从Elasticsearch聚合中的第一个和最后一个文档返回字段

Question

I have a sorted elasticsearch histogram aggregation that works fine as is.

I need to return a field from the first and last document of each aggregation bucket.

Current query (ruby-style syntax):

{
  query: {
    filtered: {
      filter: {
        bool: {
          must: [
            { term: { some_id: 'something' } },
            { range: {
                completed_at: {
                  gte: start,
                  lte: end
                }
              }
            }
          ]
        }
      },
      _cache: true,
      _cache_key: "special-query"
    }
  },
  aggs: {
    intervals: {
      histogram: {
        field: 'completed_at',
        interval: 24.hours.to_i,
        min_doc_count: 0,
        order: { _key: 'asc' }
      },
      aggs: {
        start_at: { min: { field: 'completed_at' } },
        end_at: { max: { field: 'completed_at' } },
        price_stats: { extended_stats: { field: 'rate' } },
      }
    }
  }
}

I have searched through the documentation and googled extensively but can't find any solutions. Currently I have to send a second query to the database to explicitly fetch all the values for this field but I would like to roll that functionality in to my elasticsearch query.

Answer 1

You want to return a result per aggregation right?

That's called Field Collapsing and unfortunately isn't yet available in Elasticsearch, see: https://github.com/elasticsearch/elasticsearch/issues/256

Update sept 2014

As per @bpierce in comments below: Now that Elasticsearch 1.3.x is out, field collapsing is available via the top_hits aggregation. I'm currently using it in a query similar to this one.