src/common/encconv.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        encconv.cpp
   3 // Purpose:     wxEncodingConverter class for converting between different
   4 //              font encodings
   5 // Author:      Vaclav Slavik
   6 // Copyright:   (c) 1999 Vaclav Slavik
   7 // Licence:     wxWindows Licence
   8 /////////////////////////////////////////////////////////////////////////////
   9
  10 #ifdef __GNUG__
  11 #pragma implementation "encconv.h"
  12 #endif
  13
  14 // For compilers that support precompilation, includes "wx.h".
  15 #include "wx/wxprec.h"
  16
  17 #ifdef __BORLANDC__
  18   #pragma hdrstop
  19 #endif
  20
  21 #include "wx/encconv.h"
  22
  23 #include <stdlib.h>
  24
  25 #include "unictabl.inc"
  26 // conversion tables, generated by scripts in $(WXWIN)/misc/unictabl
  27
  28
  29 static wxUint16 *GetEncTable(wxFontEncoding enc)
  30 {
  31     for (int i = 0; encodings_list[i].table != NULL; i++)
  32     {
  33         if (encodings_list[i].encoding == enc)
  34             return encodings_list[i].table;
  35     }
  36     return NULL;
  37 }
  38
  39 typedef struct {
  40     wxUint16 u;
  41     wxUint8  c;
  42 } CharsetItem;
  43
  44
  45
  46 static int CompareCharsetItems(const void *i1, const void *i2)
  47 {
  48     return ( ((CharsetItem*)i1) -> u - ((CharsetItem*)i2) -> u );
  49 }
  50
  51
  52 static CharsetItem* BuildReverseTable(wxUint16 *tbl)
  53 {
  54     CharsetItem *rev = new CharsetItem[128];
  55
  56     for (int i = 0; i < 128; i++)
  57         rev[i].c = 128 + i, rev[i].u = tbl[i];
  58
  59     qsort(rev, 128, sizeof(CharsetItem), CompareCharsetItems);
  60
  61     return rev;
  62 }
  63
  64
  65
  66 wxEncodingConverter::wxEncodingConverter()
  67 {
  68     m_Table = NULL;
  69     m_UnicodeInput = FALSE;
  70     m_JustCopy = FALSE;
  71 }
  72
  73
  74
  75 bool wxEncodingConverter::Init(wxFontEncoding input_enc, wxFontEncoding output_enc, int method)
  76 {
  77     unsigned i;
  78     wxUint16 *in_tbl = NULL, *out_tbl = NULL;
  79
  80     if (m_Table) {delete[] m_Table; m_Table = NULL;}
  81
  82 #if !wxUSE_UNICODE
  83     if (input_enc == wxFONTENCODING_UNICODE || output_enc == wxFONTENCODING_UNICODE) return FALSE;
  84 #endif
  85
  86     if (input_enc == output_enc) {m_JustCopy = TRUE; return TRUE;}
  87
  88     m_JustCopy = FALSE;
  89
  90     if (input_enc == wxFONTENCODING_UNICODE)
  91     {
  92         if ((out_tbl = GetEncTable(output_enc)) == NULL) return FALSE;
  93
  94         m_Table = new wxChar[65536];
  95         for (i = 0; i < 128; i++)  m_Table[i] = (wxChar)i; // 7bit ASCII
  96         for (i = 128; i < 65536; i++)  m_Table[i] = (wxChar)'?';
  97                 // FIXME - this should be character that means `unicode to charset' impossible, not '?'
  98
  99         if (method == wxCONVERT_SUBSTITUTE)
 100         {
 101             for (i = 0; i < encoding_unicode_fallback_count; i++)
 102                 m_Table[encoding_unicode_fallback[i].c] = (wxChar) encoding_unicode_fallback[i].s;
 103         }
 104
 105         for (i = 0; i < 128; i++)
 106             m_Table[out_tbl[i]] = (wxChar)(128 + i);
 107
 108         m_UnicodeInput = TRUE;
 109         return TRUE;
 110     }
 111
 112     else
 113     {
 114         if ((in_tbl = GetEncTable(input_enc)) == NULL) return FALSE;
 115         if (output_enc != wxFONTENCODING_UNICODE)
 116             if ((out_tbl = GetEncTable(output_enc)) == NULL) return FALSE;
 117
 118         m_UnicodeInput = FALSE;
 119
 120         m_Table = new wxChar[256];
 121         for (i = 0; i < 128; i++)  m_Table[i] = (wxChar)i; // 7bit ASCII
 122
 123         if (output_enc == wxFONTENCODING_UNICODE)
 124         {
 125             for (i = 0; i < 128; i++)  m_Table[128 + i] = (wxChar)in_tbl[i]; // wxChar is 2byte now
 126             return TRUE;
 127         }
 128         else
 129         {
 130             CharsetItem *rev = BuildReverseTable(out_tbl);
 131             CharsetItem *item, key;
 132
 133             for (i = 0; i < 128; i++)
 134             {
 135                 key.u = in_tbl[i];
 136                 item = (CharsetItem*) bsearch(&key, rev, 128, sizeof(CharsetItem), CompareCharsetItems);
 137                 if (item == NULL && method == wxCONVERT_SUBSTITUTE)
 138                     item = (CharsetItem*) bsearch(&key, encoding_unicode_fallback,
 139                                 encoding_unicode_fallback_count, sizeof(CharsetItem), CompareCharsetItems);
 140                 if (item)
 141                     m_Table[128 + i] = (wxChar)item -> c;
 142                 else
 143                     m_Table[128 + i] = 128 + i; // don't know => don't touch
 144             }
 145
 146             delete[] rev;
 147             return TRUE;
 148         }
 149     }
 150 }
 151
 152
 153 void wxEncodingConverter::Convert(const wxChar* input, wxChar* output)
 154 {
 155     if (m_JustCopy)
 156     {
 157         wxStrcpy(output, input);
 158         return;
 159     }
 160
 161     wxASSERT_MSG(m_Table != NULL, wxT("You must call wxEncodingConverter::Init() before actually converting!"));
 162
 163     const wxChar *i;
 164     wxChar *o;
 165
 166     if (m_UnicodeInput)
 167         for (i = input, o = output; *i != 0; i++, o++)
 168             *o = (wxChar)(m_Table[(wxUint16)*i]);
 169     else
 170         for (i = input, o = output; *i != 0; i++, o++)
 171             *o = (wxChar)(m_Table[(wxUint8)*i]);
 172     *o = 0;
 173 }
 174
 175
 176
 177 wxString wxEncodingConverter::Convert(const wxString& input)
 178 {
 179     if (m_JustCopy) return input;
 180
 181     wxString s;
 182     const wxChar *i;
 183
 184     wxASSERT_MSG(m_Table != NULL, wxT("You must call wxEncodingConverter::Init() before actually converting!"));
 185
 186     if (m_UnicodeInput)
 187         for (i = input.c_str(); *i != 0; i++)
 188             s << (wxChar)(m_Table[(wxUint16)*i]);
 189     else
 190         for (i = input.c_str(); *i != 0; i++)
 191             s << (wxChar)(m_Table[(wxUint8)*i]);
 192     return s;
 193 }
 194
 195
 196
 197
 198 // Following tables describe classes of encoding equivalence.
