Examples with DocumentInputStream org.apache.poi.poifs.filesystem.DocumentInputStream used on opensource projects

Example 16 with DocumentInputStream

use of org.apache.poi.poifs.filesystem.DocumentInputStream in project poi by apache.

the class CryptoAPIEncryptor method getSummaryEntries.

/**
     * Encrypt the Document-/SummaryInformation and other optionally streams.
     * Opposed to other crypto modes, cryptoapi is record based and can't be used
     * to stream-encrypt a whole file
     * 
     * @see <a href="http://msdn.microsoft.com/en-us/library/dd943321(v=office.12).aspx">2.3.5.4 RC4 CryptoAPI Encrypted Summary Stream</a>
     */
public OutputStream getSummaryEntries(DirectoryNode dir) throws IOException, GeneralSecurityException {
    // NOSONAR
    CryptoAPIDocumentOutputStream bos = new CryptoAPIDocumentOutputStream(this);
    byte[] buf = new byte[8];
    // skip header
    bos.write(buf, 0, 8);
    String[] entryNames = { SummaryInformation.DEFAULT_STREAM_NAME, DocumentSummaryInformation.DEFAULT_STREAM_NAME };
    List<StreamDescriptorEntry> descList = new ArrayList<StreamDescriptorEntry>();
    int block = 0;
    for (String entryName : entryNames) {
        if (!dir.hasEntry(entryName)) {
            continue;
        }
        StreamDescriptorEntry descEntry = new StreamDescriptorEntry();
        descEntry.block = block;
        descEntry.streamOffset = bos.size();
        descEntry.streamName = entryName;
        descEntry.flags = StreamDescriptorEntry.flagStream.setValue(0, 1);
        descEntry.reserved2 = 0;
        bos.setBlock(block);
        DocumentInputStream dis = dir.createDocumentInputStream(entryName);
        IOUtils.copy(dis, bos);
        dis.close();
        descEntry.streamSize = bos.size() - descEntry.streamOffset;
        descList.add(descEntry);
        dir.getEntry(entryName).delete();
        block++;
    }
    int streamDescriptorArrayOffset = bos.size();
    bos.setBlock(0);
    LittleEndian.putUInt(buf, 0, descList.size());
    bos.write(buf, 0, 4);
    for (StreamDescriptorEntry sde : descList) {
        LittleEndian.putUInt(buf, 0, sde.streamOffset);
        bos.write(buf, 0, 4);
        LittleEndian.putUInt(buf, 0, sde.streamSize);
        bos.write(buf, 0, 4);
        LittleEndian.putUShort(buf, 0, sde.block);
        bos.write(buf, 0, 2);
        LittleEndian.putUByte(buf, 0, (short) sde.streamName.length());
        bos.write(buf, 0, 1);
        LittleEndian.putUByte(buf, 0, (short) sde.flags);
        bos.write(buf, 0, 1);
        LittleEndian.putUInt(buf, 0, sde.reserved2);
        bos.write(buf, 0, 4);
        byte[] nameBytes = StringUtil.getToUnicodeLE(sde.streamName);
        bos.write(nameBytes, 0, nameBytes.length);
        // null-termination
        LittleEndian.putShort(buf, 0, (short) 0);
        bos.write(buf, 0, 2);
    }
    int savedSize = bos.size();
    int streamDescriptorArraySize = savedSize - streamDescriptorArrayOffset;
    LittleEndian.putUInt(buf, 0, streamDescriptorArrayOffset);
    LittleEndian.putUInt(buf, 4, streamDescriptorArraySize);
    bos.reset();
    bos.setBlock(0);
    bos.write(buf, 0, 8);
    bos.setSize(savedSize);
    dir.createDocument("EncryptedSummary", new ByteArrayInputStream(bos.getBuf(), 0, savedSize));
    DocumentSummaryInformation dsi = PropertySetFactory.newDocumentSummaryInformation();
    try {
        dsi.write(dir, DocumentSummaryInformation.DEFAULT_STREAM_NAME);
    } catch (WritingNotSupportedException e) {
        throw new IOException(e);
    }
    return bos;
}

Example 17 with DocumentInputStream

use of org.apache.poi.poifs.filesystem.DocumentInputStream in project poi by apache.

the class StandardDecryptor method getDataStream.

