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

Example 36 with DefaultHandler

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

the class RecursiveParserWrapperTest method testCharLimit.

@Test
public void testCharLimit() throws Exception {
    ParseContext context = new ParseContext();
    Metadata metadata = new Metadata();
    Parser wrapped = new AutoDetectParser();
    RecursiveParserWrapper wrapper = new RecursiveParserWrapper(wrapped, new BasicContentHandlerFactory(BasicContentHandlerFactory.HANDLER_TYPE.TEXT, 60));
    InputStream stream = RecursiveParserWrapperTest.class.getResourceAsStream("/test-documents/test_recursive_embedded.docx");
    wrapper.parse(stream, new DefaultHandler(), metadata, context);
    List<Metadata> list = wrapper.getMetadata();
    assertEquals(5, list.size());
    int wlr = 0;
    for (Metadata m : list) {
        String limitReached = m.get(RecursiveParserWrapper.WRITE_LIMIT_REACHED);
        if (limitReached != null && limitReached.equals("true")) {
            wlr++;
        }
    }
    assertEquals(1, wlr);
}

Example 37 with DefaultHandler

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

the class RecursiveParserWrapperTest method getMetadata.

private List<Metadata> getMetadata(Metadata metadata, ContentHandlerFactory contentHandlerFactory, boolean catchEmbeddedExceptions, DigestingParser.Digester digester) throws Exception {
    ParseContext context = new ParseContext();
    Parser wrapped = new AutoDetectParser();
    if (digester != null) {
        wrapped = new DigestingParser(wrapped, digester);
    }
    RecursiveParserWrapper wrapper = new RecursiveParserWrapper(wrapped, contentHandlerFactory, catchEmbeddedExceptions);
    String path = metadata.get(Metadata.RESOURCE_NAME_KEY);
    if (path == null) {
        path = "/test-documents/test_recursive_embedded.docx";
    } else {
        path = "/test-documents/" + path;
    }
    InputStream stream = null;
    try {
        stream = TikaInputStream.get(RecursiveParserWrapperTest.class.getResource(path).toURI());
        wrapper.parse(stream, new DefaultHandler(), metadata, context);
    } finally {
        IOUtils.closeQuietly(stream);
    }
    return wrapper.getMetadata();
}

Example 38 with DefaultHandler

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

the class RecursiveParserWrapperTest method testMaxEmbedded.

@Test
public void testMaxEmbedded() throws Exception {
    int maxEmbedded = 4;
    //including outer container file
    int totalNoLimit = 12;
    ParseContext context = new ParseContext();
    Metadata metadata = new Metadata();
    String limitReached = null;
    Parser wrapped = new AutoDetectParser();
    RecursiveParserWrapper wrapper = new RecursiveParserWrapper(wrapped, new BasicContentHandlerFactory(BasicContentHandlerFactory.HANDLER_TYPE.TEXT, -1));
    InputStream stream = RecursiveParserWrapperTest.class.getResourceAsStream("/test-documents/test_recursive_embedded.docx");
    wrapper.parse(stream, new DefaultHandler(), metadata, context);
    List<Metadata> list = wrapper.getMetadata();
    //test default
    assertEquals(totalNoLimit, list.size());
    limitReached = list.get(0).get(RecursiveParserWrapper.EMBEDDED_RESOURCE_LIMIT_REACHED);
    assertNull(limitReached);
    wrapper.reset();
    stream.close();
    //test setting value
    metadata = new Metadata();
    stream = RecursiveParserWrapperTest.class.getResourceAsStream("/test-documents/test_recursive_embedded.docx");
    wrapper.setMaxEmbeddedResources(maxEmbedded);
    wrapper.parse(stream, new DefaultHandler(), metadata, context);
    list = wrapper.getMetadata();
    //add 1 for outer container file
    assertEquals(maxEmbedded + 1, list.size());
    limitReached = list.get(0).get(RecursiveParserWrapper.EMBEDDED_RESOURCE_LIMIT_REACHED);
    assertEquals("true", limitReached);
    wrapper.reset();
    stream.close();
    //test setting value < 0
    metadata = new Metadata();
    stream = RecursiveParserWrapperTest.class.getResourceAsStream("/test-documents/test_recursive_embedded.docx");
    wrapper.setMaxEmbeddedResources(-2);
    wrapper.parse(stream, new DefaultHandler(), metadata, context);
    assertEquals(totalNoLimit, list.size());
    limitReached = list.get(0).get(RecursiveParserWrapper.EMBEDDED_RESOURCE_LIMIT_REACHED);
    assertNull(limitReached);
}

Example 39 with DefaultHandler

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

the class RecursiveParserWrapperTest method testPrimaryExcWEmbedded.

