Examples with XMLValue org.knime.core.data.xml.XMLValue used on opensource projects

Example 1 with XMLValue

use of org.knime.core.data.xml.XMLValue in project knime-core by knime.

the class XMLXpathCellReader method readXML.

/**
 * {@inheritDoc}
 */
@Override
public XMLValue<Document> readXML() throws IOException {
    if (!m_xpathMatcher.nodeMatches()) {
        return null;
    }
    try {
        while (m_parser.hasNext()) {
            switch(m_parser.getEventType()) {
                case XMLStreamConstants.START_ELEMENT:
                    updateBasePath();
                    updateXmlSpaceDefinition();
                    updateXmlLangDefinition();
                    pushNamespaceContext();
                    for (int i = 0; i < m_docs.size(); i++) {
                        Element element = createElement(m_docs.get(i));
                        m_currNodes.get(i).appendChild(element);
                        m_currNodes.set(i, element);
                    }
                    boolean match = m_xpathMatcher.startElement(m_parser.getName());
                    if (match) {
                        Document doc = m_builder.newDocument();
                        Element element = createElement(doc);
                        m_namespaceStack.stream().flatMap(e -> e.stream()).forEach(p -> element.setAttributeNS(XMLConstants.XMLNS_ATTRIBUTE_NS_URI, p.getFirst(), p.getSecond()));
                        // (see: http://www.w3.org/XML/1998/namespace)
                        if (!m_base.isEmpty() && null != m_base.get(0)) {
                            element.setAttributeNS(XMLConstants.XML_NS_URI, "xml:base", m_base.get(0));
                        }
                        if (!m_space.isEmpty() && null != m_space.get(0)) {
                            element.setAttributeNS(XMLConstants.XML_NS_URI, "xml:space", m_space.get(0));
                        }
                        if (!m_lang.isEmpty() && null != m_lang.get(0)) {
                            element.setAttributeNS(XMLConstants.XML_NS_URI, "xml:lang", m_lang.get(0));
                        }
                        doc.appendChild(element);
                        m_docs.add(doc);
                        m_currNodes.add(element);
                    }
                    break;
                case XMLStreamConstants.END_ELEMENT:
                    if (!m_reentrent) {
                        m_xpathMatcher.endElement();
                        if (!m_base.isEmpty()) {
                            m_base.remove(0);
                        }
                        if (!m_space.isEmpty()) {
                            m_space.remove(0);
                        }
                        if (!m_lang.isEmpty()) {
                            m_lang.remove(0);
                        }
                    }
                    for (int i = m_docs.size() - 1; i >= 0; i--) {
                        Node curr = m_currNodes.get(i);
                        Node first = m_docs.get(i).getFirstChild();
                        if (curr.isSameNode(first)) {
                            XMLValue<Document> cell = null;
                            cell = createDataCell(m_docs.get(i));
                            m_docs.remove(i);
                            m_currNodes.remove(i);
                            m_reentrent = true;
                            return cell;
                        } else {
                            m_currNodes.set(i, curr.getParentNode());
                        }
                    }
                    popNamespaceContext();
                    m_reentrent = false;
                    break;
                case XMLStreamConstants.CHARACTERS:
                    if (!m_parser.isWhiteSpace() || (m_parser.isWhiteSpace() && !m_space.isEmpty() && null != m_space.get(0) && m_space.get(0).equals("preserve"))) {
                        for (int i = 0; i < m_docs.size(); i++) {
                            String str = m_parser.getText();
                            Text text = m_docs.get(i).createTextNode(str);
                            m_currNodes.get(i).appendChild(text);
                        }
                    }
                    break;
                case XMLStreamConstants.SPACE:
                    if (!m_space.isEmpty() && null != m_space.get(0) && m_space.get(0).equals("preserve")) {
                        for (int i = 0; i < m_docs.size(); i++) {
                            String str = m_parser.getText();
                            Text text = m_docs.get(i).createTextNode(str);
                            m_currNodes.get(i).appendChild(text);
                        }
                    }
                    break;
                case XMLStreamConstants.COMMENT:
                    for (int i = 0; i < m_docs.size(); i++) {
                        String str = m_parser.getText();
                        Comment comment = m_docs.get(i).createComment(str);
                        m_currNodes.get(i).appendChild(comment);
                    }
                    break;
                case XMLStreamConstants.ENTITY_REFERENCE:
                    for (int i = 0; i < m_docs.size(); i++) {
                        String str = m_parser.getText();
                        EntityReference ref = m_docs.get(i).createEntityReference(str);
                        m_currNodes.get(i).appendChild(ref);
                    }
                    break;
                case XMLStreamConstants.PROCESSING_INSTRUCTION:
                    for (int i = 0; i < m_docs.size(); i++) {
                        String piTarget = m_parser.getPITarget();
                        String piName = m_parser.getPIData();
                        ProcessingInstruction pi = m_docs.get(i).createProcessingInstruction(piTarget, piName);
                        m_currNodes.get(i).appendChild(pi);
                    }
                    break;
                case XMLStreamConstants.DTD:
                    // not needed use DOM-Reader for full documents
                    break;
                case XMLStreamConstants.END_DOCUMENT:
                    // never called
                    m_parser.close();
                default:
                    break;
            }
            m_parser.next();
        }
        return null;
    } catch (XMLStreamException e) {
        throw new IOException(e);
    }
}

