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 #if wxUSE_WCHAR_T
  18
  19 // ----------------------------------------------------------------------------
  20 // wxConvAuto: uses BOM to automatically detect input encoding
  21 // ----------------------------------------------------------------------------
  22
  23 class WXDLLIMPEXP_BASE wxConvAuto : public wxMBConv
  24 {
  25 public:
  26     // default ctor, the real conversion will be created on demand
  27     wxConvAuto(wxFontEncoding enc = wxFONTENCODING_DEFAULT)
  28     {
  29         Init();
  30
  31         m_encDefault = enc;
  32     }
  33
  34     // copy ctor doesn't initialize anything neither as conversion can only be
  35     // deduced on first use
  36     wxConvAuto(const wxConvAuto& other) : wxMBConv()
  37     {
  38         Init();
  39
  40         m_encDefault = other.m_encDefault;
  41     }
  42
  43     virtual ~wxConvAuto()
  44     {
  45         if ( m_ownsConv )
  46             delete m_conv;
  47     }
  48
  49     // get/set the fall-back encoding used when the input text doesn't have BOM
  50     // and isn't UTF-8
  51     //
  52     // special values are wxFONTENCODING_MAX meaning not to use any fall back
  53     // at all (but just fail to convert in this case) and wxFONTENCODING_SYSTEM
  54     // meaning to use the encoding of the system locale
  55     static wxFontEncoding GetFallbackEncoding() { return ms_defaultMBEncoding; }
  56     static void SetFallbackEncoding(wxFontEncoding enc);
  57     static void DisableFallbackEncoding()
  58     {
  59         SetFallbackEncoding(wxFONTENCODING_MAX);
  60     }
  61
  62
  63     // override the base class virtual function(s) to use our m_conv
  64     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
  65                            const char *src, size_t srcLen = wxNO_LEN) const;
  66
  67     virtual size_t FromWChar(char *dst, size_t dstLen,
  68                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
  69
  70     virtual size_t GetMBNulLen() const { return m_conv->GetMBNulLen(); }
  71
  72     virtual wxMBConv *Clone() const { return new wxConvAuto(*this); }
  73
  74 private:
  75     // all currently recognized BOM values
  76     enum BOMType
  77     {
  78         BOM_Unknown = -1,
  79         BOM_None,
  80         BOM_UTF32BE,
  81         BOM_UTF32LE,
  82         BOM_UTF16BE,
  83         BOM_UTF16LE,
  84         BOM_UTF8
  85     };
  86
  87     // return the BOM type of this buffer
  88     static BOMType DetectBOM(const char *src, size_t srcLen);
  89
  90     // common part of all ctors
  91     void Init()
  92     {
  93         // no need to initialize m_bomType and m_consumedBOM here, this will be
  94         // done when m_conv is created
  95         m_conv = NULL;
  96         m_ownsConv = false;
  97     }
  98
  99     // initialize m_conv with the UTF-8 conversion
 100     void InitWithUTF8()
 101     {
 102         m_conv = &wxConvUTF8;
 103         m_ownsConv = false;
 104     }
 105
 106     // create the correct conversion object for the given BOM type
 107     void InitFromBOM(BOMType bomType);
 108
 109     // create the correct conversion object for the BOM present in the
 110     // beginning of the buffer
 111     //
 112     // return false if the buffer is too short to allow us to determine if we
 113     // have BOM or not
 114     bool InitFromInput(const char *src, size_t len);
 115
 116     // adjust src and len to skip over the BOM (identified by m_bomType) at the
 117     // start of the buffer
 118     void SkipBOM(const char **src, size_t *len) const;
 119
 120
 121     // fall-back multibyte encoding to use, may be wxFONTENCODING_SYSTEM or
 122     // wxFONTENCODING_MAX but not wxFONTENCODING_DEFAULT
 123     static wxFontEncoding ms_defaultMBEncoding;
 124
 125     // conversion object which we really use, NULL until the first call to
 126     // either ToWChar() or FromWChar()
 127     wxMBConv *m_conv;
 128
 129     // the multibyte encoding to use by default if input isn't Unicode
 130     wxFontEncoding m_encDefault;
 131
 132     // our BOM type
 133     BOMType m_bomType;
 134
 135     // true if we allocated m_conv ourselves, false if we just use an existing
 136     // global conversion
 137     bool m_ownsConv;
 138
 139     // true if we already skipped BOM when converting (and not just calculating
 140     // the size)
 141     bool m_consumedBOM;
 142
 143
 144     wxDECLARE_NO_ASSIGN_CLASS(wxConvAuto);
 145 };
 146
 147 #endif // wxUSE_WCHAR_T
 148
 149 #endif // _WX_CONVAUTO_H_
 150