Namespaces

In use-cases where it’s not feasible/convenient to create a new field in the schema for each field inside a tree. Namespace enabled fields allow efficient searching of values within nested fields. For example, this feature can be used for tags, file-system path names, or any tree-like structure for which contained names can be many, dynamic and/or not known in advance.

The _namespace parameter must be enabled during the schema field creation:

UPDATE /bank/1

{
  "style": {
    "_namespace": true,
    "clothing": {
      "pants": "khakis",
      "footwear": "casual shoes"
    },
    "hairstyle": "afro"
  }
}

The above example is the document being indexed, the parameter _namespace part enables the Namespace Queries functionality.

Because the field is namespaced, every value in the tree stays searchable by its path. Querying the full path matches the exact value:

SEARCH /bank/

{
  "_query": {
    "style.clothing.pants": "khakis"
  }
}

By default every level of the path is also indexed on its own, so a shallower, partial path matches as well (this is the behaviour the next section controls):

SEARCH /bank/

{
  "_query": {
    "style.clothing": "khakis"
  }
}

See Namespace Queries for the full set of ways to match namespaced paths.

Partial Paths

By default namespaced fields keep information for each level in the path. This behaviour can be modified by setting _partial_paths to false, when creating a field schema:

UPDATE /bank/1

{
  "hobbies": {
    "_namespace": true,
    "_partial_paths": false,
    "Observation": {
      "Indoors": {
        "Learning": 10,
        "Reading": 7
      },
      "Outdoors": {
        "Traveling": 3
      }
    },
    "Competitive": {
      "Indoors": {
        "Boxing": 4,
        "Judo": 10
      },
      "Outdoors": {
        "Baseball": 2
      }
    }
  }
}

Partial paths can be used for Namespace Queries.

Datatype

The concrete datatype for all nested objects must be of the same type as one declared in the object enabling the _namespace.