Java Code Examples for javax.xml.stream.XMLEventReader#getElementText()

@Test
public void test() {
    String xmlData = "<?xml version=\"1.0\"?><Test>Hello</Test>";
    try {
        XMLEventReader xmlReader = XMLInputFactory.newInstance().createXMLEventReader(new ByteArrayInputStream(xmlData.getBytes()));

        XMLEvent event = xmlReader.nextEvent();
        System.out.println(event.getClass());

        // xmlReader.peek(); // error in both cases with/without peek()
        event = xmlReader.nextTag(); // nextEvent() would work fine
        // nextTag() forgets to set fLastEvent
        System.out.println(event.getClass());

        String text = xmlReader.getElementText();
        System.out.println(text);
    } catch (XMLStreamException e) {
        Assert.fail(e.getMessage());
    }
}
 

/**
 * Parses an inline string from cell in XML format
 * 
 * @param xer XMLEventReader from which to read the inline string content
 * @throws XMLStreamException           in case the string item cannot be
 *                                      correctly read from the XML file
 * @throws FormatNotUnderstoodException in case a string cannot be identified in
 *                                      cell
 */
private String parseCellInlineStringText(XMLEventReader xer)
		throws XMLStreamException, FormatNotUnderstoodException {
	String result = "";
	XMLEvent xe;
	while ((xe = xer.nextTag()).isStartElement()) {
		String elementName = xe.asStartElement().getName().getLocalPart().toUpperCase();
		switch (elementName) {
		case "T": // normal text
			result = xer.getElementText();
			break;
		case "R": // rich text (returned as normal text)
			result = this.parseCellInlineStringRichText(xer);
			break;
		case "RPH": // phonetic (ignored)
		case "PHONETICPR": // phonetic properties (ignored)
			this.skipXMLElementHierarchy(xer);
			break;
		default:
			LOG.error("Unknown inline string tag: " + elementName);
			throw new FormatNotUnderstoodException("Unknown inline string tag: " + elementName);

		}
	}
	return result;
}
 

/**
 * Parses a rich text item of a shared string table and returns the unformatted
 * text
 * 
 * @param xer
 * @return unformatted text of rich text item
 * @throws FormatNotUnderstoodException
 * @throws XMLStreamException
 */
private String parseCellInlineStringRichText(XMLEventReader xer)
		throws XMLStreamException, FormatNotUnderstoodException {
	String result = "";
	XMLEvent xe;
	while ((xe = xer.nextTag()).isStartElement()) {
		String elementName = xe.asStartElement().getName().getLocalPart().toUpperCase();
		switch (elementName) {
		case "T": // normal text
			result = xer.getElementText();
			break;
		case "RPR": // run properties (ignored)
		default:
			LOG.error("Unknown rich text inline string tag: " + elementName);
			throw new FormatNotUnderstoodException("Unknown rich text inline string tag: " + elementName);

		}
	}

	return result;
}
 

/**
 * Parses a string item from a SST in XML format
 * 
 * @param xer XMLEventReader from which to read the next string item
 * @throws XMLStreamException           in case the string item cannot be
 *                                      correctly read from the XML file
 * @throws FormatNotUnderstoodException in case a string item cannot be
 *                                      identified in the shared string table
 *                                      (e.g. unknown type)
 */
private String parseSIText(XMLEventReader xer) throws XMLStreamException, FormatNotUnderstoodException {
	String result = "";
	XMLEvent xe;
	while ((xe = xer.nextTag()).isStartElement()) {
		String elementName = xe.asStartElement().getName().getLocalPart().toUpperCase();
		switch (elementName) {
		case "T": // normal text
			result = xer.getElementText();
			break;
		case "R": // rich text (returned as normal text)
			result = this.parseSIRichText(xer);
			break;
		case "RPH": // phonetic (ignored)
		case "PHONETICPR": // phonetic properties (ignored)
			this.skipXMLElementHierarchy(xer);
			break;
		default:
			LOG.error("Unknown string item in shared string table: " + elementName);
			throw new FormatNotUnderstoodException("Unknown string item in shared string table: " + elementName);

		}
	}
	return result;
}
 

/**
 * Parses a rich text item of a shared string table and returns the unformatted
 * text
 * 
 * @param xer
 * @return unformatted text of rich text item
 * @throws FormatNotUnderstoodException
 * @throws XMLStreamException
 */
private String parseSIRichText(XMLEventReader xer) throws XMLStreamException, FormatNotUnderstoodException {
	String result = "";
	XMLEvent xe;
	while ((xe = xer.nextTag()).isStartElement()) {
		String elementName = xe.asStartElement().getName().getLocalPart().toUpperCase();
		switch (elementName) {
		case "T": // normal text
			result = xer.getElementText();
			break;
		case "RPR": // run properties (ignored)
		default:
			LOG.error("Unknown rich text string item in shared string table: " + elementName);
			throw new FormatNotUnderstoodException(
					"Unknown rich text string item in shared string table: " + elementName);

