Collection<CalaisEntityOccurrence> result = new ArrayList<CalaisEntityOccurrence>();
try {
SelectQuery sQuery = (SelectQuery) QueryParser.getInstance().parse(query);
ResultSet rs = tcManager.executeSparqlQuery(sQuery, model);
while (rs.hasNext()) {
SolutionMapping row = rs.next();
CalaisEntityOccurrence occ = new CalaisEntityOccurrence();
Resource disambiguated = row.get("did");
occ.id = (disambiguated == null ? row.get("id") : disambiguated);
if (onlyNERMode) {
occ.type = row.get("type");
}
else {
occ.type = (disambiguated == null ? row.get("type") : row.get("dtype"));
}
if (calaisTypeMap != null) {
UriRef mappedType = calaisTypeMap.get(occ.type);
if (mappedType != null) {
occ.type = mappedType;
}
}
occ.name = ((Literal)row.get("name")).getLexicalForm();
occ.exact = ((Literal)row.get("exact")).getLexicalForm();
//TODO for html the offsets might not be those of the original document but refer to a cleaned up version?
occ.offset = Integer.valueOf(((Literal) row.get("offset")).getLexicalForm());
// remove brackets
occ.context = ((Literal) row.get("context")).getLexicalForm().replaceAll("[\\[\\]]", "");
occ.length = Integer.valueOf(((Literal) row.get("length")).getLexicalForm());
if (row.get("score") != null) {
occ.relevance = Double.valueOf(((Literal) row.get("score")).getLexicalForm());
}
result.add(occ);
}
} catch (ParseException e) {
throw new EngineException("Unable to parse SPARQL query for processing OpenCalais results",e);