1 // Copyright (c) 2001-2011 Hartmut Kaiser
3 // Distributed under the Boost Software License, Version 1.0. (See accompanying
4 // file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
6 #if !defined(BOOST_SPIRIT_LEXERTL_ITERATOR_TOKENISER_MARCH_22_2007_0859AM)
7 #define BOOST_SPIRIT_LEXERTL_ITERATOR_TOKENISER_MARCH_22_2007_0859AM
13 #include <boost/detail/iterator.hpp>
14 #include <boost/spirit/home/support/detail/lexer/state_machine.hpp>
15 #include <boost/spirit/home/support/detail/lexer/consts.hpp>
16 #include <boost/spirit/home/support/detail/lexer/size_t.hpp>
17 #include <boost/spirit/home/support/detail/lexer/char_traits.hpp>
20 namespace boost { namespace spirit { namespace lex { namespace lexertl
22 ///////////////////////////////////////////////////////////////////////////
23 template<typename Iterator>
24 class basic_iterator_tokeniser
27 typedef std::vector<std::size_t> size_t_vector;
28 typedef typename boost::detail::iterator_traits<Iterator>::value_type
31 static std::size_t next (
32 boost::lexer::basic_state_machine<char_type> const& state_machine_
33 , std::size_t &dfa_state_, bool& bol_, Iterator &start_token_
34 , Iterator const& end_, std::size_t& unique_id_)
36 if (start_token_ == end_)
38 unique_id_ = boost::lexer::npos;
43 boost::lexer::detail::internals const& internals_ =
44 state_machine_.data();
47 std::size_t const* lookup_ = &internals_._lookup[dfa_state_]->
49 std::size_t dfa_alphabet_ = internals_._dfa_alphabet[dfa_state_];
50 std::size_t const* dfa_ = &internals_._dfa[dfa_state_]->front ();
52 std::size_t const* ptr_ = dfa_ + dfa_alphabet_;
53 Iterator curr_ = start_token_;
54 bool end_state_ = *ptr_ != 0;
55 std::size_t id_ = *(ptr_ + boost::lexer::id_index);
56 std::size_t uid_ = *(ptr_ + boost::lexer::unique_id_index);
57 std::size_t end_start_state_ = dfa_state_;
59 Iterator end_token_ = start_token_;
63 std::size_t const BOL_state_ = ptr_[boost::lexer::bol_index];
64 std::size_t const EOL_state_ = ptr_[boost::lexer::eol_index];
66 if (BOL_state_ && bol)
68 ptr_ = &dfa_[BOL_state_ * dfa_alphabet_];
70 else if (EOL_state_ && *curr_ == '\n')
72 ptr_ = &dfa_[EOL_state_ * dfa_alphabet_];
77 boost::detail::iterator_traits<Iterator>::value_type
80 boost::lexer::char_traits<value_type>::index_type
84 boost::lexer::char_traits<value_type>::call(*curr_++);
85 bol = (index == '\n') ? true : false;
86 std::size_t const state_ = ptr_[
87 lookup_[static_cast<std::size_t>(index)]];
94 ptr_ = &dfa_[state_ * dfa_alphabet_];
100 id_ = *(ptr_ + boost::lexer::id_index);
101 uid_ = *(ptr_ + boost::lexer::unique_id_index);
102 end_start_state_ = *(ptr_ + boost::lexer::state_index);
108 std::size_t const EOL_state_ = ptr_[boost::lexer::eol_index];
110 if (EOL_state_ && curr_ == end_)
112 ptr_ = &dfa_[EOL_state_ * dfa_alphabet_];
117 id_ = *(ptr_ + boost::lexer::id_index);
118 uid_ = *(ptr_ + boost::lexer::unique_id_index);
119 end_start_state_ = *(ptr_ + boost::lexer::state_index);
126 // return longest match
127 dfa_state_ = end_start_state_;
128 start_token_ = end_token_;
141 bol_ = (*start_token_ == '\n') ? true : false;
142 id_ = boost::lexer::npos;
143 uid_ = boost::lexer::npos;
150 ///////////////////////////////////////////////////////////////////////
151 static std::size_t next (
152 boost::lexer::basic_state_machine<char_type> const& state_machine_
153 , bool& bol_, Iterator &start_token_, Iterator const& end_
154 , std::size_t& unique_id_)
156 if (start_token_ == end_)
158 unique_id_ = boost::lexer::npos;
163 std::size_t const* lookup_ = &state_machine_.data()._lookup[0]->front();
164 std::size_t dfa_alphabet_ = state_machine_.data()._dfa_alphabet[0];
165 std::size_t const* dfa_ = &state_machine_.data()._dfa[0]->front ();
166 std::size_t const* ptr_ = dfa_ + dfa_alphabet_;
168 Iterator curr_ = start_token_;
169 bool end_state_ = *ptr_ != 0;
170 std::size_t id_ = *(ptr_ + boost::lexer::id_index);
171 std::size_t uid_ = *(ptr_ + boost::lexer::unique_id_index);
172 bool end_bol_ = bol_;
173 Iterator end_token_ = start_token_;
175 while (curr_ != end_)
177 std::size_t const BOL_state_ = ptr_[boost::lexer::bol_index];
178 std::size_t const EOL_state_ = ptr_[boost::lexer::eol_index];
180 if (BOL_state_ && bol)
182 ptr_ = &dfa_[BOL_state_ * dfa_alphabet_];
184 else if (EOL_state_ && *curr_ == '\n')
186 ptr_ = &dfa_[EOL_state_ * dfa_alphabet_];
191 boost::detail::iterator_traits<Iterator>::value_type
194 boost::lexer::char_traits<value_type>::index_type
198 boost::lexer::char_traits<value_type>::call(*curr_++);
199 bol = (index == '\n') ? true : false;
200 std::size_t const state_ = ptr_[
201 lookup_[static_cast<std::size_t>(index)]];
208 ptr_ = &dfa_[state_ * dfa_alphabet_];
214 id_ = *(ptr_ + boost::lexer::id_index);
215 uid_ = *(ptr_ + boost::lexer::unique_id_index);
221 std::size_t const EOL_state_ = ptr_[boost::lexer::eol_index];
223 if (EOL_state_ && curr_ == end_)
225 ptr_ = &dfa_[EOL_state_ * dfa_alphabet_];
230 id_ = *(ptr_ + boost::lexer::id_index);
231 uid_ = *(ptr_ + boost::lexer::unique_id_index);
238 // return longest match
240 start_token_ = end_token_;
243 bol_ = *start_token_ == '\n';
244 id_ = boost::lexer::npos;
245 uid_ = boost::lexer::npos;