]>
Commit | Line | Data |
---|---|---|
6001e347 RR |
1 | /////////////////////////////////////////////////////////////////////////////// |
2 | // Name: strconv.h | |
3 | // Purpose: conversion routines for char sets any Unicode | |
4 | // Author: Robert Roebling, Ove Kaaven | |
5 | // Modified by: | |
6 | // Created: 29/01/98 | |
7 | // RCS-ID: $Id$ | |
8 | // Copyright: (c) 1998 Ove Kaaven, Robert Roebling, Vadim Zeitlin | |
65571936 | 9 | // Licence: wxWindows licence |
6001e347 RR |
10 | /////////////////////////////////////////////////////////////////////////////// |
11 | ||
12 | #ifndef _WX_WXSTRCONVH__ | |
13 | #define _WX_WXSTRCONVH__ | |
14 | ||
12028905 | 15 | #if defined(__GNUG__) && !defined(NO_GCC_PRAGMA) |
6001e347 RR |
16 | #pragma interface "strconv.h" |
17 | #endif | |
18 | ||
19 | #include "wx/defs.h" | |
20 | #include "wx/wxchar.h" | |
21 | #include "wx/buffer.h" | |
22 | ||
7db39dd6 CE |
23 | #ifdef __DIGITALMARS__ |
24 | #include "typeinfo.h" | |
25 | #endif | |
26 | ||
9dea36ef DW |
27 | #if defined(__VISAGECPP__) && __IBMCPP__ >= 400 |
28 | # undef __BSEXCPT__ | |
29 | #endif | |
dccce9ea | 30 | |
6001e347 RR |
31 | #include <stdlib.h> |
32 | ||
33 | #if wxUSE_WCHAR_T | |
34 | ||
e90c1d2a | 35 | // ---------------------------------------------------------------------------- |
bde4baac | 36 | // wxMBConv (abstract base class for conversions) |
e90c1d2a | 37 | // ---------------------------------------------------------------------------- |
6001e347 | 38 | |
bddd7a8d | 39 | class WXDLLIMPEXP_BASE wxMBConv |
6001e347 RR |
40 | { |
41 | public: | |
e90c1d2a | 42 | // the actual conversion takes place here |
bde4baac | 43 | // |
e4e3bbb4 | 44 | // note that outputSize is the size of the output buffer, not the length of input |
75736a9c DS |
45 | // (the latter is always supposed to be NUL-terminated) |
46 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const = 0; | |
47 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const = 0; | |
e90c1d2a | 48 | |
bde4baac | 49 | // MB <-> WC |
e90c1d2a VZ |
50 | const wxWCharBuffer cMB2WC(const char *psz) const; |
51 | const wxCharBuffer cWC2MB(const wchar_t *psz) const; | |
6001e347 | 52 | |
f5fb6871 RN |
53 | // MB <-> WC for strings with embedded null characters |
54 | // | |
55 | // pszLen length of the input string | |
56 | // pOutSize gets the final size of the converted string | |
57 | const wxWCharBuffer cMB2WC(const char *psz, size_t pszLen, size_t* pOutSize) const; | |
58 | const wxCharBuffer cWC2MB(const wchar_t *psz, size_t pszLen, size_t* pOutSize) const; | |
59 | ||
bde4baac | 60 | // convenience functions for converting MB or WC to/from wxWin default |
6001e347 | 61 | #if wxUSE_UNICODE |
e90c1d2a VZ |
62 | const wxWCharBuffer cMB2WX(const char *psz) const { return cMB2WC(psz); } |
63 | const wxCharBuffer cWX2MB(const wchar_t *psz) const { return cWC2MB(psz); } | |
64 | const wchar_t* cWC2WX(const wchar_t *psz) const { return psz; } | |
f6bcfd97 | 65 | const wchar_t* cWX2WC(const wchar_t *psz) const { return psz; } |
e90c1d2a VZ |
66 | #else // ANSI |
67 | const char* cMB2WX(const char *psz) const { return psz; } | |
68 | const char* cWX2MB(const char *psz) const { return psz; } | |
69 | const wxCharBuffer cWC2WX(const wchar_t *psz) const { return cWC2MB(psz); } | |
70 | const wxWCharBuffer cWX2WC(const char *psz) const { return cMB2WC(psz); } | |
71 | #endif // Unicode/ANSI | |
2b5f62a0 VZ |
72 | |
73 | // virtual dtor for any base class | |
e4a4a50b | 74 | virtual ~wxMBConv(); |
6001e347 RR |
75 | }; |
76 | ||
bde4baac VZ |
77 | // ---------------------------------------------------------------------------- |
78 | // wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for | |
79 | // conversion (hence it depends on the current locale) | |
80 | // ---------------------------------------------------------------------------- | |
81 | ||
82 | class WXDLLIMPEXP_BASE wxMBConvLibc : public wxMBConv | |
83 | { | |
84 | public: | |
