Examples with AbstractEncodingDetectorParser org.apache.tika.parser.AbstractEncodingDetectorParser used on opensource projects

Example 1 with AbstractEncodingDetectorParser

use of org.apache.tika.parser.AbstractEncodingDetectorParser in project tika by apache.

the class TikaEncodingDetectorTest method testConfigurabilityOfUserSpecified.

@Test
public void testConfigurabilityOfUserSpecified() throws Exception {
    TikaConfig tikaConfig = new TikaConfig(getResourceAsStream("/org/apache/tika/config/TIKA-2273-encoding-detector-outside-static-init.xml"));
    AutoDetectParser p = new AutoDetectParser(tikaConfig);
    //make sure that all static and non-static parsers are using the same encoding detector!
    List<Parser> parsers = new ArrayList<>();
    findEncodingDetectionParsers(p, parsers);
    assertEquals(3, parsers.size());
    for (Parser encodingDetectingParser : parsers) {
        EncodingDetector encodingDetector = ((AbstractEncodingDetectorParser) encodingDetectingParser).getEncodingDetector();
        assertTrue(encodingDetector instanceof CompositeEncodingDetector);
        assertEquals(2, ((CompositeEncodingDetector) encodingDetector).getDetectors().size());
        for (EncodingDetector child : ((CompositeEncodingDetector) encodingDetector).getDetectors()) {
            assertNotContained("cu4j", child.getClass().getCanonicalName());
        }
    }
    //also just make sure this is still true
    try {
        Metadata metadata = getXML("english.cp500.txt", p).metadata;
        fail("can't detect w/out ICU");
    } catch (TikaException e) {
        assertContains("Failed to detect", e.getMessage());
    }
}

Example 2 with AbstractEncodingDetectorParser

use of org.apache.tika.parser.AbstractEncodingDetectorParser in project tika by apache.

the class TikaEncodingDetectorTest method testNonDetectingDetectorParams.

@Test
public void testNonDetectingDetectorParams() throws Exception {
    TikaConfig tikaConfig = new TikaConfig(getResourceAsStream("/org/apache/tika/config/TIKA-2273-non-detecting-params.xml"));
    AutoDetectParser p = new AutoDetectParser(tikaConfig);
    List<Parser> parsers = new ArrayList<>();
    findEncodingDetectionParsers(p, parsers);
    assertEquals(3, parsers.size());
    EncodingDetector encodingDetector = ((AbstractEncodingDetectorParser) parsers.get(0)).getEncodingDetector();
    assertTrue(encodingDetector instanceof CompositeEncodingDetector);
    assertEquals(1, ((CompositeEncodingDetector) encodingDetector).getDetectors().size());
    EncodingDetector child = ((CompositeEncodingDetector) encodingDetector).getDetectors().get(0);
    assertTrue(child instanceof NonDetectingEncodingDetector);
    assertEquals(StandardCharsets.UTF_16LE, ((NonDetectingEncodingDetector) child).getCharset());
}

Example 3 with AbstractEncodingDetectorParser

use of org.apache.tika.parser.AbstractEncodingDetectorParser in project tika by apache.

the class TikaEncodingDetectorTest method testEncodingDetectorsAreLoaded.

@Test
public void testEncodingDetectorsAreLoaded() {
    EncodingDetector encodingDetector = ((AbstractEncodingDetectorParser) new TXTParser()).getEncodingDetector();
    assertTrue(encodingDetector instanceof CompositeEncodingDetector);
}