TB9.2 Example Applications: examples/PIPS/antiword/src/wordwin.c Source File

00001 /*
-00002  * wordwin.c
-00003  * Copyright (C) 2002-2005 A.J. van Os; Released under GPL
-00004  *
-00005  * Description:
-00006  * Deal with the WIN internals of a MS Word file
-00007  */
-00008 
-00009 #include "antiword.h"
-00010 
-00011 
-00012 /*
-00013  * bGetDocumentText - make a list of the text blocks of a Word document
-00014  *
-00015  * Return TRUE when succesful, otherwise FALSE
-00016  */
-00017 static BOOL
-00018 bGetDocumentText(FILE *pFile, const UCHAR *aucHeader)
-00019 {
-00020         text_block_type tTextBlock;
-00021         ULONG   ulBeginOfText;
-00022         ULONG   ulTextLen, ulFootnoteLen;
-00023         ULONG   ulHdrFtrLen, ulMacroLen, ulAnnotationLen;
-00024         UINT    uiQuickSaves;
-00025         USHORT  usDocStatus;
-00026         BOOL    bTemplate, bFastSaved, bEncrypted, bSuccess;
-00027 
-00028         fail(pFile == NULL);
-00029         fail(aucHeader == NULL);
-00030 
-00031         DBG_MSG("bGetDocumentText");
-00032 
-00033         /* Get the status flags from the header */
-00034         usDocStatus = usGetWord(0x0a, aucHeader);
-00035         DBG_HEX(usDocStatus);
-00036         bTemplate = (usDocStatus & BIT(0)) != 0;
-00037         DBG_MSG_C(bTemplate, "This document is a Template");
-00038         bFastSaved = (usDocStatus & BIT(2)) != 0;
-00039         uiQuickSaves = (UINT)(usDocStatus & 0x00f0) >> 4;
-00040         DBG_MSG_C(bFastSaved, "This document is Fast Saved");
-00041         DBG_DEC_C(bFastSaved, uiQuickSaves);
-00042         if (bFastSaved) {
-00043                 werr(0, "Word2: fast saved documents are not supported yet");
-00044                 return FALSE;
-00045         }
-00046         bEncrypted = (usDocStatus & BIT(8)) != 0;
-00047         if (bEncrypted) {
-00048                 werr(0, "Encrypted documents are not supported");
-00049                 return FALSE;
-00050         }
-00051 
-00052         /* Get length information */
-00053         ulBeginOfText = ulGetLong(0x18, aucHeader);
-00054         DBG_HEX(ulBeginOfText);
-00055         ulTextLen = ulGetLong(0x34, aucHeader);
-00056         ulFootnoteLen = ulGetLong(0x38, aucHeader);
-00057         ulHdrFtrLen = ulGetLong(0x3c, aucHeader);
-00058         ulMacroLen = ulGetLong(0x40, aucHeader);
-00059         ulAnnotationLen = ulGetLong(0x44, aucHeader);
-00060         DBG_DEC(ulTextLen);
-00061         DBG_DEC(ulFootnoteLen);
-00062         DBG_DEC(ulHdrFtrLen);
-00063         DBG_DEC(ulMacroLen);
-00064         DBG_DEC(ulAnnotationLen);
-00065         if (bFastSaved) {
-00066                 bSuccess = FALSE;
-00067         } else {
-00068                 tTextBlock.ulFileOffset = ulBeginOfText;
-00069                 tTextBlock.ulCharPos = ulBeginOfText;
-00070                 tTextBlock.ulLength = ulTextLen +
-00071                                 ulFootnoteLen +
-00072                                 ulHdrFtrLen + ulMacroLen + ulAnnotationLen;
-00073                 tTextBlock.bUsesUnicode = FALSE;
-00074                 tTextBlock.usPropMod = IGNORE_PROPMOD;
-00075                 bSuccess = bAdd2TextBlockList(&tTextBlock);
-00076                 DBG_HEX_C(!bSuccess, tTextBlock.ulFileOffset);
-00077                 DBG_HEX_C(!bSuccess, tTextBlock.ulCharPos);
-00078                 DBG_DEC_C(!bSuccess, tTextBlock.ulLength);
-00079                 DBG_DEC_C(!bSuccess, tTextBlock.bUsesUnicode);
-00080                 DBG_DEC_C(!bSuccess, tTextBlock.usPropMod);
-00081         }
-00082 
-00083         if (bSuccess) {
-00084                 vSplitBlockList(pFile,
-00085                                 ulTextLen,
-00086                                 ulFootnoteLen,
-00087                                 ulHdrFtrLen,
-00088                                 ulMacroLen,
-00089                                 ulAnnotationLen,
-00090                                 0,
-00091                                 0,
-00092                                 0,
-00093                                 FALSE);
-00094         } else {
-00095                 vDestroyTextBlockList();
-00096                 werr(0, "I can't find the text of this document");
-00097         }
-00098         return bSuccess;
-00099 } /* end of bGetDocumentText */
-00100 
-00101 /*
-00102  * vGetDocumentData - make a list of the data blocks of a Word document
-00103  */
-00104 static void
-00105 vGetDocumentData(FILE *pFile, const UCHAR *aucHeader)
-00106 {
-00107         data_block_type tDataBlock;
