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 class WXDLLIMPEXP_BASE wxConvAuto : public wxMBConv
  22 {
  23 public:
  24     // default ctor, the real conversion will be created on demand
  25     wxConvAuto(wxFontEncoding enc = wxFONTENCODING_DEFAULT)
  26     {
  27         Init();
  28
  29         m_encDefault = enc;
  30     }
  31
  32     // copy ctor doesn't initialize anything neither as conversion can only be
  33     // deduced on first use
  34     wxConvAuto(const wxConvAuto& other) : wxMBConv()
  35     {
  36         Init();
  37
  38         m_encDefault = other.m_encDefault;
  39     }
  40
  41     virtual ~wxConvAuto()
  42     {
  43         if ( m_ownsConv )
  44             delete m_conv;
  45     }
  46
  47     // get/set the fall-back encoding used when the input text doesn't have BOM
  48     // and isn't UTF-8
  49     //
  50     // special values are wxFONTENCODING_MAX meaning not to use any fall back
  51     // at all (but just fail to convert in this case) and wxFONTENCODING_SYSTEM
  52     // meaning to use the encoding of the system locale
  53     static wxFontEncoding GetFallbackEncoding() { return ms_defaultMBEncoding; }
  54     static void SetFallbackEncoding(wxFontEncoding enc);
  55     static void DisableFallbackEncoding()
  56     {
  57         SetFallbackEncoding(wxFONTENCODING_MAX);
  58     }
  59
  60
  61     // override the base class virtual function(s) to use our m_conv
  62     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
  63                            const char *src, size_t srcLen = wxNO_LEN) const;
  64
  65     virtual size_t FromWChar(char *dst, size_t dstLen,
  66                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
  67
  68     virtual size_t GetMBNulLen() const { return m_conv->GetMBNulLen(); }
  69
  70     virtual wxMBConv *Clone() const { return new wxConvAuto(*this); }
  71
  72 private:
  73     // all currently recognized BOM values
  74     enum BOMType
  75     {
  76         BOM_Unknown = -1,
  77         BOM_None,
  78         BOM_UTF32BE,
  79         BOM_UTF32LE,
  80         BOM_UTF16BE,
  81         BOM_UTF16LE,
  82         BOM_UTF8
  83     };
  84
  85     // return the BOM type of this buffer
  86     static BOMType DetectBOM(const char *src, size_t srcLen);
  87
  88     // common part of all ctors
  89     void Init()
  90     {
  91         // no need to initialize m_bomType and m_consumedBOM here, this will be
  92         // done when m_conv is created
  93         m_conv = NULL;
  94         m_ownsConv = false;
  95     }
  96
  97     // initialize m_conv with the UTF-8 conversion
  98     void InitWithUTF8()
  99     {
 100         m_conv = &wxConvUTF8;
 101         m_ownsConv = false;
 102     }
 103
 104     // create the correct conversion object for the given BOM type
 105     void InitFromBOM(BOMType bomType);
 106
 107     // create the correct conversion object for the BOM present in the
 108     // beginning of the buffer
 109     //
 110     // return false if the buffer is too short to allow us to determine if we
 111     // have BOM or not
 112     bool InitFromInput(const char *src, size_t len);
 113
 114     // adjust src and len to skip over the BOM (identified by m_bomType) at the
 115     // start of the buffer
 116     void SkipBOM(const char **src, size_t *len) const;
 117
 118
 119     // fall-back multibyte encoding to use, may be wxFONTENCODING_SYSTEM or
 120     // wxFONTENCODING_MAX but not wxFONTENCODING_DEFAULT
 121     static wxFontEncoding ms_defaultMBEncoding;
 122
 123     // conversion object which we really use, NULL until the first call to
 124     // either ToWChar() or FromWChar()
 125     wxMBConv *m_conv;
 126
 127     // the multibyte encoding to use by default if input isn't Unicode
 128     wxFontEncoding m_encDefault;
 129
 130     // our BOM type
 131     BOMType m_bomType;
 132
 133     // true if we allocated m_conv ourselves, false if we just use an existing
 134     // global conversion
 135     bool m_ownsConv;
 136
 137     // true if we already skipped BOM when converting (and not just calculating
 138     // the size)
 139     bool m_consumedBOM;
 140
 141
 142     wxDECLARE_NO_ASSIGN_CLASS(wxConvAuto);
 143 };
 144
 145 #endif // _WX_CONVAUTO_H_
 146