| 1 | // Copyright (c) 2001-2011 Hartmut Kaiser |
| 2 | // |
| 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) |
| 5 | |
| 6 | #if !defined(BOOST_SPIRIT_LEX_CHAR_TOKEN_DEF_MAR_28_2007_0626PM) |
| 7 | #define BOOST_SPIRIT_LEX_CHAR_TOKEN_DEF_MAR_28_2007_0626PM |
| 8 | |
| 9 | #if defined(_MSC_VER) |
| 10 | #pragma once |
| 11 | #endif |
| 12 | |
| 13 | #include <boost/spirit/home/support/common_terminals.hpp> |
| 14 | #include <boost/spirit/home/support/string_traits.hpp> |
| 15 | #include <boost/spirit/home/lex/domain.hpp> |
| 16 | #include <boost/spirit/home/lex/lexer_type.hpp> |
| 17 | #include <boost/spirit/home/lex/meta_compiler.hpp> |
| 18 | |
| 19 | namespace boost { namespace spirit |
| 20 | { |
| 21 | /////////////////////////////////////////////////////////////////////////// |
| 22 | // Enablers |
| 23 | /////////////////////////////////////////////////////////////////////////// |
| 24 | |
| 25 | // enables 'x' |
| 26 | template <> |
| 27 | struct use_terminal<lex::domain, char> |
| 28 | : mpl::true_ {}; |
| 29 | |
| 30 | // enables "x" |
| 31 | template <> |
| 32 | struct use_terminal<lex::domain, char[2]> |
| 33 | : mpl::true_ {}; |
| 34 | |
| 35 | // enables wchar_t |
| 36 | template <> |
| 37 | struct use_terminal<lex::domain, wchar_t> |
| 38 | : mpl::true_ {}; |
| 39 | |
| 40 | // enables L"x" |
| 41 | template <> |
| 42 | struct use_terminal<lex::domain, wchar_t[2]> |
| 43 | : mpl::true_ {}; |
| 44 | |
| 45 | // enables char_('x'), char_("x") |
| 46 | template <typename CharEncoding, typename A0> |
| 47 | struct use_terminal<lex::domain |
| 48 | , terminal_ex< |
| 49 | tag::char_code<tag::char_, CharEncoding> |
| 50 | , fusion::vector1<A0> > > |
| 51 | : mpl::true_ {}; |
| 52 | |
| 53 | // enables char_('x', ID), char_("x", ID) |
| 54 | template <typename CharEncoding, typename A0, typename A1> |
| 55 | struct use_terminal<lex::domain |
| 56 | , terminal_ex< |
| 57 | tag::char_code<tag::char_, CharEncoding> |
| 58 | , fusion::vector2<A0, A1> > > |
| 59 | : mpl::true_ {}; |
| 60 | }} |
| 61 | |
| 62 | namespace boost { namespace spirit { namespace lex |
| 63 | { |
| 64 | // use char_ from standard character set by default |
| 65 | #ifndef BOOST_SPIRIT_NO_PREDEFINED_TERMINALS |
| 66 | using spirit::standard::char_; |
| 67 | #endif |
| 68 | using spirit::standard::char_type; |
| 69 | |
| 70 | /////////////////////////////////////////////////////////////////////////// |
| 71 | // |
| 72 | // char_token_def |
| 73 | // represents a single character token definition |
| 74 | // |
| 75 | /////////////////////////////////////////////////////////////////////////// |
| 76 | template <typename CharEncoding = char_encoding::standard |
| 77 | , typename IdType = std::size_t> |
| 78 | struct char_token_def |
| 79 | : primitive_lexer<char_token_def<CharEncoding, IdType> > |
| 80 | { |
| 81 | typedef typename CharEncoding::char_type char_type; |
| 82 | |
| 83 | char_token_def(char_type ch, IdType const& id) |
| 84 | : ch(ch), id_(id), unique_id_(std::size_t(~0)) |
| 85 | , token_state_(std::size_t(~0)) |
| 86 | {} |
| 87 | |
| 88 | template <typename LexerDef, typename String> |
| 89 | void collect(LexerDef& lexdef, String const& state |
| 90 | , String const& targetstate) const |
| 91 | { |
| 92 | std::size_t state_id = lexdef.add_state(state.c_str()); |
| 93 | |
| 94 | // If the following assertion fires you are probably trying to use |
| 95 | // a single char_token_def instance in more than one lexer state. |
| 96 | // This is not possible. Please create a separate token_def instance |
| 97 | // from the same regular expression for each lexer state it needs |
| 98 | // to be associated with. |
| 99 | BOOST_ASSERT( |
| 100 | (std::size_t(~0) == token_state_ || state_id == token_state_) && |
| 101 | "Can't use single char_token_def with more than one lexer state" ); |
| 102 | |
| 103 | char_type const* target = targetstate.empty() ? 0 : targetstate.c_str(); |
| 104 | if (target) |
| 105 | lexdef.add_state(target); |
| 106 | |
| 107 | token_state_ = state_id; |
| 108 | unique_id_ = lexdef.add_token (state.c_str(), ch, id_, target); |
| 109 | } |
| 110 | |
| 111 | template <typename LexerDef> |
| 112 | void add_actions(LexerDef&) const {} |
| 113 | |
| 114 | IdType id() const { return id_; } |
| 115 | std::size_t unique_id() const { return unique_id_; } |
| 116 | std::size_t state() const { return token_state_; } |
