Examples with DefaultDetector org.apache.tika.detect.DefaultDetector used on opensource projects

Example 1 with DefaultDetector

use of org.apache.tika.detect.DefaultDetector in project tika by apache.

the class RFC822ParserTest method testExtractAttachments.

@Test
public void testExtractAttachments() throws Exception {
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    Parser p = new RFC822Parser();
    ParseContext context = new ParseContext();
    try (InputStream stream = getStream("test-documents/testEmailWithPNGAtt.eml")) {
        p.parse(stream, handler, metadata, context);
    }
    // Check we go the metadata
    assertEquals("Tika Test <XXXX@apache.org>", metadata.get(Metadata.MESSAGE_FROM));
    assertEquals("Test Attachment Email", metadata.get(TikaCoreProperties.TITLE));
    // Try again with attachment detecting and fetching
    final Detector detector = new DefaultDetector();
    final Parser extParser = new AutoDetectParser();
    final List<MediaType> seenTypes = new ArrayList<MediaType>();
    final List<String> seenText = new ArrayList<String>();
    EmbeddedDocumentExtractor ext = new EmbeddedDocumentExtractor() {

        @Override
        public boolean shouldParseEmbedded(Metadata metadata) {
            return true;
        }

        @Override
        public void parseEmbedded(InputStream stream, ContentHandler handler, Metadata metadata, boolean outputHtml) throws SAXException, IOException {
            seenTypes.add(detector.detect(stream, metadata));
            ContentHandler h = new BodyContentHandler();
            try {
                extParser.parse(stream, h, metadata, new ParseContext());
            } catch (TikaException e) {
                throw new RuntimeException(e);
            }
            seenText.add(h.toString());
        }
    };
    context.set(EmbeddedDocumentExtractor.class, ext);
    try (InputStream stream = getStream("test-documents/testEmailWithPNGAtt.eml")) {
        p.parse(stream, handler, metadata, context);
    }
    // Check we go the metadata
    assertEquals("Tika Test <XXXX@apache.org>", metadata.get(Metadata.MESSAGE_FROM));
    assertEquals("Test Attachment Email", metadata.get(TikaCoreProperties.TITLE));
    // Check attachments
    assertEquals(2, seenTypes.size());
    assertEquals(2, seenText.size());
    assertEquals("text/plain", seenTypes.get(0).toString());
    assertEquals("image/png", seenTypes.get(1).toString());
    assertEquals("This email has a PNG attachment included in it\n\n", seenText.get(0));
}

Example 2 with DefaultDetector

use of org.apache.tika.detect.DefaultDetector in project tika by apache.

the class OOXMLParserTest method testExcelXLSB.

@Test
public void testExcelXLSB() throws Exception {
    Detector detector = new DefaultDetector();
    AutoDetectParser parser = new AutoDetectParser();
    Metadata m = new Metadata();
    m.add(Metadata.RESOURCE_NAME_KEY, "excel.xlsb");
    // Should be detected correctly
    MediaType type;
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL.xlsb")) {
        type = detector.detect(input, m);
        assertEquals("application/vnd.ms-excel.sheet.binary.macroenabled.12", type.toString());
    }
    // OfficeParser won't handle it
    assertEquals(false, (new OfficeParser()).getSupportedTypes(new ParseContext()).contains(type));
    // OOXMLParser will (soon) handle it
    assertTrue((new OOXMLParser()).getSupportedTypes(new ParseContext()).contains(type));
    // AutoDetectParser doesn't break on it
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL.xlsb")) {
        ContentHandler handler = new BodyContentHandler(-1);
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.US);
        parser.parse(input, handler, m, context);
        String content = handler.toString();
        assertContains("This is an example spreadsheet", content);
    }
}

Example 3 with DefaultDetector

use of org.apache.tika.detect.DefaultDetector in project tika by apache.

the class Activator method start.

@Override
public void start(BundleContext context) throws Exception {
    detectorService = context.registerService(Detector.class.getName(), new DefaultDetector(Activator.class.getClassLoader()), new Properties());
    Parser parser = new DefaultParser(Activator.class.getClassLoader());
    parserService = context.registerService(Parser.class.getName(), parser, new Properties());
}

Example 4 with DefaultDetector

use of org.apache.tika.detect.DefaultDetector in project tika by apache.

the class BundleIT method testBundleDetectors.

@Test
public void testBundleDetectors() throws Exception {
    //For some reason, the detector created by OSGi has a flat
    //list of detectors, whereas the detector created by the traditional
    //service loading method has children: DefaultDetector, MimeTypes.
    //We have to flatten the service loaded DefaultDetector to get equivalence.
    //Detection behavior should all be the same.
    // Get the classes found within OSGi
    ServiceReference<Detector> detectorRef = bc.getServiceReference(Detector.class);
    DefaultDetector detectorService = (DefaultDetector) bc.getService(detectorRef);
    Set<String> osgiDetectors = new HashSet<>();
    for (Detector d : detectorService.getDetectors()) {
        osgiDetectors.add(d.getClass().getName());
    }
    // Check we did get a few, just in case...
    assertTrue("Should have several Detector names, found " + osgiDetectors.size(), osgiDetectors.size() > 3);
    // Get the raw detectors list from the traditional service loading mechanism
    DefaultDetector detector = new DefaultDetector();
    Set<String> rawDetectors = new HashSet<String>();
    for (Detector d : detector.getDetectors()) {
        if (d instanceof DefaultDetector) {
            for (Detector dChild : ((DefaultDetector) d).getDetectors()) {
                rawDetectors.add(dChild.getClass().getName());
            }
        } else {
            rawDetectors.add(d.getClass().getName());
        }
    }
    assertEquals(osgiDetectors, rawDetectors);
}

Example 5 with DefaultDetector

use of org.apache.tika.detect.DefaultDetector in project uPortal by Jasig.

the class JaxbPortalDataHandlerService method getMediaType.

protected MediaType getMediaType(BufferedInputStream inputStream, String fileName) throws IOException {
    final TikaInputStream tikaInputStreamStream = TikaInputStream.get(new CloseShieldInputStream(inputStream));
    try {
        final Detector detector = new DefaultDetector();
        final Metadata metadata = new Metadata();
        metadata.set(Metadata.RESOURCE_NAME_KEY, fileName);
        final MediaType type = detector.detect(tikaInputStreamStream, metadata);
        logger.debug("Determined '{}' for '{}'", type, fileName);
        return type;
    } catch (IOException e) {
        logger.warn("Failed to determine media type for '" + fileName + "' assuming XML", e);
        return null;
    } finally {
        IOUtils.closeQuietly(tikaInputStreamStream);
        //Reset the buffered stream to make up for anything read by the detector
        inputStream.reset();
    }
}