Examples with BodyContentHandler org.apache.tika.sax.BodyContentHandler used on opensource projects

Example 66 with BodyContentHandler

use of org.apache.tika.sax.BodyContentHandler in project tika by apache.

the class ODFParserTest method testOO3.

@Test
public void testOO3() throws Exception {
    for (Parser parser : getParsers()) {
        try (InputStream input = ODFParserTest.class.getResourceAsStream("/test-documents/testODFwithOOo3.odt")) {
            Metadata metadata = new Metadata();
            ContentHandler handler = new BodyContentHandler();
            parser.parse(input, handler, metadata, new ParseContext());
            assertEquals("application/vnd.oasis.opendocument.text", metadata.get(Metadata.CONTENT_TYPE));
            String content = handler.toString();
            assertContains("Tika is part of the Lucene project.", content);
            assertContains("Solr", content);
            assertContains("one embedded", content);
            assertContains("Rectangle Title", content);
            assertContains("a blue background and dark border", content);
        }
    }
}

Example 67 with BodyContentHandler

use of org.apache.tika.sax.BodyContentHandler in project tika by apache.

the class ArParserTest method testEmbedded.

/**
     * Tests that the ParseContext parser is correctly fired for all the
     * embedded entries.
     */
@Test
public void testEmbedded() throws Exception {
    // Should auto-detect!
    Parser parser = new AutoDetectParser();
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    try (InputStream stream = ArParserTest.class.getResourceAsStream("/test-documents/testARofText.ar")) {
        parser.parse(stream, handler, metadata, trackingContext);
    }
    assertEquals(1, tracker.filenames.size());
    assertEquals(1, tracker.mediatypes.size());
    assertEquals(1, tracker.modifiedAts.size());
    assertEquals("testTXT.txt", tracker.filenames.get(0));
    String modifiedAt = tracker.modifiedAts.get(0);
    assertTrue("Modified at " + modifiedAt, modifiedAt.startsWith("201"));
    for (String type : tracker.mediatypes) {
        assertNull(type);
    }
    for (String crt : tracker.createdAts) {
        assertNull(crt);
    }
    tracker.reset();
    try (InputStream stream = ArParserTest.class.getResourceAsStream("/test-documents/testARofSND.ar")) {
        parser.parse(stream, handler, metadata, trackingContext);
    }
    assertEquals(1, tracker.filenames.size());
    assertEquals(1, tracker.mediatypes.size());
    assertEquals(1, tracker.modifiedAts.size());
    assertEquals("testAU.au", tracker.filenames.get(0));
    modifiedAt = tracker.modifiedAts.get(0);
    assertTrue("Modified at " + modifiedAt, modifiedAt.startsWith("201"));
    for (String type : tracker.mediatypes) {
        assertNull(type);
    }
    for (String crt : tracker.createdAts) {
        assertNull(crt);
    }
}

Example 68 with BodyContentHandler

use of org.apache.tika.sax.BodyContentHandler in project tika by apache.

the class Bzip2ParserTest method testEmbedded.

/**
     * Tests that the ParseContext parser is correctly
     *  fired for all the embedded entries.
     */
@Test
public void testEmbedded() throws Exception {
    // Should auto-detect!
    Parser parser = new AutoDetectParser();
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    try (InputStream stream = ZipParserTest.class.getResourceAsStream("/test-documents/test-documents.tbz2")) {
        parser.parse(stream, handler, metadata, trackingContext);
    }
    // Should find a single entry, for the (compressed) tar file
    assertEquals(1, tracker.filenames.size());
    assertEquals(1, tracker.mediatypes.size());
    assertEquals(1, tracker.modifiedAts.size());
    assertEquals(null, tracker.filenames.get(0));
    assertEquals(null, tracker.mediatypes.get(0));
    assertEquals(null, tracker.createdAts.get(0));
    assertEquals(null, tracker.modifiedAts.get(0));
    // Tar file starts with the directory name
    assertEquals("test-documents/", new String(tracker.lastSeenStart, 0, 15, US_ASCII));
}

Example 69 with BodyContentHandler

use of org.apache.tika.sax.BodyContentHandler in project tika by apache.

the class Bzip2ParserTest method testBzip2Parsing.

@Test
public void testBzip2Parsing() throws Exception {
    // Should auto-detect!
    Parser parser = new AutoDetectParser();
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    try (InputStream stream = Bzip2ParserTest.class.getResourceAsStream("/test-documents/test-documents.tbz2")) {
        parser.parse(stream, handler, metadata, recursingContext);
    }
    assertEquals("application/x-bzip2", metadata.get(Metadata.CONTENT_TYPE));
    String content = handler.toString();
    assertContains("test-documents/testEXCEL.xls", content);
    assertContains("Sample Excel Worksheet", content);
    assertContains("test-documents/testHTML.html", content);
    assertContains("Test Indexation Html", content);
    assertContains("test-documents/testOpenOffice2.odt", content);
    assertContains("This is a sample Open Office document", content);
    assertContains("test-documents/testPDF.pdf", content);
    assertContains("Apache Tika", content);
    assertContains("test-documents/testPPT.ppt", content);
    assertContains("Sample Powerpoint Slide", content);
    assertContains("test-documents/testRTF.rtf", content);
    assertContains("indexation Word", content);
    assertContains("test-documents/testTXT.txt", content);
    assertContains("Test d'indexation de Txt", content);
    assertContains("test-documents/testWORD.doc", content);
    assertContains("This is a sample Microsoft Word Document", content);
    assertContains("test-documents/testXML.xml", content);
    assertContains("Rida Benjelloun", content);
}

Example 70 with BodyContentHandler

use of org.apache.tika.sax.BodyContentHandler in project tika by apache.

the class CompressParserTest method testEmbedded.

/**
     * Tests that the ParseContext parser is correctly
     *  fired for all the embedded entries.
     */
@Test
public void testEmbedded() throws Exception {
    // Should auto-detect!
    Parser parser = new AutoDetectParser();
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    InputStream stream = ZipParserTest.class.getResourceAsStream("/test-documents/test-documents.tar.Z");
    try {
        parser.parse(stream, handler, metadata, trackingContext);
    } finally {
        stream.close();
    }
    // Should find a single entry, for the (compressed) tar file
    assertEquals(1, tracker.filenames.size());
    assertEquals(1, tracker.mediatypes.size());
    assertEquals(1, tracker.modifiedAts.size());
    assertEquals(null, tracker.filenames.get(0));
    assertEquals(null, tracker.mediatypes.get(0));
    assertEquals(null, tracker.modifiedAts.get(0));
    // Tar file starts with the directory name
    assertEquals("test-documents/", new String(tracker.lastSeenStart, 0, 15, US_ASCII));
}