Percolating existing document api

[martijnvg]

A common use case is to percolate a document that has just been indexed. A big advantage would be to just specify the index, type and id of a document (which is returned with each index response) that has just been indexed, internally a get request can then fetch the content of the recently indexed document and then the percolate api will execute on that.

Percolate an existing document with id 1 and type tweet from index twitter:

curl -XGET 'localhost:9200/twitter/tweet/1/_percolate'

Percolate an existing document with id 2 and type tweet from index twitter-2013: (percolating 2013 tweet against 2012 queries)

curl -XGET 'localhost:9200/twitter-2013/tweet/2/_percolate?percolate_index=twitter-2012'

Additional options for percolating an existing document on top of existing percolator options:

• id - The id of the document to retrieve the source for.
• percolate_index - The index containing the percolate queries. Defaults to the index defined in the url.
• percolate_type - The percolate type (used for parsing the document). Default to type defined in the url.
• routing - The routing value to use when retrieving the document to percolate.
• preference - Which shard to prefer when retrieving the existing document.
• percolate_routing - The routing value to use when percolating the existing document.
• percolate_preference - Which shard to prefer when executing the percolate request.
• version - Enables a version check. If the fetched document's version isn't equal to the specified version then the request fails with a version conflict and the percolation request is aborted.
• version_type - Whether internal or external versioning is used. Defaults to internal versioning.

Internally the percolate api will issue a get request for fetching the_source of the document to percolate.
For this feature to work the _source for documents to be percolated need to be stored.

[phrone]

Hi! Great work, but would it be possible to use a specified document field instead of the _source?

[martijnvg]

@phrone This is actually possible via get source filtering [1], but it just needs to be exposed to the percolate rest api.

1: http://www.elasticsearch.org/guide/en/elasticsearch/reference/master/docs-get.html#get-source-filtering

[fransflippo]

Just what I need! Is this included in the Java API yet?

[martijnvg]

@fransflippo Yes, it is included in the Java api. There is a PercolateRequestBuilder#setGetRequest(GetRequest) method. The passed get request is internally used to fetch the document needed for percolating.