Full-text search skip and limit

[chrszrkl]

I need to limit the search results returned by a full-text search based on an offset and limit. That means, the first query would return the first 10 results (offset=0, limit=10) and a second query would return the next 10 results (offset=10, limit=10).

My current query uses apoc.agg.slice() to slice the result set accordingly:

CALL db.index.fulltext.queryNodes($index, $searchQuery) YIELD node, score
WITH apoc.agg.slice({node: node, score: score}, $offset, $limit) as records
RETURN collect({node: node, score: score}) as records

However, I was checking the Java implementation for db.index.fulltext.queryNodes() and noticed that there is a third parameter options which allows skip and limit: 

CALL db.index.fulltext.queryNodes($index,$searchQuery,{skip: 0, limit: 10})

This parameter seems to not be officially documented on the full-text search manual.

Therefore, my two questions are:

1. 1. What's the official recommendation by Neo4j?
2. 2. Is there a performance difference between apoc.agg.slice() and the passing options as 3rd parameter?

[steggy · Accepted solution]

Hi @chrszrkl 

Good catch. We will get this updated in the docs. This should be more performant than the apoc option, as it can push the filtering into Lucene.

edit: it's here: https://neo4j.com/docs/operations-manual/current/reference/procedures/#procedure_db_index_fulltext_q...

J

[chrszrkl]

Great, that makes the pagination for search results a little easier!