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

Example 81 with ContentHandler

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

the class OOXMLParserTest method testEncrypted.

@Test
public void testEncrypted() throws Exception {
    Map<String, String> tests = new HashMap<String, String>();
    tests.put("testWORD_protected_passtika.docx", "This is an encrypted Word 2007 File");
    tests.put("testPPT_protected_passtika.pptx", "This is an encrypted PowerPoint 2007 slide.");
    tests.put("testEXCEL_protected_passtika.xlsx", "This is an Encrypted Excel spreadsheet.");
    Parser parser = new AutoDetectParser();
    Metadata m = new Metadata();
    PasswordProvider passwordProvider = new PasswordProvider() {

        @Override
        public String getPassword(Metadata metadata) {
            return "tika";
        }
    };
    ParseContext passwordContext = new ParseContext();
    passwordContext.set(org.apache.tika.parser.PasswordProvider.class, passwordProvider);
    for (Map.Entry<String, String> e : tests.entrySet()) {
        try (InputStream is = getTestDocument(e.getKey())) {
            ContentHandler handler = new BodyContentHandler();
            parser.parse(is, handler, m, passwordContext);
            assertContains(e.getValue(), handler.toString());
        }
    }
    ParseContext context = new ParseContext();
    //now try with no password
    for (Map.Entry<String, String> e : tests.entrySet()) {
        boolean exc = false;
        try (InputStream is = getTestDocument(e.getKey())) {
            ContentHandler handler = new BodyContentHandler();
            parser.parse(is, handler, m, context);
        } catch (EncryptedDocumentException ex) {
            exc = true;
        }
        assertTrue(exc);
    }
}

Example 82 with ContentHandler

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

the class OOXMLParserTest method testWordArt.

@Test
public void testWordArt() throws Exception {
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    try (InputStream stream = OOXMLParserTest.class.getResourceAsStream("/test-documents/testWordArt.pptx")) {
        new AutoDetectParser().parse(stream, handler, metadata, new ParseContext());
    }
    String content = handler.toString();
    assertContains("Here is some red word Art", content);
}

Example 83 with ContentHandler

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

the class OOXMLParserTest method testWordFootnote.

/**
     * Test the plain text output of the Word converter
     *
     * @throws Exception
     */
@Test
public void testWordFootnote() throws Exception {
    Metadata metadata = new Metadata();
    ContentHandler handler = new BodyContentHandler();
    ParseContext context = new ParseContext();
    try (InputStream input = getTestDocument("footnotes.docx")) {
        parser.parse(input, handler, metadata, context);
        assertEquals("application/vnd.openxmlformats-officedocument.wordprocessingml.document", metadata.get(Metadata.CONTENT_TYPE));
        assertTrue(handler.toString().contains("snoska"));
    }
}

Example 84 with ContentHandler

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

the class OOXMLParserTest method testWordMissingOOXMLBeans.

//TIKA-792; with room for future missing bean tests
@Test
public void testWordMissingOOXMLBeans() throws Exception {
    //If a bean is missing, POI prints stack trace to stderr 
    String[] fileNames = new String[] { //TIKA-792
    "testWORD_missing_ooxml_bean1.docx" };
    PrintStream origErr = System.err;
    for (String fileName : fileNames) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        ParseContext context = new ParseContext();
        InputStream input = getTestDocument(fileName);
        //grab stderr
        ByteArrayOutputStream errContent = new ByteArrayOutputStream();
        System.setErr(new PrintStream(errContent, true, UTF_8.name()));
        parser.parse(input, handler, metadata, context);
        //return stderr
        System.setErr(origErr);
        String err = errContent.toString(UTF_8.name());
        assertTrue(err.length() == 0);
        input.close();
    }
}

Example 85 with ContentHandler

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

the class OOXMLParserTest method testProtectedExcelFile.

/**
     * An excel document which is password protected.
     * See TIKA-437.
     */
@Test
public void testProtectedExcelFile() throws Exception {
    Parser parser = new AutoDetectParser();
    Metadata metadata = new Metadata();
    ContentHandler handler = new BodyContentHandler();
    ParseContext context = new ParseContext();
    try (InputStream input = getTestDocument("protectedFile.xlsx")) {
        parser.parse(input, handler, metadata, context);
        assertEquals("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("true", metadata.get(TikaMetadataKeys.PROTECTED));
        String content = handler.toString();
        assertContains("Office", content);
    }
}