public void testReadWriteFromNonZeroOffset() throws Exception {
HWPFFileSystem fileSys = new HWPFFileSystem();
HWPFOutputStream tableOut = fileSys.getStream("1Table");
tableOut.write(new byte[20]); // 20 bytes of whatever at the front.
_styleSheet.writeTo(tableOut);
byte[] newTableStream = tableOut.toByteArray();
StyleSheet newStyleSheet = new StyleSheet(newTableStream, 20);
assertEquals(newStyleSheet, _styleSheet);
}
public void testReadWrite() throws Exception {
HWPFFileSystem fileSys = new HWPFFileSystem();
HWPFOutputStream tableOut = fileSys.getStream("1Table");
HWPFOutputStream mainOut = fileSys.getStream("WordDocument");
_styleSheet.writeTo(tableOut);
byte[] newTableStream = tableOut.toByteArray();
StyleSheet newStyleSheet = new StyleSheet(newTableStream, 0);
assertEquals(newStyleSheet, _styleSheet);
}
void writeTo(byte mainStream[], int offset, HWPFOutputStream tableStream) throws IOException {
int length = _fields.length / 2;
LittleEndian.putShort(mainStream, offset, (short) length);
offset += 2;
for (int x = 0; x < length; x++) {
UnhandledDataStructure ds = (UnhandledDataStructure) _unknownMap.get(Integer.valueOf(x));
if (ds != null) {
_fields[x * 2] = tableStream.getOffset();
LittleEndian.putInt(mainStream, offset, tableStream.getOffset());
offset += 4;
byte buf[] = ds.getBuf();
tableStream.write(buf);
_fields[x * 2 + 1] = buf.length;
LittleEndian.putInt(mainStream, offset, buf.length);
offset += 4;
} else {
LittleEndian.putInt(mainStream, offset, _fields[x * 2]);
offset += 4;
LittleEndian.putInt(mainStream, offset, _fields[x * 2 + 1]);
offset += 4;
}
}
}
static void write(Sttb sttb, HWPFOutputStream tableStream) throws IOException {
final int headerSize = sttb.cDataLength == 2 ? 6 : 8;
byte[] header = new byte[headerSize];
LittleEndian.putShort(header, 0, (short) 0xffff);
if (sttb.data == null || sttb.data.length == 0) {
if (sttb.cDataLength == 4) {
LittleEndian.putInt(header, 2, 0);
LittleEndian.putUShort(header, 6, sttb.cbExtra);
tableStream.write(header);
return;
}
LittleEndian.putUShort(header, 2, 0);
LittleEndian.putUShort(header, 4, sttb.cbExtra);
tableStream.write(header);
return;
}
if (sttb.cDataLength == 4) {
LittleEndian.putInt(header, 2, sttb.data.length);
LittleEndian.putUShort(header, 6, sttb.cbExtra);
tableStream.write(header);
} else {
LittleEndian.putUShort(header, 2, sttb.data.length);
LittleEndian.putUShort(header, 4, sttb.cbExtra);
tableStream.write(header);
}
for (int i = 0; i < sttb.data.length; i++) {
String entry = sttb.data[i];
if (entry == null) {
// is it correct?
tableStream.write(new byte[] {-1, 0});
continue;
}
byte[] buf = new byte[entry.length() * 2 + sttb.cbExtra + 2];
LittleEndian.putShort(buf, 0, (short) entry.length());
StringUtil.putUnicodeLE(entry, buf, 2);
if (sttb.extraData != null && i < sttb.extraData.length && sttb.extraData[i] != null)
System.arraycopy(
sttb.extraData[i],
0,
buf,
entry.length() * 2,
Math.min(sttb.extraData[i].length, sttb.cbExtra));
tableStream.write(buf);
}
}
/**
* 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;
}