@Internal
public TextPieceTable getTextTable() {
return _cft.getTextPieceTable();
}
/**
* Writes out the word file that is represented by an instance of this class.
*
* @param out The OutputStream to write to.
* @throws IOException If there is an unexpected IOException from the passed in OutputStream.
*/
public void write(OutputStream out) throws IOException {
// initialize our streams for writing.
HWPFFileSystem docSys = new HWPFFileSystem();
HWPFOutputStream wordDocumentStream = docSys.getStream(STREAM_WORD_DOCUMENT);
HWPFOutputStream tableStream = docSys.getStream(STREAM_TABLE_1);
// HWPFOutputStream dataStream = docSys.getStream("Data");
int tableOffset = 0;
// FileInformationBlock fib = (FileInformationBlock)_fib.clone();
// clear the offsets and sizes in our FileInformationBlock.
_fib.clearOffsetsSizes();
// determine the FileInformationBLock size
int fibSize = _fib.getSize();
fibSize +=
POIFSConstants.SMALLER_BIG_BLOCK_SIZE - (fibSize % POIFSConstants.SMALLER_BIG_BLOCK_SIZE);
// preserve space for the FileInformationBlock because we will be writing
// it after we write everything else.
byte[] placeHolder = new byte[fibSize];
wordDocumentStream.write(placeHolder);
int mainOffset = wordDocumentStream.getOffset();
// write out the StyleSheet.
_fib.setFcStshf(tableOffset);
_ss.writeTo(tableStream);
_fib.setLcbStshf(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
// get fcMin and fcMac because we will be writing the actual text with the
// complex table.
int fcMin = mainOffset;
/*
* clx (encoding of the sprm lists for a complex file and piece table
* for a any file) Written immediately after the end of the previously
* recorded structure. This is recorded in all Word documents
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 23 of 210
*/
// write out the Complex table, includes text.
_fib.setFcClx(tableOffset);
_cft.writeTo(wordDocumentStream, tableStream);
_fib.setLcbClx(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
int fcMac = wordDocumentStream.getOffset();
/*
* dop (document properties record) Written immediately after the end of
* the previously recorded structure. This is recorded in all Word
* documents
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 23 of 210
*/
// write out the DocumentProperties.
_fib.setFcDop(tableOffset);
_dop.writeTo(tableStream);
_fib.setLcbDop(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
/*
* plcfBkmkf (table recording beginning CPs of bookmarks) Written
* immediately after the sttbfBkmk, if the document contains bookmarks.
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 24 of 210
*/
if (_bookmarksTables != null) {
_bookmarksTables.writePlcfBkmkf(_fib, tableStream);
tableOffset = tableStream.getOffset();
}
/*
* plcfBkmkl (table recording limit CPs of bookmarks) Written
* immediately after the plcfBkmkf, if the document contains bookmarks.
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 24 of 210
*/
if (_bookmarksTables != null) {
_bookmarksTables.writePlcfBkmkl(_fib, tableStream);
tableOffset = tableStream.getOffset();
}
/*
* plcfbteChpx (bin table for CHP FKPs) Written immediately after the
* previously recorded table. This is recorded in all Word documents.
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 24 of 210
*/
// write out the CHPBinTable.
_fib.setFcPlcfbteChpx(tableOffset);
_cbt.writeTo(wordDocumentStream, tableStream, fcMin, _cft.getTextPieceTable());
_fib.setLcbPlcfbteChpx(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
/*
* plcfbtePapx (bin table for PAP FKPs) Written immediately after the
* plcfbteChpx. This is recorded in all Word documents.
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 24 of 210
*/
// write out the PAPBinTable.
_fib.setFcPlcfbtePapx(tableOffset);
_pbt.writeTo(wordDocumentStream, tableStream, _cft.getTextPieceTable());
_fib.setLcbPlcfbtePapx(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
/*
* plcfendRef (endnote reference position table) Written immediately
* after the previously recorded table if the document contains endnotes
*
* plcfendTxt (endnote text position table) Written immediately after
* the plcfendRef if the document contains endnotes
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 24 of 210
*/
_endnotesTables.writeRef(_fib, tableStream);
_endnotesTables.writeTxt(_fib, tableStream);
tableOffset = tableStream.getOffset();
/*
* plcffld*** (table of field positions and statuses for annotation
* subdocument) Written immediately after the previously recorded table,
* if the ******* subdocument contains fields.
