ansaurus

Question

Lucene.NET result using sql like LIMIT? + 1k question

Answer 1

+2 A:

The search method of the Searcher class has a parameter to limit the number of results returned for a query.

Query query = parser.parse("Apple");
TopDocs topDocs = searcher.search(query, 20);

But Lucene does not support pagination. You will have to redo your query and keep the results that fits the range that you need.

See this question.

Pascal Dimassimo 2010-09-21 21:15:22

I am now worried about what may happen if my app grows to handle hundred of thousands of docs. I'll drop the 2.9 requirement and hope there is an answer. I'll accept in a few days if there is no other solution. I'm disappointed lucene doesnt support this.

acidzombie24 2010-09-21 21:51:39

Lucene is usually really fast when redoing the same search. You should test first to see how it goes for your app.

Pascal Dimassimo 2010-09-22 00:46:39

I agree with Pascal, paging is almost always the wrong way to go with Lucene. The time complexity of finding the top n documents grows with log(n) (see http://philosophyforprogrammers.blogspot.com/2010/09/lucene-performance.html), so increasing from 20 to 40 is really a very trivial change.

Xodarap 2010-09-22 16:08:32

Interesting article Xodarap. It looks like it is fast. Ok i'll be doing this when the bounty is over.

acidzombie24 2010-10-01 03:48:19

Answer 2

+1 A:

Searcher.search, you use in second line, has also signature:

Search(Query query, Filter filter, HitCollector results)

Use HitCollector to flush temporary result into fast and temporary storage. For example if user asks first 20 - you need to return it and in background thread start caching all another. Really you need to store only document's ID, so for 1 millions result approximately 4Mb is expected.

When result is in the storage it is simple to support paging.

Dewfy 2010-10-02 19:59:09

ansaurus

tags:

views:

answers:

Lucene.NET result using sql like LIMIT? + 1k question

1k Question

related questions