]>
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 | ||
89 | // not very accurately named because it is not necessarily of type wxMBConvLibc | |
90 | // (but the name can't eb changed because of backwards compatibility) default | |
91 | // conversion | |
92 | WXDLLIMPEXP_DATA_BASE(extern wxMBConv&) wxConvLibc; | |
6001e347 | 93 | |
e90c1d2a | 94 | // ---------------------------------------------------------------------------- |
6001e347 | 95 | // wxMBConvUTF7 (for conversion using UTF7 encoding) |
e90c1d2a | 96 | // ---------------------------------------------------------------------------- |
6001e347 | 97 | |
bddd7a8d | 98 | class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv |
6001e347 RR |
99 | { |
100 | public: | |
75736a9c DS |
101 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
102 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
6001e347 RR |
103 | }; |
104 | ||
bde4baac | 105 | WXDLLIMPEXP_DATA_BASE(extern wxMBConvUTF7&) wxConvUTF7; |
6001e347 | 106 | |
e90c1d2a | 107 | // ---------------------------------------------------------------------------- |
6001e347 | 108 | // wxMBConvUTF8 (for conversion using UTF8 encoding) |
e90c1d2a | 109 | // ---------------------------------------------------------------------------- |
6001e347 | 110 | |
bddd7a8d | 111 | class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv |
6001e347 RR |
112 | { |
113 | public: | |
75736a9c DS |
114 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
115 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
6001e347 RR |
116 | }; |
117 | ||
bde4baac | 118 | WXDLLIMPEXP_DATA_BASE(extern wxMBConvUTF8&) wxConvUTF8; |
6001e347 | 119 | |
e90c1d2a | 120 | // ---------------------------------------------------------------------------- |
c91830cb VZ |
121 | // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding) |
122 | // ---------------------------------------------------------------------------- | |
123 | ||
124 | class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConv | |
125 | { | |
126 | public: | |
75736a9c DS |
127 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
128 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
c91830cb VZ |
129 | }; |
130 | ||
131 | // ---------------------------------------------------------------------------- | |
132 | // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding) | |
133 | // ---------------------------------------------------------------------------- | |
134 | ||
135 | class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConv | |
136 | { | |
137 | public: | |
75736a9c DS |
138 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
139 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
c91830cb VZ |
140 | }; |
141 | ||
142 | // ---------------------------------------------------------------------------- | |
143 | // wxMBConvUCS4LE (for conversion using UTF32 Little Endian encoding) | |
144 | // ---------------------------------------------------------------------------- | |
145 | ||
146 | class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConv | |
147 | { | |
148 | public: | |
75736a9c DS |
149 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
150 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
c91830cb VZ |
151 | }; |
152 | ||
153 | // ---------------------------------------------------------------------------- | |
154 | // wxMBConvUCS4BE (for conversion using UTF32 Big Endian encoding) | |
155 | // ---------------------------------------------------------------------------- | |
156 | ||
157 | class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConv | |
158 | { | |
159 | public: | |
75736a9c DS |
160 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
161 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
c91830cb VZ |
162 | }; |
163 | ||
164 | // ---------------------------------------------------------------------------- | |
e90c1d2a VZ |
165 | // wxCSConv (for conversion based on loadable char sets) |
166 | // ---------------------------------------------------------------------------- | |
6001e347 | 167 | |
8b04d4c4 VZ |
168 | #include "wx/fontenc.h" |
169 | ||
bddd7a8d | 170 | class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv |
6001e347 | 171 | { |
6001e347 | 172 | public: |
e95354ec VZ |
173 | // we can be created either from charset name or from an encoding constant |
174 | // but we can't have both at once | |
e90c1d2a | 175 | wxCSConv(const wxChar *charset); |
8b04d4c4 | 176 | wxCSConv(wxFontEncoding encoding); |
e95354ec | 177 | |
54380f29 | 178 | wxCSConv(const wxCSConv& conv); |
e90c1d2a VZ |
179 | virtual ~wxCSConv(); |
180 | ||
54380f29 | 181 | wxCSConv& operator=(const wxCSConv& conv); |
2b5f62a0 | 182 | |
75736a9c DS |
183 | virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const; |