@Override
@SuppressWarnings("resource")
public InputStream getDataStream(DirectoryNode dir) throws IOException {
    DocumentInputStream dis = dir.createDocumentInputStream(DEFAULT_POIFS_ENTRY);
    _length = dis.readLong();
    if (getSecretKey() == null) {
        verifyPassword(null);
    }
    // limit wrong calculated ole entries - (bug #57080)
    // standard encryption always uses aes encoding, so blockSize is always 16 
    // http://stackoverflow.com/questions/3283787/size-of-data-after-aes-encryption
    int blockSize = getEncryptionInfo().getHeader().getCipherAlgorithm().blockSize;
    long cipherLen = (_length / blockSize + 1) * blockSize;
    Cipher cipher = getCipher(getSecretKey());
    InputStream boundedDis = new BoundedInputStream(dis, cipherLen);
    return new BoundedInputStream(new CipherInputStream(boundedDis, cipher), _length);
}

Example 18 with DocumentInputStream

use of org.apache.poi.poifs.filesystem.DocumentInputStream in project tika by apache.

the class AbstractPOIFSExtractor method handleEmbeddedOfficeDoc.

/**
     * Handle an office document that's embedded at the POIFS level
     */
protected void handleEmbeddedOfficeDoc(DirectoryEntry dir, String resourceName, XHTMLContentHandler xhtml) throws IOException, SAXException, TikaException {
    if (dir.hasEntry("Package")) {
        // It's OOXML (has a ZipFile):
        Entry ooxml = dir.getEntry("Package");
        try (TikaInputStream stream = TikaInputStream.get(new DocumentInputStream((DocumentEntry) ooxml))) {
            ZipContainerDetector detector = new ZipContainerDetector();
            MediaType type = null;
            try {
                //if there's a stream error while detecting...
                type = detector.detect(stream, new Metadata());
            } catch (Exception e) {
                EmbeddedDocumentUtil.recordEmbeddedStreamException(e, parentMetadata);
                return;
            }
            handleEmbeddedResource(stream, null, dir.getName(), dir.getStorageClsid(), type.toString(), xhtml, true);
            return;
        }
    }
    // It's regular OLE2:
    // What kind of document is it?
    Metadata metadata = new Metadata();
    metadata.set(Metadata.EMBEDDED_RELATIONSHIP_ID, dir.getName());
    if (dir.getStorageClsid() != null) {
        metadata.set(Metadata.EMBEDDED_STORAGE_CLASS_ID, dir.getStorageClsid().toString());
    }
    POIFSDocumentType type = POIFSDocumentType.detectType(dir);
    TikaInputStream embedded = null;
    String rName = (resourceName == null) ? dir.getName() : resourceName;
    try {
        if (type == POIFSDocumentType.OLE10_NATIVE) {
            try {
                // Try to un-wrap the OLE10Native record:
                Ole10Native ole = Ole10Native.createFromEmbeddedOleObject((DirectoryNode) dir);
                if (ole.getLabel() != null) {
                    metadata.set(Metadata.RESOURCE_NAME_KEY, rName + '/' + ole.getLabel());
                }
                if (ole.getCommand() != null) {
                    metadata.add(TikaCoreProperties.ORIGINAL_RESOURCE_NAME, ole.getCommand());
                }
                if (ole.getFileName() != null) {
                    metadata.add(TikaCoreProperties.ORIGINAL_RESOURCE_NAME, ole.getFileName());
                }
                byte[] data = ole.getDataBuffer();
                embedded = TikaInputStream.get(data);
            } catch (Ole10NativeException ex) {
            // Not a valid OLE10Native record, skip it
            } catch (Exception e) {
                EmbeddedDocumentUtil.recordEmbeddedStreamException(e, parentMetadata);
                return;
            }
        } else if (type == POIFSDocumentType.COMP_OBJ) {
            try {
                //TODO: figure out if the equivalent of OLE 1.0's
                //getCommand() and getFileName() exist for OLE 2.0 to populate
                //TikaCoreProperties.ORIGINAL_RESOURCE_NAME
                // Grab the contents and process
                DocumentEntry contentsEntry;
                try {
                    contentsEntry = (DocumentEntry) dir.getEntry("CONTENTS");
                } catch (FileNotFoundException ioe) {
                    contentsEntry = (DocumentEntry) dir.getEntry("Contents");
                }
                DocumentInputStream inp = new DocumentInputStream(contentsEntry);
                byte[] contents = new byte[contentsEntry.getSize()];
                inp.readFully(contents);
                embedded = TikaInputStream.get(contents);
                // Try to work out what it is
                MediaType mediaType = getDetector().detect(embedded, new Metadata());
                String extension = type.getExtension();
                try {
                    MimeType mimeType = getMimeTypes().forName(mediaType.toString());
                    extension = mimeType.getExtension();
                } catch (MimeTypeException mte) {
                // No details on this type are known
                }
                // Record what we can do about it
                metadata.set(Metadata.CONTENT_TYPE, mediaType.getType().toString());
                metadata.set(Metadata.RESOURCE_NAME_KEY, rName + extension);
            } catch (Exception e) {
                EmbeddedDocumentUtil.recordEmbeddedStreamException(e, parentMetadata);
                return;
            }
        } else {
            metadata.set(Metadata.CONTENT_TYPE, type.getType().toString());
            metadata.set(Metadata.RESOURCE_NAME_KEY, rName + '.' + type.getExtension());
        }
        // Should we parse it?
        if (embeddedDocumentUtil.shouldParseEmbedded(metadata)) {
            if (embedded == null) {
                // Make a TikaInputStream that just
                // passes the root directory of the
                // embedded document, and is otherwise
                // empty (byte[0]):
                embedded = TikaInputStream.get(new byte[0]);
                embedded.setOpenContainer(dir);
            }
            embeddedDocumentUtil.parseEmbedded(embedded, xhtml, metadata, true);
        }
    } catch (IOException e) {
        EmbeddedDocumentUtil.recordEmbeddedStreamException(e, metadata);
    } finally {
        if (embedded != null) {
            embedded.close();
        }
    }
}

