Example usage for javax.xml.stream XMLEventReader peek

List of usage examples for javax.xml.stream XMLEventReader peek

  1. HOME
  2. Java
  3. javax
  4. javax.xml.stream.*
  5. XMLEventReader
  6. peek

Introduction

In this page you can find the example usage for javax.xml.stream XMLEventReader peek.

Prototype

public XMLEvent peek() throws XMLStreamException;

Source Link

Document

Check the next XMLEvent without reading it from the stream.

Usage

From source file:com.streamsets.pipeline.lib.xml.StreamingXmlParser.java

void skipIgnorable(XMLEventReader reader) throws XMLStreamException {
    while (reader.hasNext() && isIgnorable(reader.peek())) {
        reader.nextEvent();/* ww  w .ja va2 s.co  m*/
    }
}

From source file:edu.jhu.hlt.concrete.ingesters.bolt.BoltForumPostIngester.java

/**
 * Move the iterator so that a call to nextEvent will return the beginning of a post tag.
 *
 * @param rdr//from w  w  w .j  ava  2 s .  co  m
 * @throws XMLStreamException
 */
private void iterateToPosts(final XMLEventReader rdr) throws XMLStreamException {
    // Peek at the next element.
    XMLEvent fp = rdr.peek();

    // If start element and part == "post", return.
    if (fp.isStartElement()) {
        StartElement se = fp.asStartElement();
        if (se.getName().getLocalPart().equals(POST_LOCAL_NAME))
            return;
        else
            // Churn through non-post start tags.
            this.handleNonPostStartElement(rdr);
    } else
        // Drop.
        rdr.nextEvent();

    this.iterateToPosts(rdr);
}

From source file:de.tudarmstadt.ukp.dkpro.core.io.tiger.TigerXmlReader.java

@Override
public void getNext(JCas aJCas) throws IOException, CollectionException {
    Resource res = nextFile();//from w  w  w .  ja  v a 2  s. c o  m
    initCas(aJCas, res);

    posMappingProvider.configure(aJCas.getCas());

    InputStream is = null;
    try {
        is = CompressionUtils.getInputStream(res.getLocation(), res.getInputStream());

        XMLInputFactory xmlInputFactory = XMLInputFactory.newInstance();
        XMLEventReader xmlEventReader = xmlInputFactory.createXMLEventReader(is);

        JAXBContext context = JAXBContext.newInstance(Meta.class, AnnotationDecl.class, TigerSentence.class);
        Unmarshaller unmarshaller = context.createUnmarshaller();

        JCasBuilder jb = new JCasBuilder(aJCas);

        XMLEvent e = null;
        while ((e = xmlEventReader.peek()) != null) {
            if (isStartElement(e, "s")) {
                readSentence(jb, unmarshaller.unmarshal(xmlEventReader, TigerSentence.class).getValue());
            } else {
                xmlEventReader.next();
            }

        }

        jb.close();

        // Can only do that after the builder is closed, otherwise the text is not yet set in
        // the
        // CAS and we get "null" for all token strings.
        if (pennTreeEnabled) {
            for (ROOT root : select(aJCas, ROOT.class)) {
                PennTree pt = new PennTree(aJCas, root.getBegin(), root.getEnd());
                PennTreeNode rootNode = PennTreeUtils.convertPennTree(root);
                pt.setPennTree(PennTreeUtils.toPennTree(rootNode));
                pt.addToIndexes();
            }
        }
    } catch (XMLStreamException ex1) {
        throw new IOException(ex1);
    } catch (JAXBException ex2) {
        throw new IOException(ex2);
    } finally {
        closeQuietly(is);
    }
}

From source file:org.javelin.sws.ext.bind.internal.model.ElementPattern.java

@Override
public T consume(XMLEventReader eventReader, UnmarshallingContext context) throws XMLStreamException {
    // just skip element's START_ELEMENT event
    StartElement startElement = eventReader.nextEvent().asStartElement();

    // StartElement may contain attributes - these are NOT available as separate events in eventReader.nextEvent()!
    Iterator<?> attributes = startElement.getAttributes();
    List<Attribute> attrList = new LinkedList<Attribute>();
    while (attributes.hasNext()) {
        Attribute a = (Attribute) attributes.next();
        attrList.add(a);//www  .  j a va2s. co m
    }

    T value = this.nestedPattern.consumeValue(new AttributesAwareXMLEventReader(eventReader, attrList),
            context);

