public void run() throws IOException {
NPOIFSFileSystem fs = new NPOIFSFileSystem(new File(file), true);
try {
InputStream din = BiffViewer.getPOIFSInputStream(fs);
try {
RecordInputStream rinp = new RecordInputStream(din);
while (rinp.hasNextRecord()) {
int sid = rinp.getNextSid();
rinp.nextRecord();
int size = rinp.available();
Class<? extends Record> clz = RecordFactory.getRecordClass(sid);
System.out.print(formatSID(sid) + " - " + formatSize(size) + " bytes");
if (clz != null) {
System.out.print(" \t");
System.out.print(clz.getName().replace("org.apache.poi.hssf.record.", ""));
}
System.out.println();
byte[] data = rinp.readRemainder();
if (data.length > 0) {
System.out.print(" ");
System.out.println(formatData(data));
}
}
} finally {
din.close();
}
} finally {
fs.close();
}
}
public void testFactory(String file, String protectedFile, String password) throws Exception {
SlideShow<?, ?> ss;
// from file
ss = SlideShowFactory.create(fromFile(file));
assertNotNull(ss);
// from stream
ss = SlideShowFactory.create(fromStream(file));
assertNotNull(ss);
// from NPOIFS
if (!file.contains("pptx")) {
NPOIFSFileSystem npoifs = new NPOIFSFileSystem(fromFile(file));
ss = SlideShowFactory.create(npoifs);
assertNotNull(ss);
npoifs.close();
}
// from protected file
ss = SlideShowFactory.create(fromFile(protectedFile), password);
assertNotNull(ss);
// from protected stream
ss = SlideShowFactory.create(fromStream(protectedFile), password);
assertNotNull(ss);
// from protected NPOIFS
NPOIFSFileSystem npoifs = new NPOIFSFileSystem(fromFile(protectedFile));
ss = SlideShowFactory.create(npoifs, password);
assertNotNull(ss);
npoifs.close();
}
public static Workbook getWorkBook(String filePath) throws IOException {
if (!filePath.contains(".xlsx")) {
NPOIFSFileSystem fs = new NPOIFSFileSystem(new File(filePath));
HSSFWorkbook wb = new HSSFWorkbook(fs.getRoot(), false);
fs.close();
return wb;
} else {
FileInputStream fileInputStream = new FileInputStream(filePath);
XSSFWorkbook xssfWorkbook = new XSSFWorkbook(fileInputStream);
fileInputStream.close();
return xssfWorkbook;
}
}
public void testAreDocumentsIdentical() throws Exception {
POIFSFileSystem fs = new POIFSFileSystem();
DirectoryEntry dirA = fs.createDirectory("DirA");
DirectoryEntry dirB = fs.createDirectory("DirB");
DocumentEntry entryA1 = dirA.createDocument("Entry1", new ByteArrayInputStream(dataSmallA));
DocumentEntry entryA1b = dirA.createDocument("Entry1b", new ByteArrayInputStream(dataSmallA));
DocumentEntry entryA2 = dirA.createDocument("Entry2", new ByteArrayInputStream(dataSmallB));
DocumentEntry entryB1 = dirB.createDocument("Entry1", new ByteArrayInputStream(dataSmallA));
// Names must match
assertEquals(false, entryA1.getName().equals(entryA1b.getName()));
assertEquals(false, EntryUtils.areDocumentsIdentical(entryA1, entryA1b));
// Contents must match
assertEquals(false, EntryUtils.areDocumentsIdentical(entryA1, entryA2));
// Parents don't matter if contents + names are the same
assertEquals(false, entryA1.getParent().equals(entryB1.getParent()));
assertEquals(true, EntryUtils.areDocumentsIdentical(entryA1, entryB1));
// Can work with NPOIFS + POIFS
ByteArrayOutputStream tmpO = new ByteArrayOutputStream();
fs.writeFilesystem(tmpO);
ByteArrayInputStream tmpI = new ByteArrayInputStream(tmpO.toByteArray());
NPOIFSFileSystem nfs = new NPOIFSFileSystem(tmpI);
DirectoryEntry dN1 = (DirectoryEntry) nfs.getRoot().getEntry("DirA");
DirectoryEntry dN2 = (DirectoryEntry) nfs.getRoot().getEntry("DirB");
DocumentEntry eNA1 = (DocumentEntry) dN1.getEntry(entryA1.getName());
DocumentEntry eNA2 = (DocumentEntry) dN1.getEntry(entryA2.getName());
DocumentEntry eNB1 = (DocumentEntry) dN2.getEntry(entryB1.getName());
