aboutsummaryrefslogtreecommitdiff
path: root/src/3rdParty/efsw/Utf.hpp
diff options
context:
space:
mode:
Diffstat (limited to 'src/3rdParty/efsw/Utf.hpp')
-rw-r--r--[-rwxr-xr-x]src/3rdParty/efsw/Utf.hpp1442
1 files changed, 721 insertions, 721 deletions
diff --git a/src/3rdParty/efsw/Utf.hpp b/src/3rdParty/efsw/Utf.hpp
index 6e9ea71..1b042cd 100755..100644
--- a/src/3rdParty/efsw/Utf.hpp
+++ b/src/3rdParty/efsw/Utf.hpp
@@ -1,721 +1,721 @@
1/** NOTE: 1/** NOTE:
2 * This code is based on the Utf implementation from SFML2. License zlib/png ( 2 * This code is based on the Utf implementation from SFML2. License zlib/png (
3 *http://www.sfml-dev.org/license.php ) The class was modified to fit efsw own needs. This is not 3 *http://www.sfml-dev.org/license.php ) The class was modified to fit efsw own needs. This is not
4 *the original implementation from SFML2. 4 *the original implementation from SFML2.
5 * */ 5 * */
6 6
7#ifndef EFSW_UTF_HPP 7#ifndef EFSW_UTF_HPP
8#define EFSW_UTF_HPP 8#define EFSW_UTF_HPP
9 9
10//////////////////////////////////////////////////////////// 10////////////////////////////////////////////////////////////
11// Headers 11// Headers
12//////////////////////////////////////////////////////////// 12////////////////////////////////////////////////////////////
13#include <cstdlib> 13#include <cstdlib>
14#include <efsw/base.hpp> 14#include <efsw/base.hpp>
15#include <locale> 15#include <locale>
16#include <string> 16#include <string>
17 17
18namespace efsw { 18namespace efsw {
19 19
20template <unsigned int N> class Utf; 20template <unsigned int N> class Utf;
21 21
22//////////////////////////////////////////////////////////// 22////////////////////////////////////////////////////////////
23/// \brief Specialization of the Utf template for UTF-8 23/// \brief Specialization of the Utf template for UTF-8
24/// 24///
25//////////////////////////////////////////////////////////// 25////////////////////////////////////////////////////////////
26template <> class Utf<8> { 26template <> class Utf<8> {
27 public: 27 public:
28 //////////////////////////////////////////////////////////// 28 ////////////////////////////////////////////////////////////
29 /// \brief Decode a single UTF-8 character 29 /// \brief Decode a single UTF-8 character
30 /// 30 ///
31 /// Decoding a character means finding its unique 32-bits 31 /// Decoding a character means finding its unique 32-bits
32 /// code (called the codepoint) in the Unicode standard. 32 /// code (called the codepoint) in the Unicode standard.
33 /// 33 ///
34 /// \param begin Iterator pointing to the beginning of the input sequence 34 /// \param begin Iterator pointing to the beginning of the input sequence
35 /// \param end Iterator pointing to the end of the input sequence 35 /// \param end Iterator pointing to the end of the input sequence
36 /// \param output Codepoint of the decoded UTF-8 character 36 /// \param output Codepoint of the decoded UTF-8 character
37 /// \param replacement Replacement character to use in case the UTF-8 sequence is invalid 37 /// \param replacement Replacement character to use in case the UTF-8 sequence is invalid
38 /// 38 ///
39 /// \return Iterator pointing to one past the last read element of the input sequence 39 /// \return Iterator pointing to one past the last read element of the input sequence
40 /// 40 ///
41 //////////////////////////////////////////////////////////// 41 ////////////////////////////////////////////////////////////
42 template <typename In> 42 template <typename In>
43 static In Decode( In begin, In end, Uint32& output, Uint32 replacement = 0 ); 43 static In Decode( In begin, In end, Uint32& output, Uint32 replacement = 0 );
44 44
45 //////////////////////////////////////////////////////////// 45 ////////////////////////////////////////////////////////////
46 /// \brief Encode a single UTF-8 character 46 /// \brief Encode a single UTF-8 character
47 /// 47 ///
48 /// Encoding a character means converting a unique 32-bits 48 /// Encoding a character means converting a unique 32-bits
49 /// code (called the codepoint) in the target encoding, UTF-8. 49 /// code (called the codepoint) in the target encoding, UTF-8.
50 /// 50 ///
51 /// \param input Codepoint to encode as UTF-8 51 /// \param input Codepoint to encode as UTF-8
52 /// \param output Iterator pointing to the beginning of the output sequence 52 /// \param output Iterator pointing to the beginning of the output sequence
53 /// \param replacement Replacement for characters not convertible to UTF-8 (use 0 to skip them) 53 /// \param replacement Replacement for characters not convertible to UTF-8 (use 0 to skip them)
54 /// 54 ///
55 /// \return Iterator to the end of the output sequence which has been written 55 /// \return Iterator to the end of the output sequence which has been written
56 /// 56 ///
57 //////////////////////////////////////////////////////////// 57 ////////////////////////////////////////////////////////////
58 template <typename Out> static Out Encode( Uint32 input, Out output, Uint8 replacement = 0 ); 58 template <typename Out> static Out Encode( Uint32 input, Out output, Uint8 replacement = 0 );
59 59
60 //////////////////////////////////////////////////////////// 60 ////////////////////////////////////////////////////////////
61 /// \brief Advance to the next UTF-8 character 61 /// \brief Advance to the next UTF-8 character
62 /// 62 ///
63 /// This function is necessary for multi-elements encodings, as 63 /// This function is necessary for multi-elements encodings, as
64 /// a single character may use more than 1 storage element. 64 /// a single character may use more than 1 storage element.
65 /// 65 ///
66 /// \param begin Iterator pointing to the beginning of the input sequence 66 /// \param begin Iterator pointing to the beginning of the input sequence
67 /// \param end Iterator pointing to the end of the input sequence 67 /// \param end Iterator pointing to the end of the input sequence
68 /// 68 ///
69 /// \return Iterator pointing to one past the last read element of the input sequence 69 /// \return Iterator pointing to one past the last read element of the input sequence
70 /// 70 ///
71 //////////////////////////////////////////////////////////// 71 ////////////////////////////////////////////////////////////
72 template <typename In> static In Next( In begin, In end ); 72 template <typename In> static In Next( In begin, In end );
73 73
74 //////////////////////////////////////////////////////////// 74 ////////////////////////////////////////////////////////////
75 /// \brief Count the number of characters of a UTF-8 sequence 75 /// \brief Count the number of characters of a UTF-8 sequence
76 /// 76 ///
77 /// This function is necessary for multi-elements encodings, as 77 /// This function is necessary for multi-elements encodings, as
78 /// a single character may use more than 1 storage element, thus the 78 /// a single character may use more than 1 storage element, thus the
79 /// total size can be different from (begin - end). 79 /// total size can be different from (begin - end).
80 /// 80 ///
81 /// \param begin Iterator pointing to the beginning of the input sequence 81 /// \param begin Iterator pointing to the beginning of the input sequence
82 /// \param end Iterator pointing to the end of the input sequence 82 /// \param end Iterator pointing to the end of the input sequence
83 /// 83 ///
84 /// \return Iterator pointing to one past the last read element of the input sequence 84 /// \return Iterator pointing to one past the last read element of the input sequence
85 /// 85 ///
86 //////////////////////////////////////////////////////////// 86 ////////////////////////////////////////////////////////////
87 template <typename In> static std::size_t Count( In begin, In end ); 87 template <typename In> static std::size_t Count( In begin, In end );
88 88
89 //////////////////////////////////////////////////////////// 89 ////////////////////////////////////////////////////////////
90 /// \brief Convert an ANSI characters range to UTF-8 90 /// \brief Convert an ANSI characters range to UTF-8
91 /// 91 ///
92 /// The current global locale will be used by default, unless you 92 /// The current global locale will be used by default, unless you
93 /// pass a custom one in the \a locale parameter. 93 /// pass a custom one in the \a locale parameter.
