Question

I have a problem with using elastic search at the moment. When I try to perform a search and only want a subset of the fields returned, I need to specify the fields using a dot-notation if the fields are nested. Here’s a sample of my mapper json document which maps my couchDB document:

{
    "product": {
        "_type": {"store": "yes"},
        "_source": {"compress": true},
        "index_analyzer": "standard",
        "search_analyzer": "standard",
        "dynamic_date_formats": ["date_time_no_millis", "date_optional_time"],
        "properties": {
              "_id": {"type": "string", "store": "yes", "index": "not_analyzed"},
            "key": {"type": "string", "store": "yes"},
            "content": {
                "type": "object",
                "path": "just_name",
                "properties": {
                    "key": {"type": "string", "store": "yes"},
                    "name": {"type": "string", "store": "yes", "index_name": "name"},
                    "description": {"type": "string", "store": "yes", "index_name": "description"},
                    "brand": {
                        "type": "object",
                        "index_name": "brand",
                        "properties": {
                            "abbreviation": {"type": "string", "store": "yes", "index_name": "brand_abbreviation"},
                            "name": {"type": "string", "store": "yes", "index_name": "brand_name"}
                        }
                    }
                                }
                        }
                 }
          }
}

Referring to _id would just be a simple _id, but say I wanted to refer to name in the content, I would have to refer to it as content.name. The problem with this is that when the search output is out, the json output contains the field name as: “content.name”.

Is it possible to rename this as “name” and without the “content.” prefix? You can see, I tried to specify index_name, but that seemed useless.

Answer 1

You can use partial_fields to do this.

For instance, if you index a doc like this:

curl -XPUT 'http://127.0.0.1:9200/test/test/1?pretty=1'  -d '
{
   "email" : "[e-mail]",
   "name" : "john",
   "foo" : {
      "bar" : {
         "baz" : 1
      }
   }
}
'

You can include the fields/objects that you want like this:

curl -XGET 'http://127.0.0.1:9200/test/test/_search?pretty=1'  -d '
{
   "partial_fields" : {
      "doc" : {
         "include" : [
            "name",
            "foo.*"
         ]
      }
   }
}
'

Which will give you a result like this: (note the missing email field, and that field foo remains a hash – it is not flattened with dot notation)

{
   "hits" : {
      "hits" : [
         {
            "_score" : 1,
            "fields" : {
               "doc" : {
                  "name" : "john",
                  "foo" : {
                     "bar" : {
                        "baz" : 1
                     }
                  }
               }
            },
            "_index" : "test",
            "_id" : "1",
            "_type" : "test"
         }
      ],
      "max_score" : 1,
      "total" : 1
   },
   "timed_out" : false,
   "_shards" : {
      "failed" : 0,
      "successful" : 5,
      "total" : 5
   },
   "took" : 1
}

On a side note, some comments about your mapping:

• your _id field (which I assume is meant to be the elasticsearch ID, not an external ID) is at the wrong level – it should be at the same level as _type. If it IS an external ID, then it’s at the right level.
• Why are you storing all your fields? There really is no need – it just uses extra resources. Unless you have a massive _source field, it is much faster to retrieve just that field and parse it, rather than hitting the disk for each individual field.