| 1 | #ifndef BOOST_UTF8_CODECVT_FACET_HPP |
| 2 | #define BOOST_UTF8_CODECVT_FACET_HPP |
| 3 | |
| 4 | #include <boost/iostreams/detail/config/wide_streams.hpp> |
| 5 | #ifdef BOOST_IOSTREAMS_NO_WIDE_STREAMS |
| 6 | # error wide streams not supported on this platform |
| 7 | #endif |
| 8 | |
| 9 | // MS compatible compilers support #pragma once |
| 10 | #if defined(_MSC_VER) |
| 11 | # pragma once |
| 12 | #endif |
| 13 | |
| 14 | /////////1/////////2/////////3/////////4/////////5/////////6/////////7/////////8 |
| 15 | // utf8_codecvt_facet.hpp |
| 16 | |
| 17 | // Copyright (c) 2001 Ronald Garcia, Indiana University (garcia@osl.iu.edu) |
| 18 | // Andrew Lumsdaine, Indiana University (lums@osl.iu.edu). |
| 19 | // Distributed under the Boost Software License, Version 1.0. (See accompany- |
| 20 | // ing file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) |
| 21 | |
| 22 | // Note:(Robert Ramey). I have made the following alterations in the original |
| 23 | // code. |
| 24 | // a) Rendered utf8_codecvt<wchar_t, char> with using templates |
| 25 | // b) Move longer functions outside class definition to prevent inlining |
| 26 | // and make code smaller |
| 27 | // c) added on a derived class to permit translation to/from current |
| 28 | // locale to utf8 |
| 29 | |
| 30 | // See http://www.boost.org for updates, documentation, and revision history. |
| 31 | |
| 32 | // archives stored as text - note these ar templated on the basic |
| 33 | // stream templates to accommodate wide (and other?) kind of characters |
| 34 | // |
| 35 | // note the fact that on libraries without wide characters, ostream is |
| 36 | // is not a specialization of basic_ostream which in fact is not defined |
| 37 | // in such cases. So we can't use basic_ostream<OStream::char_type> but rather |
| 38 | // use two template parameters |
| 39 | // |
| 40 | // utf8_codecvt_facet |
| 41 | // This is an implementation of a std::codecvt facet for translating |
| 42 | // from UTF-8 externally to UCS-4. Note that this is not tied to |
| 43 | // any specific types in order to allow customization on platforms |
| 44 | // where wchar_t is not big enough. |
| 45 | // |
| 46 | // NOTES: The current implementation jumps through some unpleasant hoops in |
| 47 | // order to deal with signed character types. As a std::codecvt_base::result, |
| 48 | // it is necessary for the ExternType to be convertible to unsigned char. |
| 49 | // I chose not to tie the extern_type explicitly to char. But if any combination |
| 50 | // of types other than <wchar_t,char_t> is used, then std::codecvt must be |
| 51 | // specialized on those types for this to work. |
| 52 | |
| 53 | #include <locale> |
| 54 | #include <cstddef> // size_t |
| 55 | #include <cwchar> // mbstate_t |
| 56 | #include <boost/integer_traits.hpp> |
| 57 | #include <boost/iostreams/detail/config/wide_streams.hpp> |
| 58 | #include <boost/iostreams/detail/codecvt_helper.hpp> |
| 59 | |
| 60 | // maximum lenght of a multibyte string |
| 61 | #define MB_LENGTH_MAX 8 |
| 62 | |
| 63 | struct utf8_codecvt_facet_wchar_t |
| 64 | : public boost::iostreams::detail::codecvt_helper<wchar_t, char, std::mbstate_t> |
| 65 | { |
| 66 | public: |
| 67 | explicit utf8_codecvt_facet_wchar_t(std::size_t no_locale_manage = 0) |
| 68 | : boost::iostreams::detail::codecvt_helper<wchar_t, char, std::mbstate_t> |
| 69 | (no_locale_manage) |
| 70 | { } |
| 71 | protected: |
| 72 | virtual std::codecvt_base::result do_in( |
| 73 | std::mbstate_t& state, |
| 74 | const char * from, |
| 75 | const char * from_end, |
| 76 | const char * & from_next, |
| 77 | wchar_t * to, |
| 78 | wchar_t * to_end, |
| 79 | wchar_t*& to_next |
| 80 | ) const; |
| 81 | |
| 82 | virtual std::codecvt_base::result do_out( |
| 83 | std::mbstate_t & state, const wchar_t * from, |
| 84 | const wchar_t * from_end, const wchar_t* & from_next, |
