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

Example 91 with ContentHandler

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

the class ExcelParserTest method testHeaderAndFooterExtraction.

@Test
public void testHeaderAndFooterExtraction() throws Exception {
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL_headers_footers.xls")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.UK);
        new OfficeParser().parse(input, handler, metadata, context);
        assertEquals("application/vnd.ms-excel", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("Internal spreadsheet", metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Aeham Abushwashi", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("Aeham Abushwashi", metadata.get(Metadata.AUTHOR));
        String content = handler.toString();
        assertContains("John Smith1", content);
        assertContains("John Smith50", content);
        assertContains("1 Corporate HQ", content);
        assertContains("Header - Corporate Spreadsheet", content);
        assertContains("Header - For Internal Use Only", content);
        assertContains("Header - Author: John Smith", content);
        assertContains("Footer - Corporate Spreadsheet", content);
        assertContains("Footer - For Internal Use Only", content);
        assertContains("Footer - Author: John Smith", content);
    }
}

Example 92 with ContentHandler

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

the class ExcelParserTest method testWorksSpreadsheet70.

@Test
public void testWorksSpreadsheet70() throws Exception {
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testWORKSSpreadsheet7.0.xlr")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler(-1);
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.US);
        new OfficeParser().parse(input, handler, metadata, context);
        String content = handler.toString();
        assertContains("Microsoft Works", content);
    }
}

Example 93 with ContentHandler

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

the class ExcelParserTest method testExcelParserPassword.

@Test
public void testExcelParserPassword() throws Exception {
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL_protected_passtika.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);
        fail("Document is encrypted, shouldn't parse");
    } catch (EncryptedDocumentException e) {
    // Good
    }
    // Try again, this time with the password
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL_protected_passtika.xls")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.US);
        context.set(PasswordProvider.class, new PasswordProvider() {

            @Override
            public String getPassword(Metadata metadata) {
                return "tika";
            }
        });
        new OfficeParser().parse(input, handler, metadata, context);
        assertEquals("application/vnd.ms-excel", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals(null, metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Antoni", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("2011-11-25T09:52:48Z", metadata.get(TikaCoreProperties.CREATED));
        String content = handler.toString();
        assertContains("This is an Encrypted Excel spreadsheet", content);
        assertNotContained("9.0", content);
    }
}

Example 94 with ContentHandler

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

the class PowerPointParserTest method testPowerPointParser.

@Test
public void testPowerPointParser() throws Exception {
    try (InputStream input = PowerPointParserTest.class.getResourceAsStream("/test-documents/testPPT.ppt")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        new OfficeParser().parse(input, handler, metadata, new ParseContext());
        assertEquals("application/vnd.ms-powerpoint", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("Sample Powerpoint Slide", metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Keith Bennett", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
        String content = handler.toString();
        assertContains("Sample Powerpoint Slide", content);
        assertContains("Powerpoint X for Mac", content);
    }
}

Example 95 with ContentHandler

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

the class PowerPointParserTest method testMasterText.

/**
     * TIKA-712 Master Slide Text from PPT and PPTX files
     *  should be extracted too
     */
@Test
public void testMasterText() throws Exception {
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    try (InputStream stream = PowerPointParserTest.class.getResourceAsStream("/test-documents/testPPT_masterText.ppt")) {
        new OfficeParser().parse(stream, handler, metadata, new ParseContext());
    }
    String content = handler.toString();
    assertContains("Text that I added to the master slide", content);
    // Make sure boilerplate text didn't come through:
    assertEquals(-1, content.indexOf("Click to edit Master"));
    //TIKA-1171
    assertEquals(-1, content.indexOf("*"));
}