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 24 of 210
*/
if (_fieldsTables != null) {
_fieldsTables.write(_fib, tableStream);
tableOffset = tableStream.getOffset();
}
/*
* plcffndRef (footnote reference position table) Written immediately
* after the stsh if the document contains footnotes
*
* plcffndTxt (footnote text position table) Written immediately after
* the plcffndRef if the document contains footnotes
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 24 of 210
*/
_footnotesTables.writeRef(_fib, tableStream);
_footnotesTables.writeTxt(_fib, tableStream);
tableOffset = tableStream.getOffset();
/*
* plcfsed (section table) Written immediately after the previously
* recorded table. Recorded in all Word documents
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 25 of 210
*/
// write out the SectionTable.
_fib.setFcPlcfsed(tableOffset);
_st.writeTo(wordDocumentStream, tableStream);
_fib.setLcbPlcfsed(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
// write out the list tables
if (_lt != null) {
/*
* plcflst (list formats) Written immediately after the end of the
* previously recorded, if there are any lists defined in the
* document. This begins with a short count of LSTF structures
* followed by those LSTF structures. This is immediately followed
* by the allocated data hanging off the LSTFs. This data consists
* of the array of LVLs for each LSTF. (Each LVL consists of an LVLF
* followed by two grpprls and an XST.)
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 25 of 210
*/
_lt.writeListDataTo(_fib, tableStream);
tableOffset = tableStream.getOffset();
/*
* plflfo (more list formats) Written immediately after the end of
* the plcflst and its accompanying data, if there are any lists
* defined in the document. This consists first of a PL of LFO
* records, followed by the allocated data (if any) hanging off the
* LFOs. The allocated data consists of the array of LFOLVLFs for
* each LFO (and each LFOLVLF is immediately followed by some LVLs).
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 26 of 210
*/
_fib.setFcPlfLfo(tableStream.getOffset());
_lt.writeListOverridesTo(tableStream);
_fib.setLcbPlfLfo(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
}
/*
* sttbfBkmk (table of bookmark name strings) Written immediately after
* the previously recorded table, if the document contains bookmarks.
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 27 of 210
*/
if (_bookmarksTables != null) {
_bookmarksTables.writeSttbfBkmk(_fib, tableStream);
tableOffset = tableStream.getOffset();
}
/*
* sttbSavedBy (last saved by string table) Written immediately after
* the previously recorded table.
*
* Microsoft Office Word 97-2007 Binary File Format (.doc)
* Specification; Page 27 of 210
*/
// write out the saved-by table.
if (_sbt != null) {
_fib.setFcSttbSavedBy(tableOffset);
_sbt.writeTo(tableStream);
_fib.setLcbSttbSavedBy(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
}
// write out the revision mark authors table.
if (_rmat != null) {
_fib.setFcSttbfRMark(tableOffset);
_rmat.writeTo(tableStream);
_fib.setLcbSttbfRMark(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
}
// write out the FontTable.
_fib.setFcSttbfffn(tableOffset);
_ft.writeTo(tableStream);
_fib.setLcbSttbfffn(tableStream.getOffset() - tableOffset);
tableOffset = tableStream.getOffset();
// set some variables in the FileInformationBlock.
_fib.getFibBase().setFcMin(fcMin);
_fib.getFibBase().setFcMac(fcMac);
_fib.setCbMac(wordDocumentStream.getOffset());
// make sure that the table, doc and data streams use big blocks.
byte[] mainBuf = wordDocumentStream.toByteArray();
if (mainBuf.length < 4096) {
byte[] tempBuf = new byte[4096];
System.arraycopy(mainBuf, 0, tempBuf, 0, mainBuf.length);
mainBuf = tempBuf;
}
// Table1 stream will be used
_fib.getFibBase().setFWhichTblStm(true);
// write out the FileInformationBlock.
