'[1219035] cleanup: miscellaneous' and minor source cleaning.

[wxWidgets.git] / include / wx / strconv.h

diff --git a/include/wx/strconv.h b/include/wx/strconv.h
index 16be8b1aee7aeb4c7f9600f5e598fb76bc5c85e6..6919583e8d52b8ec09776340a70fc4afd75840b8 100644 (file)
--- a/include/wx/strconv.h
+++ b/include/wx/strconv.h
@@ -33,20 +33,31 @@
 #if wxUSE_WCHAR_T
 
 // ----------------------------------------------------------------------------
 #if wxUSE_WCHAR_T
 
 // ----------------------------------------------------------------------------

-// wxMBConv (base class for conversions, using libc conversion itself)
+// wxMBConv (abstract base class for conversions)

 // ----------------------------------------------------------------------------
 
 class WXDLLIMPEXP_BASE wxMBConv
 {
 public:
     // the actual conversion takes place here
 // ----------------------------------------------------------------------------
 
 class WXDLLIMPEXP_BASE wxMBConv
 {
 public:
     // the actual conversion takes place here

-    virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const;
-    virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const;
+    //
+    // note that outputSize is the size of the output buffer, not the length of input
+    // (the latter is always supposed to be NUL-terminated)
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const = 0;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const = 0;

 
 

-    // No longer inline since BC++ complains.
+    // MB <-> WC

     const wxWCharBuffer cMB2WC(const char *psz) const;
     const wxCharBuffer cWC2MB(const wchar_t *psz) const;
 
     const wxWCharBuffer cMB2WC(const char *psz) const;
     const wxCharBuffer cWC2MB(const wchar_t *psz) const;
 

+    // MB <-> WC for strings with embedded null characters
+    //
+    // pszLen length of the input string
+    // pOutSize gets the final size of the converted string
+    const wxWCharBuffer cMB2WC(const char *psz, size_t pszLen, size_t* pOutSize) const;
+    const wxCharBuffer cWC2MB(const wchar_t *psz, size_t pszLen, size_t* pOutSize) const;
+
+    // convenience functions for converting MB or WC to/from wxWin default

 #if wxUSE_UNICODE
     const wxWCharBuffer cMB2WX(const char *psz) const { return cMB2WC(psz); }
     const wxCharBuffer cWX2MB(const wchar_t *psz) const { return cWC2MB(psz); }
 #if wxUSE_UNICODE
     const wxWCharBuffer cMB2WX(const char *psz) const { return cMB2WC(psz); }
     const wxCharBuffer cWX2MB(const wchar_t *psz) const { return cWC2MB(psz); }


@@ -63,7 +74,42 @@ public:
     virtual ~wxMBConv();
 };
 
     virtual ~wxMBConv();
 };
 

-WXDLLIMPEXP_DATA_BASE(extern wxMBConv) wxConvLibc;
+// ----------------------------------------------------------------------------
+// wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for
+//              conversion (hence it depends on the current locale)
+// ----------------------------------------------------------------------------
+
+class WXDLLIMPEXP_BASE wxMBConvLibc : public wxMBConv
+{
+public:
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
+};
+
+#ifdef __UNIX__
+
+// ----------------------------------------------------------------------------
+// wxConvBrokenFileNames is made for Unix in Unicode mode when
+// files are accidentally written in an encoding which is not
+// the system encoding. Typically, the system encoding will be
+// UTF8 but there might be files stored in ISO8859-1 on disk.
+// ----------------------------------------------------------------------------
+
+class WXDLLIMPEXP_BASE wxConvBrokenFileNames : public wxMBConv
+{
+public:
+    wxConvBrokenFileNames(const wxChar *charset);
+    virtual ~wxConvBrokenFileNames() { delete m_conv; }
+
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
+
+private:
+    // the conversion object we forward to
+    wxMBConv *m_conv;
+};
+
+#endif

 
 // ----------------------------------------------------------------------------
 // wxMBConvUTF7 (for conversion using UTF7 encoding)
 
 // ----------------------------------------------------------------------------
 // wxMBConvUTF7 (for conversion using UTF7 encoding)


