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         m_conv = NULL; // the rest will be initialized later
  30         m_encDefault = enc;
  31     }
  32
  33     // copy ctor doesn't initialize anything neither as conversion can only be
  34     // deduced on first use
  35     wxConvAuto(const wxConvAuto& other) : wxMBConv()
  36     {
  37         m_conv = NULL;
  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_None,
  77         BOM_UTF32BE,
  78         BOM_UTF32LE,
  79         BOM_UTF16BE,
  80         BOM_UTF16LE,
  81         BOM_UTF8
  82     };
  83
  84     // return the BOM type of this buffer
  85     static BOMType DetectBOM(const char *src, size_t srcLen);
  86
  87     // initialize m_conv with the UTF-8 conversion
  88     void InitWithUTF8()
  89     {
  90         m_conv = &wxConvUTF8;
  91         m_ownsConv = false;
  92     }
  93
  94     // create the correct conversion object for the given BOM type
  95     void InitFromBOM(BOMType bomType);
  96
  97     // create the correct conversion object for the BOM present in the
  98     // beginning of the buffer; adjust the buffer to skip the BOM if found
  99     void InitFromInput(const char **src, size_t *len);
 100
 101     // adjust src and len to skip over the BOM (identified by m_bomType) at the
 102     // start of the buffer
 103     void SkipBOM(const char **src, size_t *len) const;
 104
 105
 106     // fall-back multibyte encoding to use, may be wxFONTENCODING_SYSTEM or
 107     // wxFONTENCODING_MAX but not wxFONTENCODING_DEFAULT
 108     static wxFontEncoding ms_defaultMBEncoding;
 109
 110     // conversion object which we really use, NULL until the first call to
 111     // either ToWChar() or FromWChar()
 112     wxMBConv *m_conv;
 113
 114     // the multibyte encoding to use by default if input isn't Unicode
 115     wxFontEncoding m_encDefault;
 116
 117     // our BOM type
 118     BOMType m_bomType;
 119
 120     // true if we allocated m_conv ourselves, false if we just use an existing
 121     // global conversion
 122     bool m_ownsConv;
 123
 124     // true if we already skipped BOM when converting (and not just calculating
 125     // the size)
 126     bool m_consumedBOM;
 127
 128
 129     DECLARE_NO_ASSIGN_CLASS(wxConvAuto)
 130 };
 131
 132 #endif // wxUSE_WCHAR_T
 133
 134 #endif // _WX_CONVAUTO_H_
 135