wordwin.c

   1 /*
   2  * wordwin.c
   3  * Copyright (C) 2002-2005 A.J. van Os; Released under GPL
   4  *
   5  * Description:
   6  * Deal with the WIN 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;
  22         ULONG   ulTextLen, ulFootnoteLen;
  23         ULONG   ulHdrFtrLen, ulMacroLen, ulAnnotationLen;
  24         UINT    uiQuickSaves;
  25         USHORT  usDocStatus;
  26         BOOL    bTemplate, bFastSaved, bEncrypted, bSuccess;
  27
  28         fail(pFile == NULL);
  29         fail(aucHeader == NULL);
  30
  31         DBG_MSG("bGetDocumentText");
  32
  33         /* Get the status flags from the header */
  34         usDocStatus = usGetWord(0x0a, aucHeader);
  35         DBG_HEX(usDocStatus);
  36         bTemplate = (usDocStatus & BIT(0)) != 0;
  37         DBG_MSG_C(bTemplate, "This document is a Template");
  38         bFastSaved = (usDocStatus & BIT(2)) != 0;
  39         uiQuickSaves = (UINT)(usDocStatus & 0x00f0) >> 4;
  40         DBG_MSG_C(bFastSaved, "This document is Fast Saved");
  41         DBG_DEC_C(bFastSaved, uiQuickSaves);
  42         if (bFastSaved) {
  43                 werr(0, "Word2: fast saved documents are not supported yet");
  44                 return FALSE;
  45         }
  46         bEncrypted = (usDocStatus & BIT(8)) != 0;
  47         if (bEncrypted) {
  48                 werr(0, "Encrypted documents are not supported");
  49                 return FALSE;
  50         }
  51
  52         /* Get length information */
  53         ulBeginOfText = ulGetLong(0x18, aucHeader);
  54         DBG_HEX(ulBeginOfText);
  55         ulTextLen = ulGetLong(0x34, aucHeader);
  56         ulFootnoteLen = ulGetLong(0x38, aucHeader);
  57         ulHdrFtrLen = ulGetLong(0x3c, aucHeader);
  58         ulMacroLen = ulGetLong(0x40, aucHeader);
  59         ulAnnotationLen = ulGetLong(0x44, aucHeader);
  60         DBG_DEC(ulTextLen);
  61         DBG_DEC(ulFootnoteLen);
  62         DBG_DEC(ulHdrFtrLen);
  63         DBG_DEC(ulMacroLen);
  64         DBG_DEC(ulAnnotationLen);
  65         if (bFastSaved) {
  66                 bSuccess = FALSE;
  67         } else {
  68                 tTextBlock.ulFileOffset = ulBeginOfText;
  69                 tTextBlock.ulCharPos = ulBeginOfText;
  70                 tTextBlock.ulLength = ulTextLen +
  71                                 ulFootnoteLen +
  72                                 ulHdrFtrLen + ulMacroLen + ulAnnotationLen;
  73                 tTextBlock.bUsesUnicode = FALSE;
  74                 tTextBlock.usPropMod = IGNORE_PROPMOD;
  75                 bSuccess = bAdd2TextBlockList(&tTextBlock);
  76                 DBG_HEX_C(!bSuccess, tTextBlock.ulFileOffset);
  77                 DBG_HEX_C(!bSuccess, tTextBlock.ulCharPos);
  78                 DBG_DEC_C(!bSuccess, tTextBlock.ulLength);
  79                 DBG_DEC_C(!bSuccess, tTextBlock.bUsesUnicode);
  80                 DBG_DEC_C(!bSuccess, tTextBlock.usPropMod);
  81         }
  82
  83         if (bSuccess) {
  84                 vSplitBlockList(pFile,
  85                                 ulTextLen,
  86                                 ulFootnoteLen,
  87                                 ulHdrFtrLen,
  88                                 ulMacroLen,
  89                                 ulAnnotationLen,
  90                                 0,
  91                                 0,
  92                                 0,
  93                                 FALSE);
  94         } else {
  95                 vDestroyTextBlockList();
  96                 werr(0, "I can't find the text of this document");
  97         }
  98         return bSuccess;
  99 } /* end of bGetDocumentText */
 100
 101 /*
 102  * vGetDocumentData - make a list of the data blocks of a Word document
 103  */
 104 static void
 105 vGetDocumentData(FILE *pFile, const UCHAR *aucHeader)
 106 {
 107         data_block_type tDataBlock;
 108         options_type    tOptions;
 109         ULONG   ulEndOfText, ulBeginCharInfo;
 110         BOOL    bFastSaved, bHasImages, bSuccess;
 111         USHORT  usDocStatus;
 112