@@ -72,12 +118,10 @@ WXDLLIMPEXP_DATA_BASE(extern wxMBConv) wxConvLibc;
 class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
 {
 public:
 class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
 {
 public:

-    virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const;
-    virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const;
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;

 };
 
 };
 

-WXDLLIMPEXP_DATA_BASE(extern wxMBConvUTF7) wxConvUTF7;
-

 // ----------------------------------------------------------------------------
 // wxMBConvUTF8 (for conversion using UTF8 encoding)
 // ----------------------------------------------------------------------------
 // ----------------------------------------------------------------------------
 // wxMBConvUTF8 (for conversion using UTF8 encoding)
 // ----------------------------------------------------------------------------


@@ -85,48 +129,158 @@ WXDLLIMPEXP_DATA_BASE(extern wxMBConvUTF7) wxConvUTF7;
 class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv
 {
 public:
 class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv
 {
 public:

-    virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const;
-    virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const;
+    enum { 
+        MAP_INVALID_UTF8_NOT = 0,
+        MAP_INVALID_UTF8_TO_PUA = 1,
+        MAP_INVALID_UTF8_TO_OCTAL = 2
+    };
+
+    wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { }
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
+    
+private:
+    int m_options;
+};
+
+// ----------------------------------------------------------------------------
+// wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
+// ----------------------------------------------------------------------------
+
+class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConv
+{
+public:
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
+};
+
+// ----------------------------------------------------------------------------
+// wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
+// ----------------------------------------------------------------------------
+
+class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConv
+{
+public:
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
+};
+
+// ----------------------------------------------------------------------------
+// wxMBConvUCS4LE (for conversion using UTF32 Little Endian encoding)
+// ----------------------------------------------------------------------------
+
+class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConv
+{
+public:
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;

 };
 
 };
 

-WXDLLIMPEXP_DATA_BASE(extern wxMBConvUTF8) wxConvUTF8;
+// ----------------------------------------------------------------------------
+// wxMBConvUCS4BE (for conversion using UTF32 Big Endian encoding)
+// ----------------------------------------------------------------------------
+
+class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConv
+{
+public:
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
+};

 
 // ----------------------------------------------------------------------------
 // wxCSConv (for conversion based on loadable char sets)
 // ----------------------------------------------------------------------------
 
 
 // ----------------------------------------------------------------------------
 // wxCSConv (for conversion based on loadable char sets)
 // ----------------------------------------------------------------------------
 

-class WXDLLIMPEXP_BASE wxCharacterSet;
+#include "wx/fontenc.h"

 
 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
 {
 public:
 
 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
 {
 public:

+    // we can be created either from charset name or from an encoding constant
+    // but we can't have both at once

     wxCSConv(const wxChar *charset);
     wxCSConv(const wxChar *charset);

+    wxCSConv(wxFontEncoding encoding);
+

     wxCSConv(const wxCSConv& conv);
     virtual ~wxCSConv();
 
     wxCSConv& operator=(const wxCSConv& conv);
 
     wxCSConv(const wxCSConv& conv);
     virtual ~wxCSConv();
 
     wxCSConv& operator=(const wxCSConv& conv);
 

-    void LoadNow();
-
-    virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const;
-    virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const;
+    virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
+    virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;

 
     void Clear() ;
 
 private:
 
     void Clear() ;
 
 private:

+    // common part of all ctors
+    void Init();
+
+    // creates m_convReal if necessary
+    void CreateConvIfNeeded() const;
+
+    // do create m_convReal (unconditionally)
+    wxMBConv *DoCreate() const;
+
+    // set the name (may be only called when m_name == NULL), makes copy of
+    // charset string

     void SetName(const wxChar *charset);
 
     void SetName(const wxChar *charset);
 

+

     // note that we can't use wxString here because of compilation
     // dependencies: we're included from wx/string.h
     wxChar *m_name;
     // note that we can't use wxString here because of compilation
     // dependencies: we're included from wx/string.h
     wxChar *m_name;

