1 /*============================================================================== 2 Copyright (c) 2001-2011 Joel de Guzman 3 Copyright (c) 2010-2011 Bryce Lelbach 4 5 Distributed under the Boost Software License, Version 1.0. (See accompanying 6 file BOOST_LICENSE_1_0.rst or copy at http://www.boost.org/LICENSE_1_0.txt) 7 ==============================================================================*/ 8 9 #if !defined(BOOST_SPIRIT_UTREE_EXAMPLE_UTF8_PARSER_HPP) 10 #define BOOST_SPIRIT_UTREE_EXAMPLE_UTF8_PARSER_HPP 11 12 #include <string> 13 14 #include <boost/cstdint.hpp> 15 16 #include <boost/regex/pending/unicode_iterator.hpp> 17 18 #include <boost/spirit/include/qi.hpp> 19 #include <boost/spirit/include/phoenix_core.hpp> 20 #include <boost/spirit/include/phoenix_container.hpp> 21 #include <boost/spirit/include/phoenix_statement.hpp> 22 #include <boost/spirit/include/phoenix_operator.hpp> 23 #include <boost/spirit/include/support_utree.hpp> 24 25 namespace utf8 26 { 27 28 namespace qi = boost::spirit::qi; 29 namespace px = boost::phoenix; 30 namespace standard = boost::spirit::standard; 31 32 using boost::spirit::utree; 33 using boost::spirit::utf8_symbol_type; 34 using boost::spirit::utf8_string_type; 35 using boost::spirit::binary_string_type; 36 37 typedef boost::uint32_t uchar; 38 39 struct push_string 40 { 41 template <typename, typename> 42 struct result 43 { 44 typedef void type; 45 }; 46 operator ()utf8::push_string47 void operator()(std::string& s, uchar code_point) const 48 { 49 typedef std::back_insert_iterator<std::string> insert_iter; 50 insert_iter out_iter(s); 51 boost::utf8_output_iterator<insert_iter> s_iter(out_iter); 52 *s_iter++ = code_point; 53 } 54 }; 55 56 struct push_escaped_string 57 { 58 template <typename, typename> 59 struct result 60 { 61 typedef void type; 62 }; 63 operator ()utf8::push_escaped_string64 void operator()(std::string& s, uchar c) const 65 { 66 switch (c) { 67 case 'b': 68 s += '\b'; 69 break; 70 case 't': 71 s += '\t'; 72 break; 73 case 'n': 74 s += '\n'; 75 break; 76 case 'f': 77 s += '\f'; 78 break; 79 case 'r': 80 s += '\r'; 81 break; 82 case '"': 83 s += '"'; 84 break; 85 case '\\': 86 s += '\\'; 87 break; 88 } 89 } 90 }; 91 92 template <typename Iterator> 93 struct parser : qi::grammar<Iterator, std::string()> 94 { 95 qi::rule<Iterator, void(std::string&)> 96 escaped; 97 98 qi::rule<Iterator, std::string()> 99 start; 100 101 px::function<push_string> 102 push_str; 103 104 px::function<push_escaped_string> 105 push_esc; 106 parserutf8::parser107 parser() : parser::base_type (start) 108 { 109 using standard::char_; 110 using qi::uint_parser; 111 using qi::_val; 112 using qi::_r1; 113 using qi::_1; 114 115 uint_parser<uchar, 16, 4, 4> hex4; 116 uint_parser<uchar, 16, 8, 8> hex8; 117 118 escaped 119 = '\\' 120 > ( ('u' > hex4) [push_str(_r1, _1)] 121 | ('U' > hex8) [push_str(_r1, _1)] 122 | char_("btnfr\\\"'") [push_esc(_r1, _1)] 123 ); 124 125 start 126 = '"' 127 > *(escaped(_val) | (~char_('"')) [_val += _1]) 128 > '"'; 129 130 escaped.name("escaped_string"); 131 start.name("string"); 132 } 133 }; 134 135 } // utf8 136 137 #endif // BOOST_SPIRIT_UTREE_EXAMPLE_UTF8_PARSER_HPP 138 139