94 /// 94 ///
95 /// \param begin Iterator pointing to the beginning of the input sequence 95 /// \param begin Iterator pointing to the beginning of the input sequence
96 /// \param end Iterator pointing to the end of the input sequence 96 /// \param end Iterator pointing to the end of the input sequence
97 /// \param output Iterator pointing to the beginning of the output sequence 97 /// \param output Iterator pointing to the beginning of the output sequence
98 /// \param locale Locale to use for conversion 98 /// \param locale Locale to use for conversion
99 /// 99 ///
100 /// \return Iterator to the end of the output sequence which has been written 100 /// \return Iterator to the end of the output sequence which has been written
101 /// 101 ///
102 //////////////////////////////////////////////////////////// 102 ////////////////////////////////////////////////////////////
103 template <typename In, typename Out> 103 template <typename In, typename Out>
104 static Out FromAnsi( In begin, In end, Out output, const std::locale& locale = std::locale() ); 104 static Out FromAnsi( In begin, In end, Out output, const std::locale& locale = std::locale() );
105 105
106 //////////////////////////////////////////////////////////// 106 ////////////////////////////////////////////////////////////
107 /// \brief Convert a wide characters range to UTF-8 107 /// \brief Convert a wide characters range to UTF-8
108 /// 108 ///
109 /// \param begin Iterator pointing to the beginning of the input sequence 109 /// \param begin Iterator pointing to the beginning of the input sequence
110 /// \param end Iterator pointing to the end of the input sequence 110 /// \param end Iterator pointing to the end of the input sequence
111 /// \param output Iterator pointing to the beginning of the output sequence 111 /// \param output Iterator pointing to the beginning of the output sequence
112 /// 112 ///
113 /// \return Iterator to the end of the output sequence which has been written 113 /// \return Iterator to the end of the output sequence which has been written
114 /// 114 ///
115 //////////////////////////////////////////////////////////// 115 ////////////////////////////////////////////////////////////
116 template <typename In, typename Out> static Out FromWide( In begin, In end, Out output ); 116 template <typename In, typename Out> static Out FromWide( In begin, In end, Out output );
117 117
118 //////////////////////////////////////////////////////////// 118 ////////////////////////////////////////////////////////////
119 /// \brief Convert a latin-1 (ISO-5589-1) characters range to UTF-8 119 /// \brief Convert a latin-1 (ISO-5589-1) characters range to UTF-8
120 /// 120 ///
121 /// \param begin Iterator pointing to the beginning of the input sequence 121 /// \param begin Iterator pointing to the beginning of the input sequence
122 /// \param end Iterator pointing to the end of the input sequence 122 /// \param end Iterator pointing to the end of the input sequence
123 /// \param output Iterator pointing to the beginning of the output sequence 123 /// \param output Iterator pointing to the beginning of the output sequence
124 /// \param locale Locale to use for conversion 124 /// \param locale Locale to use for conversion
125 /// 125 ///
126 /// \return Iterator to the end of the output sequence which has been written 126 /// \return Iterator to the end of the output sequence which has been written
127 /// 127 ///
128 //////////////////////////////////////////////////////////// 128 ////////////////////////////////////////////////////////////
129 template <typename In, typename Out> static Out FromLatin1( In begin, In end, Out output ); 129 template <typename In, typename Out> static Out FromLatin1( In begin, In end, Out output );
130 130
131 //////////////////////////////////////////////////////////// 131 ////////////////////////////////////////////////////////////
132 /// \brief Convert an UTF-8 characters range to ANSI characters 132 /// \brief Convert an UTF-8 characters range to ANSI characters
133 /// 133 ///
134 /// The current global locale will be used by default, unless you 134 /// The current global locale will be used by default, unless you
135 /// pass a custom one in the \a locale parameter. 135 /// pass a custom one in the \a locale parameter.
136 /// 136 ///
137 /// \param begin Iterator pointing to the beginning of the input sequence 137 /// \param begin Iterator pointing to the beginning of the input sequence
138 /// \param end Iterator pointing to the end of the input sequence 138 /// \param end Iterator pointing to the end of the input sequence
139 /// \param output Iterator pointing to the beginning of the output sequence 139 /// \param output Iterator pointing to the beginning of the output sequence
140 /// \param replacement Replacement for characters not convertible to ANSI (use 0 to skip them) 140 /// \param replacement Replacement for characters not convertible to ANSI (use 0 to skip them)
141 /// \param locale Locale to use for conversion 141 /// \param locale Locale to use for conversion
142 /// 142 ///
143 /// \return Iterator to the end of the output sequence which has been written 143 /// \return Iterator to the end of the output sequence which has been written
144 /// 144 ///
145 //////////////////////////////////////////////////////////// 145 ////////////////////////////////////////////////////////////
146 template <typename In, typename Out> 146 template <typename In, typename Out>
147 static Out ToAnsi( In begin, In end, Out output, char replacement = 0, 147 static Out ToAnsi( In begin, In end, Out output, char replacement = 0,
148 const std::locale& locale = std::locale() ); 148 const std::locale& locale = std::locale() );
149 149
150#ifndef EFSW_NO_WIDECHAR 150#ifndef EFSW_NO_WIDECHAR
151 //////////////////////////////////////////////////////////// 151 ////////////////////////////////////////////////////////////
152 /// \brief Convert an UTF-8 characters range to wide characters 152 /// \brief Convert an UTF-8 characters range to wide characters
153 /// 153 ///
154 /// \param begin Iterator pointing to the beginning of the input sequence 154 /// \param begin Iterator pointing to the beginning of the input sequence
155 /// \param end Iterator pointing to the end of the input sequence 155 /// \param end Iterator pointing to the end of the input sequence
156 /// \param output Iterator pointing to the beginning of the output sequence 156 /// \param output Iterator pointing to the beginning of the output sequence
157 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them) 157 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them)
158 /// 158 ///
159 /// \return Iterator to the end of the output sequence which has been written 159 /// \return Iterator to the end of the output sequence which has been written
160 /// 160 ///
161 //////////////////////////////////////////////////////////// 161 ////////////////////////////////////////////////////////////
162 template <typename In, typename Out> 162 template <typename In, typename Out>
163 static Out ToWide( In begin, In end, Out output, wchar_t replacement = 0 ); 163 static Out ToWide( In begin, In end, Out output, wchar_t replacement = 0 );
164#endif 164#endif
165 165
166 //////////////////////////////////////////////////////////// 166 ////////////////////////////////////////////////////////////
167 /// \brief Convert an UTF-8 characters range to latin-1 (ISO-5589-1) characters 167 /// \brief Convert an UTF-8 characters range to latin-1 (ISO-5589-1) characters
168 /// 168 ///
169 /// \param begin Iterator pointing to the beginning of the input sequence 169 /// \param begin Iterator pointing to the beginning of the input sequence
170 /// \param end Iterator pointing to the end of the input sequence 170 /// \param end Iterator pointing to the end of the input sequence
171 /// \param output Iterator pointing to the beginning of the output sequence 171 /// \param output Iterator pointing to the beginning of the output sequence
172 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them) 172 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them)
173 /// 173 ///
174 /// \return Iterator to the end of the output sequence which has been written 174 /// \return Iterator to the end of the output sequence which has been written
175 /// 175 ///
176 //////////////////////////////////////////////////////////// 176 ////////////////////////////////////////////////////////////
177 template <typename In, typename Out> 177 template <typename In, typename Out>
178 static Out ToLatin1( In begin, In end, Out output, char replacement = 0 ); 178 static Out ToLatin1( In begin, In end, Out output, char replacement = 0 );
179 179
180 //////////////////////////////////////////////////////////// 180 ////////////////////////////////////////////////////////////
181 /// \brief Convert a UTF-8 characters range to UTF-8 181 /// \brief Convert a UTF-8 characters range to UTF-8
182 /// 182 ///
183 /// This functions does nothing more than a direct copy; 183 /// This functions does nothing more than a direct copy;
184 /// it is defined only to provide the same interface as other 184 /// it is defined only to provide the same interface as other
185 /// specializations of the efsw::Utf<> template, and allow 185 /// specializations of the efsw::Utf<> template, and allow
186 /// generic code to be written on top of it. 186 /// generic code to be written on top of it.
