| 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_None, |
| 79 | BOM_UTF32BE, |
| 80 | BOM_UTF32LE, |
| 81 | BOM_UTF16BE, |
| 82 | BOM_UTF16LE, |
| 83 | BOM_UTF8 |
| 84 | }; |
| 85 | |
| 86 | // return the BOM type of this buffer |
| 87 | static BOMType DetectBOM(const char *src, size_t srcLen); |
| 88 | |
| 89 | // common part of all ctors |
| 90 | void Init() |
| 91 | { |
| 92 | // no need to initialize m_bomType and m_consumedBOM here, this will be |
| 93 | // done when m_conv is created |
| 94 | m_conv = NULL; |
| 95 | m_ownsConv = false; |
| 96 | } |
| 97 | |
| 98 | // initialize m_conv with the UTF-8 conversion |
| 99 | void InitWithUTF8() |
| 100 | { |
| 101 | m_conv = &wxConvUTF8; |
| 102 | m_ownsConv = false; |
| 103 | } |
| 104 | |
| 105 | // create the correct conversion object for the given BOM type |
| 106 | void InitFromBOM(BOMType bomType); |
| 107 | |
| 108 | // create the correct conversion object for the BOM present in the |
| 109 | // beginning of the buffer; adjust the buffer to skip the BOM if found |
| 110 | void InitFromInput(const char **src, size_t *len); |
| 111 | |
| 112 | // adjust src and len to skip over the BOM (identified by m_bomType) at the |
| 113 | // start of the buffer |
| 114 | void SkipBOM(const char **src, size_t *len) const; |
| 115 | |
| 116 | |
| 117 | // fall-back multibyte encoding to use, may be wxFONTENCODING_SYSTEM or |
| 118 | // wxFONTENCODING_MAX but not wxFONTENCODING_DEFAULT |
| 119 | static wxFontEncoding ms_defaultMBEncoding; |
| 120 | |
| 121 | // conversion object which we really use, NULL until the first call to |
| 122 | // either ToWChar() or FromWChar() |
| 123 | wxMBConv *m_conv; |
| 124 | |
| 125 | // the multibyte encoding to use by default if input isn't Unicode |
| 126 | wxFontEncoding m_encDefault; |
| 127 | |
| 128 | // our BOM type |
| 129 | BOMType m_bomType; |
| 130 | |
| 131 | // true if we allocated m_conv ourselves, false if we just use an existing |
| 132 | // global conversion |
| 133 | bool m_ownsConv; |
| 134 | |
| 135 | // true if we already skipped BOM when converting (and not just calculating |
| 136 | // the size) |
| 137 | bool m_consumedBOM; |
| 138 | |
| 139 | |
| 140 | DECLARE_NO_ASSIGN_CLASS(wxConvAuto) |
| 141 | }; |
| 142 | |
| 143 | #endif // wxUSE_WCHAR_T |
| 144 | |
| 145 | #endif // _WX_CONVAUTO_H_ |
| 146 | |