    // skip element's END_ELEMENT event
    while (eventReader.peek() != null) {
        XMLEvent ev = eventReader.nextEvent();
        if (ev.getEventType() == XMLStreamConstants.END_ELEMENT)
            break;
    }

    return value;
}

From source file:org.javelin.sws.ext.bind.internal.model.ComplexTypePattern.java

@Override
public T consumeValue(XMLEventReader eventReader, UnmarshallingContext context) throws XMLStreamException {
    // first create an object to be filled (using PropertyAccessors - direct or bean) according to the content model
    T object = BeanUtils.instantiate(this.getJavaType());

    // the order is dictated by incoming events, not by the mode
    // TODO: create a property to enable strict unmarshalling - dictated by content model
    // only this (ContentModel) pattern iterates over XML Events
    XMLEvent event = null;/*  w  w  w. ja  va  2  s. co  m*/
    PropertyMetadataValue<T, ?> pmv = null;

    // this loop will only handle first level of start elements and only single end element
    // deeper levels will be handled by nested patterns
    while (true) {
        boolean end = false;
        event = eventReader.peek();
        pmv = null;

        switch (event.getEventType()) {
        case XMLStreamConstants.ATTRIBUTE:
            pmv = this.consumeNestedAttribute(eventReader, context);
            break;
        case XMLStreamConstants.CDATA:
        case XMLStreamConstants.CHARACTERS:
            // TODO: XMLEvent.ENTITY_REFERENCE?
            if (this.simpleContent != null) {
                pmv = this.consumeSimpleContent(eventReader, context);
                break;
            }
        case XMLStreamConstants.COMMENT:
        case XMLStreamConstants.DTD:
        case XMLStreamConstants.SPACE:
        case XMLStreamConstants.ENTITY_DECLARATION:
        case XMLStreamConstants.NOTATION_DECLARATION:
        case XMLStreamConstants.PROCESSING_INSTRUCTION:
            eventReader.nextEvent();
            break;
        case XMLStreamConstants.ENTITY_REFERENCE:
            // TODO: XMLEvent.ENTITY_REFERENCE?
            eventReader.nextEvent();
            break;
        case XMLStreamConstants.START_DOCUMENT:
            // strange
            break;
        case XMLStreamConstants.START_ELEMENT:
            pmv = this.consumeNestedElement(eventReader, context);
            break;
        case XMLStreamConstants.END_ELEMENT:
            // TODO: in mixed content there will be more than one end element it this content model's level
        case XMLStreamConstants.END_DOCUMENT:
            end = true;
            break;
        }

        if (end)
            break;

        if (pmv != null)
            pmv.getMetadata().setValue(object, pmv.getValue());
    }

    return (T) object;
}

From source file:de.dfki.km.leech.parser.wikipedia.WikipediaDumpParser.java

public MultiValueHashMap<String, String> getPageTitle2Redirects(InputStream sWikipediaDump)
        throws FileNotFoundException, XMLStreamException {
    // <text xml:space="preserve">#REDIRECT [[Autopoiesis]]</text>
    // <text xml:space="preserve">#REDIRECT:[[Hans Leo Haler]]</text>
    // <text xml:space="preserve">#redirect [[Weier Hai]]</text>
    // #weiterleitung
    // <page>
    // <title>Autopoiesis</title>

    Logger.getLogger(WikipediaDumpParser.class.getName()).info("will collect redirects from wikipedia dump...");

    MultiValueHashMap<String, String> hsPageTitle2Redirects = new MultiValueBalancedTreeMap<String, String>();

    String strCurrentTitle = "";
    XMLInputFactory xmlInputFactory = XMLInputFactory.newInstance();

    XMLEventReader xmlEventReader = xmlInputFactory.createXMLEventReader(sWikipediaDump, "Utf-8");
    int iTitlesRead = 0;
    while (xmlEventReader.hasNext()) {
        XMLEvent xmlEvent = xmlEventReader.nextEvent();

        if (!xmlEvent.isStartElement())
            continue;
        // wenn wir einen Title haben, dann merken wir uns den, falls wir ihn brauchen
        if (xmlEvent.asStartElement().getName().getLocalPart().equals("title")) {
            strCurrentTitle = readNextCharEventsText(xmlEventReader);

            iTitlesRead++;/*from  w w  w. j  av  a2  s .co m*/
            if (iTitlesRead % 200000 == 0)
                Logger.getLogger(WikipediaDumpParser.class.getName())
                        .info("read doc #" + StringUtils.beautifyNumber(iTitlesRead));

            continue;
        }

        if (!xmlEvent.asStartElement().getName().getLocalPart().equals("text"))
            continue;

