| 1 | // |
| 2 | // Copyright (c) 2009-2011 Artyom Beilis (Tonkikh) |
| 3 | // |
| 4 | // Distributed under the Boost Software License, Version 1.0. |
| 5 | // https://www.boost.org/LICENSE_1_0.txt |
| 6 | |
| 7 | #ifndef BOOST_LOCALE_IMPL_ICONV_CODEPAGE_HPP |
| 8 | #define BOOST_LOCALE_IMPL_ICONV_CODEPAGE_HPP |
| 9 | |
| 10 | #include <boost/locale/encoding.hpp> |
| 11 | #include "boost/locale/util/encoding.hpp" |
| 12 | #include "boost/locale/util/iconv.hpp" |
| 13 | #include <boost/assert.hpp> |
| 14 | #include <cerrno> |
| 15 | #include <string> |
| 16 | |
| 17 | namespace boost { namespace locale { namespace conv { namespace impl { |
| 18 | |
| 19 | class iconverter_base { |
| 20 | public: |
| 21 | bool do_open(const char* to, const char* from, method_type how) |
| 22 | { |
| 23 | cvt_ = iconv_open(tocode: to, fromcode: from); |
| 24 | how_ = how; |
| 25 | return static_cast<bool>(cvt_); |
| 26 | } |
| 27 | |
| 28 | template<typename OutChar, typename InChar> |
| 29 | std::basic_string<OutChar> real_convert(const InChar* ubegin, const InChar* uend) |
| 30 | { |
| 31 | std::basic_string<OutChar> sresult; |
| 32 | |
| 33 | sresult.reserve(uend - ubegin); |
| 34 | |
| 35 | OutChar tmp_buf[64]; |
| 36 | |
| 37 | char* out_start = reinterpret_cast<char*>(tmp_buf); |
| 38 | const char* begin = reinterpret_cast<const char*>(ubegin); |
| 39 | const char* end = reinterpret_cast<const char*>(uend); |
| 40 | |
| 41 | bool is_unshifting = false; |
| 42 | |
| 43 | for(;;) { |
| 44 | size_t in_left = end - begin; |
| 45 | size_t out_left = sizeof(tmp_buf); |
| 46 | char* out_ptr = out_start; |
| 47 | |
| 48 | if(in_left == 0) |
| 49 | is_unshifting = true; |
| 50 | |
| 51 | const auto old_in_left = in_left; |
| 52 | const size_t res = (!is_unshifting) ? conv(inbuf: &begin, inchar_left: &in_left, outbuf: &out_ptr, outchar_left: &out_left) : |
| 53 | conv(inbuf: nullptr, inchar_left: nullptr, outbuf: &out_ptr, outchar_left: &out_left); |
| 54 | |
| 55 | if(res != 0 && res != (size_t)(-1)) { |
| 56 | if(how_ == stop) |
| 57 | throw conversion_error(); |
| 58 | } |
| 59 | |
| 60 | const size_t output_count = (out_ptr - out_start) / sizeof(OutChar); |
| 61 | sresult.append(tmp_buf, output_count); |
| 62 | |
| 63 | if(res == (size_t)(-1)) { |
| 64 | const int err = errno; |
| 65 | BOOST_ASSERT_MSG(err == EILSEQ || err == EINVAL || err == E2BIG, "Invalid error code from IConv" ); |
| 66 | if(err == EILSEQ || err == EINVAL) { |
| 67 | if(how_ == stop) |
| 68 | throw conversion_error(); |
| 69 | |
| 70 | if(begin != end) { |
| 71 | begin += sizeof(InChar); |
| 72 | if(begin >= end) |
| 73 | break; |
| 74 | } else |
| 75 | break; |
| 76 | } else if(err == E2BIG) { |
| 77 | if(in_left != old_in_left || out_ptr != out_start) // Check to avoid infinite loop |
| 78 | continue; |
| 79 | throw std::runtime_error("No progress, IConv is faulty!" ); // LCOV_EXCL_LINE |
| 80 | } else // Invalid error code, shouldn't ever happen or iconv has a bug |
| 81 | throw conversion_error(); // LCOV_EXCL_LINE |
| 82 | } |
| 83 | if(is_unshifting) |
| 84 | break; |
| 85 | } |
| 86 | return sresult; |
| 87 | } |
| 88 | |
| 89 | private: |
| 90 | size_t conv(const char** inbuf, size_t* inchar_left, char** outbuf, size_t* outchar_left) |
| 91 | { |
| 92 | return call_iconv(d: cvt_, in: inbuf, insize: inchar_left, out: outbuf, outsize: outchar_left); |
| 93 | } |
| 94 | |
| 95 | iconv_handle cvt_; |
| 96 | method_type how_; |
| 97 | }; |
| 98 | |
| 99 | template<typename CharType> |
| 100 | class iconv_from_utf final : public detail::utf_decoder<CharType> { |
| 101 | public: |
| 102 | bool open(const std::string& charset, method_type how) |
| 103 | { |
| 104 | return self_.do_open(to: charset.c_str(), from: util::utf_name<CharType>(), how); |
| 105 | } |
| 106 | |
| 107 | std::string convert(const CharType* ubegin, const CharType* uend) override |
| 108 | { |
| 109 | return self_.template real_convert<char>(ubegin, uend); |
| 110 | } |
| 111 | |
| 112 | private: |
| 113 | iconverter_base self_; |
| 114 | }; |
| 115 | |
| 116 | class iconv_between final : public detail::narrow_converter { |
| 117 | public: |
| 118 | bool open(const std::string& to_charset, const std::string& from_charset, method_type how) |
| 119 | { |
| 120 | return self_.do_open(to: to_charset.c_str(), from: from_charset.c_str(), how); |
| 121 | } |
| 122 | std::string convert(const char* begin, const char* end) override |
| 123 | { |
| 124 | return self_.real_convert<char, char>(ubegin: begin, uend: end); |
| 125 | } |
| 126 | |
| 127 | private: |
| 128 | iconverter_base self_; |
| 129 | }; |
| 130 | |
| 131 | template<typename CharType> |
| 132 | class iconv_to_utf final : public detail::utf_encoder<CharType> { |
| 133 | public: |
| 134 | bool open(const std::string& charset, method_type how) |
| 135 | { |
| 136 | return self_.do_open(to: util::utf_name<CharType>(), from: charset.c_str(), how); |
| 137 | } |
| 138 | |
| 139 | std::basic_string<CharType> convert(const char* begin, const char* end) override |
| 140 | { |
| 141 | return self_.template real_convert<CharType>(begin, end); |
| 142 | } |
| 143 | |
| 144 | private: |
| 145 | iconverter_base self_; |
| 146 | }; |
| 147 | |
| 148 | }}}} // namespace boost::locale::conv::impl |
| 149 | |
| 150 | #endif |
| 151 | |