dlls/richedit/text-writer.c

   1 /*
   2  * text-writer -- RTF-to-text translation writer code.
   3  *
   4  * Read RTF input, write text of document (text extraction).
   5  *
   6  * Wrapper must call WriterInit() once before processing any files,
   7  * then set up input and call BeginFile() for each input file.
   8  *
   9  * This installs callbacks for the text and control token classes.
  10  * The control class is necessary so that special characters such as
  11  * \par, \tab, \sect, etc.  can be converted.
  12  *
  13  * It's problematic what to do with text in headers and footers, and
  14  * what to do about tables.
  15  *
  16  * This really is quite a stupid program, for instance, it could keep
  17  * track of the current leader character and dump that out when a tab
  18  * is encountered.
  19  *
  20  * 04 Feb 91    Paul DuBois     dubois@primate.wisc.edu
  21  *
  22  * This software may be redistributed without restriction and used for
  23  * any purpose whatsoever.
  24  *
  25  * 04 Feb 91
  26  * -Created.
  27  * 27 Feb 91
  28  * - Updated for distribution 1.05.
  29  * 13 Jul 93
  30  * - Updated to compile under THINK C 6.0.
  31  * 31 Aug 93
  32  * - Added Mike Sendall's entries for Macintosh char map.
  33  * 07 Sep 93
  34  * - Uses charset map and output sequence map for character translation.
  35  * 11 Mar 94
  36  * - Updated for 1.10 distribution.
  37  */
  38
  39 #include <stdio.h>
  40
  41 #include "rtf.h"
  42 #include "rtf2text.h"
  43 #include "wine/debug.h"
  44
  45 WINE_DEFAULT_DEBUG_CHANNEL(richedit);
  46
  47 static void     TextClass (RTF_Info *info);
  48 static void     ControlClass (RTF_Info *info);
  49 static void     Destination (RTF_Info *info);
  50 static void     SpecialChar (RTF_Info *info);
  51 static void     PutStdChar (RTF_Info *info, int stdCode);
  52 static void     PutLitChar (RTF_Info *info, int c);
  53 static void     PutLitStr (RTF_Info *info, char *s);
  54
  55 /*
  56  * Initialize the writer.
  57  */
  58
  59 void
  60 WriterInit (RTF_Info *info )
  61 {
  62         RTFReadOutputMap (info, info->outMap,1);
  63 }
  64
  65
  66 int
  67 BeginFile (RTF_Info *info )
  68 {
  69         /* install class callbacks */
  70
  71         RTFSetClassCallback (info, rtfText, TextClass);
  72         RTFSetClassCallback (info, rtfControl, ControlClass);
  73
  74         return (1);
  75 }
  76
  77
  78 /*
  79  * Write out a character.  rtfMajor contains the input character, rtfMinor
  80  * contains the corresponding standard character code.
  81  *
  82  * If the input character isn't in the charset map, try to print some
  83  * representation of it.
  84  */
  85
  86 static void
  87 TextClass (RTF_Info *info)
  88 {
  89 char    buf[rtfBufSiz];
  90
  91         TRACE("\n");
  92
  93         if (info->rtfFormat == SF_TEXT)
  94                 PutLitChar (info, info->rtfMajor);
  95         else if (info->rtfMinor != rtfSC_nothing)
  96                 PutStdChar (info, info->rtfMinor);
  97         else
  98         {
  99                 if (info->rtfMajor < 128)       /* in ASCII range */
 100                         sprintf (buf, "[[%c]]", info->rtfMajor);
 101                 else
 102                         sprintf (buf, "[[\\'%02x]]", info->rtfMajor);
 103                 PutLitStr (info, buf);
 104         }
 105 }
 106
 107
 108 static void
 109 ControlClass (RTF_Info *info)
 110 {
 111         TRACE("\n");
 112         switch (info->rtfMajor)
 113         {
 114         case rtfDestination:
 115                 Destination (info);
 116                 break;
 117         case rtfSpecialChar:
 118                 SpecialChar (info);
 119                 break;
 120         }
 121 }
 122
 123
 124 /*
 125  * This function notices destinations that should be ignored
 126  * and skips to their ends.  This keeps, for instance, picture
 127  * data from being considered as plain text.