        // jetzt haben wir ein text-tag. Wir schauen, ob jetzt ein redirect kommt
        // entweder kommt ein charEvent oder ein EndEvent. Leere Texte gibts wohl auch
        XMLEvent nextEvent = xmlEventReader.peek();

        if (!nextEvent.isCharacters())
            continue;

        String strCharEventData = readNextCharEventsText(xmlEventReader);
        if (strCharEventData == null)
            continue;

        strCharEventData = strCharEventData.trim();

        boolean bRedirect = false;

        if (strCharEventData.length() >= 9 && strCharEventData.substring(0, 9).equalsIgnoreCase("#redirect"))
            bRedirect = true;
        if (!bRedirect && strCharEventData.length() >= 8
                && strCharEventData.substring(0, 8).equalsIgnoreCase("redirect")
                && !strCharEventData.contains("\n"))
            bRedirect = true;
        if (!bRedirect && strCharEventData.length() >= 14
                && strCharEventData.substring(0, 14).equalsIgnoreCase("#weiterleitung"))
            bRedirect = true;
        if (!bRedirect && strCharEventData.length() >= 13
                && strCharEventData.substring(0, 13).equalsIgnoreCase("weiterleitung")
                && !strCharEventData.contains("\n"))
            bRedirect = true;

        if (!bRedirect)
            continue;

        // wir haben einen redirect - der wird in unsere Datenstruktur eingetragen
        int iStart = strCharEventData.indexOf("[[");
        int iEnd = strCharEventData.indexOf("]]");
        if (iStart < 0 || iEnd < 0)
            continue;
        if (iEnd <= iStart)
            continue;
        if ((iStart + 2) > strCharEventData.length() || iEnd > strCharEventData.length())
            continue;

        String strRedirectTarget = strCharEventData.substring(iStart + 2, iEnd).trim();
        hsPageTitle2Redirects.add(strRedirectTarget, strCurrentTitle);

        // if("Venceslav Konstantinov".equalsIgnoreCase(strCurrentTitle) || "Venceslav Konstantinov".equalsIgnoreCase(strRedirectTarget))
        // System.out.println("redirect found: (" + hsPageTitle2Redirects.keySize() + ") " + strCurrentTitle + " => '" + strRedirectTarget + "'");

    }

    Logger.getLogger(WikipediaDumpParser.class.getName())
            .info("Redirects found: " + StringUtils.beautifyNumber(hsPageTitle2Redirects.valueSize()));

    return hsPageTitle2Redirects;

}

From source file:org.alex73.osm.converters.bel.Convert.java

public static void main(String[] args) throws Exception {
    loadStreetNamesForHouses();/* w w  w .  ja va2 s. c  om*/

    InputStream in = new BZip2CompressorInputStream(
            new BufferedInputStream(new FileInputStream("tmp/belarus-latest.osm.bz2"), BUFFER_SIZE));

    // create xml event reader for input stream
    XMLEventFactory eventFactory = XMLEventFactory.newInstance();
    XMLEvent newLine = eventFactory.createCharacters("\n");
    XMLInputFactory xif = XMLInputFactory.newInstance();
    XMLOutputFactory xof = XMLOutputFactory.newInstance();
    XMLEventReader reader = xif.createXMLEventReader(in);
    XMLEventWriter wrCyr = xof.createXMLEventWriter(
            new BufferedOutputStream(new FileOutputStream("tmp/belarus-bel.osm"), BUFFER_SIZE));
    XMLEventWriter wrInt = xof.createXMLEventWriter(
            new BufferedOutputStream(new FileOutputStream("tmp/belarus-intl.osm"), BUFFER_SIZE));

    // initialize jaxb
    JAXBContext jaxbCtx = JAXBContext.newInstance(Node.class, Way.class, Relation.class);
    Unmarshaller um = jaxbCtx.createUnmarshaller();
    Marshaller m = jaxbCtx.createMarshaller();
    m.setProperty(Marshaller.JAXB_FRAGMENT, true);
    m.setProperty(Marshaller.JAXB_FORMATTED_OUTPUT, Boolean.TRUE);