 199 //
 200
 201 #define STOP wxFONTENCODING_SYSTEM
 202
 203 #define NUM_OF_PLATFORMS  4 /*must conform to enum wxPLATFORM_XXXX !!!*/
 204 #define ENC_PER_PLATFORM  3
 205            // max no. of encodings for one language used on one platform
 206            // Anybody thinks 3 is not enough? ;-)
 207
 208 static wxFontEncoding
 209     EquivalentEncodings[][NUM_OF_PLATFORMS][ENC_PER_PLATFORM+1] = {
 210
 211     // West European (Latin1)
 212     {
 213         /* unix    */ {wxFONTENCODING_ISO8859_1, wxFONTENCODING_ISO8859_15, STOP},
 214         /* windows */ {wxFONTENCODING_CP1252, STOP},
 215         /* os2     */ {STOP},
 216         /* mac     */ {STOP}
 217     },
 218
 219     // Central European (Latin2)
 220     {
 221         /* unix    */ {wxFONTENCODING_ISO8859_2, STOP},
 222         /* windows */ {wxFONTENCODING_CP1250, STOP},
 223         /* os2     */ {STOP},
 224         /* mac     */ {STOP}
 225     },
 226
 227     {{STOP},{STOP},{STOP},{STOP}} /* Terminator */
 228     /* no, _not_ Arnold! */
 229 };
 230
 231
 232
 233
 234 wxFontEncodingArray wxEncodingConverter::GetPlatformEquivalents(wxFontEncoding enc, int platform)
 235 {
 236     if (platform == wxPLATFORM_CURRENT)
 237     {
 238 #if defined(__WXMSW__)
 239         platform = wxPLATFORM_WINDOWS;
 240 #elif defined(__WXGTK__) || defined(__WXMOTIF__)
 241         platform = wxPLATFORM_UNIX;
 242 #elif defined(__WXOS2__)
 243         platform = wxPLATFORM_OS2;
 244 #elif defined(__WXMAC__)
 245         platform = wxPLATFORM_MAC;
 246 #endif
 247     }
 248
 249     int i, clas, e ;
 250     wxFontEncoding *f;
 251     wxFontEncodingArray arr;
 252
 253     clas = 0;
 254     while (EquivalentEncodings[clas][0][0] != STOP)
 255     {
 256         for (i = 0; i < NUM_OF_PLATFORMS; i++)
 257             for (e = 0; EquivalentEncodings[clas][i][e] != STOP; e++)
 258                 if (EquivalentEncodings[clas][i][e] == enc)
 259                 {
 260                     for (f = EquivalentEncodings[clas][platform]; *f != STOP; f++)
 261                         if (arr.Index(*f) == wxNOT_FOUND) arr.Add(*f);
 262                     i = NUM_OF_PLATFORMS/*hack*/; break;
 263                 }
 264         clas++;
 265     }
 266
 267     return arr;
 268 }
 269
 270
 271
 272 wxFontEncodingArray wxEncodingConverter::GetAllEquivalents(wxFontEncoding enc)
 273 {
 274     int i, clas, e, j ;
 275     wxFontEncoding *f;
 276     wxFontEncodingArray arr;
 277
 278     clas = 0;
 279     while (EquivalentEncodings[clas][0][0] != STOP)
 280     {
 281         for (i = 0; i < NUM_OF_PLATFORMS; i++)
 282             for (e = 0; EquivalentEncodings[clas][i][e] != STOP; e++)
 283                 if (EquivalentEncodings[clas][i][e] == enc)
 284                 {
 285                     for (j = 0; j < NUM_OF_PLATFORMS; j++)
 286                         for (f = EquivalentEncodings[clas][j]; *f != STOP; f++)
 287                             if (arr.Index(*f) == wxNOT_FOUND) arr.Add(*f);
 288                     i = NUM_OF_PLATFORMS/*hack*/; break;
 289                 }
 290         clas++;
 291     }
 292
 293     return arr;
 294 }