wordmac.c

   1 /*
   2  * wordmac.c
   3  * Copyright (C) 2002-2004 A.J. van Os; Released under GNU GPL
   4  *
   5  * Description:
   6  * Deal with the MAC 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, const UCHAR *aucHeader)
  19 {
  20         text_block_type tTextBlock;
  21         ULONG   ulBeginOfText, ulEndOfText;
  22         ULONG   ulTextLen;
  23         UCHAR   ucDocStatus;
  24         BOOL    bFastSaved;
  25
  26         fail(pFile == NULL);
  27         fail(aucHeader == NULL);
  28
  29         DBG_MSG("bGetDocumentText");
  30
  31         NO_DBG_PRINT_BLOCK(aucHeader, 0x20);
  32
  33         /* Get the status flags from the header */
  34         ucDocStatus = ucGetByte(0x0a, aucHeader);
  35         DBG_HEX(ucDocStatus);
  36         bFastSaved = (ucDocStatus & BIT(5)) != 0;
  37         DBG_MSG_C(bFastSaved, "This document is Fast Saved");
  38         if (bFastSaved) {
  39                 werr(0, "MacWord: fast saved documents are not supported yet");
  40                 return FALSE;
  41         }
  42
  43         /* Get length information */
  44         ulBeginOfText = ulGetLongBE(0x14, aucHeader);
  45         DBG_HEX(ulBeginOfText);
  46         ulEndOfText = ulGetLongBE(0x18, aucHeader);
  47         DBG_HEX(ulEndOfText);
  48         ulTextLen = ulEndOfText - ulBeginOfText;
  49         DBG_DEC(ulTextLen);
  50         tTextBlock.ulFileOffset = ulBeginOfText;
  51         tTextBlock.ulCharPos = ulBeginOfText;
  52         tTextBlock.ulLength = ulTextLen;
  53         tTextBlock.bUsesUnicode = FALSE;
  54         tTextBlock.usPropMod = IGNORE_PROPMOD;
  55         if (!bAdd2TextBlockList(&tTextBlock)) {
  56                 DBG_HEX(tTextBlock.ulFileOffset);
  57                 DBG_HEX(tTextBlock.ulCharPos);
  58                 DBG_DEC(tTextBlock.ulLength);
  59                 DBG_DEC(tTextBlock.bUsesUnicode);
  60                 DBG_DEC(tTextBlock.usPropMod);
  61                 return FALSE;
  62         }
  63         return TRUE;
  64 } /* end of bGetDocumentText */
  65
  66 /*
  67  * iInitDocumentMAC - initialize an MAC document
  68  *
  69  * Returns the version of Word that made the document or -1
  70  */
  71 int
  72 iInitDocumentMAC(FILE *pFile, long lFilesize)
  73 {
  74         int     iWordVersion;
  75         BOOL    bSuccess;
  76         USHORT  usIdent;
  77         UCHAR   aucHeader[256];
  78
  79         fail(pFile == NULL);
  80
  81         if (lFilesize < 256) {
  82                 return -1;
  83         }
  84
  85         /* Read the headerblock */
  86         if (!bReadBytes(aucHeader, 256, 0x00, pFile)) {
  87                 return -1;
  88         }
  89         /* Get the "magic number" from the header */
  90         usIdent = usGetWord(0x00, aucHeader);
  91         DBG_HEX(usIdent);
  92         fail(usIdent != 0x37fe);        /* MacWord 4 and 5 */
  93         iWordVersion = iGetVersionNumber(aucHeader);
  94         if (iWordVersion != 4 && iWordVersion != 5) {
  95                 werr(0, "This file is not from ''Mac Word 4 or 5'.");
  96                 return -1;
  97         }
  98         bSuccess = bGetDocumentText(pFile, aucHeader);
  99         if (bSuccess) {
 100                 vGetPropertyInfo(pFile, NULL,
 101                                 NULL, 0, NULL, 0,
 102                                 aucHeader, iWordVersion);
 103                 vSetDefaultTabWidth(pFile, NULL,
 104                                 NULL, 0, NULL, 0,
 105                                 aucHeader, iWordVersion);
 106         }
 107         return bSuccess ? iWordVersion : -1;
 108 } /* end of iInitDocumentMAC */