Examples with Tika org.apache.tika.Tika used on opensource projects

Example 46 with Tika

use of org.apache.tika.Tika in project tika by apache.

the class AudioParserTest method testAIFF.

@Test
public void testAIFF() throws Exception {
    String path = "/test-documents/testAIFF.aif";
    Metadata metadata = new Metadata();
    String content = new Tika().parseToString(AudioParserTest.class.getResourceAsStream(path), metadata);
    assertEquals("audio/x-aiff", metadata.get(Metadata.CONTENT_TYPE));
    assertEquals("44100.0", metadata.get("samplerate"));
    assertEquals("2", metadata.get("channels"));
    assertEquals("16", metadata.get("bits"));
    assertEquals("PCM_SIGNED", metadata.get("encoding"));
    assertEquals("", content);
}

Example 47 with Tika

use of org.apache.tika.Tika in project tika by apache.

the class ObjectRecognitionParserTest method jpegTesorflowTest.

@Ignore("If tensorflow not available Ignore")
@Test
public void jpegTesorflowTest() throws IOException, TikaException, SAXException {
    try (InputStream stream = loader.getResourceAsStream(CONFIG_FILE)) {
        assert stream != null;
        Tika tika = new Tika(new TikaConfig(stream));
        Metadata metadata = new Metadata();
        try (InputStream imageStream = loader.getResourceAsStream(CAT_IMAGE)) {
            Reader reader = tika.parse(imageStream, metadata);
            List<String> lines = IOUtils.readLines(reader);
            String text = StringUtils.join(lines, " ");
            String[] expectedObjects = { "Egyptian cat", "tabby, tabby cat" };
            String metaValues = StringUtils.join(metadata.getValues(ObjectRecognitionParser.MD_KEY), " ");
            for (String expectedObject : expectedObjects) {
                String message = "'" + expectedObject + "' must have been detected";
                Assert.assertTrue(message, text.contains(expectedObject));
                Assert.assertTrue(message, metaValues.contains(expectedObject));
            }
        }
    }
}

Example 48 with Tika

use of org.apache.tika.Tika in project tika by apache.

the class ObjectRecognitionParserTest method testREST.

@Ignore("Configure Rest API service")
@Test
public void testREST() throws Exception {
    try (InputStream stream = loader.getResourceAsStream(CONFIG_REST_FILE)) {
        assert stream != null;
        Tika tika = new Tika(new TikaConfig(stream));
        Metadata metadata = new Metadata();
        try (InputStream imageStream = loader.getResourceAsStream(CAT_IMAGE)) {
            Reader reader = tika.parse(imageStream, metadata);
            String text = IOUtils.toString(reader);
            String[] expectedObjects = { "Egyptian cat", "tabby, tabby cat" };
            String metaValues = StringUtils.join(metadata.getValues(ObjectRecognitionParser.MD_KEY), " ");
            for (String expectedObject : expectedObjects) {
                String message = "'" + expectedObject + "' must have been detected";
                Assert.assertTrue(message, text.contains(expectedObject));
                Assert.assertTrue(message, metaValues.contains(expectedObject));
            }
        }
    }
}

Example 49 with Tika

use of org.apache.tika.Tika in project tika by apache.

the class FLVParserTest method testFLV.

@Test
public void testFLV() throws Exception {
    String path = "/test-documents/testFLV.flv";
    Metadata metadata = new Metadata();
    String content = new Tika().parseToString(FLVParserTest.class.getResourceAsStream(path), metadata);
    assertEquals("", content);
    assertEquals("video/x-flv", metadata.get(Metadata.CONTENT_TYPE));
    assertEquals("true", metadata.get("hasVideo"));
    assertEquals("false", metadata.get("stereo"));
    assertEquals("true", metadata.get("hasAudio"));
    assertEquals("120.0", metadata.get("height"));
    assertEquals("16.0", metadata.get("audiosamplesize"));
}

Example 50 with Tika

use of org.apache.tika.Tika in project ddf by codice.

the class URLResourceReader method getMimeType.

private String getMimeType(URI resourceURI, String productName) throws MimeTypeResolutionException, IOException {
    // Determine the mime type in a hierarchical fashion. The hierarchy is based on the
    // most accurate mime type resolution being used and lesser accurate approaches being
    // used
    // if a mime type is not resolved.
    // The approaches, in order, are:
    // 1. Try using the DDF MimeTypeMapper so that custom MimeTypeResolvers are used
    // 2. Try using Apache Tika directly on the URL
    String mimeType = null;
    if (mimeTypeMapper == null) {
        LOGGER.debug("mimeTypeMapper is NULL");
    } else {
        // Extract the file extension (if any) from the URL's filename
        String fileExtension = FilenameUtils.getExtension(productName);
        mimeType = mimeTypeMapper.getMimeTypeForFileExtension(fileExtension);
    }
    // mime type resolution than just file extension mime type mapping
    if ((mimeType == null || mimeType.isEmpty() || mimeType.equals(DEFAULT_MIME_TYPE)) && URL_FILE_SCHEME.equalsIgnoreCase(resourceURI.getScheme())) {
        // Use Apache Tika to detect mime type from URL
        Tika tika = new Tika();
        mimeType = tika.detect(resourceURI.toURL());
        LOGGER.debug("Tika determined mimeType for url = {}", mimeType);
    } else {
        LOGGER.debug("mimeType = {} set by MimeTypeMapper", mimeType);
    }
    // never be returned.
    if (mimeType == null || mimeType.equals("content/unknown")) {
        mimeType = "application/unknown";
    }
    LOGGER.debug("mimeType set to: {}", mimeType);
    return mimeType;
}