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

Example 6 with Detector

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

the class TikaDetectors method detectorAsHTML.

private void detectorAsHTML(Detector d, StringBuffer html, int level) {
    html.append("<h");
    html.append(level);
    html.append(">");
    String name = d.getClass().getName();
    html.append(name.substring(name.lastIndexOf('.') + 1));
    html.append("</h");
    html.append(level);
    html.append(">");
    html.append("<p>Class: ");
    html.append(name);
    html.append("</p>");
    if (d instanceof CompositeDetector) {
        html.append("<p>Composite Detector</p>");
        for (Detector cd : ((CompositeDetector) d).getDetectors()) {
            detectorAsHTML(cd, html, level + 1);
        }
    }
}

Example 7 with Detector

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

the class TikaDetectors method renderDetector.

private void renderDetector(Detector d, StringBuffer text, int indent) {
    boolean isComposite = (d instanceof CompositeDetector);
    String name = d.getClass().getName();
    for (int i = 0; i < indent; i++) {
        text.append("  ");
    }
    text.append(name);
    if (isComposite) {
        text.append(" (Composite Detector):\n");
        List<Detector> subDetectors = ((CompositeDetector) d).getDetectors();
        for (Detector sd : subDetectors) {
            renderDetector(sd, text, indent + 1);
        }
    } else {
        text.append("\n");
    }
}

Example 8 with Detector

use of org.apache.tika.detect.Detector 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 9 with Detector

use of org.apache.tika.detect.Detector in project ddf by codice.

the class OperationsMetacardSupport method guessMimeType.

// package-private for unit testing
String guessMimeType(String mimeTypeRaw, String fileName, Path tmpContentPath) throws IOException {
    if (ContentItem.DEFAULT_MIME_TYPE.equals(mimeTypeRaw)) {
        try (InputStream inputStreamMessageCopy = com.google.common.io.Files.asByteSource(tmpContentPath.toFile()).openStream()) {
            String mimeTypeGuess = frameworkProperties.getMimeTypeMapper().guessMimeType(inputStreamMessageCopy, FilenameUtils.getExtension(fileName));
            if (StringUtils.isNotEmpty(mimeTypeGuess)) {
                mimeTypeRaw = mimeTypeGuess;
            }
        } catch (MimeTypeResolutionException e) {
            LOGGER.debug(MIME_TYPE_MSG, e);
        }
        if (ContentItem.DEFAULT_MIME_TYPE.equals(mimeTypeRaw)) {
            Detector detector = new DefaultProbDetector();
            try (InputStream inputStreamMessageCopy = TikaInputStream.get(tmpContentPath)) {
                MediaType mediaType = detector.detect(inputStreamMessageCopy, new Metadata());
                mimeTypeRaw = mediaType.toString();
            } catch (IOException e) {
                LOGGER.debug(MIME_TYPE_MSG, e);
            }
        }
        if (mimeTypeRaw.equals("text/plain")) {
            try (InputStream inputStreamMessageCopy = com.google.common.io.Files.asByteSource(tmpContentPath.toFile()).openStream();
                BufferedReader bufferedReader = new BufferedReader(new InputStreamReader(inputStreamMessageCopy, Charset.forName("UTF-8")))) {
                String line = bufferedReader.lines().map(String::trim).filter(StringUtils::isNotEmpty).findFirst().orElse("");
                if (line.startsWith("<")) {
                    mimeTypeRaw = "text/xml";
                } else if (line.startsWith("{") || line.startsWith("[")) {
                    mimeTypeRaw = "application/json";
                }
            } catch (IOException e) {
                LOGGER.debug(MIME_TYPE_MSG, e);
            }
        }
    }
    return mimeTypeRaw;
}

Example 10 with Detector

use of org.apache.tika.detect.Detector in project winery by eclipse.

the class BackendUtils method getMimeType.

/**
 * Detect the mime type of the stream. The stream is marked at the beginning and reset at the end
 *
 * @param bis the stream
 * @param fn  the fileName of the file belonging to the stream
 */
public static MediaType getMimeType(BufferedInputStream bis, String fn) throws IOException {
    AutoDetectParser parser = new AutoDetectParser();
    Detector detector = parser.getDetector();
    Metadata md = new Metadata();
    md.add(Metadata.RESOURCE_NAME_KEY, fn);
    return detector.detect(bis, md);
}