Examples with ContentHandler org.xml.sax.ContentHandler used on opensource projects

Example 56 with ContentHandler

use of org.xml.sax.ContentHandler in project tika by apache.

the class FontParsersTest method testTTFParsing.

@Test
public void testTTFParsing() throws Exception {
    // Should auto-detect!
    Parser parser = new AutoDetectParser();
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    ParseContext context = new ParseContext();
    try (TikaInputStream stream = TikaInputStream.get(FontParsersTest.class.getResource("/test-documents/testTrueType3.ttf"))) {
        parser.parse(stream, handler, metadata, context);
    }
    assertEquals("application/x-font-ttf", metadata.get(Metadata.CONTENT_TYPE));
    assertEquals("Open Sans Bold", metadata.get(TikaCoreProperties.TITLE));
    assertEquals("2010-12-30T11:04:00Z", metadata.get(Metadata.CREATION_DATE));
    assertEquals("2010-12-30T11:04:00Z", metadata.get(TikaCoreProperties.CREATED));
    assertEquals("2011-05-05T12:37:53Z", metadata.get(TikaCoreProperties.MODIFIED));
    assertEquals("Open Sans Bold", metadata.get(MET_FONT_NAME));
    assertEquals("Open Sans", metadata.get(MET_FONT_FAMILY_NAME));
    assertEquals("Bold", metadata.get(MET_FONT_SUB_FAMILY_NAME));
    assertEquals("OpenSans-Bold", metadata.get(MET_PS_NAME));
    assertEquals("Digitized", metadata.get("Copyright").substring(0, 9));
    assertEquals("Open Sans", metadata.get("Trademark").substring(0, 9));
    // Not extracted
    assertEquals(null, metadata.get(MET_FONT_FULL_NAME));
    assertEquals(null, metadata.get(MET_FONT_WEIGHT));
    assertEquals(null, metadata.get(MET_FONT_VERSION));
    // Currently, the parser doesn't extract any contents
    String content = handler.toString();
    assertEquals("", content);
}

Example 57 with ContentHandler

use of org.xml.sax.ContentHandler in project tika by apache.

the class HDFParserTest method testHDF4.

@Test
public void testHDF4() throws Exception {
    if (System.getProperty("java.version").startsWith("1.5")) {
        return;
    }
    Parser parser = new HDFParser();
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    /*
       * this is a publicly available HDF4 file from the HD4 examples:
       * 
       * http://www.hdfgroup.org/training/hdf4_chunking/Chunkit/bin/input54kmdata.hdf
       */
    try (InputStream stream = HDFParser.class.getResourceAsStream("/test-documents/test.hdf")) {
        parser.parse(stream, handler, metadata, new ParseContext());
    }
    assertNotNull(metadata);
    assertEquals("Direct read of HDF4 file through CDM library", metadata.get("_History"));
    assertEquals("Ascending", metadata.get("Pass"));
    assertEquals("Hierarchical Data Format, version 4", metadata.get("File-Type-Description"));
}

Example 58 with ContentHandler

use of org.xml.sax.ContentHandler in project tika by apache.

the class GribParserTest method testParseGlobalMetadata.

@Test
public void testParseGlobalMetadata() throws Exception {
    Parser parser = new GribParser();
    Metadata metadata = new Metadata();
    ContentHandler handler = new BodyContentHandler();
    try (InputStream stream = GribParser.class.getResourceAsStream("/test-documents/gdas1.forecmwf.2014062612.grib2")) {
        parser.parse(stream, handler, metadata, new ParseContext());
    }
    assertNotNull(metadata);
    String content = handler.toString();
    assertTrue(content.contains("dimensions:"));
    assertTrue(content.contains("variables:"));
}

Example 59 with ContentHandler

use of org.xml.sax.ContentHandler in project tika by apache.

the class DWGParserTest method testParser.

@SuppressWarnings("deprecation")
private void testParser(InputStream input) throws Exception {
    try {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        new DWGParser().parse(input, handler, metadata);
        assertEquals("image/vnd.dwg", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("The quick brown fox jumps over the lazy dog", metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Gym class featuring a brown fox and lazy dog", metadata.get(TikaCoreProperties.DESCRIPTION));
        assertEquals("Gym class featuring a brown fox and lazy dog", metadata.get(Metadata.SUBJECT));
        assertEquals("Nevin Nollop", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("Pangram, fox, dog", metadata.get(TikaCoreProperties.KEYWORDS));
        assertEquals("Lorem ipsum", metadata.get(TikaCoreProperties.COMMENTS).substring(0, 11));
        assertEquals("http://www.alfresco.com", metadata.get(TikaCoreProperties.RELATION));
        // Check some of the old style metadata too
        assertEquals("The quick brown fox jumps over the lazy dog", metadata.get(Metadata.TITLE));
        assertEquals("Gym class featuring a brown fox and lazy dog", metadata.get(Metadata.SUBJECT));
        String content = handler.toString();
        assertContains("The quick brown fox jumps over the lazy dog", content);
        assertContains("Gym class", content);
        assertContains("www.alfresco.com", content);
    } finally {
        input.close();
    }
}

Example 60 with ContentHandler

use of org.xml.sax.ContentHandler in project tika by apache.

the class DWGParserTest method testDWG2010CustomPropertiesParser.

@Test
public void testDWG2010CustomPropertiesParser() throws Exception {
    // Check that standard parsing works
    InputStream testInput = DWGParserTest.class.getResourceAsStream("/test-documents/testDWG2010_custom_props.dwg");
    testParser(testInput);
    // Check that custom properties with alternate padding work
    try (InputStream input = DWGParserTest.class.getResourceAsStream("/test-documents/testDWG2010_custom_props.dwg")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        new DWGParser().parse(input, handler, metadata, null);
        assertEquals("valueforcustomprop1", metadata.get("customprop1"));
        assertEquals("valueforcustomprop2", metadata.get("customprop2"));
    }
}