// _fib.serialize(mainBuf, 0);
_fib.writeTo(mainBuf, tableStream);
byte[] tableBuf = tableStream.toByteArray();
if (tableBuf.length < 4096) {
byte[] tempBuf = new byte[4096];
System.arraycopy(tableBuf, 0, tempBuf, 0, tableBuf.length);
tableBuf = tempBuf;
}
byte[] dataBuf = _dataStream;
if (dataBuf == null) {
dataBuf = new byte[4096];
}
if (dataBuf.length < 4096) {
byte[] tempBuf = new byte[4096];
System.arraycopy(dataBuf, 0, tempBuf, 0, dataBuf.length);
dataBuf = tempBuf;
}
// create new document preserving order of entries
POIFSFileSystem pfs = new POIFSFileSystem();
boolean docWritten = false;
boolean dataWritten = false;
boolean objectPoolWritten = false;
boolean tableWritten = false;
boolean propertiesWritten = false;
for (Iterator<Entry> iter = directory.getEntries(); iter.hasNext(); ) {
Entry entry = iter.next();
if (entry.getName().equals(STREAM_WORD_DOCUMENT)) {
if (!docWritten) {
pfs.createDocument(new ByteArrayInputStream(mainBuf), STREAM_WORD_DOCUMENT);
docWritten = true;
}
} else if (entry.getName().equals(STREAM_OBJECT_POOL)) {
if (!objectPoolWritten) {
_objectPool.writeTo(pfs.getRoot());
objectPoolWritten = true;
}
} else if (entry.getName().equals(STREAM_TABLE_0) || entry.getName().equals(STREAM_TABLE_1)) {
if (!tableWritten) {
pfs.createDocument(new ByteArrayInputStream(tableBuf), STREAM_TABLE_1);
tableWritten = true;
}
} else if (entry.getName().equals(SummaryInformation.DEFAULT_STREAM_NAME)
|| entry.getName().equals(DocumentSummaryInformation.DEFAULT_STREAM_NAME)) {
if (!propertiesWritten) {
writeProperties(pfs);
propertiesWritten = true;
}
} else if (entry.getName().equals(STREAM_DATA)) {
if (!dataWritten) {
pfs.createDocument(new ByteArrayInputStream(dataBuf), STREAM_DATA);
dataWritten = true;
}
} else {
EntryUtils.copyNodeRecursively(entry, pfs.getRoot());
}
}
if (!docWritten) pfs.createDocument(new ByteArrayInputStream(mainBuf), STREAM_WORD_DOCUMENT);
if (!tableWritten) pfs.createDocument(new ByteArrayInputStream(tableBuf), STREAM_TABLE_1);
if (!propertiesWritten) writeProperties(pfs);
if (!dataWritten) pfs.createDocument(new ByteArrayInputStream(dataBuf), STREAM_DATA);
if (!objectPoolWritten) _objectPool.writeTo(pfs.getRoot());
pfs.writeFilesystem(out);
this.directory = pfs.getRoot();
/*
* since we updated all references in FIB and etc, using new arrays to
* access data
*/
this.directory = pfs.getRoot();
this._tableStream = tableStream.toByteArray();
this._dataStream = dataBuf;
}
/**
* This constructor loads a Word document from a specific point in a POIFSFileSystem, probably not
* the default. Used typically to open embeded documents.
*
* @param directory The DirectoryNode that contains the Word document.
* @throws IOException If there is an unexpected IOException from the passed in POIFSFileSystem.
*/
public HWPFDocument(DirectoryNode directory) throws IOException {
// Load the main stream and FIB
// Also handles HPSF bits
super(directory);
// Is this document too old for us?
if (_fib.getFibBase().getNFib() < 106) {
throw new OldWordFileFormatException(
"The document is too old - Word 95 or older. Try HWPFOldDocument instead?");
}
// use the fib to determine the name of the table stream.
String name = STREAM_TABLE_0;
if (_fib.getFibBase().isFWhichTblStm()) {
name = STREAM_TABLE_1;
}
// Grab the table stream.
DocumentEntry tableProps;
try {
tableProps = (DocumentEntry) directory.getEntry(name);
} catch (FileNotFoundException fnfe) {
throw new IllegalStateException(
"Table Stream '"
+ name
+ "' wasn't found - Either the document is corrupt, or is Word95 (or earlier)");
}
// read in the table stream.
_tableStream = new byte[tableProps.getSize()];
directory.createDocumentInputStream(name).read(_tableStream);
_fib.fillVariableFields(_mainStream, _tableStream);
// read in the data stream.
try {
DocumentEntry dataProps = (DocumentEntry) directory.getEntry(STREAM_DATA);
_dataStream = new byte[dataProps.getSize()];
directory.createDocumentInputStream(STREAM_DATA).read(_dataStream);
} catch (java.io.FileNotFoundException e) {
_dataStream = new byte[0];
}
// Get the cp of the start of text in the main stream
// The latest spec doc says this is always zero!
int fcMin = 0;
// fcMin = _fib.getFcMin()
// Start to load up our standard structures.
