worddos.c

   1 /*
   2  * worddos.c
   3  * Copyright (C) 2002-2005 A.J. van Os; Released under GNU GPL
   4  *
   5  * Description:
   6  * Deal with the DOS internals of a MS Word file
   7  */
   8
   9 #include "antiword.h"
  10
  11
  12 /*
  13  * bGetDocumentText - make a list of the text blocks of a Word document
  14  *
  15  * Return TRUE when succesful, otherwise FALSE
  16  */
  17 static BOOL
  18 bGetDocumentText(FILE *pFile, long lFilesize, const UCHAR *aucHeader)
  19 {
  20         text_block_type tTextBlock;
  21         ULONG   ulTextLen;
  22         BOOL    bFastSaved;
  23         UCHAR   ucDocStatus, ucVersion;
  24
  25         fail(pFile == NULL);
  26         fail(lFilesize < 128);
  27         fail(aucHeader == NULL);
  28
  29         /* Get the status flags from the header */
  30         ucDocStatus = ucGetByte(0x75, aucHeader);
  31         DBG_HEX(ucDocStatus);
  32         bFastSaved = (ucDocStatus & BIT(1)) != 0;
  33         DBG_MSG_C(bFastSaved, "This document is Fast Saved");
  34         ucVersion = ucGetByte(0x74, aucHeader);
  35         DBG_DEC(ucVersion);
  36         DBG_MSG_C(ucVersion == 0, "Written by Word 4.0 or earlier");
  37         DBG_MSG_C(ucVersion == 3, "Word 5.0 format, but not written by Word");
  38         DBG_MSG_C(ucVersion == 4, "Written by Word 5.x");
  39         if (bFastSaved) {
  40                 werr(0, "Word for DOS: autosave documents are not supported");
  41                 return FALSE;
  42         }
  43
  44         /* Get length information */
  45         ulTextLen = ulGetLong(0x0e, aucHeader);
  46         DBG_HEX(ulTextLen);
  47         ulTextLen -= 128;
  48         DBG_DEC(ulTextLen);
  49         tTextBlock.ulFileOffset = 128;
  50         tTextBlock.ulCharPos = 128;
  51         tTextBlock.ulLength = ulTextLen;
  52         tTextBlock.bUsesUnicode = FALSE;
  53         tTextBlock.usPropMod = IGNORE_PROPMOD;
  54         if (!bAdd2TextBlockList(&tTextBlock)) {
  55                 DBG_HEX(tTextBlock.ulFileOffset);
  56                 DBG_HEX(tTextBlock.ulCharPos);
  57                 DBG_DEC(tTextBlock.ulLength);
  58                 DBG_DEC(tTextBlock.bUsesUnicode);
  59                 DBG_DEC(tTextBlock.usPropMod);
  60                 return FALSE;
  61         }
  62         return TRUE;
  63 } /* end of bGetDocumentText */
  64
  65 /*
  66  * iInitDocumentDOS - initialize an DOS document
  67  *
  68  * Returns the version of Word that made the document or -1
  69  */
  70 int
  71 iInitDocumentDOS(FILE *pFile, long lFilesize)
  72 {
  73         int     iWordVersion;
  74         BOOL    bSuccess;
  75         USHORT  usIdent;
  76         UCHAR   aucHeader[128];
  77
  78         fail(pFile == NULL);
  79
  80         if (lFilesize < 128) {
  81                 return -1;
  82         }
  83
  84         /* Read the headerblock */
  85         if (!bReadBytes(aucHeader, 128, 0x00, pFile)) {
  86                 return -1;
  87         }
  88         /* Get the "magic number" from the header */
  89         usIdent = usGetWord(0x00, aucHeader);
  90         DBG_HEX(usIdent);
  91         fail(usIdent != 0xbe31);        /* Word for DOS */
  92         iWordVersion = iGetVersionNumber(aucHeader);
  93         if (iWordVersion != 0) {
  94                 werr(0, "This file is not from 'Word for DOS'.");
  95                 return -1;
  96         }
  97         bSuccess = bGetDocumentText(pFile, lFilesize, aucHeader);
  98         if (bSuccess) {
  99                 vGetPropertyInfo(pFile, NULL,
 100                                 NULL, 0, NULL, 0,
 101                                 aucHeader, iWordVersion);
 102                 vSetDefaultTabWidth(pFile, NULL,
 103                                 NULL, 0, NULL, 0,
 104                                 aucHeader, iWordVersion);
 105                 vGetNotesInfo(pFile, NULL,
 106                                 NULL, 0, NULL, 0,
 107                                 aucHeader, iWordVersion);
 108         }
 109         return bSuccess ? iWordVersion : -1;
 110 } /* end of iInitDocumentDOS */