On Mar 23, 2010, at 12:58 AM, Benoit Mercier wrote: > Hi, > > I would like to write a query composed of a BooleanQuery (several clauses) > and a SpanQuery (SpanNearQuery), where both are mandatory. Sounds simple > but I have to work on spans returned by this query. > > I know that I could use a Filter, but my goal is to get the spans from the « > combined » query : BooleanQuery + SpanQuery. Even if I filter my > BooleanQuery with the SpanQuery, spans returned by the > SpanQuery.getSpans(reader) are not « filtered » by the BooleanQuery. Since > executing the query is not needed to get spans from a SpanQuery I understand > this behaviour. > > My current implementation first runs the BooleanQuery filtered by the > SpanQuery. I then get the spans from the SpanQuery and remove from them all > docs that are not in the score docs returned by the filtered BooleanQuery. Is > there a more efficient, simple or clever way to reach the same goal? > > Thank you very much in advance for your advices.
If you are 3.x: I think maybe you could reverse this around. Get a filter from your BooleanQuery and get the DocIdSet and then advance through the Spans and the DocIdSetIterator, as they will both be forward facing. For each span, check to see whether that doc is in the filter or not. In 2.x, I think on the filter you can get the BitSet and then just directly look up to see if the current span is in the bit set. In either case, I don't think this will be that big of a performance hit as it is all a forward facing iteration. -Grant --------------------------------------------------------------------- To unsubscribe, e-mail: java-user-unsubscr...@lucene.apache.org For additional commands, e-mail: java-user-h...@lucene.apache.org