187 /// 187 ///
188 /// \param begin Iterator pointing to the beginning of the input sequence 188 /// \param begin Iterator pointing to the beginning of the input sequence
189 /// \param end Iterator pointing to the end of the input sequence 189 /// \param end Iterator pointing to the end of the input sequence
190 /// \param output Iterator pointing to the beginning of the output sequence 190 /// \param output Iterator pointing to the beginning of the output sequence
191 /// 191 ///
192 /// \return Iterator to the end of the output sequence which has been written 192 /// \return Iterator to the end of the output sequence which has been written
193 /// 193 ///
194 //////////////////////////////////////////////////////////// 194 ////////////////////////////////////////////////////////////
195 template <typename In, typename Out> static Out toUtf8( In begin, In end, Out output ); 195 template <typename In, typename Out> static Out toUtf8( In begin, In end, Out output );
196 196
197 //////////////////////////////////////////////////////////// 197 ////////////////////////////////////////////////////////////
198 /// \brief Convert a UTF-8 characters range to UTF-16 198 /// \brief Convert a UTF-8 characters range to UTF-16
199 /// 199 ///
200 /// \param begin Iterator pointing to the beginning of the input sequence 200 /// \param begin Iterator pointing to the beginning of the input sequence
201 /// \param end Iterator pointing to the end of the input sequence 201 /// \param end Iterator pointing to the end of the input sequence
202 /// \param output Iterator pointing to the beginning of the output sequence 202 /// \param output Iterator pointing to the beginning of the output sequence
203 /// 203 ///
204 /// \return Iterator to the end of the output sequence which has been written 204 /// \return Iterator to the end of the output sequence which has been written
205 /// 205 ///
206 //////////////////////////////////////////////////////////// 206 ////////////////////////////////////////////////////////////
207 template <typename In, typename Out> static Out ToUtf16( In begin, In end, Out output ); 207 template <typename In, typename Out> static Out ToUtf16( In begin, In end, Out output );
208 208
209 //////////////////////////////////////////////////////////// 209 ////////////////////////////////////////////////////////////
210 /// \brief Convert a UTF-8 characters range to UTF-32 210 /// \brief Convert a UTF-8 characters range to UTF-32
211 /// 211 ///
212 /// \param begin Iterator pointing to the beginning of the input sequence 212 /// \param begin Iterator pointing to the beginning of the input sequence
213 /// \param end Iterator pointing to the end of the input sequence 213 /// \param end Iterator pointing to the end of the input sequence
214 /// \param output Iterator pointing to the beginning of the output sequence 214 /// \param output Iterator pointing to the beginning of the output sequence
215 /// 215 ///
216 /// \return Iterator to the end of the output sequence which has been written 216 /// \return Iterator to the end of the output sequence which has been written
217 /// 217 ///
218 //////////////////////////////////////////////////////////// 218 ////////////////////////////////////////////////////////////
219 template <typename In, typename Out> static Out ToUtf32( In begin, In end, Out output ); 219 template <typename In, typename Out> static Out ToUtf32( In begin, In end, Out output );
220}; 220};
221 221
222//////////////////////////////////////////////////////////// 222////////////////////////////////////////////////////////////
223/// \brief Specialization of the Utf template for UTF-16 223/// \brief Specialization of the Utf template for UTF-16
224/// 224///
225//////////////////////////////////////////////////////////// 225////////////////////////////////////////////////////////////
226template <> class Utf<16> { 226template <> class Utf<16> {
227 public: 227 public:
228 //////////////////////////////////////////////////////////// 228 ////////////////////////////////////////////////////////////
229 /// \brief Decode a single UTF-16 character 229 /// \brief Decode a single UTF-16 character
230 /// 230 ///
231 /// Decoding a character means finding its unique 32-bits 231 /// Decoding a character means finding its unique 32-bits
232 /// code (called the codepoint) in the Unicode standard. 232 /// code (called the codepoint) in the Unicode standard.
233 /// 233 ///
234 /// \param begin Iterator pointing to the beginning of the input sequence 234 /// \param begin Iterator pointing to the beginning of the input sequence
235 /// \param end Iterator pointing to the end of the input sequence 235 /// \param end Iterator pointing to the end of the input sequence
236 /// \param output Codepoint of the decoded UTF-16 character 236 /// \param output Codepoint of the decoded UTF-16 character
237 /// \param replacement Replacement character to use in case the UTF-8 sequence is invalid 237 /// \param replacement Replacement character to use in case the UTF-8 sequence is invalid
238 /// 238 ///
239 /// \return Iterator pointing to one past the last read element of the input sequence 239 /// \return Iterator pointing to one past the last read element of the input sequence
240 /// 240 ///
241 //////////////////////////////////////////////////////////// 241 ////////////////////////////////////////////////////////////
242 template <typename In> 242 template <typename In>
243 static In Decode( In begin, In end, Uint32& output, Uint32 replacement = 0 ); 243 static In Decode( In begin, In end, Uint32& output, Uint32 replacement = 0 );
244 244
245 //////////////////////////////////////////////////////////// 245 ////////////////////////////////////////////////////////////
246 /// \brief Encode a single UTF-16 character 246 /// \brief Encode a single UTF-16 character
247 /// 247 ///
248 /// Encoding a character means converting a unique 32-bits 248 /// Encoding a character means converting a unique 32-bits
249 /// code (called the codepoint) in the target encoding, UTF-16. 249 /// code (called the codepoint) in the target encoding, UTF-16.
250 /// 250 ///
251 /// \param input Codepoint to encode as UTF-16 251 /// \param input Codepoint to encode as UTF-16
252 /// \param output Iterator pointing to the beginning of the output sequence 252 /// \param output Iterator pointing to the beginning of the output sequence
253 /// \param replacement Replacement for characters not convertible to UTF-16 (use 0 to skip them) 253 /// \param replacement Replacement for characters not convertible to UTF-16 (use 0 to skip them)
254 /// 254 ///
255 /// \return Iterator to the end of the output sequence which has been written 255 /// \return Iterator to the end of the output sequence which has been written
256 /// 256 ///
257 //////////////////////////////////////////////////////////// 257 ////////////////////////////////////////////////////////////
258 template <typename Out> static Out Encode( Uint32 input, Out output, Uint16 replacement = 0 ); 258 template <typename Out> static Out Encode( Uint32 input, Out output, Uint16 replacement = 0 );
259 259
260 //////////////////////////////////////////////////////////// 260 ////////////////////////////////////////////////////////////
261 /// \brief Advance to the next UTF-16 character 261 /// \brief Advance to the next UTF-16 character
262 /// 262 ///
263 /// This function is necessary for multi-elements encodings, as 263 /// This function is necessary for multi-elements encodings, as
264 /// a single character may use more than 1 storage element. 264 /// a single character may use more than 1 storage element.
