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

Example 86 with ContentHandler

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

the class TNEFParserTest method testMetadata.

@Test
public void testMetadata() throws Exception {
    TikaInputStream stream = getTestFile(file);
    Metadata metadata = new Metadata();
    ContentHandler handler = new BodyContentHandler();
    TNEFParser tnef = new TNEFParser();
    tnef.parse(stream, handler, metadata, new ParseContext());
    assertEquals("This is a test message", metadata.get(TikaCoreProperties.TITLE));
    assertEquals("This is a test message", metadata.get(Metadata.SUBJECT));
}

Example 87 with ContentHandler

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

the class WordParserTest method testCustomProperties.

/**
     * Ensures that custom OLE2 (HPSF) properties are extracted
     */
@Test
public void testCustomProperties() throws Exception {
    Metadata metadata = new Metadata();
    try (InputStream input = WordParserTest.class.getResourceAsStream("/test-documents/testWORD_custom_props.doc")) {
        ContentHandler handler = new BodyContentHandler(-1);
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.US);
        new OfficeParser().parse(input, handler, metadata, context);
    }
    assertEquals("application/msword", metadata.get(Metadata.CONTENT_TYPE));
    assertEquals("EJ04325S", metadata.get(TikaCoreProperties.CREATOR));
    assertEquals("Etienne Jouvin", metadata.get(TikaCoreProperties.MODIFIER));
    assertEquals("Etienne Jouvin", metadata.get(Metadata.LAST_AUTHOR));
    assertEquals("2012-01-03T22:14:00Z", metadata.get(TikaCoreProperties.MODIFIED));
    assertEquals("2012-01-03T22:14:00Z", metadata.get(Metadata.DATE));
    assertEquals("2010-10-05T09:03:00Z", metadata.get(TikaCoreProperties.CREATED));
    assertEquals("2010-10-05T09:03:00Z", metadata.get(Metadata.CREATION_DATE));
    assertEquals("Microsoft Office Word", metadata.get(OfficeOpenXMLExtended.APPLICATION));
    assertEquals("1", metadata.get(Office.PAGE_COUNT));
    assertEquals("2", metadata.get(Office.WORD_COUNT));
    assertEquals("My Title", metadata.get(TikaCoreProperties.TITLE));
    assertEquals("My Keyword", metadata.get(TikaCoreProperties.KEYWORDS));
    assertEquals("Normal.dotm", metadata.get(OfficeOpenXMLExtended.TEMPLATE));
    assertEquals("My Comments", metadata.get(TikaCoreProperties.COMMENTS));
    // TODO: Move to OO subject in Tika 2.0
    assertEquals("My subject", metadata.get(Metadata.SUBJECT));
    assertEquals("My subject", metadata.get(OfficeOpenXMLCore.SUBJECT));
    assertEquals("EDF-DIT", metadata.get(OfficeOpenXMLExtended.COMPANY));
    assertEquals("MyStringValue", metadata.get("custom:MyCustomString"));
    assertEquals("2010-12-30T23:00:00Z", metadata.get("custom:MyCustomDate"));
}

Example 88 with ContentHandler

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

the class WordParserTest method testNoFormat.

/**
     * TIKA-1044 - Handle documents where parts of the
     *  text have no formatting or styles applied to them
     */
@Test
public void testNoFormat() throws Exception {
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    try (InputStream stream = WordParserTest.class.getResourceAsStream("/test-documents/testWORD_no_format.doc")) {
        new OfficeParser().parse(stream, handler, metadata, new ParseContext());
    }
    String content = handler.toString();
    assertContains("Will generate an exception", content);
}

Example 89 with ContentHandler

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

the class WordParserTest method testWordParser.

@Test
public void testWordParser() throws Exception {
    try (InputStream input = WordParserTest.class.getResourceAsStream("/test-documents/testWORD.doc")) {
        ContentHandler handler = new BodyContentHandler();
        Metadata metadata = new Metadata();
        new OfficeParser().parse(input, handler, metadata, new ParseContext());
        assertEquals("application/msword", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("Sample Word Document", metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Keith Bennett", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
        assertContains("Sample Word Document", handler.toString());
    }
}

Example 90 with ContentHandler

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

the class ExcelParserTest method testExcelParser.

@Test
// Checks legacy Tika-1.0 style metadata keys
@SuppressWarnings("deprecation")
public void testExcelParser() throws Exception {
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL.xls")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.US);
        new OfficeParser().parse(input, handler, metadata, context);
        assertEquals("application/vnd.ms-excel", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("Simple Excel document", metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Keith Bennett", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
        // Mon Oct 01 17:13:56 BST 2007
        assertEquals("2007-10-01T16:13:56Z", metadata.get(TikaCoreProperties.CREATED));
        assertEquals("2007-10-01T16:13:56Z", metadata.get(Metadata.CREATION_DATE));
        // Mon Oct 01 17:31:43 BST 2007
        assertEquals("2007-10-01T16:31:43Z", metadata.get(TikaCoreProperties.MODIFIED));
        assertEquals("2007-10-01T16:31:43Z", metadata.get(Metadata.DATE));
        String content = handler.toString();
        assertContains("Sample Excel Worksheet", content);
        assertContains("Numbers and their Squares", content);
        assertContains("\t\tNumber\tSquare", content);
        assertContains("9", content);
        assertNotContained("9.0", content);
        assertContains("196", content);
        assertNotContained("196.0", content);
    }
}