public NtcirTopicReader(File topicFile)
throws ParserConfigurationException, IOException, SAXException {
DocumentBuilder documentBuilder = XMLHelper.getDocumentBuilder(true);
topics = documentBuilder.parse(topicFile);
// TODO: Find out how this code duplication can be avoided in Java.
queryGenerator = new XQueryGenerator(topics);
}
/**
* Splits the given NTCIR query file into individual queries, converts each query into an XQuery
* using XQueryGenerator, and returns the result as a list of NtcirPatterns for each individual
* query.
*
* @return List of NtcirPatterns for each query
* @throws XPathExpressionException Thrown if xpaths fail to compile or fail to evaluate +
*/
public final List<NtcirPattern> extractPatterns() throws XPathExpressionException {
final XPath xpath = XMLHelper.namespaceAwareXpath("t", NS_NII);
final XPathExpression xNum = xpath.compile("./t:num");
final XPathExpression xFormula = xpath.compile("./t:query/t:formula");
final NonWhitespaceNodeList topicList =
new NonWhitespaceNodeList(topics.getElementsByTagNameNS(NS_NII, "topic"));
for (final Node node : topicList) {
final String num = xNum.evaluate(node);
final NonWhitespaceNodeList formulae =
new NonWhitespaceNodeList((NodeList) xFormula.evaluate(node, XPathConstants.NODESET));
for (final Node formula : formulae) {
final String id = formula.getAttributes().getNamedItem("id").getTextContent();
final Node mathMLNode = getFirstChild(formula);
queryGenerator.setMainElement(getFirstChild(mathMLNode));
patterns.add(new NtcirPattern(num, id, queryGenerator.toString(), mathMLNode));
}
}
return patterns;
}