265 /// 265 ///
266 /// \param begin Iterator pointing to the beginning of the input sequence 266 /// \param begin Iterator pointing to the beginning of the input sequence
267 /// \param end Iterator pointing to the end of the input sequence 267 /// \param end Iterator pointing to the end of the input sequence
268 /// 268 ///
269 /// \return Iterator pointing to one past the last read element of the input sequence 269 /// \return Iterator pointing to one past the last read element of the input sequence
270 /// 270 ///
271 //////////////////////////////////////////////////////////// 271 ////////////////////////////////////////////////////////////
272 template <typename In> static In Next( In begin, In end ); 272 template <typename In> static In Next( In begin, In end );
273 273
274 //////////////////////////////////////////////////////////// 274 ////////////////////////////////////////////////////////////
275 /// \brief Count the number of characters of a UTF-16 sequence 275 /// \brief Count the number of characters of a UTF-16 sequence
276 /// 276 ///
277 /// This function is necessary for multi-elements encodings, as 277 /// This function is necessary for multi-elements encodings, as
278 /// a single character may use more than 1 storage element, thus the 278 /// a single character may use more than 1 storage element, thus the
279 /// total size can be different from (begin - end). 279 /// total size can be different from (begin - end).
280 /// 280 ///
281 /// \param begin Iterator pointing to the beginning of the input sequence 281 /// \param begin Iterator pointing to the beginning of the input sequence
282 /// \param end Iterator pointing to the end of the input sequence 282 /// \param end Iterator pointing to the end of the input sequence
283 /// 283 ///
284 /// \return Iterator pointing to one past the last read element of the input sequence 284 /// \return Iterator pointing to one past the last read element of the input sequence
285 /// 285 ///
286 //////////////////////////////////////////////////////////// 286 ////////////////////////////////////////////////////////////
287 template <typename In> static std::size_t Count( In begin, In end ); 287 template <typename In> static std::size_t Count( In begin, In end );
288 288
289 //////////////////////////////////////////////////////////// 289 ////////////////////////////////////////////////////////////
290 /// \brief Convert an ANSI characters range to UTF-16 290 /// \brief Convert an ANSI characters range to UTF-16
291 /// 291 ///
292 /// The current global locale will be used by default, unless you 292 /// The current global locale will be used by default, unless you
293 /// pass a custom one in the \a locale parameter. 293 /// pass a custom one in the \a locale parameter.
294 /// 294 ///
295 /// \param begin Iterator pointing to the beginning of the input sequence 295 /// \param begin Iterator pointing to the beginning of the input sequence
296 /// \param end Iterator pointing to the end of the input sequence 296 /// \param end Iterator pointing to the end of the input sequence
297 /// \param output Iterator pointing to the beginning of the output sequence 297 /// \param output Iterator pointing to the beginning of the output sequence
298 /// \param locale Locale to use for conversion 298 /// \param locale Locale to use for conversion
299 /// 299 ///
300 /// \return Iterator to the end of the output sequence which has been written 300 /// \return Iterator to the end of the output sequence which has been written
301 /// 301 ///
302 //////////////////////////////////////////////////////////// 302 ////////////////////////////////////////////////////////////
303 template <typename In, typename Out> 303 template <typename In, typename Out>
304 static Out FromAnsi( In begin, In end, Out output, const std::locale& locale = std::locale() ); 304 static Out FromAnsi( In begin, In end, Out output, const std::locale& locale = std::locale() );
305 305
306 //////////////////////////////////////////////////////////// 306 ////////////////////////////////////////////////////////////
307 /// \brief Convert a wide characters range to UTF-16 307 /// \brief Convert a wide characters range to UTF-16
308 /// 308 ///
309 /// \param begin Iterator pointing to the beginning of the input sequence 309 /// \param begin Iterator pointing to the beginning of the input sequence
310 /// \param end Iterator pointing to the end of the input sequence 310 /// \param end Iterator pointing to the end of the input sequence
311 /// \param output Iterator pointing to the beginning of the output sequence 311 /// \param output Iterator pointing to the beginning of the output sequence
312 /// 312 ///
313 /// \return Iterator to the end of the output sequence which has been written 313 /// \return Iterator to the end of the output sequence which has been written
314 /// 314 ///
315 //////////////////////////////////////////////////////////// 315 ////////////////////////////////////////////////////////////
316 template <typename In, typename Out> static Out FromWide( In begin, In end, Out output ); 316 template <typename In, typename Out> static Out FromWide( In begin, In end, Out output );
317 317
318 //////////////////////////////////////////////////////////// 318 ////////////////////////////////////////////////////////////
319 /// \brief Convert a latin-1 (ISO-5589-1) characters range to UTF-16 319 /// \brief Convert a latin-1 (ISO-5589-1) characters range to UTF-16
320 /// 320 ///
321 /// \param begin Iterator pointing to the beginning of the input sequence 321 /// \param begin Iterator pointing to the beginning of the input sequence
322 /// \param end Iterator pointing to the end of the input sequence 322 /// \param end Iterator pointing to the end of the input sequence
323 /// \param output Iterator pointing to the beginning of the output sequence 323 /// \param output Iterator pointing to the beginning of the output sequence
324 /// \param locale Locale to use for conversion 324 /// \param locale Locale to use for conversion
325 /// 325 ///
326 /// \return Iterator to the end of the output sequence which has been written 326 /// \return Iterator to the end of the output sequence which has been written
327 /// 327 ///
328 //////////////////////////////////////////////////////////// 328 ////////////////////////////////////////////////////////////
329 template <typename In, typename Out> static Out FromLatin1( In begin, In end, Out output ); 329 template <typename In, typename Out> static Out FromLatin1( In begin, In end, Out output );
330 330
331 //////////////////////////////////////////////////////////// 331 ////////////////////////////////////////////////////////////
332 /// \brief Convert an UTF-16 characters range to ANSI characters 332 /// \brief Convert an UTF-16 characters range to ANSI characters
333 /// 333 ///
334 /// The current global locale will be used by default, unless you 334 /// The current global locale will be used by default, unless you
335 /// pass a custom one in the \a locale parameter. 335 /// pass a custom one in the \a locale parameter.