Example 19 with DocumentInputStream

use of org.apache.poi.poifs.filesystem.DocumentInputStream in project tika by apache.

the class SummaryExtractor method parseSummaryEntryIfExists.

private void parseSummaryEntryIfExists(DirectoryNode root, String entryName) throws IOException, TikaException {
    try {
        DocumentEntry entry = (DocumentEntry) root.getEntry(entryName);
        PropertySet properties = new PropertySet(new DocumentInputStream(entry));
        if (properties.isSummaryInformation()) {
            parse(new SummaryInformation(properties));
        }
        if (properties.isDocumentSummaryInformation()) {
            parse(new DocumentSummaryInformation(properties));
        }
    } catch (FileNotFoundException e) {
    // entry does not exist, just skip it
    } catch (NoPropertySetStreamException e) {
    // no property stream, just skip it
    } catch (UnexpectedPropertySetTypeException e) {
        throw new TikaException("Unexpected HPSF document", e);
    } catch (MarkUnsupportedException e) {
        throw new TikaException("Invalid DocumentInputStream", e);
    } catch (Exception e) {
        LOG.warn("Ignoring unexpected exception while parsing summary entry {}", entryName, e);
    }
}

Example 20 with DocumentInputStream

use of org.apache.poi.poifs.filesystem.DocumentInputStream in project poi by apache.

the class TestWriteWellKnown method getSummaryInformation.

private static SummaryInformation getSummaryInformation(NPOIFSFileSystem poifs) throws Exception {
    DocumentInputStream dis = poifs.createDocumentInputStream(SummaryInformation.DEFAULT_STREAM_NAME);
    PropertySet ps = new PropertySet(dis);
    SummaryInformation si = new SummaryInformation(ps);
    dis.close();
    return si;
}