include/wx/convauto.h

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        wx/convauto.h
   3 // Purpose:     wxConvAuto class declaration
   4 // Author:      Vadim Zeitlin
   5 // Created:     2006-04-03
   6 // RCS-ID:      $Id$
   7 // Copyright:   (c) 2006 Vadim Zeitlin
   8 // Licence:     wxWindows licence
   9 ///////////////////////////////////////////////////////////////////////////////
  10
  11 #ifndef _WX_CONVAUTO_H_
  12 #define _WX_CONVAUTO_H_
  13
  14 #include "wx/strconv.h"
  15 #include "wx/fontenc.h"
  16
  17 // ----------------------------------------------------------------------------
  18 // wxConvAuto: uses BOM to automatically detect input encoding
  19 // ----------------------------------------------------------------------------
  20
  21 // All currently recognized BOM values.
  22 enum wxBOM
  23 {
  24     wxBOM_Unknown = -1,
  25     wxBOM_None,
  26     wxBOM_UTF32BE,
  27     wxBOM_UTF32LE,
  28     wxBOM_UTF16BE,
  29     wxBOM_UTF16LE,
  30     wxBOM_UTF8
  31 };
  32
  33 class WXDLLIMPEXP_BASE wxConvAuto : public wxMBConv
  34 {
  35 public:
  36     // default ctor, the real conversion will be created on demand
  37     wxConvAuto(wxFontEncoding enc = wxFONTENCODING_DEFAULT)
  38     {
  39         Init();
  40
  41         m_encDefault = enc;
  42     }
  43
  44     // copy ctor doesn't initialize anything neither as conversion can only be
  45     // deduced on first use
  46     wxConvAuto(const wxConvAuto& other) : wxMBConv()
  47     {
  48         Init();
  49
  50         m_encDefault = other.m_encDefault;
  51     }
  52
  53     virtual ~wxConvAuto()
  54     {
  55         if ( m_ownsConv )
  56             delete m_conv;
  57     }
  58
  59     // get/set the fall-back encoding used when the input text doesn't have BOM
  60     // and isn't UTF-8
  61     //
  62     // special values are wxFONTENCODING_MAX meaning not to use any fall back
  63     // at all (but just fail to convert in this case) and wxFONTENCODING_SYSTEM
  64     // meaning to use the encoding of the system locale
  65     static wxFontEncoding GetFallbackEncoding() { return ms_defaultMBEncoding; }
  66     static void SetFallbackEncoding(wxFontEncoding enc);
  67     static void DisableFallbackEncoding()
  68     {
  69         SetFallbackEncoding(wxFONTENCODING_MAX);
  70     }
  71
  72
  73     // override the base class virtual function(s) to use our m_conv
  74     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
  75                            const char *src, size_t srcLen = wxNO_LEN) const;
  76
  77     virtual size_t FromWChar(char *dst, size_t dstLen,
  78                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
  79
  80     virtual size_t GetMBNulLen() const { return m_conv->GetMBNulLen(); }
  81
  82     virtual wxMBConv *Clone() const { return new wxConvAuto(*this); }
  83
  84     // return the BOM type of this buffer
  85     static wxBOM DetectBOM(const char *src, size_t srcLen);
  86
  87     // return the characters composing the given BOM.
  88     static const char* GetBOMChars(wxBOM bomType, size_t* count);
  89
  90     wxBOM GetBOM() const
  91     {
  92         return m_bomType;
  93     }
  94
  95 private:
  96     // common part of all ctors
  97     void Init()
  98     {
  99         // We don't initialize m_encDefault here as different ctors do it
 100         // differently.
 101         m_conv = NULL;
 102         m_bomType = wxBOM_Unknown;
 103         m_ownsConv = false;
 104         m_consumedBOM = false;
 105     }
 106
 107     // initialize m_conv with the UTF-8 conversion
 108     void InitWithUTF8()
 109     {
 110         m_conv = &wxConvUTF8;
 111         m_ownsConv = false;
 112     }
 113
 114     // create the correct conversion object for the given BOM type
 115     void InitFromBOM(wxBOM bomType);
 116
 117     // create the correct conversion object for the BOM present in the
 118     // beginning of the buffer
 119     //
 120     // return false if the buffer is too short to allow us to determine if we
 121     // have BOM or not
 122     bool InitFromInput(const char *src, size_t len);
 123
 124     // adjust src and len to skip over the BOM (identified by m_bomType) at the
 125     // start of the buffer
 126     void SkipBOM(const char **src, size_t *len) const;
 127
 128
 129     // fall-back multibyte encoding to use, may be wxFONTENCODING_SYSTEM or
 130     // wxFONTENCODING_MAX but not wxFONTENCODING_DEFAULT
 131     static wxFontEncoding ms_defaultMBEncoding;
 132
 133     // conversion object which we really use, NULL until the first call to
 134     // either ToWChar() or FromWChar()
 135     wxMBConv *m_conv;
 136
 137     // the multibyte encoding to use by default if input isn't Unicode
 138     wxFontEncoding m_encDefault;
 139
 140     // our BOM type
 141     wxBOM m_bomType;
 142
 143     // true if we allocated m_conv ourselves, false if we just use an existing
 144     // global conversion
 145     bool m_ownsConv;
 146
 147     // true if we already skipped BOM when converting (and not just calculating
 148     // the size)
 149     bool m_consumedBOM;
 150
 151
 152     wxDECLARE_NO_ASSIGN_CLASS(wxConvAuto);
 153 };
 154
 155 #endif // _WX_CONVAUTO_H_
 156