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

Example 26 with DirectoryNode

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

the class TestEncryptor method listDir.

@SuppressWarnings("unused")
private void listDir(DirectoryNode dn, String ext, String path) throws IOException {
    path += "\\" + dn.getName().replace('', '_');
    System.out.println(ext + ": " + path + " (" + dn.getStorageClsid() + ")");
    Iterator<Entry> iter = dn.getEntries();
    while (iter.hasNext()) {
        Entry ent = iter.next();
        if (ent instanceof DirectoryNode) {
            listDir((DirectoryNode) ent, ext, path);
        } else {
            listEntry((DocumentNode) ent, ext, path);
        }
    }
}

Example 27 with DirectoryNode

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

the class QPWTextExtractor method extract.

@SuppressWarnings("resource")
public void extract(InputStream input, XHTMLContentHandler xhtml, Metadata metadata) throws IOException, SAXException, TikaException {
    POIFSFileSystem pfs = new POIFSFileSystem(input);
    DirectoryNode rootNode = pfs.getRoot();
    if (rootNode == null || !rootNode.hasEntry(OLE_DOCUMENT_NAME)) {
        throw new UnsupportedFormatException("Unsupported QuattroPro file format. " + "Looking for OLE entry \"" + OLE_DOCUMENT_NAME + "\". Found: " + (rootNode == null ? "null" : rootNode.getEntryNames()));
    }
    //TODO shall we validate and throw warning/error if the file does not 
    //start with a BOF and ends with a EOF?
    xhtml.startElement("p");
    try (WPInputStream in = new WPInputStream(pfs.createDocumentInputStream(OLE_DOCUMENT_NAME))) {
        Context ctx = new Context(in, xhtml, metadata);
        while (hasNext(in)) {
            ctx.type = in.readWPShort();
            ctx.bodyLength = in.readWPShort();
            Extractor extractor = EXTRACTORS.get(ctx.type);
            if (extractor != null) {
                extractor.extract(ctx);
            } else {
                // Use DEBUG to find out what we are ignoring
                //                    Extractor.DEBUG.extract(ctx);
                Extractor.IGNORE.extract(ctx);
            }
        }
    }
    xhtml.endElement("p");
}

Example 28 with DirectoryNode

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

the class RTFObjDataParser method handleEmbeddedPOIFS.

//will throw IOException if not actually POIFS
//can return null byte[]
private byte[] handleEmbeddedPOIFS(InputStream is, Metadata metadata, AtomicInteger unknownFilenameCount) throws IOException {
    byte[] ret = null;
    try (NPOIFSFileSystem 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");
                }
                try (DocumentInputStream inp = new DocumentInputStream(contentsEntry)) {
                    ret = new byte[contentsEntry.getSize()];
                    inp.readFully(ret);
                }
            } 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());
            }
        }
    }
    return ret;
}

Example 29 with DirectoryNode

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

the class POIFSContainerDetector method detect.

public MediaType detect(InputStream input, Metadata metadata) throws IOException {
    // Check if we have access to the document
    if (input == null) {
        return MediaType.OCTET_STREAM;
    }
    // If this is a TikaInputStream wrapping an already
    // parsed NPOIFileSystem/DirectoryNode, just get the
    // names from the root:
    TikaInputStream tis = TikaInputStream.cast(input);
    Set<String> names = null;
    if (tis != null) {
        Object container = tis.getOpenContainer();
        if (container instanceof NPOIFSFileSystem) {
            names = getTopLevelNames(((NPOIFSFileSystem) container).getRoot());
        } else if (container instanceof DirectoryNode) {
            names = getTopLevelNames((DirectoryNode) container);
        }
    }
    if (names == null) {
        // Check if the document starts with the OLE header
        input.mark(8);
        try {
            if (input.read() != 0xd0 || input.read() != 0xcf || input.read() != 0x11 || input.read() != 0xe0 || input.read() != 0xa1 || input.read() != 0xb1 || input.read() != 0x1a || input.read() != 0xe1) {
                return MediaType.OCTET_STREAM;
            }
        } finally {
            input.reset();
        }
    }
    // We can only detect the exact type when given a TikaInputStream
    if (names == null && tis != null) {
        // Look for known top level entry names to detect the document type
        names = getTopLevelNames(tis);
    }
    // Detect based on the names (as available)
    if (tis != null && tis.getOpenContainer() != null && tis.getOpenContainer() instanceof NPOIFSFileSystem) {
        return detect(names, ((NPOIFSFileSystem) tis.getOpenContainer()).getRoot());
    } else {
        return detect(names, null);
    }
}

Example 30 with DirectoryNode

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

the class AbstractOOXMLExtractor method handleEmbeddedOLE.

/**
     * Handles an embedded OLE object in the document
     */
private void handleEmbeddedOLE(PackagePart part, ContentHandler handler, String rel, Metadata parentMetadata) throws IOException, SAXException {
    // A POIFSFileSystem needs to be at least 3 blocks big to be valid
    if (part.getSize() >= 0 && part.getSize() < 512 * 3) {
        // Too small, skip
        return;
    }
    InputStream is = part.getInputStream();
    // Open the POIFS (OLE2) structure and process
    POIFSFileSystem fs = null;
    try {
        fs = new POIFSFileSystem(part.getInputStream());
    } catch (Exception e) {
        EmbeddedDocumentUtil.recordEmbeddedStreamException(e, parentMetadata);
        return;
    }
    TikaInputStream stream = null;
    try {
        Metadata metadata = new Metadata();
        metadata.set(Metadata.EMBEDDED_RELATIONSHIP_ID, rel);
        DirectoryNode root = fs.getRoot();
        POIFSDocumentType type = POIFSDocumentType.detectType(root);
        if (root.hasEntry("CONTENTS") && root.hasEntry("
Ole") && root.hasEntry("
CompObj")) {
            // TIKA-704: OLE 2.0 embedded non-Office document?
            //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
            stream = TikaInputStream.get(fs.createDocumentInputStream("CONTENTS"));
            if (embeddedExtractor.shouldParseEmbedded(metadata)) {
                embeddedExtractor.parseEmbedded(stream, new EmbeddedContentHandler(handler), metadata, false);
            }
        } else if (POIFSDocumentType.OLE10_NATIVE == type) {
            // TIKA-704: OLE 1.0 embedded document
            Ole10Native ole = Ole10Native.createFromEmbeddedOleObject(fs);
            if (ole.getLabel() != null) {
                metadata.set(Metadata.RESOURCE_NAME_KEY, 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();
            if (data != null) {
                stream = TikaInputStream.get(data);
            }
            if (stream != null && embeddedExtractor.shouldParseEmbedded(metadata)) {
                embeddedExtractor.parseEmbedded(stream, new EmbeddedContentHandler(handler), metadata, false);
            }
        } else {
            handleEmbeddedFile(part, handler, rel);
        }
    } catch (FileNotFoundException e) {
    // There was no CONTENTS entry, so skip this part
    } catch (Ole10NativeException e) {
    // Could not process an OLE 1.0 entry, so skip this part
    } catch (IOException e) {
        EmbeddedDocumentUtil.recordEmbeddedStreamException(e, parentMetadata);
    } finally {
        if (fs != null) {
            fs.close();
        }
        if (stream != null) {
            stream.close();
        }
    }
}