos;
#endif
   case re_detail::RegExData::type_copy:
      {
      std::map<int, std::ptrdiff_t, std::less<int> >::iterator pos = pdata->positions.find(i);
      if(pos == pdata->positions.end())
         return RegEx::npos;
      return (*pos).second;
      }
   }
   return RegEx::npos;
}

std::size_t RegEx::Marks()const
{
   return pdata->e.mark_count();
}


std::size_t RegEx::Length(int i)const
{
   switch(pdata->t)
   {
   case re_detail::RegExData::type_pc:
      return pdata->m[i].matched ? pdata->m[i].second - pdata->m[i].first : RegEx::npos;
#ifndef BOOST_REGEX_NO_FILEITER
   case re_detail::RegExData::type_pf:
      return pdata->fm[i].matched ? pdata->fm[i].second - pdata->fm[i].first : RegEx::npos;
#endif
   case re_detail::RegExData::type_copy:
      {
      std::map<int, std::string, std::less<int> >::iterator pos = pdata->strings.find(i);
      if(pos == pdata->strings.end())
         return RegEx::npos;
      return (*pos).second.size();
      }
   }
   return RegEx::npos;
}

bool RegEx::Matched(int i)const
{
   switch(pdata->t)
   {
   case re_detail::RegExData::type_pc:
      return pdata->m[i].matched;
#ifndef BOOST_REGEX_NO_FILEITER
   case re_detail::RegExData::type_pf:
      return pdata->fm[i].matched;
#endif      
   case re_detail::RegExData::type_copy:
      {
      std::map<int, std::string, std::less<int> >::iterator pos = pdata->strings.find(i);
      if(pos == pdata->strings.end())
         return false;
      return true;
      }
   }
   return false;
}


std::string RegEx::What(int i)const
{
   std::string result;
   switch(pdata->t)
   {
   case re_detail::RegExData::type_pc:
      if(pdata->m[i].matched) 
         result.assign(pdata->m[i].first, pdata->m[i].second);
      break;
   case re_detail::RegExData::type_pf:
      if(pdata->m[i].matched) 
         result.assign(to_string(pdata->m[i].first, pdata->m[i].second));
      break;
   case re_detail::RegExData::type_copy:
      {
      std::map<int, std::string, std::less<int> >::iterator pos = pdata->strings.find(i);
      if(pos != pdata->strings.end())
         result = (*pos).second;
      break;
      }
   }
   return result;
}

#ifdef BOOST_HAS_LONG_LONG
const std::size_t RegEx::npos = static_cast<std::size_t>(~0ULL);
#else
const std::size_t RegEx::npos = static_cast<std::size_t>(~0UL);
#endif

} // namespace boost

#if defined(__BORLANDC__) && (__BORLANDC__ >= 0x550) && (__BORLANDC__ <= 0x551) && !defined(_RWSTD_COMPILE_INSTANTIATE)
//
// this is an ugly hack to work around an ugly problem:
// by default this file will produce unresolved externals during
// linking unless _RWSTD_COMPILE_INSTANTIATE is defined (Borland bug).
// However if _RWSTD_COMPILE_INSTANTIATE is defined then we get separate
// copies of basic_string's static data in the RTL and this DLL, this messes
// with basic_string's memory management and results in run-time crashes,
// Oh sweet joy of Catch 22....
//
namespace std{
template<> template<>
basic_string<char>&
basic_string<char>::replace<const char*>(char* f1, char* f2, const char* i1, const char* i2)
{
   unsigned insert_pos = f1 - begin();
   unsigned remove_len = f2 - f1;
   unsigned insert_len = i2 - i1;
   unsigned org_size = size();
   if(insert_len > remove_len)
   {
      append(insert_len-remove_len, ' ');
      std::copy_backward(begin() + insert_pos + remove_len, begin() + org_size, end());
      std::copy(i1, i2, begin() + insert_pos);
   }
   else
   {
      std::copy(begin() + insert_pos + remove_len, begin() + org_size, begin() + insert_pos + insert_len);
      std::copy(i1, i2, begin() + insert_pos);
      erase(size() + insert_len - remove_len);
   }
   return *this;
}
}
#endif

#endif