-    wxCharacterSet *m_cset;
+    wxFontEncoding m_encoding;
+
+    // use CreateConvIfNeeded() before accessing m_convReal!
+    wxMBConv *m_convReal;

     bool m_deferred;
 };
 
     bool m_deferred;
 };
 

-#define wxConvFile wxConvLocal
-WXDLLIMPEXP_DATA_BASE(extern wxCSConv) wxConvLocal;
-WXDLLIMPEXP_DATA_BASE(extern wxCSConv) wxConvISO8859_1;
-WXDLLIMPEXP_DATA_BASE(extern wxMBConv *) wxConvCurrent;
+
+// ----------------------------------------------------------------------------
+// declare predefined conversion objects
+// ----------------------------------------------------------------------------
+
+// conversion to be used with all standard functions affected by locale, e.g.
+// strtol(), strftime(), ...
+extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc;
+
+// conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
+extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1;
+extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7;
+extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8;
+
+// conversion used for the file names on the systems where they're not Unicode
+// (basically anything except Windows)
+//
+// this is used by all file functions, can be changed by the application
+//
+// by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
+// under Windows normally)
+extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName;
+
+// backwards compatible define
+#define wxConvFile (*wxConvFileName)
+
+// the current conversion object, may be set to any conversion, is used by
+// default in a couple of places inside wx (initially same as wxConvLibc)
+extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
+
+// ???
+extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal;
+
+
+// ----------------------------------------------------------------------------
+// endianness-dependent conversions
+// ----------------------------------------------------------------------------
+
+#ifdef WORDS_BIGENDIAN
+    typedef wxMBConvUTF16BE wxMBConvUTF16;
+    typedef wxMBConvUTF32BE wxMBConvUTF32;
+#else
+    typedef wxMBConvUTF16LE wxMBConvUTF16;
+    typedef wxMBConvUTF32LE wxMBConvUTF32;
+#endif

 
 // ----------------------------------------------------------------------------
 // filename conversion macros
 
 // ----------------------------------------------------------------------------
 // filename conversion macros


@@ -140,15 +294,18 @@ WXDLLIMPEXP_DATA_BASE(extern wxMBConv *) wxConvCurrent;
 #endif
 
 #if wxMBFILES && wxUSE_UNICODE
 #endif
 
 #if wxMBFILES && wxUSE_UNICODE

-    #define wxFNCONV(name) wxConvFile.cWX2MB(name)
+    #define wxFNCONV(name) wxConvFileName->cWX2MB(name)

     #define wxFNSTRINGCAST wxMBSTRINGCAST
     #define wxFNSTRINGCAST wxMBSTRINGCAST

+#else
+#if defined( __WXOSX__ ) && wxMBFILES
+    #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )

 #else
     #define wxFNCONV(name) name
 #else
     #define wxFNCONV(name) name

+#endif

     #define wxFNSTRINGCAST WXSTRINGCAST
 #endif
 
     #define wxFNSTRINGCAST WXSTRINGCAST
 #endif
 

-#else
-  // !wxUSE_WCHAR_T
+#else // !wxUSE_WCHAR_T

 
 // ----------------------------------------------------------------------------
 // stand-ins in absence of wchar_t
 
 // ----------------------------------------------------------------------------
 // stand-ins in absence of wchar_t


@@ -161,8 +318,13 @@ public:
     const char* cWX2MB(const char *psz) const { return psz; }
 };
 
     const char* cWX2MB(const char *psz) const { return psz; }
 };
 

-WXDLLIMPEXP_DATA_BASE(extern wxMBConv) wxConvLibc, wxConvFile, wxConvLocal, wxConvISO8859_1, wxConvUTF8;
-WXDLLIMPEXP_DATA_BASE(extern wxMBConv *) wxConvCurrent;
+#define wxConvFile wxConvLocal
+
+extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
+                                       wxConvLocal,
+                                       wxConvISO8859_1,
+                                       wxConvUTF8;
+extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;

 
 #define wxFNCONV(name) name
 #define wxFNSTRINGCAST WXSTRINGCAST
 
 #define wxFNCONV(name) name
 #define wxFNSTRINGCAST WXSTRINGCAST