75736a9c DS |
85 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
86 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
bde4baac VZ |
87 | }; |
88 | ||
5576edf8 RR |
89 | #ifdef __UNIX__ |
90 | ||
91 | // ---------------------------------------------------------------------------- | |
92 | // wxConvBrokenFileNames is made for Unix in Unicode mode when | |
93 | // files are accidentally written in an encoding which is not | |
94 | // the system encoding. Typically, the system encoding will be | |
95 | // UTF8 but there might be files stored in ISO8859-1 on disk. | |
96 | // ---------------------------------------------------------------------------- | |
97 | ||
98 | class WXDLLIMPEXP_BASE wxConvBrokenFileNames : public wxMBConv | |
99 | { | |
100 | public: | |
845905d5 | 101 | wxConvBrokenFileNames(const wxChar *charset); |
5576edf8 RR |
102 | virtual ~wxConvBrokenFileNames() { delete m_conv; } |
103 | ||
104 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; | |
105 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
106 | ||
107 | private: | |
108 | // the conversion object we forward to | |
109 | wxMBConv *m_conv; | |
110 | }; | |
111 | ||
112 | #endif | |
113 | ||
e90c1d2a | 114 | // ---------------------------------------------------------------------------- |
6001e347 | 115 | // wxMBConvUTF7 (for conversion using UTF7 encoding) |
e90c1d2a | 116 | // ---------------------------------------------------------------------------- |
6001e347 | 117 | |
bddd7a8d | 118 | class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv |
6001e347 RR |
119 | { |
120 | public: | |
75736a9c DS |
121 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
122 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
6001e347 RR |
123 | }; |
124 | ||
e90c1d2a | 125 | // ---------------------------------------------------------------------------- |
6001e347 | 126 | // wxMBConvUTF8 (for conversion using UTF8 encoding) |
e90c1d2a | 127 | // ---------------------------------------------------------------------------- |
6001e347 | 128 | |
bddd7a8d | 129 | class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv |
6001e347 RR |
130 | { |
131 | public: | |
ea8ce907 RR |
132 | enum { |
133 | MAP_INVALID_UTF8_NOT = 0, | |
134 | MAP_INVALID_UTF8_TO_PUA = 1, | |
135 | MAP_INVALID_UTF8_TO_OCTAL = 2 | |
136 | }; | |
137 | ||
138 | wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { } | |
75736a9c DS |
139 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
140 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
ea8ce907 RR |
141 | |
142 | private: | |
143 | int m_options; | |
6001e347 RR |
144 | }; |
145 | ||
e90c1d2a | 146 | // ---------------------------------------------------------------------------- |
c91830cb VZ |
147 | // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding) |
148 | // ---------------------------------------------------------------------------- | |
149 | ||
150 | class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConv | |
151 | { | |
152 | public: | |
75736a9c DS |
153 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
154 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
c91830cb VZ |
155 | }; |
156 | ||
157 | // ---------------------------------------------------------------------------- | |
158 | // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding) | |
159 | // ---------------------------------------------------------------------------- | |
160 | ||
161 | class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConv | |
162 | { | |
163 | public: | |
75736a9c DS |
164 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
165 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
c91830cb VZ |
166 | }; |
167 | ||
168 | // ---------------------------------------------------------------------------- | |
169 | // wxMBConvUCS4LE (for conversion using UTF32 Little Endian encoding) | |
170 | // ---------------------------------------------------------------------------- | |