 128  */
 129
 130 static void
 131 Destination (RTF_Info *info)
 132 {
 133
 134         TRACE("\n");
 135
 136         switch (info->rtfMinor)
 137         {
 138         case rtfPict:
 139         case rtfFNContSep:
 140         case rtfFNContNotice:
 141         case rtfInfo:
 142         case rtfIndexRange:
 143         case rtfITitle:
 144         case rtfISubject:
 145         case rtfIAuthor:
 146         case rtfIOperator:
 147         case rtfIKeywords:
 148         case rtfIComment:
 149         case rtfIVersion:
 150         case rtfIDoccomm:
 151                 RTFSkipGroup (info);
 152                 break;
 153         }
 154 }
 155
 156
 157 /*
 158  * The reason these use the rtfSC_xxx thingies instead of just writing
 159  * out ' ', '-', '"', etc., is so that the mapping for these characters
 160  * can be controlled by the text-map file.
 161  */
 162
 163 void SpecialChar (RTF_Info *info)
 164 {
 165
 166         TRACE("\n");
 167
 168         switch (info->rtfMinor)
 169         {
 170         case rtfPage:
 171         case rtfSect:
 172         case rtfRow:
 173         case rtfLine:
 174         case rtfPar:
 175                 PutLitChar (info, '\n');
 176                 break;
 177         case rtfCell:
 178                 PutStdChar (info, rtfSC_space); /* make sure cells are separated */
 179                 break;
 180         case rtfNoBrkSpace:
 181                 PutStdChar (info, rtfSC_nobrkspace);
 182                 break;
 183         case rtfTab:
 184                 PutLitChar (info, '\t');
 185                 break;
 186         case rtfNoBrkHyphen:
 187                 PutStdChar (info, rtfSC_nobrkhyphen);
 188                 break;
 189         case rtfBullet:
 190                 PutStdChar (info, rtfSC_bullet);
 191                 break;
 192         case rtfEmDash:
 193                 PutStdChar (info, rtfSC_emdash);
 194                 break;
 195         case rtfEnDash:
 196                 PutStdChar (info, rtfSC_endash);
 197                 break;
 198         case rtfLQuote:
 199                 PutStdChar (info, rtfSC_quoteleft);
 200                 break;
 201         case rtfRQuote:
 202                 PutStdChar (info, rtfSC_quoteright);
 203                 break;
 204         case rtfLDblQuote:
 205                 PutStdChar (info, rtfSC_quotedblleft);
 206                 break;
 207         case rtfRDblQuote:
 208                 PutStdChar (info, rtfSC_quotedblright);
 209                 break;
 210         }
 211 }
 212
 213
 214 /*
 215  * Eventually this should keep track of the destination of the
 216  * current state and only write text when in the initial state.
 217  *
 218  * If the output sequence is unspecified in the output map, write
 219  * the character's standard name instead.  This makes map deficiencies
 220  * obvious and provides incentive to fix it. :-)
 221  */
 222
 223 void PutStdChar (RTF_Info *info, int stdCode)
 224 {
 225
 226   char  *oStr = (char *) NULL;
 227   char  buf[rtfBufSiz];
 228
 229 /*      if (stdCode == rtfSC_nothing)
 230                 RTFPanic ("Unknown character code, logic error\n");
 231 */
 232         TRACE("\n");
 233
 234         oStr = info->outMap[stdCode];
 235         if (oStr == (char *) NULL)      /* no output sequence in map */
 236         {
 237                 sprintf (buf, "[[%s]]", RTFStdCharName (info, stdCode));
 238                 oStr = buf;
 239         }
 240         PutLitStr (info, oStr);
 241 }
 242
 243 void PutLitChar (RTF_Info *info, int c)
 244 {
 245     if( info->dwOutputCount >= ( sizeof info->OutputBuffer - 1 ) )
 246         RTFFlushOutputBuffer( info );
 247     info->OutputBuffer[info->dwOutputCount++] = c;
 248 }
 249
 250 void RTFFlushOutputBuffer( RTF_Info *info )
 251 {
 252     info->OutputBuffer[info->dwOutputCount] = 0;
 253     SendMessageA( info->hwndEdit, EM_REPLACESEL, FALSE, (LPARAM) info->OutputBuffer );
 254     info->dwOutputCount = 0;
 255 }
 256
 257 static void PutLitStr (RTF_Info *info, char *str )
 258 {
 259     int len = strlen( str );
 260     if( ( len + info->dwOutputCount + 1 ) > sizeof info->OutputBuffer )
 261         RTFFlushOutputBuffer( info );
 262     if( ( len + 1 ) >= sizeof info->OutputBuffer )
 263     {
 264         SendMessageA( info->hwndEdit, EM_REPLACESEL, FALSE, (LPARAM) str );
 265         return;
 266     }
 267     strcpy( &info->OutputBuffer[info->dwOutputCount], str );
 268     info->dwOutputCount += len;
 269 }