assertEquals(false, EntryUtils.areDocumentsIdentical(eNA1, eNA2));
assertEquals(true, EntryUtils.areDocumentsIdentical(eNA1, eNB1));
assertEquals(false, EntryUtils.areDocumentsIdentical(eNA1, entryA1b));
assertEquals(false, EntryUtils.areDocumentsIdentical(eNA1, entryA2));
assertEquals(true, EntryUtils.areDocumentsIdentical(eNA1, entryA1));
assertEquals(true, EntryUtils.areDocumentsIdentical(eNA1, entryB1));
}
private static Set<String> getTopLevelNames(TikaInputStream stream) throws IOException {
// Force the document stream to a (possibly temporary) file
// so we don't modify the current position of the stream
File file = stream.getFile();
try {
NPOIFSFileSystem fs = new NPOIFSFileSystem(file, true);
// Optimize a possible later parsing process by keeping
// a reference to the already opened POI file system
stream.setOpenContainer(fs);
return getTopLevelNames(fs.getRoot());
} catch (IOException e) {
// Parse error in POI, so we don't know the file type
return Collections.emptySet();
} catch (RuntimeException e) {
// Another problem in POI
return Collections.emptySet();
}
}
/**
* read in a file and write it back out again
*
* @param args names of the files; arg[ 0 ] is the input file, arg[ 1 ] is the output file
* @exception IOException
*/
public static void main(String args[]) throws IOException {
if (args.length != 2) {
System.err.println("two arguments required: input filename and output filename");
System.exit(1);
}
FileInputStream istream = new FileInputStream(args[0]);
try {
FileOutputStream ostream = new FileOutputStream(args[1]);
try {
NPOIFSFileSystem fs = new NPOIFSFileSystem(istream);
try {
fs.writeFilesystem(ostream);
} finally {
fs.close();
}
} finally {
ostream.close();
}
} finally {
istream.close();
}
}
/**
* @param bytes
* @param metadata incoming metadata
* @param unknownFilenameCount
* @return byte[] for contents of obj data
* @throws IOException
*/
protected byte[] parse(byte[] bytes, Metadata metadata, AtomicInteger unknownFilenameCount)
throws IOException {
ByteArrayInputStream is = new ByteArrayInputStream(bytes);
long version = readUInt(is);
metadata.add(RTFMetadata.EMB_APP_VERSION, Long.toString(version));
long formatId = readUInt(is);
// 2 is an embedded object. 1 is a link.
if (formatId != 2L) {
return null;
}
String className = readLengthPrefixedAnsiString(is).trim();
String topicName = readLengthPrefixedAnsiString(is).trim();
String itemName = readLengthPrefixedAnsiString(is).trim();
if (className != null && className.length() > 0) {
metadata.add(RTFMetadata.EMB_CLASS, className);
}
if (topicName != null && topicName.length() > 0) {
metadata.add(RTFMetadata.EMB_TOPIC, topicName);
}
if (itemName != null && itemName.length() > 0) {
metadata.add(RTFMetadata.EMB_ITEM, itemName);
}
long dataSz = readUInt(is);
// readBytes tests for reading too many bytes
byte[] embObjBytes = readBytes(is, dataSz);
if (className.toLowerCase(Locale.ROOT).equals("package")) {
return handlePackage(embObjBytes, metadata);
} else if (className.toLowerCase(Locale.ROOT).equals("pbrush")) {
// simple bitmap bytes
return embObjBytes;
} else {
ByteArrayInputStream embIs = new ByteArrayInputStream(embObjBytes);
if (NPOIFSFileSystem.hasPOIFSHeader(embIs)) {
try {
return handleEmbeddedPOIFS(embIs, metadata, unknownFilenameCount);
} catch (IOException e) {
// swallow
}
}
}
return embObjBytes;
}
/** Creates a Workbook from the given NPOIFSFileSystem, which may be password protected */
private static Workbook create(NPOIFSFileSystem fs, String password)
throws IOException, InvalidFormatException {
DirectoryNode root = fs.getRoot();
// Encrypted OOXML files go inside OLE2 containers, is this one?