171 | ||
172 | class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConv | |
173 | { | |
174 | public: | |
75736a9c DS |
175 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
176 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
c91830cb VZ |
177 | }; |
178 | ||
179 | // ---------------------------------------------------------------------------- | |
180 | // wxMBConvUCS4BE (for conversion using UTF32 Big Endian encoding) | |
181 | // ---------------------------------------------------------------------------- | |
182 | ||
183 | class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConv | |
184 | { | |
185 | public: | |
75736a9c DS |
186 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
187 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
c91830cb VZ |
188 | }; |
189 | ||
190 | // ---------------------------------------------------------------------------- | |
e90c1d2a VZ |
191 | // wxCSConv (for conversion based on loadable char sets) |
192 | // ---------------------------------------------------------------------------- | |
6001e347 | 193 | |
8b04d4c4 VZ |
194 | #include "wx/fontenc.h" |
195 | ||
bddd7a8d | 196 | class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv |
6001e347 | 197 | { |
6001e347 | 198 | public: |
e95354ec VZ |
199 | // we can be created either from charset name or from an encoding constant |
200 | // but we can't have both at once | |
e90c1d2a | 201 | wxCSConv(const wxChar *charset); |
8b04d4c4 | 202 | wxCSConv(wxFontEncoding encoding); |
e95354ec | 203 | |
54380f29 | 204 | wxCSConv(const wxCSConv& conv); |
e90c1d2a VZ |
205 | virtual ~wxCSConv(); |
206 | ||
54380f29 | 207 | wxCSConv& operator=(const wxCSConv& conv); |
2b5f62a0 | 208 | |
75736a9c DS |
209 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
210 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
e90c1d2a | 211 | |
65e50848 JS |
212 | void Clear() ; |
213 | ||
e90c1d2a | 214 | private: |
8b04d4c4 VZ |
215 | // common part of all ctors |
216 | void Init(); | |
217 | ||
e95354ec VZ |
218 | // creates m_convReal if necessary |
219 | void CreateConvIfNeeded() const; | |
220 | ||
221 | // do create m_convReal (unconditionally) | |
222 | wxMBConv *DoCreate() const; | |
223 | ||
bda3d86a VZ |
224 | // set the name (may be only called when m_name == NULL), makes copy of |
225 | // charset string | |
e90c1d2a VZ |
226 | void SetName(const wxChar *charset); |
227 | ||
e95354ec | 228 | |
dccce9ea VZ |
229 | // note that we can't use wxString here because of compilation |
230 | // dependencies: we're included from wx/string.h | |
e90c1d2a | 231 | wxChar *m_name; |
8b04d4c4 | 232 | wxFontEncoding m_encoding; |
e95354ec VZ |
233 | |
234 | // use CreateConvIfNeeded() before accessing m_convReal! | |
235 | wxMBConv *m_convReal; | |
e90c1d2a | 236 | bool m_deferred; |
6001e347 RR |
237 | }; |
238 | ||
c3c1a9a9 | 239 | |
f5a1953b VZ |
240 | // ---------------------------------------------------------------------------- |
241 | // declare predefined conversion objects | |
242 | // ---------------------------------------------------------------------------- | |
d5c8817c | 243 | |
f5a1953b VZ |
244 | // conversion to be used with all standard functions affected by locale, e.g. |
245 | // strtol(), strftime(), ... | |
246 | extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc; | |
247 | ||
248 | // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t | |
16cba29d | 249 | extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1; |
f5a1953b VZ |
250 | extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7; |
251 | extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8; | |
252 | ||
253 | // conversion used for the file names on the systems where they're not Unicode | |
254 | // (basically anything except Windows) | |
255 | // | |
256 | // this is used by all file functions, can be changed by the application | |
257 | // | |
258 | // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used | |