    XMLEvent e = null;
    while ((e = reader.peek()) != null) {
        boolean processed = false;
        if (e.isStartElement()) {
            StartElement se = (StartElement) e;
            switch (se.getName().getLocalPart()) {
            case "way":
                Way way = um.unmarshal(reader, Way.class).getValue();
                if (way.getId() == 25439425) {
                    System.out.println();
                }
                fixBel(way.getTag(), "name:be", "name");
                String nameBeHouse = houseStreetBe.get(way.getId());
                if (nameBeHouse != null) {
                    setTag(way.getTag(), "addr:street", nameBeHouse);
                }
                m.marshal(way, wrCyr);
                fixInt(way.getTag());
                m.marshal(way, wrInt);
                wrCyr.add(newLine);
                wrInt.add(newLine);
                processed = true;
                break;
            case "node":
                Node node = um.unmarshal(reader, Node.class).getValue();
                fixBel(node.getTag(), "name:be", "name");
                // fixBel(node.getTag(),"addr:street:be","addr:street");
                m.marshal(node, wrCyr);
                fixInt(node.getTag());
                m.marshal(node, wrInt);
                wrCyr.add(newLine);
                wrInt.add(newLine);
                processed = true;
                break;
            case "relation":
                Relation relation = um.unmarshal(reader, Relation.class).getValue();
                fixBel(relation.getTag(), "name:be", "name");
                // fixBel(relation.getTag(),"addr:street:be","addr:street");
                m.marshal(relation, wrCyr);
                fixInt(relation.getTag());
                m.marshal(relation, wrInt);
                wrCyr.add(newLine);
                wrInt.add(newLine);
                processed = true;
                break;
            }
        }
        if (!processed) {
            wrCyr.add(e);
            wrInt.add(e);
        }
        reader.next();
    }

    wrCyr.flush();
    wrCyr.close();
    wrInt.flush();
    wrInt.close();
    System.out.println("UniqueTranslatedTags: " + uniqueTranslatedTags);
}

From source file:org.apache.olingo.client.core.serialization.AtomDeserializer.java

private PropertyType guessPropertyType(final XMLEventReader reader, final EdmTypeInfo typeInfo)
        throws XMLStreamException {

    XMLEvent child = null;/*from  w  w  w  .  j a va 2  s.  co  m*/
    while (reader.hasNext() && child == null) {
        final XMLEvent event = reader.peek();
        if (event.isCharacters() && event.asCharacters().isWhiteSpace()) {
            reader.nextEvent();
        } else {
            child = event;
        }
    }

    final PropertyType type;
    if (child == null) {
        type = typeInfo == null || typeInfo.isPrimitiveType() ? PropertyType.PRIMITIVE : PropertyType.ENUM;
    } else {
        if (child.isStartElement()) {
            if (Constants.NS_GML.equals(child.asStartElement().getName().getNamespaceURI())) {
                type = PropertyType.PRIMITIVE;
            } else if (elementQName.equals(child.asStartElement().getName())) {
                type = PropertyType.COLLECTION;
            } else {
                type = PropertyType.COMPLEX;
            }
        } else if (child.isCharacters()) {
            type = typeInfo == null || typeInfo.isPrimitiveType() ? PropertyType.PRIMITIVE : PropertyType.ENUM;
        } else {
            type = PropertyType.EMPTY;
        }
    }

    return type;
}

From source file:org.apache.olingo.client.core.serialization.AtomDeserializer.java

private StartElement getStartElement(final XMLEventReader reader) throws XMLStreamException {
    while (reader.hasNext()) {
        final XMLEvent innerEvent = reader.peek();
        if (innerEvent.isCharacters() && innerEvent.asCharacters().isWhiteSpace()) {
            reader.nextEvent();//from   ww  w . j a  v  a  2  s  .c  om
        } else if (innerEvent.isStartElement()) {
            return innerEvent.asStartElement();
        } else if (innerEvent.isEndElement() && inlineQName.equals(innerEvent.asEndElement().getName())) {
            return null;
        }
    }
    return null;
}

From source file:org.apache.olingo.commons.core.data.AtomPropertyDeserializer.java

private ODataPropertyType guessPropertyType(final XMLEventReader reader) throws XMLStreamException {
    XMLEvent child = null;//  w w w  . j a  v a  2  s.  co  m
    while (reader.hasNext() && child == null) {
        final XMLEvent event = reader.peek();
        if (event.isCharacters() && event.asCharacters().isWhiteSpace()) {
            reader.nextEvent();
        } else {
            child = event;
        }
    }

    final ODataPropertyType type;
    if (child == null) {
        type = ODataPropertyType.PRIMITIVE;
    } else {
        if (child.isStartElement()) {
            if (Constants.NS_GML.equals(child.asStartElement().getName().getNamespaceURI())) {
                type = ODataPropertyType.PRIMITIVE;
            } else if (elementQName.equals(child.asStartElement().getName())) {
                type = ODataPropertyType.COLLECTION;
            } else {
                type = ODataPropertyType.COMPLEX;
            }
        } else if (child.isCharacters()) {
            type = ODataPropertyType.PRIMITIVE;
        } else {
            type = ODataPropertyType.EMPTY;
        }
    }

    return type;
}