perl_matcher.hpp

C++的一个好库。。。现在很流行

         while(p->id != id)
            p = p->next;
         count = p->count;
         start_pos = p->start_pos;
      }
   }
   ~repeater_count()
   {
      *stack = next;
   }
   std::size_t get_count() { return count; }
   int get_id() { return id; }
   std::size_t operator++() { return ++count; }
   bool check_null_repeat(const BidiIterator& pos, std::size_t max)
   {
      // this is called when we are about to start a new repeat,
      // if the last one was NULL move our count to max,
      // otherwise save the current position.
      bool result = (count == 0) ? false : (pos == start_pos);
      if(result)
         count = max;
      else
         start_pos = pos;
      return result;
   }
};

struct saved_state;

enum saved_state_type
{
   saved_type_end = 0,
   saved_type_paren = 1,
   saved_type_recurse = 2,
   saved_type_assertion = 3,
   saved_state_alt = 4,
   saved_state_repeater_count = 5,
   saved_state_extra_block = 6,
   saved_state_greedy_single_repeat = 7,
   saved_state_rep_slow_dot = 8,
   saved_state_rep_fast_dot = 9,
   saved_state_rep_char = 10,
   saved_state_rep_short_set = 11,
   saved_state_rep_long_set = 12,
   saved_state_non_greedy_long_repeat = 13, 
   saved_state_count = 14
};

#ifdef BOOST_MSVC
#pragma warning(push)
#pragma warning(disable : 4251 4231 4660)
#endif

template <class BidiIterator, class Allocator, class traits>
class perl_matcher
{
public:
   typedef typename traits::char_type char_type;
   typedef perl_matcher<BidiIterator, Allocator, traits> self_type;
   typedef bool (self_type::*matcher_proc_type)(void);
   typedef typename traits::size_type traits_size_type;
   typedef typename is_byte<char_type>::width_type width_type;
   typedef typename regex_iterator_traits<BidiIterator>::difference_type difference_type;

   perl_matcher(BidiIterator first, BidiIterator end, 
      match_results<BidiIterator, Allocator>& what, 
      const basic_regex<char_type, traits>& e,
      match_flag_type f,
      BidiIterator base);

   bool match();
   bool find();

   void setf(match_flag_type f)
   { m_match_flags |= f; }
   void unsetf(match_flag_type f)
   { m_match_flags &= ~f; }

private:
   void construct_init(const basic_regex<char_type, traits>& e, match_flag_type f);
   bool find_imp();
   bool match_imp();
#ifdef BOOST_REGEX_HAS_MS_STACK_GUARD
   typedef bool (perl_matcher::*protected_proc_type)();
   bool protected_call(protected_proc_type);
#endif
   void estimate_max_state_count(std::random_access_iterator_tag*);
   void estimate_max_state_count(void*);
   bool match_prefix();
   bool match_all_states();

   // match procs, stored in s_match_vtable:
   bool match_startmark();
   bool match_endmark();
   bool match_literal();
   bool match_start_line();
   bool match_end_line();
   bool match_wild();
   bool match_match();
   bool match_word_boundary();
   bool match_within_word();
   bool match_word_start();
   bool match_word_end();
   bool match_buffer_start();
   bool match_buffer_end();
   bool match_backref();
   bool match_long_set();
   bool match_set();
   bool match_jump();
   bool match_alt();
   bool match_rep();
   bool match_combining();
   bool match_soft_buffer_end();
   bool match_restart_continue();
   bool match_long_set_repeat();
   bool match_set_repeat();
   bool match_char_repeat();
   bool match_dot_repeat_fast();
   bool match_dot_repeat_slow();
   bool match_backstep();
   bool match_assert_backref();
   bool match_toggle_case();
#ifdef BOOST_REGEX_RECURSIVE
   bool backtrack_till_match(std::size_t count);
#endif

