Examples with TikaConfig org.apache.tika.config.TikaConfig used on opensource projects

Example 6 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class TikaConfigTest method testDynamicServiceLoaderFromConfig.

@Test
public void testDynamicServiceLoaderFromConfig() throws Exception {
    URL url = TikaConfigTest.class.getResource("TIKA-1700-dynamic.xml");
    TikaConfig config = new TikaConfig(url);
    DummyParser parser = (DummyParser) config.getParser();
    ServiceLoader loader = parser.getLoader();
    boolean dynamicValue = loader.isDynamic();
    assertTrue("Dynamic Service Loading Should be true", dynamicValue);
}

Example 7 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class MyFirstTika method main.

public static void main(String[] args) throws Exception {
    String filename = args[0];
    TikaConfig tikaConfig = TikaConfig.getDefaultConfig();
    Metadata metadata = new Metadata();
    String text = parseUsingComponents(filename, tikaConfig, metadata);
    System.out.println("Parsed Metadata: ");
    System.out.println(metadata);
    System.out.println("Parsed Text: ");
    System.out.println(text);
    System.out.println("-------------------------");
    metadata = new Metadata();
    text = parseUsingAutoDetect(filename, tikaConfig, metadata);
    System.out.println("Parsed Metadata: ");
    System.out.println(metadata);
    System.out.println("Parsed Text: ");
    System.out.println(text);
}

Example 8 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class MimeUtilTest method assertResult.

private void assertResult(String contentType, String expected) throws MimeTypeException {
    TikaConfig tikaConfig = TikaConfig.getDefaultConfig();
    MimeTypes r = tikaConfig.getMimeRepository();
    MimeType mt = r.forName(contentType);
    //        String ext = MimeUtil.getExtension(contentType, config);
    assertEquals(expected, mt.getExtension());
}

Example 9 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class TikaCLI method dumpConfig.

private void dumpConfig(TikaConfigSerializer.Mode mode) throws Exception {
    TikaConfig localConfig = (config == null) ? TikaConfig.getDefaultConfig() : config;
    TikaConfigSerializer.serialize(localConfig, mode, new OutputStreamWriter(System.out, UTF_8), UTF_8);
}

Example 10 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class TikaCLI method compareFileMagic.

/**
     * Compares our mime types registry with the File(1) tool's 
     *  directory of (uncompiled) Magic entries. 
     * (Well, those with mimetypes anyway)
     * @param magicDir Path to the magic directory
     */
private void compareFileMagic(String magicDir) throws Exception {
    Set<String> tikaLacking = new TreeSet<String>();
    Set<String> tikaNoMagic = new TreeSet<String>();
    // Sanity check
    File dir = new File(magicDir);
    if ((new File(dir, "elf")).exists() && (new File(dir, "mime")).exists() && (new File(dir, "vorbis")).exists()) {
    // Looks plausible
    } else {
        throw new IllegalArgumentException(magicDir + " doesn't seem to hold uncompressed file magic entries");
    }
    // Find all the mimetypes in the directory
    Set<String> fileMimes = new HashSet<String>();
    for (File mf : dir.listFiles()) {
        if (mf.isFile()) {
            BufferedReader r = new BufferedReader(new InputStreamReader(new FileInputStream(mf), UTF_8));
            String line;
            while ((line = r.readLine()) != null) {
                if (line.startsWith("!:mime") || line.startsWith("#!:mime")) {
                    String mime = line.substring(7).trim();
                    fileMimes.add(mime);
                }
            }
            r.close();
        }
    }
    // See how those compare to the Tika ones
    TikaConfig config = TikaConfig.getDefaultConfig();
    MimeTypes mimeTypes = config.getMimeRepository();
    MediaTypeRegistry registry = config.getMediaTypeRegistry();
    for (String mime : fileMimes) {
        try {
            final MimeType type = mimeTypes.getRegisteredMimeType(mime);
            if (type == null) {
                // Tika doesn't know about this one
                tikaLacking.add(mime);
            } else {
                // Tika knows about this one!
                // Does Tika have magic for it?
                boolean hasMagic = type.hasMagic();
                // How about the children?
                if (!hasMagic) {
                    for (MediaType child : registry.getChildTypes(type.getType())) {
                        MimeType childType = mimeTypes.getRegisteredMimeType(child.toString());
                        if (childType != null && childType.hasMagic()) {
                            hasMagic = true;
                        }
                    }
                }
                // How about the parents?
                MimeType parentType = type;
                while (parentType != null && !hasMagic) {
                    if (parentType.hasMagic()) {
                        // Has magic, fine
                        hasMagic = true;
                    } else {
                        // Check the parent next
                        MediaType parent = registry.getSupertype(type.getType());
                        if (parent == MediaType.APPLICATION_XML || parent == MediaType.TEXT_PLAIN || parent == MediaType.OCTET_STREAM) {
                            // Stop checking parents if we hit a top level type
                            parent = null;
                        }
                        if (parent != null) {
                            parentType = mimeTypes.getRegisteredMimeType(parent.toString());
                        } else {
                            parentType = null;
                        }
                    }
                }
                if (!hasMagic) {
                    tikaNoMagic.add(mime);
                }
            }
        } catch (MimeTypeException e) {
        // Broken entry in the file magic directory
        // Silently skip
        }
    }
    // Check how many tika knows about
    int tikaTypes = 0;
    int tikaAliases = 0;
    for (MediaType type : registry.getTypes()) {
        tikaTypes++;
        tikaAliases += registry.getAliases(type).size();
    }
    // Report
    System.out.println("Tika knows about " + tikaTypes + " unique mime types");
    System.out.println("Tika knows about " + (tikaTypes + tikaAliases) + " mime types including aliases");
    System.out.println("The File Magic directory knows about " + fileMimes.size() + " unique mime types");
    System.out.println();
    System.out.println("The following mime types are known to File but not Tika:");
    for (String mime : tikaLacking) {
        System.out.println("  " + mime);
    }
    System.out.println();
    System.out.println("The following mime types from File have no Tika magic (but their children might):");
    for (String mime : tikaNoMagic) {
        System.out.println("  " + mime);
    }
}