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

Example 46 with ContentHandler

use of org.xml.sax.ContentHandler in project spring-framework by spring-projects.

the class AbstractStaxXMLReaderTestCase method contentHandlerNoNamespacesPrefixes.

@Test
public void contentHandlerNoNamespacesPrefixes() throws Exception {
    standardReader.setFeature("http://xml.org/sax/features/namespaces", false);
    standardReader.setFeature("http://xml.org/sax/features/namespace-prefixes", true);
    standardReader.parse(new InputSource(createTestInputStream()));
    AbstractStaxXMLReader staxXmlReader = createStaxXmlReader(createTestInputStream());
    ContentHandler contentHandler = mockContentHandler();
    staxXmlReader.setFeature("http://xml.org/sax/features/namespaces", false);
    staxXmlReader.setFeature("http://xml.org/sax/features/namespace-prefixes", true);
    staxXmlReader.setContentHandler(contentHandler);
    staxXmlReader.parse(new InputSource());
    verifyIdenticalInvocations(standardContentHandler, contentHandler);
}

Example 47 with ContentHandler

use of org.xml.sax.ContentHandler in project spring-framework by spring-projects.

the class AbstractStaxXMLReaderTestCase method mockContentHandler.

protected final ContentHandler mockContentHandler() throws Exception {
    ContentHandler contentHandler = mock(ContentHandler.class);
    willAnswer(new CopyCharsAnswer()).given(contentHandler).characters(any(char[].class), anyInt(), anyInt());
    willAnswer(new CopyCharsAnswer()).given(contentHandler).ignorableWhitespace(any(char[].class), anyInt(), anyInt());
    willAnswer(new Answer<Object>() {

        @Override
        public Object answer(InvocationOnMock invocation) throws Throwable {
            invocation.getArguments()[3] = new AttributesImpl((Attributes) invocation.getArguments()[3]);
            return null;
        }
    }).given(contentHandler).startElement(anyString(), anyString(), anyString(), any(Attributes.class));
    return contentHandler;
}

Example 48 with ContentHandler

use of org.xml.sax.ContentHandler in project spring-framework by spring-projects.

the class AbstractStaxXMLReaderTestCase method contentHandlerNamespacesPrefixes.

@Test
public void contentHandlerNamespacesPrefixes() throws Exception {
    standardReader.setFeature("http://xml.org/sax/features/namespaces", true);
    standardReader.setFeature("http://xml.org/sax/features/namespace-prefixes", true);
    standardReader.parse(new InputSource(createTestInputStream()));
    AbstractStaxXMLReader staxXmlReader = createStaxXmlReader(createTestInputStream());
    ContentHandler contentHandler = mockContentHandler();
    staxXmlReader.setFeature("http://xml.org/sax/features/namespaces", true);
    staxXmlReader.setFeature("http://xml.org/sax/features/namespace-prefixes", true);
    staxXmlReader.setContentHandler(contentHandler);
    staxXmlReader.parse(new InputSource());
    verifyIdenticalInvocations(standardContentHandler, contentHandler);
}

Example 49 with ContentHandler

use of org.xml.sax.ContentHandler in project ddf by codice.

the class TikaInputTransformer method transform.

@Override
public Metacard transform(InputStream input, String id) throws IOException, CatalogTransformerException {
    LOGGER.debug("Transforming input stream using Tika.");
    if (input == null) {
        throw new CatalogTransformerException("Cannot transform null input.");
    }
    try (TemporaryFileBackedOutputStream fileBackedOutputStream = new TemporaryFileBackedOutputStream()) {
        try {
            IOUtils.copy(input, fileBackedOutputStream);
        } catch (IOException e) {
            throw new CatalogTransformerException("Could not copy bytes of content message.", e);
        }
        Parser parser = new AutoDetectParser();
        ToXMLContentHandler xmlContentHandler = new ToXMLContentHandler();
        ToTextContentHandler textContentHandler = null;
        ContentHandler contentHandler;
        if (!contentMetadataExtractors.isEmpty()) {
            textContentHandler = new ToTextContentHandler();
            contentHandler = new TeeContentHandler(xmlContentHandler, textContentHandler);
        } else {
            contentHandler = xmlContentHandler;
        }
        TikaMetadataExtractor tikaMetadataExtractor = new TikaMetadataExtractor(parser, contentHandler);
        Metadata metadata;
        try (InputStream inputStreamCopy = fileBackedOutputStream.asByteSource().openStream()) {
            metadata = tikaMetadataExtractor.parseMetadata(inputStreamCopy, new ParseContext());
        }
        String metadataText = xmlContentHandler.toString();
        if (templates != null) {
            metadataText = transformToXml(metadataText);
        }
        String metacardContentType = metadata.get(Metadata.CONTENT_TYPE);
        MetacardType metacardType = getMetacardTypeFromMimeType(metacardContentType);
        if (metacardType == null) {
            metacardType = commonTikaMetacardType;
        }
        Metacard metacard;
        if (textContentHandler != null) {
            String plainText = textContentHandler.toString();
            Set<AttributeDescriptor> attributes = contentMetadataExtractors.values().stream().map(ContentMetadataExtractor::getMetacardAttributes).flatMap(Collection::stream).collect(Collectors.toSet());
            MetacardTypeImpl extendedMetacardType = new MetacardTypeImpl(metacardType.getName(), metacardType, attributes);
            metacard = MetacardCreator.createMetacard(metadata, id, metadataText, extendedMetacardType);
            for (ContentMetadataExtractor contentMetadataExtractor : contentMetadataExtractors.values()) {
                contentMetadataExtractor.process(plainText, metacard);
            }
        } else {
            metacard = MetacardCreator.createMetacard(metadata, id, metadataText, metacardType);
        }
        if (StringUtils.isNotBlank(metacardContentType)) {
            metacard.setAttribute(new AttributeImpl(Core.DATATYPE, getDatatype(metacardContentType)));
        }
        if (StringUtils.startsWith(metacardContentType, "image")) {
            try (InputStream inputStreamCopy = fileBackedOutputStream.asByteSource().openStream()) {
                createThumbnail(inputStreamCopy, metacard);
            }
        }
        LOGGER.debug("Finished transforming input stream using Tika.");
        return metacard;
    }
}

Example 50 with ContentHandler

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

the class ParserPostProcessor method parse.

/**
     * Forwards the call to the delegated parser and post-processes the
     * results as described above.
     */
public void parse(InputStream stream, ContentHandler handler, Metadata metadata, ParseContext context) throws IOException, SAXException, TikaException {
    ContentHandler body = new BodyContentHandler();
    ContentHandler tee = new TeeContentHandler(handler, body);
    super.parse(stream, tee, metadata, context);
    String content = body.toString();
    metadata.set("fulltext", content);
    int length = Math.min(content.length(), 500);
    metadata.set("summary", content.substring(0, length));
    for (String link : RegexUtils.extractLinks(content)) {
        metadata.add("outlinks", link);
    }
}