   // find procs stored in s_find_vtable:
   bool find_restart_any();
   bool find_restart_word();
   bool find_restart_line();
   bool find_restart_buf();
   bool find_restart_lit();

private:
   // final result structure to be filled in:
   match_results<BidiIterator, Allocator>& m_result;
   // temporary result for POSIX matches:
   scoped_ptr<match_results<BidiIterator, Allocator> > m_temp_match;
   // pointer to actual result structure to fill in:
   match_results<BidiIterator, Allocator>* m_presult;
   // start of sequence being searched:
   BidiIterator base;
   // end of sequence being searched:
   BidiIterator last; 
   // current character being examined:
   BidiIterator position;
   // where to restart next search after failed match attempt:
   BidiIterator restart;
   // where the current search started from, acts as base for $` during grep:
   BidiIterator search_base;
   // how far we can go back when matching lookbehind:
   BidiIterator backstop;
   // the expression being examined:
   const basic_regex<char_type, traits>& re;
   // the expression's traits class:
   const ::boost::regex_traits_wrapper<traits>& traits_inst;
   // the next state in the machine being matched:
   const re_syntax_base* pstate;
   // matching flags in use:
   match_flag_type m_match_flags;
   // how many states we have examined so far:
   difference_type state_count;
   // max number of states to examine before giving up:
   difference_type max_state_count;
   // whether we should ignore case or not:
   bool icase;
   // set to true when (position == last), indicates that we may have a partial match:
   bool m_has_partial_match;
   // set to true whenever we get a match:
   bool m_has_found_match;
   // set to true whenever we're inside an independent sub-expression:
   bool m_independent;
   // the current repeat being examined:
   repeater_count<BidiIterator>* next_count;
   // the first repeat being examined (top of linked list):
   repeater_count<BidiIterator> rep_obj;
   // the mask to pass when matching word boundaries:
   typename traits::char_class_type m_word_mask;
   // the bitmask to use when determining whether a match_any matches a newline or not:
   unsigned char match_any_mask;

#ifdef BOOST_REGEX_NON_RECURSIVE
   //
   // additional members for non-recursive version:
   //
   typedef bool (self_type::*unwind_proc_type)(bool);

   void extend_stack();
   bool unwind(bool);
   bool unwind_end(bool);
   bool unwind_paren(bool);
   bool unwind_recursion_stopper(bool);
   bool unwind_assertion(bool);
   bool unwind_alt(bool);
   bool unwind_repeater_counter(bool);
   bool unwind_extra_block(bool);
   bool unwind_greedy_single_repeat(bool);
   bool unwind_slow_dot_repeat(bool);
   bool unwind_fast_dot_repeat(bool);
   bool unwind_char_repeat(bool);
   bool unwind_short_set_repeat(bool);
   bool unwind_long_set_repeat(bool);
   bool unwind_non_greedy_repeat(bool);
   void destroy_single_repeat();
   void push_matched_paren(int index, const sub_match<BidiIterator>& sub);
   void push_recursion_stopper();
   void push_assertion(const re_syntax_base* ps, bool positive);
   void push_alt(const re_syntax_base* ps);
   void push_repeater_count(int i, repeater_count<BidiIterator>** s);
   void push_single_repeat(std::size_t c, const re_repeat* r, BidiIterator last_position, int id);
   void push_non_greedy_repeat(const re_syntax_base* ps);


   // pointer to base of stack:
   saved_state* m_stack_base;
   // pointer to current stack position:
   saved_state* m_backup_state;
   // determines what value to return when unwinding from recursion,
   // allows for mixed recursive/non-recursive algorithm:
   bool m_recursive_result;
   // how many memory blocks have we used up?:
   unsigned used_block_count;
#endif

   // these operations aren't allowed, so are declared private,
   // bodies are provided to keep explicit-instantiation requests happy:
   perl_matcher& operator=(const perl_matcher&)
   {
      return *this;
   }
   perl_matcher(const perl_matcher& that)
      : m_result(that.m_result), re(that.re), traits_inst(that.traits_inst), rep_obj(0) {}
};

#ifdef BOOST_MSVC
#pragma warning(pop)
#endif

} // namespace re_detail

#ifdef BOOST_HAS_ABI_HEADERS
#  include BOOST_ABI_SUFFIX
#endif

} // namespace boost

//
// include the implementation of perl_matcher:
//
#ifdef BOOST_REGEX_RECURSIVE
#include <boost/regex/v4/perl_matcher_recursive.hpp>
#else
#include <boost/regex/v4/perl_matcher_non_recursive.hpp>
#endif
// this one has to be last:
#include <boost/regex/v4/perl_matcher_common.hpp>

#endif