| 117 | |
| 118 | char_type ch; |
| 119 | mutable IdType id_; |
| 120 | mutable std::size_t unique_id_; |
| 121 | mutable std::size_t token_state_; |
| 122 | }; |
| 123 | |
| 124 | /////////////////////////////////////////////////////////////////////////// |
| 125 | // Lexer generators: make_xxx function (objects) |
| 126 | /////////////////////////////////////////////////////////////////////////// |
| 127 | namespace detail |
| 128 | { |
| 129 | template <typename CharEncoding> |
| 130 | struct basic_literal |
| 131 | { |
| 132 | typedef char_token_def<CharEncoding> result_type; |
| 133 | |
| 134 | template <typename Char> |
| 135 | result_type operator()(Char ch, unused_type) const |
| 136 | { |
| 137 | return result_type(ch, ch); |
| 138 | } |
| 139 | |
| 140 | template <typename Char> |
| 141 | result_type operator()(Char const* str, unused_type) const |
| 142 | { |
| 143 | return result_type(str[0], str[0]); |
| 144 | } |
| 145 | }; |
| 146 | } |
| 147 | |
| 148 | // literals: 'x', "x" |
| 149 | template <typename Modifiers> |
| 150 | struct make_primitive<char, Modifiers> |
| 151 | : detail::basic_literal<char_encoding::standard> {}; |
| 152 | |
| 153 | template <typename Modifiers> |
| 154 | struct make_primitive<char const(&)[2], Modifiers> |
| 155 | : detail::basic_literal<char_encoding::standard> {}; |
| 156 | |
| 157 | // literals: L'x', L"x" |
| 158 | template <typename Modifiers> |
| 159 | struct make_primitive<wchar_t, Modifiers> |
| 160 | : detail::basic_literal<char_encoding::standard_wide> {}; |
| 161 | |
| 162 | template <typename Modifiers> |
| 163 | struct make_primitive<wchar_t const(&)[2], Modifiers> |
| 164 | : detail::basic_literal<char_encoding::standard_wide> {}; |
| 165 | |
| 166 | // handle char_('x') |
| 167 | template <typename CharEncoding, typename Modifiers, typename A0> |
| 168 | struct make_primitive< |
| 169 | terminal_ex< |
| 170 | tag::char_code<tag::char_, CharEncoding> |
| 171 | , fusion::vector1<A0> |
| 172 | > |
| 173 | , Modifiers> |
| 174 | { |
| 175 | typedef char_token_def<CharEncoding> result_type; |
| 176 | |
| 177 | template <typename Terminal> |
| 178 | result_type operator()(Terminal const& term, unused_type) const |
| 179 | { |
| 180 | return result_type(fusion::at_c<0>(term.args), fusion::at_c<0>(term.args)); |
| 181 | } |
| 182 | }; |
| 183 | |
| 184 | // handle char_("x") |
| 185 | template <typename CharEncoding, typename Modifiers, typename Char> |
| 186 | struct make_primitive< |
| 187 | terminal_ex< |
| 188 | tag::char_code<tag::char_, CharEncoding> |
| 189 | , fusion::vector1<Char(&)[2]> // single char strings |
| 190 | > |
| 191 | , Modifiers> |
| 192 | { |
| 193 | typedef char_token_def<CharEncoding> result_type; |
| 194 | |
| 195 | template <typename Terminal> |
| 196 | result_type operator()(Terminal const& term, unused_type) const |
| 197 | { |
| 198 | Char ch = fusion::at_c<0>(term.args)[0]; |
| 199 | return result_type(ch, ch); |
| 200 | } |
| 201 | }; |
| 202 | |
| 203 | // handle char_('x', ID) |
| 204 | template <typename CharEncoding, typename Modifiers, typename A0, typename A1> |
| 205 | struct make_primitive< |
| 206 | terminal_ex< |
| 207 | tag::char_code<tag::char_, CharEncoding> |
| 208 | , fusion::vector2<A0, A1> |
| 209 | > |
| 210 | , Modifiers> |
| 211 | { |
| 212 | typedef char_token_def<CharEncoding> result_type; |
| 213 | |
| 214 | template <typename Terminal> |
| 215 | result_type operator()(Terminal const& term, unused_type) const |
| 216 | { |
| 217 | return result_type( |
| 218 | fusion::at_c<0>(term.args), fusion::at_c<1>(term.args)); |
| 219 | } |
| 220 | }; |
| 221 | |
| 222 | // handle char_("x", ID) |
| 223 | template <typename CharEncoding, typename Modifiers, typename Char, typename A1> |
| 224 | struct make_primitive< |
| 225 | terminal_ex< |
| 226 | tag::char_code<tag::char_, CharEncoding> |
| 227 | , fusion::vector2<Char(&)[2], A1> // single char strings |
| 228 | > |
| 229 | , Modifiers> |
| 230 | { |
| 231 | typedef char_token_def<CharEncoding> result_type; |
| 232 | |
| 233 | template <typename Terminal> |
| 234 | result_type operator()(Terminal const& term, unused_type) const |
| 235 | { |
| 236 | return result_type( |
| 237 | fusion::at_c<0>(term.args)[0], fusion::at_c<1>(term.args)); |
| 238 | } |
| 239 | }; |
| 240 | }}} // namespace boost::spirit::lex |
| 241 | |
| 242 | #endif |
| 243 | |