 113         /* Get the options */
 114         vGetOptions(&tOptions);
 115
 116         /* Get the status flags from the header */
 117         usDocStatus = usGetWord(0x0a, aucHeader);
 118         DBG_HEX(usDocStatus);
 119         bFastSaved = (usDocStatus & BIT(2)) != 0;
 120         bHasImages = (usDocStatus & BIT(3)) != 0;
 121
 122         if (!bHasImages ||
 123             tOptions.eConversionType == conversion_text ||
 124             tOptions.eConversionType == conversion_fmt_text ||
 125             tOptions.eConversionType == conversion_xml ||
 126             tOptions.eImageLevel == level_no_images) {
 127                 /*
 128                  * No images in the document or text-only output or
 129                  * no images wanted, so no data blocks will be needed
 130                  */
 131                 vDestroyDataBlockList();
 132                 return;
 133         }
 134
 135         if (bFastSaved) {
 136                 bSuccess = FALSE;
 137         } else {
 138                 /* This datablock is too big, but it contains all images */
 139                 ulEndOfText = ulGetLong(0x1c, aucHeader);
 140                 DBG_HEX(ulEndOfText);
 141                 ulBeginCharInfo = ulGetLong(0xa0, aucHeader);
 142                 DBG_HEX(ulBeginCharInfo);
 143                 if (ulBeginCharInfo > ulEndOfText) {
 144                         tDataBlock.ulFileOffset = ulEndOfText;
 145                         tDataBlock.ulDataPos = ulEndOfText;
 146                         tDataBlock.ulLength = ulBeginCharInfo - ulEndOfText;
 147                         bSuccess = bAdd2DataBlockList(&tDataBlock);
 148                         DBG_HEX_C(!bSuccess, tDataBlock.ulFileOffset);
 149                         DBG_HEX_C(!bSuccess, tDataBlock.ulDataPos);
 150                         DBG_DEC_C(!bSuccess, tDataBlock.ulLength);
 151                 } else {
 152                         bSuccess = ulBeginCharInfo == ulEndOfText;
 153                 }
 154         }
 155
 156         if (!bSuccess) {
 157                 vDestroyDataBlockList();
 158                 werr(0, "I can't find the data of this document");
 159         }
 160 } /* end of vGetDocumentData */
 161
 162 /*
 163  * iInitDocumentWIN - initialize an WIN document
 164  *
 165  * Returns the version of Word that made the document or -1
 166  */
 167 int
 168 iInitDocumentWIN(FILE *pFile, long lFilesize)
 169 {
 170         int     iWordVersion;
 171         BOOL    bSuccess;
 172         USHORT  usIdent;
 173         UCHAR   aucHeader[384];
 174
 175         fail(pFile == NULL);
 176
 177         if (lFilesize < 384) {
 178                 return -1;
 179         }
 180
 181         /* Read the headerblock */
 182         if (!bReadBytes(aucHeader, 384, 0x00, pFile)) {
 183                 return -1;
 184         }
 185         /* Get the "magic number" from the header */
 186         usIdent = usGetWord(0x00, aucHeader);
 187         DBG_HEX(usIdent);
 188         fail(usIdent != 0xa59b &&       /* WinWord 1.x */
 189                 usIdent != 0xa5db);     /* WinWord 2.0 */
 190         iWordVersion = iGetVersionNumber(aucHeader);
 191         if (iWordVersion != 1 && iWordVersion != 2) {
 192                 werr(0, "This file is not from ''Win Word 1 or 2'.");
 193                 return -1;
 194         }
 195         bSuccess = bGetDocumentText(pFile, aucHeader);
 196         if (bSuccess) {
 197                 vGetDocumentData(pFile, aucHeader);
 198                 vGetPropertyInfo(pFile, NULL,
 199                                 NULL, 0, NULL, 0,
 200                                 aucHeader, iWordVersion);
 201                 vSetDefaultTabWidth(pFile, NULL,
 202                                 NULL, 0, NULL, 0,
 203                                 aucHeader, iWordVersion);
 204                 vGetNotesInfo(pFile, NULL,
 205                                 NULL, 0, NULL, 0,
 206                                 aucHeader, iWordVersion);
 207         }
 208         return bSuccess ? iWordVersion : -1;
 209 } /* end of iInitDocumentWIN */