259 | // under Windows normally) | |
260 | extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName; | |
261 | ||
262 | // backwards compatible define | |
263 | #define wxConvFile (*wxConvFileName) | |
264 | ||
265 | // the current conversion object, may be set to any conversion, is used by | |
266 | // default in a couple of places inside wx (initially same as wxConvLibc) | |
16cba29d | 267 | extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent; |
6001e347 | 268 | |
f5a1953b VZ |
269 | // ??? |
270 | extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal; | |
271 | ||
272 | ||
e95354ec VZ |
273 | // ---------------------------------------------------------------------------- |
274 | // endianness-dependent conversions | |
275 | // ---------------------------------------------------------------------------- | |
276 | ||
277 | #ifdef WORDS_BIGENDIAN | |
278 | typedef wxMBConvUTF16BE wxMBConvUTF16; | |
279 | typedef wxMBConvUTF32BE wxMBConvUTF32; | |
280 | #else | |
281 | typedef wxMBConvUTF16LE wxMBConvUTF16; | |
282 | typedef wxMBConvUTF32LE wxMBConvUTF32; | |
283 | #endif | |
284 | ||
e90c1d2a | 285 | // ---------------------------------------------------------------------------- |
6001e347 | 286 | // filename conversion macros |
e90c1d2a | 287 | // ---------------------------------------------------------------------------- |
6001e347 RR |
288 | |
289 | // filenames are multibyte on Unix and probably widechar on Windows? | |
c4e41ce3 | 290 | #if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ ) |
e90c1d2a | 291 | #define wxMBFILES 1 |
6001e347 | 292 | #else |
e90c1d2a | 293 | #define wxMBFILES 0 |
6001e347 RR |
294 | #endif |
295 | ||
80df4d31 | 296 | #if wxMBFILES && wxUSE_UNICODE |
f5a1953b | 297 | #define wxFNCONV(name) wxConvFileName->cWX2MB(name) |
e90c1d2a | 298 | #define wxFNSTRINGCAST wxMBSTRINGCAST |
d5c8817c SC |
299 | #else |
300 | #if defined( __WXOSX__ ) && wxMBFILES | |
f5a1953b | 301 | #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) ) |
6001e347 | 302 | #else |
e90c1d2a | 303 | #define wxFNCONV(name) name |
d5c8817c | 304 | #endif |
e90c1d2a | 305 | #define wxFNSTRINGCAST WXSTRINGCAST |
6001e347 RR |
306 | #endif |
307 | ||
f5a1953b | 308 | #else // !wxUSE_WCHAR_T |
6001e347 | 309 | |
e90c1d2a | 310 | // ---------------------------------------------------------------------------- |
6001e347 | 311 | // stand-ins in absence of wchar_t |
e90c1d2a | 312 | // ---------------------------------------------------------------------------- |
6001e347 | 313 | |
bddd7a8d | 314 | class WXDLLIMPEXP_BASE wxMBConv |
6001e347 RR |
315 | { |
316 | public: | |
e90c1d2a VZ |
317 | const char* cMB2WX(const char *psz) const { return psz; } |
318 | const char* cWX2MB(const char *psz) const { return psz; } | |
6001e347 | 319 | }; |
e90c1d2a | 320 | |
bde4baac VZ |
321 | #define wxConvFile wxConvLocal |
322 | ||
16cba29d | 323 | extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc, |
8b04d4c4 VZ |
324 | wxConvLocal, |
325 | wxConvISO8859_1, | |
326 | wxConvUTF8; | |
16cba29d | 327 | extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent; |
6001e347 RR |
328 | |
329 | #define wxFNCONV(name) name | |
e90c1d2a | 330 | #define wxFNSTRINGCAST WXSTRINGCAST |
6001e347 RR |
331 | |
332 | #endif | |
333 | // wxUSE_WCHAR_T | |
334 | ||
e90c1d2a VZ |
335 | // ---------------------------------------------------------------------------- |
336 | // macros for the most common conversions | |
337 | // ---------------------------------------------------------------------------- | |
338 | ||
339 | #if wxUSE_UNICODE | |
340 | #define wxConvertWX2MB(s) wxConvCurrent->cWX2MB(s) | |
341 | #define wxConvertMB2WX(s) wxConvCurrent->cMB2WX(s) | |
342 | #else // ANSI | |
343 | // no conversions to do | |
344 | #define wxConvertWX2MB(s) (s) | |
345 | #define wxConvertMB2WX(s) (s) | |
346 | #endif // Unicode/ANSI | |
347 | ||
348 | #endif | |
6001e347 RR |
349 | // _WX_WXSTRCONVH__ |
350 |