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 "charlist.h"
  44 #include "wine/debug.h"
  45
  46 WINE_DEFAULT_DEBUG_CHANNEL(richedit);
  47
  48 static void     TextClass (RTF_Info *info);
  49 static void     ControlClass (RTF_Info *info);
  50 static void     Destination (RTF_Info *info);
  51 static void     SpecialChar (RTF_Info *info);
  52 static void     PutStdChar (RTF_Info *info, int stdCode);
  53 static void     PutLitChar (RTF_Info *info, int c);
  54 static void     PutLitStr (RTF_Info *info, char *s);
  55
  56 #if 0
  57 static char     *outMap[rtfSC_MaxChar];
  58
  59 static CHARLIST charlist = {0, NULL, NULL};
  60 #endif
  61
  62 /*int RTFToBuffer(char* pBuffer, int nBufferSize); */
  63 int RTFToBuffer(RTF_Info *info, char* pBuffer, int nBufferSize)
  64 {
  65
  66    /* check if the buffer is big enough to hold all characters  */
  67    /* we require one more for the '\0'                          */
  68
  69    TRACE("\n");
  70
  71    if(nBufferSize < info->charlist.nCount + 1) {
  72         return info->charlist.nCount + CHARLIST_CountChar(&info->charlist, '\n') + 1;
  73    }
  74
  75    while(info->charlist.nCount)
  76    {
  77        *pBuffer = CHARLIST_Dequeue(&info->charlist);
  78        if(*pBuffer=='\n')
  79        {
  80          *pBuffer = '\r';
  81          pBuffer++;
  82          *pBuffer = '\n';
  83        }
  84        pBuffer++;
  85    }
  86    *pBuffer = '\0';
  87
  88    return 0;
  89 }
  90
  91
  92 /*
  93  * Initialize the writer.
  94  */
  95
  96 void
  97 WriterInit (RTF_Info *info )
  98 {
  99         RTFReadOutputMap (info, info->outMap,1);
 100 }
 101
 102
 103 int
 104 BeginFile (RTF_Info *info )
 105 {
 106         /* install class callbacks */
 107
 108         RTFSetClassCallback (info, rtfText, TextClass);
 109         RTFSetClassCallback (info, rtfControl, ControlClass);
 110
 111         return (1);
 112 }
 113
 114
 115 /*
 116  * Write out a character.  rtfMajor contains the input character, rtfMinor
 117  * contains the corresponding standard character code.
 118  *
 119  * If the input character isn't in the charset map, try to print some
 120  * representation of it.
 121  */
 122
 123 static void
 124 TextClass (RTF_Info *info)
 125 {
 126 char    buf[rtfBufSiz];
 127
 128         TRACE("\n");
 129
 130         if (info->rtfFormat == SF_TEXT)
 131                 PutLitChar (info, info->rtfMajor);
 132         else if (info->rtfMinor != rtfSC_nothing)
 133                 PutStdChar (info, info->rtfMinor);
 134         else
 135         {
 136                 if (info->rtfMajor < 128)       /* in ASCII range */
 137                         sprintf (buf, "[[%c]]", info->rtfMajor);
 138                 else
 139                         sprintf (buf, "[[\\'%02x]]", info->rtfMajor);
 140                 PutLitStr (info, buf);
 141         }
 142 }
 143
 144
 145 static void
 146 ControlClass (RTF_Info *info)
 147 {
 148         TRACE("\n");
 149         switch (info->rtfMajor)
 150         {
 151         case rtfDestination:
 152                 Destination (info);
 153                 break;
 154         case rtfSpecialChar:
 155                 SpecialChar (info);
 156                 break;
 157         }
 158 }
 159
 160
 161 /*
 162  * This function notices destinations that should be ignored
 163  * and skips to their ends.  This keeps, for instance, picture
 164  * data from being considered as plain text.