		}
	}

	return result;
}
 

/**
 * Parse the module-option element
 * @param xmlEventReader
 * @return
 * @throws XMLStreamException
 */
public Map<String, Object> parse(XMLEventReader xmlEventReader) throws XMLStreamException
{
   Map<String, Object> options = new HashMap<String,Object>();
   
   //See if there are options
   while(true)
   {
      XMLEvent xmlEvent = xmlEventReader.peek();
      if(xmlEvent instanceof EndElement) break;
      StartElement peekedStartElement = (StartElement) xmlEvent;
      if(xmlEvent == null)
         break; //no module options
      
      String peekedStartElementName = StaxParserUtil.getStartElementName(peekedStartElement);
      
      if("module-option".equals(peekedStartElementName))
      {
         xmlEvent = xmlEventReader.nextEvent();
         Attribute attribute = (Attribute) peekedStartElement.getAttributes().next();
         
         //Sometime, there may be embedded xml in the option. We cannot use peek
         //next event here because the event reader jumps to the next module option
         //in the presence of a text (and not embedded xml). Since embedded xml is rare,
         //we are going to rely on exceptions as a mode of control. The issue is that
         //we have used an event filter on the XMLEventReader for convenience
         Object val = null;
         try
         {
            val = xmlEventReader.getElementText();
         }
         catch(XMLStreamException xse)
         {
            //Look for embedded xml
            XMLEvent embeddedOrText = xmlEventReader.peek();
            if(embeddedOrText.getEventType() == XMLStreamConstants.START_ELEMENT)
            { 
               val = embeddedXMLParsing(xmlEventReader); 
            }   
         } 
         options.put(attribute.getValue(), val );
      }
      else break; 
   }
   return options;
}
 

private void handleMetrics(XMLEventReader reader, StartElement element)
    throws XMLStreamException
  {
    DatatypeFactory dtFactory;
    try {
      dtFactory = DatatypeFactory.newInstance();
    } catch (DatatypeConfigurationException dce) {
      throw new MarkLogicIOException("Cannot instantiate datatypeFactory", dce);
    }

    Calendar now = Calendar.getInstance();

    QName queryName    = new QName(SEARCH_NS, "query-resolution-time");
    QName facetName    = new QName(SEARCH_NS, "facet-resolution-time");
    QName snippetName  = new QName(SEARCH_NS, "snippet-resolution-time");
    QName metadataName = new QName(SEARCH_NS, "metadata-resolution-time");
    QName extractName  = new QName(SEARCH_NS, "extract-resolution-time");
    QName totalName    = new QName(SEARCH_NS, "total-time");

    long qrTime = -1;
    long frTime = -1;
    long srTime = -1;
    long mrTime = -1;
    long erTime  = -1;
    long tTime  = -1;

    QName metricsName = element.getName();
    events: while (reader.hasNext()) {
      XMLEvent event = reader.nextEvent();

      int eventType = event.getEventType();
      switch (eventType) {
        case XMLStreamConstants.START_ELEMENT:
          StartElement startElement = event.asStartElement();
          QName startName = startElement.getName();
          String readerValue = reader.getElementText();
   if (readerValue != null && readerValue.length() > 0) {
if (queryName.equals(startName)) {
	qrTime = parseTime(dtFactory, now, readerValue);
} else if (facetName.equals(startName)) {
	frTime = parseTime(dtFactory, now, readerValue);
} else if (snippetName.equals(startName)) {
	srTime = parseTime(dtFactory, now, readerValue);
} else if (metadataName.equals(startName)) {
	mrTime = parseTime(dtFactory, now, readerValue);
} else if (extractName.equals(startName)) {
	erTime = parseTime(dtFactory, now, readerValue);
} else if (totalName.equals(startName)) {
	tTime = parseTime(dtFactory, now, readerValue);
} else {
	logger.warn("Unexpected metrics element " + startName.toString());
}
   }
          break;
        case XMLStreamConstants.END_ELEMENT:
          if (metricsName.equals(event.asEndElement().getName())) {
            break events;
          }
          break;
      }
    }

    tempMetrics = new SearchMetricsImpl(qrTime, frTime, srTime, mrTime, erTime, tTime);
  }