1
///////////////////////////////////////////////////////////////////////////////
3
// Purpose: conversion routines for char sets any Unicode
4
// Author: Robert Roebling, Ove Kaaven
7
// RCS-ID: $Id: strconv.h,v 1.43 2005/08/08 22:04:56 MW Exp $
8
// Copyright: (c) 1998 Ove Kaaven, Robert Roebling, Vadim Zeitlin
9
// Licence: wxWindows licence
10
///////////////////////////////////////////////////////////////////////////////
12
#ifndef _WX_WXSTRCONVH__
13
#define _WX_WXSTRCONVH__
15
#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)
16
#pragma interface "strconv.h"
20
#include "wx/wxchar.h"
21
#include "wx/buffer.h"
23
#ifdef __DIGITALMARS__
27
#if defined(__VISAGECPP__) && __IBMCPP__ >= 400
35
// ----------------------------------------------------------------------------
36
// wxMBConv (abstract base class for conversions)
37
// ----------------------------------------------------------------------------
39
class WXDLLIMPEXP_BASE wxMBConv
42
// the actual conversion takes place here
44
// note that outputSize is the size of the output buffer, not the length of input
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;
50
const wxWCharBuffer cMB2WC(const char *psz) const;
51
const wxCharBuffer cWC2MB(const wchar_t *psz) const;
53
// MB <-> WC for strings with embedded null characters
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;
60
// convenience functions for converting MB or WC to/from wxWin default
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; }
65
const wchar_t* cWX2WC(const wchar_t *psz) const { return psz; }
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
73
// virtual dtor for any base class
77
// ----------------------------------------------------------------------------
78
// wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for
79
// conversion (hence it depends on the current locale)
80
// ----------------------------------------------------------------------------
82
class WXDLLIMPEXP_BASE wxMBConvLibc : public wxMBConv
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;
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
// ----------------------------------------------------------------------------
98
class WXDLLIMPEXP_BASE wxConvBrokenFileNames : public wxMBConv
101
wxConvBrokenFileNames(const wxChar *charset);
102
virtual ~wxConvBrokenFileNames() { delete m_conv; }
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;
108
// the conversion object we forward to
114
// ----------------------------------------------------------------------------
115
// wxMBConvUTF7 (for conversion using UTF7 encoding)
116
// ----------------------------------------------------------------------------
118
class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
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;
125
// ----------------------------------------------------------------------------
126
// wxMBConvUTF8 (for conversion using UTF8 encoding)
127
// ----------------------------------------------------------------------------
129
class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv
133
MAP_INVALID_UTF8_NOT = 0,
134
MAP_INVALID_UTF8_TO_PUA = 1,
135
MAP_INVALID_UTF8_TO_OCTAL = 2
138
wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { }
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;
146
// ----------------------------------------------------------------------------
147
// wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
148
// ----------------------------------------------------------------------------
150
class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConv
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;
157
// ----------------------------------------------------------------------------
158
// wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
159
// ----------------------------------------------------------------------------
161
class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConv
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;
168
// ----------------------------------------------------------------------------
169
// wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding)
170
// ----------------------------------------------------------------------------
172
class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConv
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;
179
// ----------------------------------------------------------------------------
180
// wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding)
181
// ----------------------------------------------------------------------------
183
class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConv
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;
190
// ----------------------------------------------------------------------------
191
// wxCSConv (for conversion based on loadable char sets)
192
// ----------------------------------------------------------------------------
194
#include "wx/fontenc.h"
196
class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
199
// we can be created either from charset name or from an encoding constant
200
// but we can't have both at once
201
wxCSConv(const wxChar *charset);
202
wxCSConv(wxFontEncoding encoding);
204
wxCSConv(const wxCSConv& conv);
207
wxCSConv& operator=(const wxCSConv& conv);
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;
215
// common part of all ctors
218
// creates m_convReal if necessary
219
void CreateConvIfNeeded() const;
221
// do create m_convReal (unconditionally)
222
wxMBConv *DoCreate() const;
224
// set the name (may be only called when m_name == NULL), makes copy of
226
void SetName(const wxChar *charset);
229
// note that we can't use wxString here because of compilation
230
// dependencies: we're included from wx/string.h
232
wxFontEncoding m_encoding;
234
// use CreateConvIfNeeded() before accessing m_convReal!
235
wxMBConv *m_convReal;
240
// ----------------------------------------------------------------------------
241
// declare predefined conversion objects
242
// ----------------------------------------------------------------------------
244
// conversion to be used with all standard functions affected by locale, e.g.
245
// strtol(), strftime(), ...
246
extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc;
248
// conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
249
extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1;
250
extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7;
251
extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8;
253
// conversion used for the file names on the systems where they're not Unicode
254
// (basically anything except Windows)
256
// this is used by all file functions, can be changed by the application
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;
262
// backwards compatible define
263
#define wxConvFile (*wxConvFileName)
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)
267
extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
270
extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal;
273
// ----------------------------------------------------------------------------
274
// endianness-dependent conversions
275
// ----------------------------------------------------------------------------
277
#ifdef WORDS_BIGENDIAN
278
typedef wxMBConvUTF16BE wxMBConvUTF16;
279
typedef wxMBConvUTF32BE wxMBConvUTF32;
281
typedef wxMBConvUTF16LE wxMBConvUTF16;
282
typedef wxMBConvUTF32LE wxMBConvUTF32;
285
// ----------------------------------------------------------------------------
286
// filename conversion macros
287
// ----------------------------------------------------------------------------
289
// filenames are multibyte on Unix and probably widechar on Windows?
290
#if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ )
296
#if wxMBFILES && wxUSE_UNICODE
297
#define wxFNCONV(name) wxConvFileName->cWX2MB(name)
298
#define wxFNSTRINGCAST wxMBSTRINGCAST
300
#if defined( __WXOSX__ ) && wxMBFILES
301
#define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
303
#define wxFNCONV(name) name
305
#define wxFNSTRINGCAST WXSTRINGCAST
308
#else // !wxUSE_WCHAR_T
310
// ----------------------------------------------------------------------------
311
// stand-ins in absence of wchar_t
312
// ----------------------------------------------------------------------------
314
class WXDLLIMPEXP_BASE wxMBConv
317
const char* cMB2WX(const char *psz) const { return psz; }
318
const char* cWX2MB(const char *psz) const { return psz; }
321
#define wxConvFile wxConvLocal
323
extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
327
extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
329
#define wxFNCONV(name) name
330
#define wxFNSTRINGCAST WXSTRINGCAST
335
// ----------------------------------------------------------------------------
336
// macros for the most common conversions
337
// ----------------------------------------------------------------------------
340
#define wxConvertWX2MB(s) wxConvCurrent->cWX2MB(s)
341
#define wxConvertMB2WX(s) wxConvCurrent->cMB2WX(s)
343
// no conversions to do
344
#define wxConvertWX2MB(s) (s)
345
#define wxConvertMB2WX(s) (s)
346
#endif // Unicode/ANSI