336 /// 336 ///
337 /// \param begin Iterator pointing to the beginning of the input sequence 337 /// \param begin Iterator pointing to the beginning of the input sequence
338 /// \param end Iterator pointing to the end of the input sequence 338 /// \param end Iterator pointing to the end of the input sequence
339 /// \param output Iterator pointing to the beginning of the output sequence 339 /// \param output Iterator pointing to the beginning of the output sequence
340 /// \param replacement Replacement for characters not convertible to ANSI (use 0 to skip them) 340 /// \param replacement Replacement for characters not convertible to ANSI (use 0 to skip them)
341 /// \param locale Locale to use for conversion 341 /// \param locale Locale to use for conversion
342 /// 342 ///
343 /// \return Iterator to the end of the output sequence which has been written 343 /// \return Iterator to the end of the output sequence which has been written
344 /// 344 ///
345 //////////////////////////////////////////////////////////// 345 ////////////////////////////////////////////////////////////
346 template <typename In, typename Out> 346 template <typename In, typename Out>
347 static Out ToAnsi( In begin, In end, Out output, char replacement = 0, 347 static Out ToAnsi( In begin, In end, Out output, char replacement = 0,
348 const std::locale& locale = std::locale() ); 348 const std::locale& locale = std::locale() );
349 349
350#ifndef EFSW_NO_WIDECHAR 350#ifndef EFSW_NO_WIDECHAR
351 //////////////////////////////////////////////////////////// 351 ////////////////////////////////////////////////////////////
352 /// \brief Convert an UTF-16 characters range to wide characters 352 /// \brief Convert an UTF-16 characters range to wide characters
353 /// 353 ///
354 /// \param begin Iterator pointing to the beginning of the input sequence 354 /// \param begin Iterator pointing to the beginning of the input sequence
355 /// \param end Iterator pointing to the end of the input sequence 355 /// \param end Iterator pointing to the end of the input sequence
356 /// \param output Iterator pointing to the beginning of the output sequence 356 /// \param output Iterator pointing to the beginning of the output sequence
357 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them) 357 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them)
358 /// 358 ///
359 /// \return Iterator to the end of the output sequence which has been written 359 /// \return Iterator to the end of the output sequence which has been written
360 /// 360 ///
361 //////////////////////////////////////////////////////////// 361 ////////////////////////////////////////////////////////////
362 template <typename In, typename Out> 362 template <typename In, typename Out>
363 static Out ToWide( In begin, In end, Out output, wchar_t replacement = 0 ); 363 static Out ToWide( In begin, In end, Out output, wchar_t replacement = 0 );
364#endif 364#endif
365 365
366 //////////////////////////////////////////////////////////// 366 ////////////////////////////////////////////////////////////
367 /// \brief Convert an UTF-16 characters range to latin-1 (ISO-5589-1) characters 367 /// \brief Convert an UTF-16 characters range to latin-1 (ISO-5589-1) characters
368 /// 368 ///
369 /// \param begin Iterator pointing to the beginning of the input sequence 369 /// \param begin Iterator pointing to the beginning of the input sequence
370 /// \param end Iterator pointing to the end of the input sequence 370 /// \param end Iterator pointing to the end of the input sequence
371 /// \param output Iterator pointing to the beginning of the output sequence 371 /// \param output Iterator pointing to the beginning of the output sequence
372 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them) 372 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them)
373 /// 373 ///
374 /// \return Iterator to the end of the output sequence which has been written 374 /// \return Iterator to the end of the output sequence which has been written
375 /// 375 ///
376 //////////////////////////////////////////////////////////// 376 ////////////////////////////////////////////////////////////
377 template <typename In, typename Out> 377 template <typename In, typename Out>
378 static Out ToLatin1( In begin, In end, Out output, char replacement = 0 ); 378 static Out ToLatin1( In begin, In end, Out output, char replacement = 0 );
379 379
380 //////////////////////////////////////////////////////////// 380 ////////////////////////////////////////////////////////////
381 /// \brief Convert a UTF-16 characters range to UTF-8 381 /// \brief Convert a UTF-16 characters range to UTF-8
382 /// 382 ///
383 /// \param begin Iterator pointing to the beginning of the input sequence 383 /// \param begin Iterator pointing to the beginning of the input sequence
384 /// \param end Iterator pointing to the end of the input sequence 384 /// \param end Iterator pointing to the end of the input sequence
385 /// \param output Iterator pointing to the beginning of the output sequence 385 /// \param output Iterator pointing to the beginning of the output sequence
386 /// 386 ///
387 /// \return Iterator to the end of the output sequence which has been written 387 /// \return Iterator to the end of the output sequence which has been written
388 /// 388 ///
389 //////////////////////////////////////////////////////////// 389 ////////////////////////////////////////////////////////////
390 template <typename In, typename Out> static Out toUtf8( In begin, In end, Out output ); 390 template <typename In, typename Out> static Out toUtf8( In begin, In end, Out output );
391 391
392 //////////////////////////////////////////////////////////// 392 ////////////////////////////////////////////////////////////
393 /// \brief Convert a UTF-16 characters range to UTF-16 393 /// \brief Convert a UTF-16 characters range to UTF-16
394 /// 394 ///
395 /// This functions does nothing more than a direct copy; 395 /// This functions does nothing more than a direct copy;
396 /// it is defined only to provide the same interface as other 396 /// it is defined only to provide the same interface as other
397 /// specializations of the efsw::Utf<> template, and allow 397 /// specializations of the efsw::Utf<> template, and allow
398 /// generic code to be written on top of it. 398 /// generic code to be written on top of it.
399 /// 399 ///
400 /// \param begin Iterator pointing to the beginning of the input sequence 400 /// \param begin Iterator pointing to the beginning of the input sequence
401 /// \param end Iterator pointing to the end of the input sequence 401 /// \param end Iterator pointing to the end of the input sequence
402 /// \param output Iterator pointing to the beginning of the output sequence 402 /// \param output Iterator pointing to the beginning of the output sequence
403 /// 403 ///
404 /// \return Iterator to the end of the output sequence which has been written 404 /// \return Iterator to the end of the output sequence which has been written
405 /// 405 ///
406 //////////////////////////////////////////////////////////// 406 ////////////////////////////////////////////////////////////
407 template <typename In, typename Out> static Out ToUtf16( In begin, In end, Out output ); 407 template <typename In, typename Out> static Out ToUtf16( In begin, In end, Out output );
408 408
409 //////////////////////////////////////////////////////////// 409 ////////////////////////////////////////////////////////////
410 /// \brief Convert a UTF-16 characters range to UTF-32 410 /// \brief Convert a UTF-16 characters range to UTF-32
411 /// 411 ///
412 /// \param begin Iterator pointing to the beginning of the input sequence 412 /// \param begin Iterator pointing to the beginning of the input sequence
413 /// \param end Iterator pointing to the end of the input sequence 413 /// \param end Iterator pointing to the end of the input sequence
414 /// \param output Iterator pointing to the beginning of the output sequence 414 /// \param output Iterator pointing to the beginning of the output sequence
415 /// 415 ///
416 /// \return Iterator to the end of the output sequence which has been written 416 /// \return Iterator to the end of the output sequence which has been written
417 /// 417 ///
418 //////////////////////////////////////////////////////////// 418 ////////////////////////////////////////////////////////////
419 template <typename In, typename Out> static Out ToUtf32( In begin, In end, Out output ); 419 template <typename In, typename Out> static Out ToUtf32( In begin, In end, Out output );
420}; 420};
421 421
422//////////////////////////////////////////////////////////// 422////////////////////////////////////////////////////////////
423/// \brief Specialization of the Utf template for UTF-32 423/// \brief Specialization of the Utf template for UTF-32
424/// 424///
425//////////////////////////////////////////////////////////// 425////////////////////////////////////////////////////////////
426template <> class Utf<32> { 426template <> class Utf<32> {
427 public: 427 public:
428 //////////////////////////////////////////////////////////// 428 ////////////////////////////////////////////////////////////
429 /// \brief Decode a single UTF-32 character 429 /// \brief Decode a single UTF-32 character
430 /// 430 ///
431 /// Decoding a character means finding its unique 32-bits 431 /// Decoding a character means finding its unique 32-bits
432 /// code (called the codepoint) in the Unicode standard. 432 /// code (called the codepoint) in the Unicode standard.
433 /// For UTF-32, the character value is the same as the codepoint. 433 /// For UTF-32, the character value is the same as the codepoint.
434 /// 434 ///
435 /// \param begin Iterator pointing to the beginning of the input sequence 435 /// \param begin Iterator pointing to the beginning of the input sequence
436 /// \param end Iterator pointing to the end of the input sequence 436 /// \param end Iterator pointing to the end of the input sequence
437 /// \param output Codepoint of the decoded UTF-32 character 437 /// \param output Codepoint of the decoded UTF-32 character
438 /// \param replacement Replacement character to use in case the UTF-8 sequence is invalid 438 /// \param replacement Replacement character to use in case the UTF-8 sequence is invalid
439 /// 439 ///
440 /// \return Iterator pointing to one past the last read element of the input sequence 440 /// \return Iterator pointing to one past the last read element of the input sequence
441 /// 441 ///
442 //////////////////////////////////////////////////////////// 442 ////////////////////////////////////////////////////////////
443 template <typename In> 443 template <typename In>
444 static In Decode( In begin, In end, Uint32& output, Uint32 replacement = 0 ); 444 static In Decode( In begin, In end, Uint32& output, Uint32 replacement = 0 );
445 445
446 //////////////////////////////////////////////////////////// 446 ////////////////////////////////////////////////////////////
447 /// \brief Encode a single UTF-32 character 447 /// \brief Encode a single UTF-32 character
448 /// 448 ///
449 /// Encoding a character means converting a unique 32-bits 449 /// Encoding a character means converting a unique 32-bits
450 /// code (called the codepoint) in the target encoding, UTF-32. 450 /// code (called the codepoint) in the target encoding, UTF-32.
