1 | // Copyright Vladimir Prus 2004. |
2 | // Distributed under the Boost Software License, Version 1.0. |
3 | // (See accompanying file LICENSE_1_0.txt |
4 | // or copy at http://www.boost.org/LICENSE_1_0.txt) |
5 | |
6 | #include <fstream> |
7 | #include <locale.h> |
8 | #include <locale> |
9 | #include <iostream> |
10 | #include <string> |
11 | #include <locale> |
12 | #include <stdexcept> |
13 | |
14 | #include <boost/config.hpp> |
15 | |
16 | #ifndef BOOST_PROGRAM_OPTIONS_SOURCE |
17 | # define BOOST_PROGRAM_OPTIONS_SOURCE |
18 | #endif |
19 | #include <boost/program_options/config.hpp> |
20 | #include <boost/program_options/detail/convert.hpp> |
21 | #include <boost/program_options/detail/utf8_codecvt_facet.hpp> |
22 | #include <boost/throw_exception.hpp> |
23 | |
24 | #include <boost/bind/bind.hpp> |
25 | |
26 | using namespace std; |
27 | using namespace boost::placeholders; |
28 | |
29 | namespace boost { namespace detail { |
30 | |
31 | /* Internal function to actually perform conversion. |
32 | The logic in from_8_bit and to_8_bit function is exactly |
33 | the same, except that one calls 'in' method of codecvt and another |
34 | calls the 'out' method, and that syntax difference makes straightforward |
35 | template implementation impossible. |
36 | |
37 | This functions takes a 'fun' argument, which should have the same |
38 | parameters and return type and the in/out methods. The actual converting |
39 | function will pass functional objects created with boost::bind. |
40 | Experiments show that the performance loss is less than 10%. |
41 | */ |
42 | template<class ToChar, class FromChar, class Fun> |
43 | std::basic_string<ToChar> |
44 | convert(const std::basic_string<FromChar>& s, Fun fun) |
45 | |
46 | { |
47 | std::basic_string<ToChar> result; |
48 | |
49 | std::mbstate_t state = std::mbstate_t(); |
50 | |
51 | const FromChar* from = s.data(); |
52 | const FromChar* from_end = s.data() + s.size(); |
53 | // The interface of cvt is not really iterator-like, and it's |
54 | // not possible the tell the required output size without the conversion. |
55 | // All we can is convert data by pieces. |
56 | while(from != from_end) { |
57 | |
58 | // std::basic_string does not provide non-const pointers to the data, |
59 | // so converting directly into string is not possible. |
60 | ToChar buffer[32]; |
61 | |
62 | ToChar* to_next = buffer; |
63 | // Need variable because boost::bind doesn't work with rvalues. |
64 | ToChar* to_end = buffer + 32; |
65 | std::codecvt_base::result r = |
66 | fun(state, from, from_end, from, buffer, to_end, to_next); |
67 | |
68 | if (r == std::codecvt_base::error) |
69 | boost::throw_exception( |
70 | e: std::logic_error("character conversion failed" )); |
71 | // 'partial' is not an error, it just means not all source |
72 | // characters were converted. However, we need to check that at |
73 | // least one new target character was produced. If not, it means |
74 | // the source data is incomplete, and since we don't have extra |
75 | // data to add to source, it's error. |
76 | if (to_next == buffer) |
77 | boost::throw_exception( |
78 | e: std::logic_error("character conversion failed" )); |
79 | |
80 | // Add converted characters |
81 | result.append(buffer, to_next); |
82 | } |
83 | |
84 | return result; |
85 | } |
86 | }} |
87 | |
88 | namespace boost { |
89 | |
90 | #ifndef BOOST_NO_STD_WSTRING |
91 | BOOST_PROGRAM_OPTIONS_DECL std::wstring |
92 | from_8_bit(const std::string& s, |
93 | const std::codecvt<wchar_t, char, std::mbstate_t>& cvt) |
94 | { |
95 | return detail::convert<wchar_t>( |
96 | s, |
97 | fun: boost::bind(f: &std::codecvt<wchar_t, char, mbstate_t>::in, |
98 | a1: &cvt, |
99 | a2: _1, a3: _2, a4: _3, a5: _4, a6: _5, a7: _6, a8: _7)); |
100 | } |
101 | |
102 | BOOST_PROGRAM_OPTIONS_DECL std::string |
103 | to_8_bit(const std::wstring& s, |
104 | const std::codecvt<wchar_t, char, std::mbstate_t>& cvt) |
105 | { |
106 | return detail::convert<char>( |
107 | s, |
108 | fun: boost::bind(f: &codecvt<wchar_t, char, mbstate_t>::out, |
109 | a1: &cvt, |
110 | a2: _1, a3: _2, a4: _3, a5: _4, a6: _5, a7: _6, a8: _7)); |
111 | } |
112 | |
113 | |
114 | namespace { |
115 | boost::program_options::detail::utf8_codecvt_facet |
116 | utf8_facet; |
117 | } |
118 | |
119 | BOOST_PROGRAM_OPTIONS_DECL std::wstring |
120 | from_utf8(const std::string& s) |
121 | { |
122 | return from_8_bit(s, cvt: utf8_facet); |
123 | } |
124 | |
125 | BOOST_PROGRAM_OPTIONS_DECL std::string |
126 | to_utf8(const std::wstring& s) |
127 | { |
128 | return to_8_bit(s, cvt: utf8_facet); |
129 | } |
130 | |
131 | BOOST_PROGRAM_OPTIONS_DECL std::wstring |
132 | from_local_8_bit(const std::string& s) |
133 | { |
134 | typedef codecvt<wchar_t, char, mbstate_t> facet_type; |
135 | return from_8_bit(s, |
136 | BOOST_USE_FACET(facet_type, locale())); |
137 | } |
138 | |
139 | BOOST_PROGRAM_OPTIONS_DECL std::string |
140 | to_local_8_bit(const std::wstring& s) |
141 | { |
142 | typedef codecvt<wchar_t, char, mbstate_t> facet_type; |
143 | return to_8_bit(s, |
144 | BOOST_USE_FACET(facet_type, locale())); |
145 | } |
146 | #endif |
147 | |
148 | namespace program_options |
149 | { |
150 | BOOST_PROGRAM_OPTIONS_DECL std::string to_internal(const std::string& s) |
151 | { |
152 | return s; |
153 | } |
154 | |
155 | #ifndef BOOST_NO_STD_WSTRING |
156 | BOOST_PROGRAM_OPTIONS_DECL std::string to_internal(const std::wstring& s) |
157 | { |
158 | return to_utf8(s); |
159 | } |
160 | #endif |
161 | } |
162 | |
163 | |
164 | } |
165 | |