Examples with DefaultHandler org.xml.sax.helpers.DefaultHandler used on opensource projects

Example 51 with DefaultHandler

use of org.xml.sax.helpers.DefaultHandler in project tika by apache.

the class DcXMLParserTest method testXMLParserNonAsciiChars.

@Test
public void testXMLParserNonAsciiChars() throws Exception {
    try (InputStream input = DcXMLParserTest.class.getResourceAsStream("/test-documents/testXML.xml")) {
        Metadata metadata = new Metadata();
        new DcXMLParser().parse(input, new DefaultHandler(), metadata);
        final String expected = "Archimède et Lius à Châteauneuf testing chars en été";
        assertEquals(expected, metadata.get(TikaCoreProperties.RIGHTS));
    }
}

Example 52 with DefaultHandler

use of org.xml.sax.helpers.DefaultHandler in project tika by apache.

the class RFC822ParserTest method getDate.

private Date getDate(String dateString) throws Exception {
    String mail = "From: dev@tika.apache.org\n" + "Date: " + dateString + "\n";
    Parser p = new RFC822Parser();
    Metadata m = new Metadata();
    try (InputStream is = TikaInputStream.get(mail.getBytes(StandardCharsets.UTF_8))) {
        p.parse(is, new DefaultHandler(), m, new ParseContext());
    }
    return m.getDate(TikaCoreProperties.CREATED);
}

Example 53 with DefaultHandler

use of org.xml.sax.helpers.DefaultHandler in project tika by apache.

the class HtmlParserTest method testParseAscii.

@Test
public void testParseAscii() throws Exception {
    String path = "/test-documents/testHTML.html";
    final StringWriter href = new StringWriter();
    final StringWriter name = new StringWriter();
    ContentHandler body = new BodyContentHandler();
    Metadata metadata = new Metadata();
    try (InputStream stream = HtmlParserTest.class.getResourceAsStream(path)) {
        ContentHandler link = new DefaultHandler() {

            @Override
            public void startElement(String u, String l, String n, Attributes a) throws SAXException {
                if ("a".equals(l)) {
                    if (a.getValue("href") != null) {
                        href.append(a.getValue("href"));
                    } else if (a.getValue("name") != null) {
                        name.append(a.getValue("name"));
                    }
                }
            }
        };
        new HtmlParser().parse(stream, new TeeContentHandler(body, link), metadata, new ParseContext());
    }
    assertEquals("Title : Test Indexation Html", metadata.get(TikaCoreProperties.TITLE));
    assertEquals("Tika Developers", metadata.get("Author"));
    assertEquals("5", metadata.get("refresh"));
    assertEquals("51.2312", metadata.get(Geographic.LATITUDE));
    assertEquals("-5.1987", metadata.get(Geographic.LONGITUDE));
    assertEquals("http://www.apache.org/", href.toString());
    assertEquals("test-anchor", name.toString());
    String content = body.toString();
    assertTrue("Did not contain expected text:" + "Test Indexation Html", content.contains("Test Indexation Html"));
    assertTrue("Did not contain expected text:" + "Indexation du fichier", content.contains("Indexation du fichier"));
}

Example 54 with DefaultHandler

use of org.xml.sax.helpers.DefaultHandler in project tika by apache.

the class BPGParserTest method testBPG_Geo.

