Examples with XmlDocumentHandler gate.xml.XmlDocumentHandler used on opensource projects

Example 1 with XmlDocumentHandler

use of gate.xml.XmlDocumentHandler in project gate-core by GateNLP.

the class TikaFormat method unpackMarkup.

@Override
public void unpackMarkup(Document doc, RepositioningInfo repInfo, RepositioningInfo ampCodingInfo) throws DocumentFormatException {
    if (doc == null || doc.getSourceUrl() == null) {
        throw new DocumentFormatException("GATE document is null or no content found. Nothing to parse!");
    }
    // End if
    // Create a status listener
    StatusListener statusListener = new StatusListener() {

        @Override
        public void statusChanged(String text) {
            // This is implemented in DocumentFormat.java and inherited here
            fireStatusChanged(text);
        }
    };
    XmlDocumentHandler ch = new XmlDocumentHandler(doc, this.markupElementsMap, this.element2StringMap);
    Metadata metadata = extractParserTips(doc);
    ch.addStatusListener(statusListener);
    ch.setRepositioningInfo(repInfo);
    // set the object with ampersand coding positions
    ch.setAmpCodingInfo(ampCodingInfo);
    InputStream input = null;
    try {
        Parser tikaParser = new TikaConfig().getParser();
        input = doc.getSourceUrl().openStream();
        tikaParser.parse(input, ch, metadata, new ParseContext());
        setDocumentFeatures(metadata, doc);
    } catch (IOException e) {
        throw new DocumentFormatException(e);
    } catch (SAXException e) {
        throw new DocumentFormatException(e);
    } catch (TikaException e) {
        throw new DocumentFormatException(e);
    } finally {
        // null safe
        IOUtils.closeQuietly(input);
        ch.removeStatusListener(statusListener);
    }
    if (doc instanceof DocumentImpl) {
        ((DocumentImpl) doc).setNextAnnotationId(ch.getCustomObjectsId());
    }
}

Example 2 with XmlDocumentHandler

use of gate.xml.XmlDocumentHandler in project gate-core by GateNLP.

the class SgmlDocumentFormat method unpackMarkup.

/**
 * Unpack the markup in the document. This converts markup from the
 * native format (e.g. SGML) into annotations in GATE format.
 * Uses the markupElementsMap to determine which elements to convert, and
 * what annotation type names to use.
 * The doc's content is first converted to a wel formed XML.
 * If this succeddes then the document is saved into a temp file and parsed
 * as an XML document.
 *
 * @param doc The gate document you want to parse.
 */
@Override
public void unpackMarkup(Document doc) throws DocumentFormatException {
    if ((doc == null) || (doc.getSourceUrl() == null && doc.getContent() == null)) {
        throw new DocumentFormatException("GATE document is null or no content found. Nothing to parse!");
    }
    // End if
    // Create a status listener
    StatusListener statusListener = new StatusListener() {

        @Override
        public void statusChanged(String text) {
            fireStatusChanged(text);
        }
    };
    XmlDocumentHandler xmlDocHandler = null;
    try {
        Sgml2Xml sgml2Xml = new Sgml2Xml(doc);
        fireStatusChanged("Performing SGML to XML...");
        // convert the SGML document
        String xmlUri = sgml2Xml.convert();
        fireStatusChanged("DONE !");
        // Out.println("Conversion done..." + xmlUri);
        // Out.println(sgml2Xml.convert());
        // Get a parser factory.
        SAXParserFactory saxParserFactory = SAXParserFactory.newInstance();
        // Set up the factory to create the appropriate type of parser
        // Set up the factory to create the appropriate type of parser
        // non validating one
        saxParserFactory.setValidating(false);
        // non namesapace aware one
        saxParserFactory.setNamespaceAware(true);
        // Create a SAX parser
        SAXParser parser = saxParserFactory.newSAXParser();
        // use it
        // create a new Xml document handler
        xmlDocHandler = new XmlDocumentHandler(doc, this.markupElementsMap, this.element2StringMap);
        // register a status listener with it
        xmlDocHandler.addStatusListener(statusListener);
        parser.parse(xmlUri, xmlDocHandler);
        ((DocumentImpl) doc).setNextAnnotationId(xmlDocHandler.getCustomObjectsId());
    } catch (ParserConfigurationException e) {
        throw new DocumentFormatException("XML parser configuration exception ", e);
    } catch (SAXException e) {
        throw new DocumentFormatException(e);
    } catch (IOException e) {
        throw new DocumentFormatException("I/O exception for " + doc.getSourceUrl().toString());
    } finally {
        if (xmlDocHandler != null)
            xmlDocHandler.removeStatusListener(statusListener);
    }
// End try
}

Example 3 with XmlDocumentHandler

use of gate.xml.XmlDocumentHandler in project gate-core by GateNLP.

the class XmlDocumentFormat method unpackGeneralXmlMarkup.

