@Override
public final List<String[]> analyzeSandboxed(final Document document)
throws XPathExpressionException {
Preconditions.checkNotNull(document);
final NodeList mainNodes = ParserUtil.evaluateXPath(document, this.getXPath());
if (mainNodes.getLength() == 0) {
return null;
}
final Node mainNode = mainNodes.item(0);
this.assertBeforeProcessingBegins();
this.enterProcessingStage();
final String[] token = new String[this.tokenLenght];
token[0] = this.getVenueNameFromNode(mainNode);
token[0] = VenueUtil.cleanupVenueName(token[0]);
token[1] = this.getVenueAddressFromNode(mainNode);
token[1] = CleanupUtil.cleanUpName(token[1]);
if (!this.isTokenValid(token)) {
this.endProcessingOfToken();
return Lists.newArrayList();
}
this.assertDuringProcessing();
this.endProcessingOfToken();
final List<String[]> tokenInList = Lists.<String[]>newArrayList();
tokenInList.add(token);
Preconditions.checkNotNull(tokenInList);
return tokenInList;
}
@Override
protected final String[] analyzeNodeInternal(final Node node) {
final String[] token = new String[this.tokenLenght];
// 0 - event full name
token[0] = this.getStringFromNode(node);
token[0] = CleanupUtil.cleanUpName(token[0]);
// 1 - name of band/artist
token[1] = BandUtil.cleanBandName(token[0]);
// 2 - name of venue
token[2] = this.getVenueFromNode(node.getParentNode().getNextSibling().getNextSibling());
token[2] = CleanupUtil.cleanUpName(token[2]);
token[2] = VenueCleanupComponent.extractKnownVenueNameFromLargerText(token[2]);
// 3 - date
token[3] = this.getDateFromNode(node.getParentNode().getNextSibling().getNextSibling());
token[3] = DateAnalysisUtil.cleanupDate(token[3]);
return token;
}
/**
* - note: the assumption that once the XPath reaches the elements, everything is one big list
* with no interruptions is wrong <br>
*/
@Override
protected final String[] analyzeNodeInternal(final Node node) {
final String[] token = new String[this.tokenLenght];
String text = null;
text = this.getStringFromNode(node);
if (text == null) { // skipping over noise
return null;
}
text = text.trim();
token[0] = this.getStreetType(text);
if (token[0] != null) {
token[0] = CleanupUtil.cleanUpName(token[0]);
}
token[1] = this.getStreetName(text);
if (token[1] != null) {
token[1] = StreetUtil.cleanStreetName(token[1]);
}
return token;
}