/**
     * Tests a file with geographic information in it
     */
@Test
public void testBPG_Geo() throws Exception {
    Metadata metadata = new Metadata();
    metadata.set(Metadata.CONTENT_TYPE, "image/x-bpg");
    InputStream stream = getClass().getResourceAsStream("/test-documents/testBPG_GEO.bpg");
    parser.parse(stream, new DefaultHandler(), metadata, new ParseContext());
    assertEquals("100", metadata.get(Metadata.IMAGE_WIDTH));
    assertEquals("68", metadata.get(Metadata.IMAGE_LENGTH));
    assertEquals("10", metadata.get(Metadata.BITS_PER_SAMPLE));
    assertEquals("YCbCr Colour", metadata.get(Photoshop.COLOR_MODE));
    // TODO Get the geographic data to be properly extracted, see TIKA-1495
    if (false) {
        assertEquals("12.54321", metadata.get(Metadata.LATITUDE));
        assertEquals("-54.1234", metadata.get(Metadata.LONGITUDE));
    }
    // TODO Get the exif data to be properly extracted, see TIKA-1495
    if (false) {
        // 1/1600
        assertEquals("6.25E-4", metadata.get(Metadata.EXPOSURE_TIME));
        assertEquals("5.6", metadata.get(Metadata.F_NUMBER));
        assertEquals("false", metadata.get(Metadata.FLASH_FIRED));
        assertEquals("194.0", metadata.get(Metadata.FOCAL_LENGTH));
        assertEquals("400", metadata.get(Metadata.ISO_SPEED_RATINGS));
        assertEquals("Canon", metadata.get(Metadata.EQUIPMENT_MAKE));
        assertEquals("Canon EOS 40D", metadata.get(Metadata.EQUIPMENT_MODEL));
        assertEquals("Adobe Photoshop CS3 Macintosh", metadata.get(Metadata.SOFTWARE));
        assertEquals("240.0", metadata.get(Metadata.RESOLUTION_HORIZONTAL));
        assertEquals("240.0", metadata.get(Metadata.RESOLUTION_VERTICAL));
        assertEquals("Inch", metadata.get(Metadata.RESOLUTION_UNIT));
    }
}

Example 55 with DefaultHandler

use of org.xml.sax.helpers.DefaultHandler in project tika by apache.

the class BPGParserTest method testBPG_Commented.

/**
     * Tests a file with comments
     */
@Test
public void testBPG_Commented() throws Exception {
    Metadata metadata = new Metadata();
    metadata.set(Metadata.CONTENT_TYPE, "image/x-bpg");
    InputStream stream = getClass().getResourceAsStream("/test-documents/testBPG_commented.bpg");
    parser.parse(stream, new DefaultHandler(), metadata, new ParseContext());
    assertEquals("103", metadata.get(Metadata.IMAGE_WIDTH));
    assertEquals("77", metadata.get(Metadata.IMAGE_LENGTH));
    assertEquals("10", metadata.get(Metadata.BITS_PER_SAMPLE));
    assertEquals("YCbCr Colour", metadata.get(Photoshop.COLOR_MODE));
    // TODO Get the exif comment data to be properly extracted, see TIKA-1495
    if (false) {
        assertEquals("Tosteberga Ängar", metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Bird site in north eastern Skåne, Sweden.\n(new line)", metadata.get(TikaCoreProperties.DESCRIPTION));
        List<String> keywords = Arrays.asList(metadata.getValues(Metadata.KEYWORDS));
        assertTrue(keywords.contains("coast"));
        assertTrue(keywords.contains("bird watching"));
        assertEquals(keywords, Arrays.asList(metadata.getValues(TikaCoreProperties.KEYWORDS)));
    }
    // TODO Get the exif data to be properly extracted, see TIKA-1495
    if (false) {
        // 1/1000000
        assertEquals("1.0E-6", metadata.get(Metadata.EXPOSURE_TIME));
        assertEquals("2.8", metadata.get(Metadata.F_NUMBER));
        assertEquals("4.6", metadata.get(Metadata.FOCAL_LENGTH));
        assertEquals("114", metadata.get(Metadata.ISO_SPEED_RATINGS));
        assertEquals(null, metadata.get(Metadata.EQUIPMENT_MAKE));
        assertEquals(null, metadata.get(Metadata.EQUIPMENT_MODEL));
        assertEquals(null, metadata.get(Metadata.SOFTWARE));
        assertEquals("1", metadata.get(Metadata.ORIENTATION));
        assertEquals("300.0", metadata.get(Metadata.RESOLUTION_HORIZONTAL));
        assertEquals("300.0", metadata.get(Metadata.RESOLUTION_VERTICAL));
        assertEquals("Inch", metadata.get(Metadata.RESOLUTION_UNIT));
    }
}