451 /// For UTF-32, the codepoint is the same as the character value. 451 /// For UTF-32, the codepoint is the same as the character value.
452 /// 452 ///
453 /// \param input Codepoint to encode as UTF-32 453 /// \param input Codepoint to encode as UTF-32
454 /// \param output Iterator pointing to the beginning of the output sequence 454 /// \param output Iterator pointing to the beginning of the output sequence
455 /// \param replacement Replacement for characters not convertible to UTF-32 (use 0 to skip them) 455 /// \param replacement Replacement for characters not convertible to UTF-32 (use 0 to skip them)
456 /// 456 ///
457 /// \return Iterator to the end of the output sequence which has been written 457 /// \return Iterator to the end of the output sequence which has been written
458 /// 458 ///
459 //////////////////////////////////////////////////////////// 459 ////////////////////////////////////////////////////////////
460 template <typename Out> static Out Encode( Uint32 input, Out output, Uint32 replacement = 0 ); 460 template <typename Out> static Out Encode( Uint32 input, Out output, Uint32 replacement = 0 );
461 461
462 //////////////////////////////////////////////////////////// 462 ////////////////////////////////////////////////////////////
463 /// \brief Advance to the next UTF-32 character 463 /// \brief Advance to the next UTF-32 character
464 /// 464 ///
465 /// This function is trivial for UTF-32, which can store 465 /// This function is trivial for UTF-32, which can store
466 /// every character in a single storage element. 466 /// every character in a single storage element.
467 /// 467 ///
468 /// \param begin Iterator pointing to the beginning of the input sequence 468 /// \param begin Iterator pointing to the beginning of the input sequence
469 /// \param end Iterator pointing to the end of the input sequence 469 /// \param end Iterator pointing to the end of the input sequence
470 /// 470 ///
471 /// \return Iterator pointing to one past the last read element of the input sequence 471 /// \return Iterator pointing to one past the last read element of the input sequence
472 /// 472 ///
473 //////////////////////////////////////////////////////////// 473 ////////////////////////////////////////////////////////////
474 template <typename In> static In Next( In begin, In end ); 474 template <typename In> static In Next( In begin, In end );
475 475
476 //////////////////////////////////////////////////////////// 476 ////////////////////////////////////////////////////////////
477 /// \brief Count the number of characters of a UTF-32 sequence 477 /// \brief Count the number of characters of a UTF-32 sequence
478 /// 478 ///
479 /// This function is trivial for UTF-32, which can store 479 /// This function is trivial for UTF-32, which can store
480 /// every character in a single storage element. 480 /// every character in a single storage element.
481 /// 481 ///
482 /// \param begin Iterator pointing to the beginning of the input sequence 482 /// \param begin Iterator pointing to the beginning of the input sequence
483 /// \param end Iterator pointing to the end of the input sequence 483 /// \param end Iterator pointing to the end of the input sequence
484 /// 484 ///
485 /// \return Iterator pointing to one past the last read element of the input sequence 485 /// \return Iterator pointing to one past the last read element of the input sequence
486 /// 486 ///
487 //////////////////////////////////////////////////////////// 487 ////////////////////////////////////////////////////////////
488 template <typename In> static std::size_t Count( In begin, In end ); 488 template <typename In> static std::size_t Count( In begin, In end );
489 489
490 //////////////////////////////////////////////////////////// 490 ////////////////////////////////////////////////////////////
491 /// \brief Convert an ANSI characters range to UTF-32 491 /// \brief Convert an ANSI characters range to UTF-32
492 /// 492 ///
493 /// The current global locale will be used by default, unless you 493 /// The current global locale will be used by default, unless you
494 /// pass a custom one in the \a locale parameter. 494 /// pass a custom one in the \a locale parameter.
495 /// 495 ///
496 /// \param begin Iterator pointing to the beginning of the input sequence 496 /// \param begin Iterator pointing to the beginning of the input sequence
497 /// \param end Iterator pointing to the end of the input sequence 497 /// \param end Iterator pointing to the end of the input sequence
498 /// \param output Iterator pointing to the beginning of the output sequence 498 /// \param output Iterator pointing to the beginning of the output sequence
499 /// \param locale Locale to use for conversion 499 /// \param locale Locale to use for conversion
500 /// 500 ///
501 /// \return Iterator to the end of the output sequence which has been written 501 /// \return Iterator to the end of the output sequence which has been written
502 /// 502 ///
503 //////////////////////////////////////////////////////////// 503 ////////////////////////////////////////////////////////////
504 template <typename In, typename Out> 504 template <typename In, typename Out>
505 static Out FromAnsi( In begin, In end, Out output, const std::locale& locale = std::locale() ); 505 static Out FromAnsi( In begin, In end, Out output, const std::locale& locale = std::locale() );
506 506
507 //////////////////////////////////////////////////////////// 507 ////////////////////////////////////////////////////////////
508 /// \brief Convert a wide characters range to UTF-32 508 /// \brief Convert a wide characters range to UTF-32
509 /// 509 ///
510 /// \param begin Iterator pointing to the beginning of the input sequence 510 /// \param begin Iterator pointing to the beginning of the input sequence
511 /// \param end Iterator pointing to the end of the input sequence 511 /// \param end Iterator pointing to the end of the input sequence
512 /// \param output Iterator pointing to the beginning of the output sequence 512 /// \param output Iterator pointing to the beginning of the output sequence
513 /// 513 ///
514 /// \return Iterator to the end of the output sequence which has been written 514 /// \return Iterator to the end of the output sequence which has been written
515 /// 515 ///
516 //////////////////////////////////////////////////////////// 516 ////////////////////////////////////////////////////////////
517 template <typename In, typename Out> static Out FromWide( In begin, In end, Out output ); 517 template <typename In, typename Out> static Out FromWide( In begin, In end, Out output );
518 518
519 //////////////////////////////////////////////////////////// 519 ////////////////////////////////////////////////////////////
520 /// \brief Convert a latin-1 (ISO-5589-1) characters range to UTF-32 520 /// \brief Convert a latin-1 (ISO-5589-1) characters range to UTF-32
521 /// 521 ///
522 /// \param begin Iterator pointing to the beginning of the input sequence 522 /// \param begin Iterator pointing to the beginning of the input sequence
523 /// \param end Iterator pointing to the end of the input sequence 523 /// \param end Iterator pointing to the end of the input sequence
524 /// \param output Iterator pointing to the beginning of the output sequence 524 /// \param output Iterator pointing to the beginning of the output sequence
525 /// \param locale Locale to use for conversion 525 /// \param locale Locale to use for conversion
526 /// 526 ///
527 /// \return Iterator to the end of the output sequence which has been written 527 /// \return Iterator to the end of the output sequence which has been written
528 /// 528 ///
529 //////////////////////////////////////////////////////////// 529 ////////////////////////////////////////////////////////////
530 template <typename In, typename Out> static Out FromLatin1( In begin, In end, Out output ); 530 template <typename In, typename Out> static Out FromLatin1( In begin, In end, Out output );
531 531
532 //////////////////////////////////////////////////////////// 532 ////////////////////////////////////////////////////////////
533 /// \brief Convert an UTF-32 characters range to ANSI characters 533 /// \brief Convert an UTF-32 characters range to ANSI characters
534 /// 534 ///
535 /// The current global locale will be used by default, unless you 535 /// The current global locale will be used by default, unless you
536 /// pass a custom one in the \a locale parameter. 536 /// pass a custom one in the \a locale parameter.