@Test
public void testPrimaryExcWEmbedded() throws Exception {
    //if embedded content is handled and then
    //the parser hits an exception in the container document,
    //that the first element of the returned list is the container document
    //and the second is the embedded content
    Metadata metadata = new Metadata();
    metadata.set(Metadata.RESOURCE_NAME_KEY, "embedded_then_npe.xml");
    ParseContext context = new ParseContext();
    Parser wrapped = new AutoDetectParser();
    RecursiveParserWrapper wrapper = new RecursiveParserWrapper(wrapped, new BasicContentHandlerFactory(BasicContentHandlerFactory.HANDLER_TYPE.TEXT, -1), true);
    String path = "/test-documents/mock/embedded_then_npe.xml";
    InputStream stream = null;
    boolean npe = false;
    try {
        stream = RecursiveParserWrapperTest.class.getResourceAsStream(path);
        wrapper.parse(stream, new DefaultHandler(), metadata, context);
    } catch (TikaException e) {
        if (e.getCause().getClass().equals(NullPointerException.class)) {
            npe = true;
        }
    } finally {
        IOUtils.closeQuietly(stream);
    }
    assertTrue("npe", npe);
    List<Metadata> metadataList = wrapper.getMetadata();
    assertEquals(2, metadataList.size());
    Metadata outerMetadata = metadataList.get(0);
    Metadata embeddedMetadata = metadataList.get(1);
    assertContains("main_content", outerMetadata.get(RecursiveParserWrapper.TIKA_CONTENT));
    assertEquals("embedded_then_npe.xml", outerMetadata.get(TikaMetadataKeys.RESOURCE_NAME_KEY));
    assertEquals("Nikolai Lobachevsky", outerMetadata.get("author"));
    assertContains("some_embedded_content", embeddedMetadata.get(RecursiveParserWrapper.TIKA_CONTENT));
    assertEquals("embed1.xml", embeddedMetadata.get(TikaMetadataKeys.RESOURCE_NAME_KEY));
    assertEquals("embeddedAuthor", embeddedMetadata.get("author"));
}

Example 40 with DefaultHandler

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

the class UnpackerResource method process.

private Map<String, byte[]> process(InputStream is, @Context HttpHeaders httpHeaders, @Context UriInfo info, boolean saveAll) throws Exception {
    Metadata metadata = new Metadata();
    ParseContext pc = new ParseContext();
    Parser parser = TikaResource.createParser();
    if (parser instanceof DigestingParser) {
        //no need to digest for unwrapping
        parser = ((DigestingParser) parser).getWrappedParser();
    }
    TikaResource.fillMetadata(parser, metadata, pc, httpHeaders.getRequestHeaders());
    TikaResource.logRequest(LOG, info, metadata);
    ContentHandler ch;
    ByteArrayOutputStream text = new ByteArrayOutputStream();
    if (saveAll) {
        ch = new BodyContentHandler(new RichTextContentHandler(new OutputStreamWriter(text, UTF_8)));
    } else {
        ch = new DefaultHandler();
    }
    Map<String, byte[]> files = new HashMap<>();
    MutableInt count = new MutableInt();
    pc.set(EmbeddedDocumentExtractor.class, new MyEmbeddedDocumentExtractor(count, files));
    TikaResource.parse(parser, LOG, info.getPath(), is, ch, metadata, pc);
    if (count.intValue() == 0 && !saveAll) {
        throw new WebApplicationException(Response.Status.NO_CONTENT);
    }
    if (saveAll) {
        files.put(TEXT_FILENAME, text.toByteArray());
        ByteArrayOutputStream metaStream = new ByteArrayOutputStream();
        metadataToCsv(metadata, metaStream);
        files.put(META_FILENAME, metaStream.toByteArray());
    }
    return files;
}