_dop = new DocumentProperties(_tableStream, _fib.getFcDop(), _fib.getLcbDop());
_cft = new ComplexFileTable(_mainStream, _tableStream, _fib.getFcClx(), fcMin);
TextPieceTable _tpt = _cft.getTextPieceTable();
// Now load the rest of the properties, which need to be adjusted
// for where text really begin
_cbt =
new CHPBinTable(
_mainStream, _tableStream, _fib.getFcPlcfbteChpx(), _fib.getLcbPlcfbteChpx(), _tpt);
_pbt =
new PAPBinTable(
_mainStream,
_tableStream,
_dataStream,
_fib.getFcPlcfbtePapx(),
_fib.getLcbPlcfbtePapx(),
_tpt);
_text = _tpt.getText();
/*
* in this mode we preserving PAPX/CHPX structure from file, so text may
* miss from output, and text order may be corrupted
*/
boolean preserveBinTables = false;
try {
preserveBinTables = Boolean.parseBoolean(System.getProperty(PROPERTY_PRESERVE_BIN_TABLES));
} catch (Exception exc) {
// ignore;
}
if (!preserveBinTables) {
_cbt.rebuild(_cft);
_pbt.rebuild(_text, _cft);
}
/*
* Property to disable text rebuilding. In this mode changing the text
* will lead to unpredictable behavior
*/
boolean preserveTextTable = false;
try {
preserveTextTable = Boolean.parseBoolean(System.getProperty(PROPERTY_PRESERVE_TEXT_TABLE));
} catch (Exception exc) {
// ignore;
}
if (!preserveTextTable) {
_cft = new ComplexFileTable();
_tpt = _cft.getTextPieceTable();
final TextPiece textPiece = new SinglentonTextPiece(_text);
_tpt.add(textPiece);
_text = textPiece.getStringBuilder();
}
// Read FSPA and Escher information
// _fspa = new FSPATable(_tableStream, _fib.getFcPlcspaMom(),
// _fib.getLcbPlcspaMom(), getTextTable().getTextPieces());
_fspaHeaders = new FSPATable(_tableStream, _fib, FSPADocumentPart.HEADER);
_fspaMain = new FSPATable(_tableStream, _fib, FSPADocumentPart.MAIN);
if (_fib.getFcDggInfo() != 0) {
_escherRecordHolder =
new EscherRecordHolder(_tableStream, _fib.getFcDggInfo(), _fib.getLcbDggInfo());
} else {
_escherRecordHolder = new EscherRecordHolder();
}
// read in the pictures stream
_pictures = new PicturesTable(this, _dataStream, _mainStream, _fspaMain, _escherRecordHolder);
// And the art shapes stream
_officeArts = new ShapesTable(_tableStream, _fib);
// And escher pictures
_officeDrawingsHeaders = new OfficeDrawingsImpl(_fspaHeaders, _escherRecordHolder, _mainStream);
_officeDrawingsMain = new OfficeDrawingsImpl(_fspaMain, _escherRecordHolder, _mainStream);
_st =
new SectionTable(
_mainStream,
_tableStream,
_fib.getFcPlcfsed(),
_fib.getLcbPlcfsed(),
fcMin,
_tpt,
_fib.getSubdocumentTextStreamLength(SubdocumentType.MAIN));
_ss = new StyleSheet(_tableStream, _fib.getFcStshf());
_ft = new FontTable(_tableStream, _fib.getFcSttbfffn(), _fib.getLcbSttbfffn());
int listOffset = _fib.getFcPlcfLst();
int lfoOffset = _fib.getFcPlfLfo();
if (listOffset != 0 && _fib.getLcbPlcfLst() != 0) {
_lt = new ListTables(_tableStream, _fib.getFcPlcfLst(), _fib.getFcPlfLfo());
}
int sbtOffset = _fib.getFcSttbSavedBy();
int sbtLength = _fib.getLcbSttbSavedBy();
if (sbtOffset != 0 && sbtLength != 0) {
_sbt = new SavedByTable(_tableStream, sbtOffset, sbtLength);
}
int rmarkOffset = _fib.getFcSttbfRMark();
int rmarkLength = _fib.getLcbSttbfRMark();
if (rmarkOffset != 0 && rmarkLength != 0) {
_rmat = new RevisionMarkAuthorTable(_tableStream, rmarkOffset, rmarkLength);
}
_bookmarksTables = new BookmarksTables(_tableStream, _fib);
_bookmarks = new BookmarksImpl(_bookmarksTables);
_endnotesTables = new NotesTables(NoteType.ENDNOTE, _tableStream, _fib);
_endnotes = new NotesImpl(_endnotesTables);
_footnotesTables = new NotesTables(NoteType.FOOTNOTE, _tableStream, _fib);
_footnotes = new NotesImpl(_footnotesTables);
_fieldsTables = new FieldsTables(_tableStream, _fib);
_fields = new FieldsImpl(_fieldsTables);
}