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

Example 1 with CompositeDetector

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

the class TikaDetectorConfigTest method testPSTDetectionWithoutZipDetector.

/**
     * TIKA-1708 - If the Zip detector is disabled, either explicitly,
     *  or via giving a list of detectors that it isn't part of, ensure
     *  that detection of PST files still works
     */
@Test
public void testPSTDetectionWithoutZipDetector() throws Exception {
    // Check the one with an exclude
    TikaConfig configWX = getConfig("TIKA-1708-detector-default.xml");
    assertNotNull(configWX.getParser());
    assertNotNull(configWX.getDetector());
    CompositeDetector detectorWX = (CompositeDetector) configWX.getDetector();
    // Check it has the POIFS one, but not the zip one
    assertDetectors(detectorWX, true, false);
    // Check the one with an explicit list
    TikaConfig configCL = getConfig("TIKA-1708-detector-composite.xml");
    assertNotNull(configCL.getParser());
    assertNotNull(configCL.getDetector());
    CompositeDetector detectorCL = (CompositeDetector) configCL.getDetector();
    assertEquals(2, detectorCL.getDetectors().size());
    // Check it also has the POIFS one, but not the zip one
    assertDetectors(detectorCL, true, false);
    // Check that both detectors have a mimetypes with entries
    assertTrue("Not enough mime types: " + configWX.getMediaTypeRegistry().getTypes().size(), configWX.getMediaTypeRegistry().getTypes().size() > 100);
    assertTrue("Not enough mime types: " + configCL.getMediaTypeRegistry().getTypes().size(), configCL.getMediaTypeRegistry().getTypes().size() > 100);
    // Now check they detect PST files correctly
    TikaInputStream stream = TikaInputStream.get(getResourceAsFile("/test-documents/testPST.pst"));
    assertEquals(OutlookPSTParser.MS_OUTLOOK_PST_MIMETYPE, detectorWX.detect(stream, new Metadata()));
    assertEquals(OutlookPSTParser.MS_OUTLOOK_PST_MIMETYPE, detectorCL.detect(stream, new Metadata()));
}

Example 2 with CompositeDetector

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

the class TikaCLI method displayDetector.

private void displayDetector(Detector d, int i) {
    boolean isComposite = (d instanceof CompositeDetector);
    String name = d.getClass().getName();
    System.out.println(indent(i) + name + (isComposite ? " (Composite Detector):" : ""));
    if (isComposite) {
        List<Detector> subDetectors = ((CompositeDetector) d).getDetectors();
        for (Detector sd : subDetectors) {
            displayDetector(sd, i + 2);
        }
    }
}

Example 3 with CompositeDetector

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

the class DumpTikaConfigExampleTest method testDump.

@Test
public void testDump() throws Exception {
    DumpTikaConfigExample ex = new DumpTikaConfigExample();
    for (Charset charset : new Charset[] { UTF_8, UTF_16LE }) {
        for (TikaConfigSerializer.Mode mode : TikaConfigSerializer.Mode.values()) {
            Writer writer = new OutputStreamWriter(new FileOutputStream(configFile), charset);
            TikaConfigSerializer.serialize(TikaConfig.getDefaultConfig(), mode, writer, charset);
            writer.flush();
            writer.close();
            TikaConfig c = new TikaConfig(configFile);
            assertTrue(c.getParser().toString(), c.getParser() instanceof CompositeParser);
            assertTrue(c.getDetector().toString(), c.getDetector() instanceof CompositeDetector);
            CompositeParser p = (CompositeParser) c.getParser();
            assertTrue("enough parsers?", p.getParsers().size() > 130);
            CompositeDetector d = (CompositeDetector) c.getDetector();
            assertTrue("enough detectors?", d.getDetectors().size() > 3);
            //just try to load it into autodetect to make sure no errors are thrown
            Parser auto = new AutoDetectParser(c);
            assertNotNull(auto);
        }
    }
}

Example 4 with CompositeDetector

use of org.apache.tika.detect.CompositeDetector 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 5 with CompositeDetector

use of org.apache.tika.detect.CompositeDetector 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");
    }
}