@Override
public void process() throws ProcessingException {
if (!enabled) {
return;
}
if (query == null) {
query = "";
}
// Create regexp patterns for each query word
final String[] queryTerms =
querySanitizePatternCompiled.matcher(query).replaceAll("").split("\\s+");
Pattern queryPattern = null;
List<String> patterns = Lists.newArrayList();
for (String queryTerm : queryTerms) {
if (Strings.isNullOrEmpty(queryTerm)) {
continue;
}
if (dontHighlightPatternCompiled != null
&& dontHighlightPatternCompiled.matcher(queryTerm).matches()) {
continue;
}
patterns.add("(" + Pattern.quote(escapeLtGt(queryTerm)) + ")");
}
if (patterns.size() > 0) {
queryPattern =
Pattern.compile(
Joiner.on("|").join(patterns), Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE);
}
// As we're going to modify documents, we need to copy them to
// avoid ConcurrentModificationExceptions.
final List<Document> inputDocuments = documents;
final List<Document> outputDocuments = Lists.newArrayListWithCapacity(inputDocuments.size());
for (Document document : inputDocuments) {
final Document clonedDocument = document.clone();
for (String fieldName : fields) {
highlightQueryTerms(clonedDocument, fieldName, queryPattern);
}
outputDocuments.add(clonedDocument);
}
documents = outputDocuments;
}
@Test
public void testRemoteUrl() throws MalformedURLException {
String base = System.getProperty("carrot2.xml.feed.url.base");
assumeTrue("carrot2.xml.feed.url.base property undefined.", !Strings.isNullOrEmpty(base));
IResource xml = new URLResourceWithParams(new URL(base + "&q=${query}&results=${results}"));
final String query = "apple computer";
processingAttributes.put(AttributeUtils.getKey(XmlDocumentSource.class, "xml"), xml);
processingAttributes.put(AttributeNames.QUERY, query);
processingAttributes.put(AttributeNames.RESULTS, 50);
final int documentCount = runQuery();
assertEquals(50, documentCount);
assertEquals(query, resultAttributes.get(AttributeNames.QUERY));
}