if (root.hasEntry(Decryptor.DEFAULT_POIFS_ENTRY)) {
EncryptionInfo info = new EncryptionInfo(fs);
Decryptor d = Decryptor.getInstance(info);
boolean passwordCorrect = false;
InputStream stream = null;
try {
if (password != null && d.verifyPassword(password)) {
passwordCorrect = true;
}
if (!passwordCorrect && d.verifyPassword(Decryptor.DEFAULT_PASSWORD)) {
passwordCorrect = true;
}
if (passwordCorrect) {
stream = d.getDataStream(root);
}
} catch (GeneralSecurityException e) {
throw new IOException(e);
}
if (!passwordCorrect) {
if (password != null) throw new EncryptedDocumentException("Password incorrect");
else
throw new EncryptedDocumentException(
"The supplied spreadsheet is protected, but no password was supplied");
}
OPCPackage pkg = OPCPackage.open(stream);
return create(pkg);
}
// If we get here, it isn't an encrypted XLSX file
// So, treat it as a regular HSSF XLS one
if (password != null) {
Biff8EncryptionKey.setCurrentUserPassword(password);
}
Workbook wb = new HSSFWorkbook(root, true);
Biff8EncryptionKey.setCurrentUserPassword(null);
return wb;
}
public InputStream getDataStream(NPOIFSFileSystem fs)
throws IOException, GeneralSecurityException {
return getDataStream(fs.getRoot());
}
// will throw IOException if not actually POIFS
// can return null byte[]
private byte[] handleEmbeddedPOIFS(
InputStream is, Metadata metadata, AtomicInteger unknownFilenameCount) throws IOException {
NPOIFSFileSystem fs = null;
byte[] ret = null;
try {
fs = new NPOIFSFileSystem(is);
DirectoryNode root = fs.getRoot();
if (root == null) {
return ret;
}
if (root.hasEntry("Package")) {
Entry ooxml = root.getEntry("Package");
TikaInputStream stream =
TikaInputStream.get(new DocumentInputStream((DocumentEntry) ooxml));
ByteArrayOutputStream out = new ByteArrayOutputStream();
IOUtils.copy(stream, out);
ret = out.toByteArray();
} else {
// try poifs
POIFSDocumentType type = POIFSDocumentType.detectType(root);
if (type == POIFSDocumentType.OLE10_NATIVE) {
try {
// Try to un-wrap the OLE10Native record:
Ole10Native ole = Ole10Native.createFromEmbeddedOleObject(root);
ret = ole.getDataBuffer();
} catch (Ole10NativeException ex) {
// Not a valid OLE10Native record, skip it
}
} else if (type == POIFSDocumentType.COMP_OBJ) {
DocumentEntry contentsEntry;
try {
contentsEntry = (DocumentEntry) root.getEntry("CONTENTS");
} catch (FileNotFoundException ioe) {
contentsEntry = (DocumentEntry) root.getEntry("Contents");
}
DocumentInputStream inp = null;
try {
inp = new DocumentInputStream(contentsEntry);
ret = new byte[contentsEntry.getSize()];
inp.readFully(ret);
} finally {
if (inp != null) {
inp.close();
}
}
} else {
ByteArrayOutputStream out = new ByteArrayOutputStream();
is.reset();
IOUtils.copy(is, out);
ret = out.toByteArray();
metadata.set(
Metadata.RESOURCE_NAME_KEY,
"file_" + unknownFilenameCount.getAndIncrement() + "." + type.getExtension());
metadata.set(Metadata.CONTENT_TYPE, type.getType().toString());
}
}
} finally {
if (fs != null) {
fs.close();
}
}
return ret;
}
