CODEWITHSUNDEEP
elasticsearch
Oliver H
Oliver H

Reputation: 133

Limiting results by multiple attributes

I've got data with record, version, title and expires fields. Record is a non-unique field, and each record can have a number of versions.

In search results, I need to only show one of each version.

So is it possible to search by title for articles that on/before a certain date, but only return one result for each version.

For example, given this data: 

{"record": 1, "version": 1, "title": "Hello", "expires": "2011-08-17 00:00:00"},
{"record": 1, "version": 2, "title": "Hello", "expires": "2012-08-17 00:00:00"},
{"record": 2, "version": 1, "title": "Hello world", "expires": "2010-08-17 00:00:00"},
{"record": 2, "version": 2, "title": "Hello world", "expires": "2011-08-17 00:00:00"},
{"record": 2, "version": 3, "title": "Hello world", "expires": "2012-08-17 00:00:00"},

searching for documents containing "Hello" in the title, that expired on/before 2012-08-18, should return: 

{"record": 1, "version": 2, "title": "Hello", "expires": "2012-08-17 00:00:00"},
{"record": 2, "version": 3, "title": "Hello world", "expires": "2012-08-17 00:00:00"}

(the most recent 'version' of each record).

Any ideas?

Will I have to iterate over the results outside of ES? Thanks for reading!

Upvotes: 1

Views: 84

Answers (1)

Constantijn Visinescu
Constantijn Visinescu

Reputation: 752

What you want is called field collapsing and it's one of the few features Apache Solr has and ElasticSearch doesn't.

http://wiki.apache.org/solr/FieldCollapsing

There's a lot of requests for this feature in ElasticSearch but it's not implemented yet.

Upvotes: 1

Related Questions