-00108         options_type    tOptions;
-00109         ULONG   ulEndOfText, ulBeginCharInfo;
-00110         BOOL    bFastSaved, bHasImages, bSuccess;
-00111         USHORT  usDocStatus;
-00112 
-00113         /* Get the options */
-00114         vGetOptions(&tOptions);
-00115 
-00116         /* Get the status flags from the header */
-00117         usDocStatus = usGetWord(0x0a, aucHeader);
-00118         DBG_HEX(usDocStatus);
-00119         bFastSaved = (usDocStatus & BIT(2)) != 0;
-00120         bHasImages = (usDocStatus & BIT(3)) != 0;
-00121 
-00122         if (!bHasImages ||
-00123             tOptions.eConversionType == conversion_text ||
-00124             tOptions.eConversionType == conversion_fmt_text ||
-00125             tOptions.eConversionType == conversion_xml ||
-00126             tOptions.eImageLevel == level_no_images) {
-00127                 /*
-00128                  * No images in the document or text-only output or
-00129                  * no images wanted, so no data blocks will be needed
-00130                  */
-00131                 vDestroyDataBlockList();
-00132                 return;
-00133         }
-00134 
-00135         if (bFastSaved) {
-00136                 bSuccess = FALSE;
-00137         } else {
-00138                 /* This datablock is too big, but it contains all images */
-00139                 ulEndOfText = ulGetLong(0x1c, aucHeader);
-00140                 DBG_HEX(ulEndOfText);
-00141                 ulBeginCharInfo = ulGetLong(0xa0, aucHeader);
-00142                 DBG_HEX(ulBeginCharInfo);
-00143                 if (ulBeginCharInfo > ulEndOfText) {
-00144                         tDataBlock.ulFileOffset = ulEndOfText;
-00145                         tDataBlock.ulDataPos = ulEndOfText;
-00146                         tDataBlock.ulLength = ulBeginCharInfo - ulEndOfText;
-00147                         bSuccess = bAdd2DataBlockList(&tDataBlock);
-00148                         DBG_HEX_C(!bSuccess, tDataBlock.ulFileOffset);
-00149                         DBG_HEX_C(!bSuccess, tDataBlock.ulDataPos);
-00150                         DBG_DEC_C(!bSuccess, tDataBlock.ulLength);
-00151                 } else {
-00152                         bSuccess = ulBeginCharInfo == ulEndOfText;
-00153                 }
-00154         }
-00155 
-00156         if (!bSuccess) {
-00157                 vDestroyDataBlockList();
-00158                 werr(0, "I can't find the data of this document");
-00159         }
-00160 } /* end of vGetDocumentData */
-00161 
-00162 /*
-00163  * iInitDocumentWIN - initialize an WIN document
-00164  *
-00165  * Returns the version of Word that made the document or -1
-00166  */
-00167 int
-00168 iInitDocumentWIN(FILE *pFile, long lFilesize)
-00169 {
-00170         int     iWordVersion;
-00171         BOOL    bSuccess;
-00172         USHORT  usIdent;
-00173         UCHAR   aucHeader[384];
-00174 
-00175         fail(pFile == NULL);
-00176 
-00177         if (lFilesize < 384) {
-00178                 return -1;
-00179         }
-00180 
-00181         /* Read the headerblock */
-00182         if (!bReadBytes(aucHeader, 384, 0x00, pFile)) {
-00183                 return -1;
-00184         }
-00185         /* Get the "magic number" from the header */
-00186         usIdent = usGetWord(0x00, aucHeader);
-00187         DBG_HEX(usIdent);
-00188         fail(usIdent != 0xa59b &&       /* WinWord 1.x */
-00189                 usIdent != 0xa5db);     /* WinWord 2.0 */
-00190         iWordVersion = iGetVersionNumber(aucHeader);
-00191         if (iWordVersion != 1 && iWordVersion != 2) {
-00192                 werr(0, "This file is not from ''Win Word 1 or 2'.");
-00193                 return -1;
-00194         }
-00195         bSuccess = bGetDocumentText(pFile, aucHeader);
-00196         if (bSuccess) {
-00197                 vGetDocumentData(pFile, aucHeader);
-00198                 vGetPropertyInfo(pFile, NULL,
-00199                                 NULL, 0, NULL, 0,
-00200                                 aucHeader, iWordVersion);
-00201                 vSetDefaultTabWidth(pFile, NULL,
-00202                                 NULL, 0, NULL, 0,
-00203                                 aucHeader, iWordVersion);
-00204                 vGetNotesInfo(pFile, NULL,
-00205                                 NULL, 0, NULL, 0,
-00206                                 aucHeader, iWordVersion);
-00207         }
-00208         return bSuccess ? iWordVersion : -1;
-00209 } /* end of iInitDocumentWIN */
-