Example 2 with XMLValue

use of org.knime.core.data.xml.XMLValue in project knime-core by knime.

the class XML2PMMLNodeModel method createColRearranger.

private ColumnRearranger createColRearranger(final DataTableSpec spec) throws InvalidSettingsException {
    if (m_xmlColumnName.getStringValue() == null) {
        guessDefaultXMLColumn(spec);
    }
    String xmlColName = m_xmlColumnName.getStringValue();
    String newColName = m_newColumnName.getStringValue();
    final int colIndex = spec.findColumnIndex(xmlColName);
    CheckUtils.checkSetting(colIndex >= 0, "Column: '%s' does not exist anymore.", xmlColName);
    final DataColumnSpec colSpec = spec.getColumnSpec(colIndex);
    CheckUtils.checkSetting(colSpec.getType().isCompatible(StringValue.class), "Selected column '%s' is not string/xml-compatible", xmlColName);
    DataColumnSpecCreator colSpecCreator;
    if (newColName != null && !m_replaceColumn.getBooleanValue()) {
        String newName = DataTableSpec.getUniqueColumnName(spec, newColName);
        colSpecCreator = new DataColumnSpecCreator(newName, PMMLCell.TYPE);
    } else {
        colSpecCreator = new DataColumnSpecCreator(colSpec);
        colSpecCreator.setType(PMMLCell.TYPE);
        colSpecCreator.removeAllHandlers();
        colSpecCreator.setDomain(null);
    }
    DataColumnSpec outColumnSpec = colSpecCreator.createSpec();
    ColumnRearranger rearranger = new ColumnRearranger(spec);
    CellFactory fac = new SingleCellFactory(outColumnSpec) {

        @Override
        public DataCell getCell(final DataRow row) {
            DataCell cell = row.getCell(colIndex);
            if (cell.isMissing()) {
                return DataType.getMissingCell();
            } else {
                PMMLDocument pmmlDoc = null;
                String failure = null;
                XmlObject xmlDoc;
                try (LockedSupplier<Document> supplier = ((XMLValue<Document>) cell).getDocumentSupplier()) {
                    xmlDoc = XmlObject.Factory.parse(supplier.get().cloneNode(true));
                    if (xmlDoc instanceof PMMLDocument) {
                        pmmlDoc = (PMMLDocument) xmlDoc;
                    } else if (PMMLUtils.isOldKNIMEPMML(xmlDoc) || PMMLUtils.is4_1PMML(xmlDoc)) {
                        String updatedPMML = PMMLUtils.getUpdatedVersionAndNamespace(xmlDoc);
                        /* Parse the modified document and assign it to a
                                 * PMMLDocument.*/
                        pmmlDoc = PMMLDocument.Factory.parse(updatedPMML);
                    } else {
                        failure = "No valid PMML v 3.x/4.0/4.1 document";
                    }
                } catch (XmlException e) {
                    if (!m_failOnInvalid.getBooleanValue()) {
                        LOGGER.error("Invalid PMML in row " + row.getKey() + ": " + e.getMessage(), e);
                    }
                    failure = e.getMessage();
                }
                if (failure != null) {
                    m_failCounter.incrementAndGet();
                    if (m_failOnInvalid.getBooleanValue()) {
                        throw new RuntimeException("Invalid PMML in row " + row.getKey() + ": " + failure);
                    } else {
                        return new MissingCell(failure);
                    }
                } else {
                    try {
                        return PMMLCellFactory.create(pmmlDoc.toString());
                    } catch (Exception e) {
                        return new MissingCell(e.getMessage());
                    }
                }
            }
        }
    };
    if (m_replaceColumn.getBooleanValue()) {
        rearranger.replace(fac, colIndex);
    } else {
        rearranger.append(fac);
    }
    return rearranger;
}

Example 3 with XMLValue

use of org.knime.core.data.xml.XMLValue in project knime-core by knime.

the class XMLDOMCellReader method readXML.

/**
 * {@inheritDoc}
 */
@SuppressWarnings("unchecked")
@Override
public XMLValue<Document> readXML() throws IOException {
    if (m_first) {
        m_first = false;
        Document doc;
        try {
            doc = m_builder.parse(m_in);
        } catch (SAXException e) {
            throw new IOException(e);
        }
        removeEmptyTextRecursive(doc, new LinkedList<Boolean>());
        return (XMLValue<Document>) XMLCellFactory.create(doc);
    } else {
        return null;
    }
}