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