184 | virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const; | |
e90c1d2a | 185 | |
65e50848 JS |
186 | void Clear() ; |
187 | ||
e90c1d2a | 188 | private: |
8b04d4c4 VZ |
189 | // common part of all ctors |
190 | void Init(); | |
191 | ||
e95354ec VZ |
192 | // creates m_convReal if necessary |
193 | void CreateConvIfNeeded() const; | |
194 | ||
195 | // do create m_convReal (unconditionally) | |
196 | wxMBConv *DoCreate() const; | |
197 | ||
bda3d86a VZ |
198 | // set the name (may be only called when m_name == NULL), makes copy of |
199 | // charset string | |
e90c1d2a VZ |
200 | void SetName(const wxChar *charset); |
201 | ||
e95354ec | 202 | |
dccce9ea VZ |
203 | // note that we can't use wxString here because of compilation |
204 | // dependencies: we're included from wx/string.h | |
e90c1d2a | 205 | wxChar *m_name; |
8b04d4c4 | 206 | wxFontEncoding m_encoding; |
e95354ec VZ |
207 | |
208 | // use CreateConvIfNeeded() before accessing m_convReal! | |
209 | wxMBConv *m_convReal; | |
e90c1d2a | 210 | bool m_deferred; |
6001e347 RR |
211 | }; |
212 | ||
d5c8817c SC |
213 | #ifdef __WXOSX__ |
214 | #define wxConvFile wxConvUTF8 | |
215 | #else | |
b1ac3b56 | 216 | #define wxConvFile wxConvLocal |
d5c8817c SC |
217 | #endif |
218 | ||
bde4baac VZ |
219 | WXDLLIMPEXP_DATA_BASE(extern wxCSConv&) wxConvLocal; |
220 | WXDLLIMPEXP_DATA_BASE(extern wxCSConv&) wxConvISO8859_1; | |
bddd7a8d | 221 | WXDLLIMPEXP_DATA_BASE(extern wxMBConv *) wxConvCurrent; |
6001e347 | 222 | |
e95354ec VZ |
223 | // ---------------------------------------------------------------------------- |
224 | // endianness-dependent conversions | |
225 | // ---------------------------------------------------------------------------- | |
226 | ||
227 | #ifdef WORDS_BIGENDIAN | |
228 | typedef wxMBConvUTF16BE wxMBConvUTF16; | |
229 | typedef wxMBConvUTF32BE wxMBConvUTF32; | |
230 | #else | |
231 | typedef wxMBConvUTF16LE wxMBConvUTF16; | |
232 | typedef wxMBConvUTF32LE wxMBConvUTF32; | |
233 | #endif | |
234 | ||
e90c1d2a | 235 | // ---------------------------------------------------------------------------- |
6001e347 | 236 | // filename conversion macros |
e90c1d2a | 237 | // ---------------------------------------------------------------------------- |
6001e347 RR |
238 | |
239 | // filenames are multibyte on Unix and probably widechar on Windows? | |
c4e41ce3 | 240 | #if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ ) |
e90c1d2a | 241 | #define wxMBFILES 1 |
6001e347 | 242 | #else |
e90c1d2a | 243 | #define wxMBFILES 0 |
6001e347 RR |
244 | #endif |
245 | ||
80df4d31 | 246 | #if wxMBFILES && wxUSE_UNICODE |
e90c1d2a VZ |
247 | #define wxFNCONV(name) wxConvFile.cWX2MB(name) |
248 | #define wxFNSTRINGCAST wxMBSTRINGCAST | |
d5c8817c SC |
249 | #else |
250 | #if defined( __WXOSX__ ) && wxMBFILES | |
251 | #define wxFNCONV(name) wxConvFile.cWC2MB( wxConvLocal.cWX2WC(name) ) | |
6001e347 | 252 | #else |
e90c1d2a | 253 | #define wxFNCONV(name) name |
d5c8817c | 254 | #endif |
e90c1d2a | 255 | #define wxFNSTRINGCAST WXSTRINGCAST |
6001e347 RR |
256 | #endif |
257 | ||
258 | #else | |
259 | // !wxUSE_WCHAR_T | |
260 | ||
e90c1d2a | 261 | // ---------------------------------------------------------------------------- |
6001e347 | 262 | // stand-ins in absence of wchar_t |
e90c1d2a | 263 | // ---------------------------------------------------------------------------- |
6001e347 | 264 | |
bddd7a8d | 265 | class WXDLLIMPEXP_BASE wxMBConv |
6001e347 RR |
266 | { |
267 | public: | |
e90c1d2a VZ |
268 | const char* cMB2WX(const char *psz) const { return psz; } |
269 | const char* cWX2MB(const char *psz) const { return psz; } | |
6001e347 | 270 | }; |
e90c1d2a | 271 | |
bde4baac VZ |
272 | #define wxConvFile wxConvLocal |
273 | ||
8b04d4c4 | 274 | WXDLLIMPEXP_DATA_BASE(extern wxMBConv) wxConvLibc, |
8b04d4c4 VZ |
275 | wxConvLocal, |
276 | wxConvISO8859_1, | |
277 | wxConvUTF8; | |
bddd7a8d | 278 | WXDLLIMPEXP_DATA_BASE(extern wxMBConv *) wxConvCurrent; |
6001e347 RR |
279 | |
280 | #define wxFNCONV(name) name | |
e90c1d2a | 281 | #define wxFNSTRINGCAST WXSTRINGCAST |
6001e347 RR |
282 | |
283 | #endif | |
284 | // wxUSE_WCHAR_T | |
285 | ||
e90c1d2a VZ |
286 | // ---------------------------------------------------------------------------- |
287 | // macros for the most common conversions | |
288 | // ---------------------------------------------------------------------------- | |
289 | ||
290 | #if wxUSE_UNICODE | |
291 | #define wxConvertWX2MB(s) wxConvCurrent->cWX2MB(s) | |
292 | #define wxConvertMB2WX(s) wxConvCurrent->cMB2WX(s) | |
293 | #else // ANSI | |
294 | // no conversions to do | |
295 | #define wxConvertWX2MB(s) (s) | |
296 | #define wxConvertMB2WX(s) (s) | |
297 | #endif // Unicode/ANSI | |
298 | ||
299 | #endif | |
6001e347 RR |
300 | // _WX_WXSTRCONVH__ |
301 |