Retrieve results with aggregations

[caufab]

I'm reopening this issue as I was unable to reply in time.
I need to retrieve the results of a search (made with a mix of search/where methods) along with bucket aggregations. This is needed as my Laravel app is exposing APIs to a frontend consumer that will show results and filter facets (taken from buckets).

I tried the same query that this package builds, directly into Elasticsearch and it works fine: it returns results and the aggregation buckets.

The issue
Let's say I make a search like this:

$builder = Product::query();
$builder->search('notebook');
$builder->whereNotNull('name');
....
// other where stuff
...
$builder->size(10);

This returns correctly the 10 elements requested, but when next to size() I add this:

$options = [ 'field' => 'name', 'size' => 50 ]
$builder->bucket('name', 'terms', $options);

I notice that I get the aggregation buckets for name but 0 results, and If a print the query with

$builder->toDsl();

I can see the query has size: 0.
I saw that this is being done in the Grammar class - if I understand correctly - here, and I can't find a way to re-set the size to 10, so I'm currently forced to make two queries, one for the results with size 10, one identical with aggregations (and size 0) and merge them into the response.

My question
Is it actually necessary to reset size when using bucket aggregations?
If it is for performance reasons, does it perform better even than making two queries?

[pdphilip]

Hey @caufab, as asked before, please give me the full example that I can recreate.

The size of Aggs is essential to making groupBy() and distinct() work correctly, which uses bucket Aggs under the hood.

It'll help me to see your real-world bucket example outside of groupBy() and distinct() - also share your work around please. Thanks

[caufab]

Hello @pdphilip, yes, here is a full example:

class DocumentController extends Controller
{
    public function index(Request $request)
    {

        // Search the first 10 documents from 2020 and order by datestamp descending
        // Build query
        $builder = Document::query();
        $builder->where('datestamp', '>=', '2020-01-01');
        $builder->orderBy('datestamp', 'desc');
        // Set page and size
        $builder->offset(0);
        $builder->limit(10);

        // set bucket aggregations by topic
        $bucketOptions = [
            'field' => 'topic.id',
            'size' => 2,
        ];
        $builder->bucket('topic', 'terms', $bucketOptions);

        $documents = $builder->getRaw();

        $aggregations = $builder->query()->getRawAggregationResults();

        return response()->json([
            'documents' => $documents,
            'aggregations' => $aggregations,
            'query' => $builder->toDsl(),
        ]);

    }

}

This returns this json:

{
  "documents": {
    "took": 346,
    "timed_out": false,
    "_shards": { "total": 1, "successful": 1, "skipped": 0, "failed": 0 },
    "hits": {
      "total": { "value": 2297476, "relation": "eq" },
      "max_score": null,
      "hits": []
    },
    "aggregations": {
      "topic": {
        "doc_count_error_upper_bound": 0,
        "sum_other_doc_count": 2933637,
        "buckets": [
          {
            "key": "hist",
            "doc_count": 417389
          },
          {
            "key": "musiq",
            "doc_count": 405667
          }
        ]
      }
    }
  },
  "aggregations": {
    "topic": {
      "doc_count_error_upper_bound": 0,
      "sum_other_doc_count": 2933637,
      "buckets": [
        {
          "key": "hist",
          "doc_count": 417389
        },
        {
          "key": "musiq",
          "doc_count": 405667
        }
      ]
    }
  },
  "query": {
    "index": "documents_index",
    "body": {
      "query": {
        "range": {
          "datestamp": {
            "gte": "2020-01-01"
          }
        }
      },
      "sort": [
        {
          "datestamp": {
            "order": "desc"
          }
        }
      ],
      "size": 0,
      "track_total_hits": true,
      "aggs": {
        "topic": {
          "terms": {
            "field": "topic.id",
            "size": 2
          }
        }
      }
    }
  }
}

First thing is that when I set bucket aggregations if I call ->get() method it returns an array of objects of the aggs ({ "topic": "hist" }, { "topic": "musiq" }), so as you can see I used getRaw (which made useless using getRawAggregationResults later, but no problem).

The main problem is that hits are empty because the size param in the body of the query has been set to 0.
If I comment/remove the bucket(), getRawAggregationResults() methods,the hits are shown and size is of course 10.

[pdphilip]

Hey @caufab, haven't forgotten about this one. Only a time issue since it's not a quick fix.

Will let you know

[pdphilip]

@caufab - fixed with v5.2.0