| 85 | char * to, char * to_end, char * & to_next |
| 86 | ) const; |
| 87 | |
| 88 | bool invalid_continuing_octet(unsigned char octet_1) const { |
| 89 | return (octet_1 < 0x80|| 0xbf< octet_1); |
| 90 | } |
| 91 | |
| 92 | bool invalid_leading_octet(unsigned char octet_1) const { |
| 93 | return (0x7f < octet_1 && octet_1 < 0xc0) || |
| 94 | (octet_1 > 0xfd); |
| 95 | } |
| 96 | |
| 97 | // continuing octets = octets except for the leading octet |
| 98 | static unsigned int get_cont_octet_count(unsigned char lead_octet) { |
| 99 | return get_octet_count(lead_octet) - 1; |
| 100 | } |
| 101 | |
| 102 | static unsigned int get_octet_count(unsigned char lead_octet); |
| 103 | |
| 104 | // How many "continuing octets" will be needed for this word |
| 105 | // == total octets - 1. |
| 106 | int get_cont_octet_out_count(wchar_t word) const ; |
| 107 | |
| 108 | virtual bool do_always_noconv() const throw() { return false; } |
| 109 | |
| 110 | // UTF-8 isn't really stateful since we rewind on partial conversions |
| 111 | virtual std::codecvt_base::result do_unshift( |
| 112 | std::mbstate_t&, |
| 113 | char * from, |
| 114 | char * /* to */, |
| 115 | char * & next |
| 116 | ) const{ |
| 117 | next = from; |
| 118 | return ok; |
| 119 | } |
| 120 | |
| 121 | virtual int do_encoding() const throw() { |
| 122 | const int variable_byte_external_encoding=0; |
| 123 | return variable_byte_external_encoding; |
| 124 | } |
| 125 | |
| 126 | // How many char objects can I process to get <= max_limit |
| 127 | // wchar_t objects? |
| 128 | virtual int do_length( |
| 129 | BOOST_IOSTREAMS_CODECVT_CV_QUALIFIER std::mbstate_t &, |
| 130 | const char * from, |
| 131 | const char * from_end, |
| 132 | std::size_t max_limit |
| 133 | ) const throw(); |
| 134 | |
| 135 | // Largest possible value do_length(state,from,from_end,1) could return. |
| 136 | virtual int do_max_length() const throw () { |
| 137 | return 6; // largest UTF-8 encoding of a UCS-4 character |
| 138 | } |
| 139 | }; |
| 140 | |
| 141 | #if 0 // not used - incorrect in any case |
| 142 | // Robert Ramey - use the above to make a code converter from multi-byte |
| 143 | // char strings to utf8 encoding |
| 144 | struct utf8_codecvt_facet_char : public utf8_codecvt_facet_wchar_t |
| 145 | { |
| 146 | typedef utf8_codecvt_facet_wchar_t base_class; |
| 147 | public: |
| 148 | explicit utf8_codecvt_facet_char(std::size_t no_locale_manage=0) |
| 149 | : base_class(no_locale_manage) |
| 150 | {} |
| 151 | protected: |
| 152 | virtual std::codecvt_base::result do_in( |
| 153 | std::mbstate_t & state, |
| 154 | const char * from, |
| 155 | const char * from_end, |
| 156 | const char * & from_next, |
| 157 | char * to, |
| 158 | char * to_end, |
| 159 | char * & to_next |
| 160 | ) const; |
| 161 | |
| 162 | virtual std::codecvt_base::result do_out( |
| 163 | std::mbstate_t & state, |
| 164 | const char * from, |
| 165 | const char * from_end, |
| 166 | const char* & from_next, |
| 167 | char * to, |
| 168 | char * to_end, |
| 169 | char * & to_next |
| 170 | ) const; |
| 171 | |
| 172 | // How many char objects can I process to get <= max_limit |
| 173 | // char objects? |
| 174 | virtual int do_length( |
| 175 | const std::mbstate_t&, |
| 176 | const char * from, |
| 177 | const char * from_end, |
| 178 | std::size_t max_limit |
| 179 | ) const; |
| 180 | }; |
| 181 | #endif |
| 182 | |
| 183 | template<class Internal, class External> |
| 184 | struct utf8_codecvt_facet |
| 185 | {}; |
| 186 | |
| 187 | template<> |
| 188 | struct utf8_codecvt_facet<wchar_t, char> |
| 189 | : public utf8_codecvt_facet_wchar_t |
| 190 | {}; |
| 191 | |
| 192 | #if 0 |
| 193 | template<> |
| 194 | struct utf8_codecvt_facet<char, char> |
| 195 | : public utf8_codecvt_facet_char |
| 196 | {}; |
| 197 | #endif |
| 198 | |
| 199 | #endif // BOOST_UTF8_CODECVT_FACET_HPP |
| 200 | |
| 201 | |