537 /// 537 ///
538 /// \param begin Iterator pointing to the beginning of the input sequence 538 /// \param begin Iterator pointing to the beginning of the input sequence
539 /// \param end Iterator pointing to the end of the input sequence 539 /// \param end Iterator pointing to the end of the input sequence
540 /// \param output Iterator pointing to the beginning of the output sequence 540 /// \param output Iterator pointing to the beginning of the output sequence
541 /// \param replacement Replacement for characters not convertible to ANSI (use 0 to skip them) 541 /// \param replacement Replacement for characters not convertible to ANSI (use 0 to skip them)
542 /// \param locale Locale to use for conversion 542 /// \param locale Locale to use for conversion
543 /// 543 ///
544 /// \return Iterator to the end of the output sequence which has been written 544 /// \return Iterator to the end of the output sequence which has been written
545 /// 545 ///
546 //////////////////////////////////////////////////////////// 546 ////////////////////////////////////////////////////////////
547 template <typename In, typename Out> 547 template <typename In, typename Out>
548 static Out ToAnsi( In begin, In end, Out output, char replacement = 0, 548 static Out ToAnsi( In begin, In end, Out output, char replacement = 0,
549 const std::locale& locale = std::locale() ); 549 const std::locale& locale = std::locale() );
550 550
551#ifndef EFSW_NO_WIDECHAR 551#ifndef EFSW_NO_WIDECHAR
552 //////////////////////////////////////////////////////////// 552 ////////////////////////////////////////////////////////////
553 /// \brief Convert an UTF-32 characters range to wide characters 553 /// \brief Convert an UTF-32 characters range to wide characters
554 /// 554 ///
555 /// \param begin Iterator pointing to the beginning of the input sequence 555 /// \param begin Iterator pointing to the beginning of the input sequence
556 /// \param end Iterator pointing to the end of the input sequence 556 /// \param end Iterator pointing to the end of the input sequence
557 /// \param output Iterator pointing to the beginning of the output sequence 557 /// \param output Iterator pointing to the beginning of the output sequence
558 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them) 558 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them)
559 /// 559 ///
560 /// \return Iterator to the end of the output sequence which has been written 560 /// \return Iterator to the end of the output sequence which has been written
561 /// 561 ///
562 //////////////////////////////////////////////////////////// 562 ////////////////////////////////////////////////////////////
563 template <typename In, typename Out> 563 template <typename In, typename Out>
564 static Out ToWide( In begin, In end, Out output, wchar_t replacement = 0 ); 564 static Out ToWide( In begin, In end, Out output, wchar_t replacement = 0 );
565#endif 565#endif
566 566
567 //////////////////////////////////////////////////////////// 567 ////////////////////////////////////////////////////////////
568 /// \brief Convert an UTF-16 characters range to latin-1 (ISO-5589-1) characters 568 /// \brief Convert an UTF-16 characters range to latin-1 (ISO-5589-1) characters
569 /// 569 ///
570 /// \param begin Iterator pointing to the beginning of the input sequence 570 /// \param begin Iterator pointing to the beginning of the input sequence
571 /// \param end Iterator pointing to the end of the input sequence 571 /// \param end Iterator pointing to the end of the input sequence
572 /// \param output Iterator pointing to the beginning of the output sequence 572 /// \param output Iterator pointing to the beginning of the output sequence
573 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them) 573 /// \param replacement Replacement for characters not convertible to wide (use 0 to skip them)
574 /// 574 ///
575 /// \return Iterator to the end of the output sequence which has been written 575 /// \return Iterator to the end of the output sequence which has been written
576 /// 576 ///
577 //////////////////////////////////////////////////////////// 577 ////////////////////////////////////////////////////////////
578 template <typename In, typename Out> 578 template <typename In, typename Out>
579 static Out ToLatin1( In begin, In end, Out output, char replacement = 0 ); 579 static Out ToLatin1( In begin, In end, Out output, char replacement = 0 );
580 580
581 //////////////////////////////////////////////////////////// 581 ////////////////////////////////////////////////////////////
582 /// \brief Convert a UTF-32 characters range to UTF-8 582 /// \brief Convert a UTF-32 characters range to UTF-8
583 /// 583 ///
584 /// \param begin Iterator pointing to the beginning of the input sequence 584 /// \param begin Iterator pointing to the beginning of the input sequence
585 /// \param end Iterator pointing to the end of the input sequence 585 /// \param end Iterator pointing to the end of the input sequence
586 /// \param output Iterator pointing to the beginning of the output sequence 586 /// \param output Iterator pointing to the beginning of the output sequence
587 /// 587 ///
588 /// \return Iterator to the end of the output sequence which has been written 588 /// \return Iterator to the end of the output sequence which has been written
589 /// 589 ///
590 //////////////////////////////////////////////////////////// 590 ////////////////////////////////////////////////////////////
591 template <typename In, typename Out> static Out toUtf8( In begin, In end, Out output ); 591 template <typename In, typename Out> static Out toUtf8( In begin, In end, Out output );
592 592
593 //////////////////////////////////////////////////////////// 593 ////////////////////////////////////////////////////////////
594 /// \brief Convert a UTF-32 characters range to UTF-16 594 /// \brief Convert a UTF-32 characters range to UTF-16
595 /// 595 ///
596 /// \param begin Iterator pointing to the beginning of the input sequence 596 /// \param begin Iterator pointing to the beginning of the input sequence
597 /// \param end Iterator pointing to the end of the input sequence 597 /// \param end Iterator pointing to the end of the input sequence
598 /// \param output Iterator pointing to the beginning of the output sequence 598 /// \param output Iterator pointing to the beginning of the output sequence
599 /// 599 ///
600 /// \return Iterator to the end of the output sequence which has been written 600 /// \return Iterator to the end of the output sequence which has been written
601 /// 601 ///
602 //////////////////////////////////////////////////////////// 602 ////////////////////////////////////////////////////////////
603 template <typename In, typename Out> static Out ToUtf16( In begin, In end, Out output ); 603 template <typename In, typename Out> static Out ToUtf16( In begin, In end, Out output );
604 604
605 //////////////////////////////////////////////////////////// 605 ////////////////////////////////////////////////////////////
606 /// \brief Convert a UTF-32 characters range to UTF-32 606 /// \brief Convert a UTF-32 characters range to UTF-32
607 /// 607 ///
608 /// This functions does nothing more than a direct copy; 608 /// This functions does nothing more than a direct copy;
609 /// it is defined only to provide the same interface as other 609 /// it is defined only to provide the same interface as other
610 /// specializations of the efsw::Utf<> template, and allow 610 /// specializations of the efsw::Utf<> template, and allow
611 /// generic code to be written on top of it. 611 /// generic code to be written on top of it.
612 /// 612 ///
613 /// \param begin Iterator pointing to the beginning of the input sequence 613 /// \param begin Iterator pointing to the beginning of the input sequence
614 /// \param end Iterator pointing to the end of the input sequence 614 /// \param end Iterator pointing to the end of the input sequence
615 /// \param output Iterator pointing to the beginning of the output sequence 615 /// \param output Iterator pointing to the beginning of the output sequence
616 /// 616 ///
617 /// \return Iterator to the end of the output sequence which has been written 617 /// \return Iterator to the end of the output sequence which has been written
618 /// 618 ///
619 //////////////////////////////////////////////////////////// 619 ////////////////////////////////////////////////////////////
620 template <typename In, typename Out> static Out ToUtf32( In begin, In end, Out output ); 620 template <typename In, typename Out> static Out ToUtf32( In begin, In end, Out output );
621 621
622 //////////////////////////////////////////////////////////// 622 ////////////////////////////////////////////////////////////
623 /// \brief Decode a single ANSI character to UTF-32 623 /// \brief Decode a single ANSI character to UTF-32
624 /// 624 ///
625 /// This function does not exist in other specializations 625 /// This function does not exist in other specializations
626 /// of efsw::Utf<>, it is defined for convenience (it is used by 626 /// of efsw::Utf<>, it is defined for convenience (it is used by
627 /// several other conversion functions). 627 /// several other conversion functions).