/**
 * Unpack markup from any XML format. The XML elements are translated
 * to annotations on the Original markups annotation set.
 *
 * @param doc the document to process
 * @throws DocumentFormatException
 */
private void unpackGeneralXmlMarkup(Document doc, RepositioningInfo repInfo, RepositioningInfo ampCodingInfo, StatusListener statusListener) throws DocumentFormatException {
    boolean docHasContentButNoValidURL = hasContentButNoValidUrl(doc);
    XmlDocumentHandler xmlDocHandler = null;
    try {
        // use Xerces XML parser with JAXP
        // System.setProperty("javax.xml.parsers.SAXParserFactory",
        // "org.apache.xerces.jaxp.SAXParserFactoryImpl");
        // Get a parser factory.
        SAXParserFactory saxParserFactory = SAXParserFactory.newInstance();
        // Set up the factory to create the appropriate type of parser
        // non validating one
        saxParserFactory.setValidating(false);
        // non namesapace aware one
        saxParserFactory.setNamespaceAware(true);
        // create it
        SAXParser xmlParser = saxParserFactory.newSAXParser();
        // Create a new Xml document handler
        xmlDocHandler = new XmlDocumentHandler(doc, this.markupElementsMap, this.element2StringMap);
        // Register a status listener with it
        xmlDocHandler.addStatusListener(statusListener);
        // set repositioning object
        xmlDocHandler.setRepositioningInfo(repInfo);
        // set the object with ampersand coding positions
        xmlDocHandler.setAmpCodingInfo(ampCodingInfo);
        org.xml.sax.XMLReader newxmlParser = xmlParser.getXMLReader();
        // Set up the factory to create the appropriate type of parser
        // non validating one
        // http://xml.org/sax/features/validation set to false
        newxmlParser.setFeature("http://xml.org/sax/features/validation", false);
        // namesapace aware one
        // http://xml.org/sax/features/namespaces set to true
        newxmlParser.setFeature("http://xml.org/sax/features/namespaces", true);
        newxmlParser.setFeature("http://xml.org/sax/features/namespace-prefixes", true);
        newxmlParser.setContentHandler(xmlDocHandler);
        newxmlParser.setErrorHandler(xmlDocHandler);
        newxmlParser.setDTDHandler(xmlDocHandler);
        newxmlParser.setEntityResolver(xmlDocHandler);
        // Parse the XML Document with the appropriate encoding
        Reader docReader = null;
        try {
            InputSource is;
            if (docHasContentButNoValidURL) {
                // no URL, so parse from string
                is = new InputSource(new StringReader(doc.getContent().toString()));
            } else if (doc instanceof TextualDocument) {
                // textual document - load with user specified encoding
                String docEncoding = ((TextualDocument) doc).getEncoding();
                // don't strip BOM on XML.
                docReader = new InputStreamReader(doc.getSourceUrl().openStream(), docEncoding);
                is = new InputSource(docReader);
                // must set system ID to allow relative URLs (e.g. to a DTD) to
                // work
                is.setSystemId(doc.getSourceUrl().toString());
            } else {
                // let the parser decide the encoding
                is = new InputSource(doc.getSourceUrl().toString());
            }
            newxmlParser.parse(is);
        } finally {
            // make sure the open streams are closed
            if (docReader != null)
                docReader.close();
        }
        // Angel - end
        ((DocumentImpl) doc).setNextAnnotationId(xmlDocHandler.getCustomObjectsId());
    } catch (ParserConfigurationException e) {
        throw new DocumentFormatException("XML parser configuration exception ", e);
    } catch (SAXException e) {
        doc.getFeatures().put("parsingError", Boolean.TRUE);
        Boolean bThrow = (Boolean) doc.getFeatures().get(GateConstants.THROWEX_FORMAT_PROPERTY_NAME);
        if (bThrow != null && bThrow.booleanValue()) {
            // error
            throw new DocumentFormatException(e);
        } else {
            Out.println("Warning: Document remains unparsed. \n" + "\n  Stack Dump: ");
            e.printStackTrace(Out.getPrintWriter());
        }
    // if
    } catch (IOException e) {
        throw new DocumentFormatException("I/O exception for " + doc.getSourceUrl(), e);
    } finally {
        if (xmlDocHandler != null)
            xmlDocHandler.removeStatusListener(statusListener);
    }
// End if else try
}