 165  */
 166
 167 static void
 168 Destination (RTF_Info *info)
 169 {
 170
 171         TRACE("\n");
 172
 173         switch (info->rtfMinor)
 174         {
 175         case rtfPict:
 176         case rtfFNContSep:
 177         case rtfFNContNotice:
 178         case rtfInfo:
 179         case rtfIndexRange:
 180         case rtfITitle:
 181         case rtfISubject:
 182         case rtfIAuthor:
 183         case rtfIOperator:
 184         case rtfIKeywords:
 185         case rtfIComment:
 186         case rtfIVersion:
 187         case rtfIDoccomm:
 188                 RTFSkipGroup (info);
 189                 break;
 190         }
 191 }
 192
 193
 194 /*
 195  * The reason these use the rtfSC_xxx thingies instead of just writing
 196  * out ' ', '-', '"', etc., is so that the mapping for these characters
 197  * can be controlled by the text-map file.
 198  */
 199
 200 void SpecialChar (RTF_Info *info)
 201 {
 202
 203         TRACE("\n");
 204
 205         switch (info->rtfMinor)
 206         {
 207         case rtfPage:
 208         case rtfSect:
 209         case rtfRow:
 210         case rtfLine:
 211         case rtfPar:
 212                 PutLitChar (info, '\n');
 213                 break;
 214         case rtfCell:
 215                 PutStdChar (info, rtfSC_space); /* make sure cells are separated */
 216                 break;
 217         case rtfNoBrkSpace:
 218                 PutStdChar (info, rtfSC_nobrkspace);
 219                 break;
 220         case rtfTab:
 221                 PutLitChar (info, '\t');
 222                 break;
 223         case rtfNoBrkHyphen:
 224                 PutStdChar (info, rtfSC_nobrkhyphen);
 225                 break;
 226         case rtfBullet:
 227                 PutStdChar (info, rtfSC_bullet);
 228                 break;
 229         case rtfEmDash:
 230                 PutStdChar (info, rtfSC_emdash);
 231                 break;
 232         case rtfEnDash:
 233                 PutStdChar (info, rtfSC_endash);
 234                 break;
 235         case rtfLQuote:
 236                 PutStdChar (info, rtfSC_quoteleft);
 237                 break;
 238         case rtfRQuote:
 239                 PutStdChar (info, rtfSC_quoteright);
 240                 break;
 241         case rtfLDblQuote:
 242                 PutStdChar (info, rtfSC_quotedblleft);
 243                 break;
 244         case rtfRDblQuote:
 245                 PutStdChar (info, rtfSC_quotedblright);
 246                 break;
 247         }
 248 }
 249
 250
 251 /*
 252  * Eventually this should keep track of the destination of the
 253  * current state and only write text when in the initial state.
 254  *
 255  * If the output sequence is unspecified in the output map, write
 256  * the character's standard name instead.  This makes map deficiencies
 257  * obvious and provides incentive to fix it. :-)
 258  */
 259
 260 void PutStdChar (RTF_Info *info, int stdCode)
 261 {
 262
 263   char  *oStr = (char *) NULL;
 264   char  buf[rtfBufSiz];
 265
 266 /*      if (stdCode == rtfSC_nothing)
 267                 RTFPanic ("Unknown character code, logic error\n");
 268 */
 269         TRACE("\n");
 270
 271         oStr = info->outMap[stdCode];
 272         if (oStr == (char *) NULL)      /* no output sequence in map */
 273         {
 274                 sprintf (buf, "[[%s]]", RTFStdCharName (info, stdCode));
 275                 oStr = buf;
 276         }
 277         PutLitStr (info, oStr);
 278 }
 279
 280
 281 void PutLitChar (RTF_Info *info, int c)
 282 {
 283         CHARLIST_Enqueue(&info->charlist, (char) c);
 284         /* fputc (c, ostream); */
 285 }
 286
 287
 288 static void PutLitStr (RTF_Info *info, char     *s)
 289 {
 290         for(;*s;s++)
 291         {
 292           CHARLIST_Enqueue(&info->charlist, *s);
 293         }
 294         /* fputs (s, ostream); */
 295 }