628 /// 628 ///
629 /// \param input Input ANSI character 629 /// \param input Input ANSI character
630 /// \param locale Locale to use for conversion 630 /// \param locale Locale to use for conversion
631 /// 631 ///
632 /// \return Converted character 632 /// \return Converted character
633 /// 633 ///
634 //////////////////////////////////////////////////////////// 634 ////////////////////////////////////////////////////////////
635 template <typename In> 635 template <typename In>
636 static Uint32 DecodeAnsi( In input, const std::locale& locale = std::locale() ); 636 static Uint32 DecodeAnsi( In input, const std::locale& locale = std::locale() );
637 637
638 //////////////////////////////////////////////////////////// 638 ////////////////////////////////////////////////////////////
639 /// \brief Decode a single wide character to UTF-32 639 /// \brief Decode a single wide character to UTF-32
640 /// 640 ///
641 /// This function does not exist in other specializations 641 /// This function does not exist in other specializations
642 /// of efsw::Utf<>, it is defined for convenience (it is used by 642 /// of efsw::Utf<>, it is defined for convenience (it is used by
643 /// several other conversion functions). 643 /// several other conversion functions).
644 /// 644 ///
645 /// \param input Input wide character 645 /// \param input Input wide character
646 /// 646 ///
647 /// \return Converted character 647 /// \return Converted character
648 /// 648 ///
649 //////////////////////////////////////////////////////////// 649 ////////////////////////////////////////////////////////////
650 template <typename In> static Uint32 DecodeWide( In input ); 650 template <typename In> static Uint32 DecodeWide( In input );
651 651
652 //////////////////////////////////////////////////////////// 652 ////////////////////////////////////////////////////////////
653 /// \brief Encode a single UTF-32 character to ANSI 653 /// \brief Encode a single UTF-32 character to ANSI
654 /// 654 ///
655 /// This function does not exist in other specializations 655 /// This function does not exist in other specializations
656 /// of efsw::Utf<>, it is defined for convenience (it is used by 656 /// of efsw::Utf<>, it is defined for convenience (it is used by
657 /// several other conversion functions). 657 /// several other conversion functions).
658 /// 658 ///
659 /// \param codepoint Iterator pointing to the beginning of the input sequence 659 /// \param codepoint Iterator pointing to the beginning of the input sequence
660 /// \param output Iterator pointing to the beginning of the output sequence 660 /// \param output Iterator pointing to the beginning of the output sequence
661 /// \param replacement Replacement if the input character is not convertible to ANSI (use 0 to 661 /// \param replacement Replacement if the input character is not convertible to ANSI (use 0 to
662 /// skip it) \param locale Locale to use for conversion 662 /// skip it) \param locale Locale to use for conversion
663 /// 663 ///
664 /// \return Iterator to the end of the output sequence which has been written 664 /// \return Iterator to the end of the output sequence which has been written
665 /// 665 ///
666 //////////////////////////////////////////////////////////// 666 ////////////////////////////////////////////////////////////
667 template <typename Out> 667 template <typename Out>
668 static Out EncodeAnsi( Uint32 codepoint, Out output, char replacement = 0, 668 static Out EncodeAnsi( Uint32 codepoint, Out output, char replacement = 0,
669 const std::locale& locale = std::locale() ); 669 const std::locale& locale = std::locale() );
670 670
671#ifndef EFSW_NO_WIDECHAR 671#ifndef EFSW_NO_WIDECHAR
672 //////////////////////////////////////////////////////////// 672 ////////////////////////////////////////////////////////////
673 /// \brief Encode a single UTF-32 character to wide 673 /// \brief Encode a single UTF-32 character to wide
674 /// 674 ///
675 /// This function does not exist in other specializations 675 /// This function does not exist in other specializations
676 /// of efsw::Utf<>, it is defined for convenience (it is used by 676 /// of efsw::Utf<>, it is defined for convenience (it is used by
677 /// several other conversion functions). 677 /// several other conversion functions).
678 /// 678 ///
679 /// \param codepoint Iterator pointing to the beginning of the input sequence 679 /// \param codepoint Iterator pointing to the beginning of the input sequence
680 /// \param output Iterator pointing to the beginning of the output sequence 680 /// \param output Iterator pointing to the beginning of the output sequence
681 /// \param replacement Replacement if the input character is not convertible to wide (use 0 to 681 /// \param replacement Replacement if the input character is not convertible to wide (use 0 to
682 /// skip it) 682 /// skip it)
683 /// 683 ///
684 /// \return Iterator to the end of the output sequence which has been written 684 /// \return Iterator to the end of the output sequence which has been written
685 /// 685 ///
686 //////////////////////////////////////////////////////////// 686 ////////////////////////////////////////////////////////////
687 template <typename Out> 687 template <typename Out>
688 static Out EncodeWide( Uint32 codepoint, Out output, wchar_t replacement = 0 ); 688 static Out EncodeWide( Uint32 codepoint, Out output, wchar_t replacement = 0 );
689#endif 689#endif
690}; 690};
691 691
692#include "Utf.inl" 692#include "Utf.inl"
693 693
694// Make typedefs to get rid of the template syntax 694// Make typedefs to get rid of the template syntax
695typedef Utf<8> Utf8; 695typedef Utf<8> Utf8;
696typedef Utf<16> Utf16; 696typedef Utf<16> Utf16;
697typedef Utf<32> Utf32; 697typedef Utf<32> Utf32;
698 698
699} // namespace efsw 699} // namespace efsw
700#endif 700#endif
701 701
702//////////////////////////////////////////////////////////// 702////////////////////////////////////////////////////////////
703/// \class efsw::Utf 703/// \class efsw::Utf
704/// \ingroup system 704/// \ingroup system
705/// 705///
706/// Utility class providing generic functions for UTF conversions. 706/// Utility class providing generic functions for UTF conversions.
707/// 707///
708/// efsw::Utf is a low-level, generic interface for counting, iterating, 708/// efsw::Utf is a low-level, generic interface for counting, iterating,
709/// encoding and decoding Unicode characters and strings. It is able 709/// encoding and decoding Unicode characters and strings. It is able
710/// to handle ANSI, wide, UTF-8, UTF-16 and UTF-32 encodings. 710/// to handle ANSI, wide, UTF-8, UTF-16 and UTF-32 encodings.
711/// 711///
712/// efsw::Utf<X> functions are all static, these classes are not meant to 712/// efsw::Utf<X> functions are all static, these classes are not meant to
713/// be instanciated. All the functions are template, so that you 713/// be instanciated. All the functions are template, so that you
714/// can use any character / string type for a given encoding. 714/// can use any character / string type for a given encoding.
715/// 715///
716/// It has 3 specializations: 716/// It has 3 specializations:
717/// \li efsw::Utf<8> (typedef'd to efsw::Utf8) 717/// \li efsw::Utf<8> (typedef'd to efsw::Utf8)
718/// \li efsw::Utf<16> (typedef'd to efsw::Utf16) 718/// \li efsw::Utf<16> (typedef'd to efsw::Utf16)
719/// \li efsw::Utf<32> (typedef'd to efsw::Utf32) 719/// \li efsw::Utf<32> (typedef'd to efsw::Utf32)
720/// 720///
721//////////////////////////////////////////////////////////// 721////////////////////////////////////////////////////////////
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-04-25 17:57:29 +0000