# Elasticsearch

The Elasticsearch plugin provides Rosetta with functionality relating to the Elasticsearch search engine.

<dependency>
  <groupId>com.k-int.rosetta</groupId>
  <artifactId>rosetta-elasticsearch</artifactId>
  <version>3.0.0</version>
</dependency>

The elasticsearch Provider allows Rosetta to communicate with Elasticsearch clusters. Specifically, it interprets the object contained in the request field of a DataServiceRequest as an Elasticsearch search request, which is then executed against the _search endpoint for a specific index, or for all indices if none is specified.

The response is processed such that:

• The _source field of each hit is registered as a provider result
• The hits.total.value field is used as the total value within the provider statistics
• Aggregations are processed into a custom provider field, aggregations, which is a list of Rosetta Aggregation objects.

name: my-provider
type: elasticsearch
properties:
  protocol: http
  host: localhost
  port: 9200
  path_prefix: /es/base/path/
  index: my-index
  skip_ping: true
  username: rosetta
  password: password
  client_timeout: 60000
  aggregations_key: aggs
  date_range_format: {from} to {to}

The elastic-generic-search Glyph transforms a GenericSearchRequest into an Elasticsearch search request.

The top-level query generated by this Glyph is a boolean Elasticsearch query. Each GenericSearchRequest parameter is converted as follows:

• The from and size parameters are interpreted literally.
• Each value in queries is used to generate one query for each of the fields configured in the search_config.search_fields property.
  A query_string query is generated if query string syntax elements are detected (e.g. "OR" or "AND") in the value, or a match query is generated otherwise.
  All queries generated from the queries list are combined in a should clause of a boolean query, which is in turn combined with queries resulting from other parameters using a must clause of the top-level boolean query.
• Each value in ids is used to generate one query for each of the fields configured in the search_config.id_fields property.
  A match query is generated in each case. All queries generated from the ids list are combined in a should clause of a boolean query, which is in turn combined with queries resulting from other parameters using a must clause of the top-level boolean query.
• Each entry in the filters map is treated as a separate filter whose key is interpreted as either a field alias or a literal Elasticsearch field (see the Properties table) and the value is a list of values against which that field should match. Each filter is combined in the filter list of the top-level boolean query (so uses AND-like logic) and each filter value is combined in a the should clause of a nested boolean query (OR-like logic).
• Each entry in the fields map generates a separate query where the key interpreted as either a field alias or a literal Elasticsearch field and the value either generates a query_string query or a match query, depending on the presence of query string syntax (like with queries). The queries generated by each entry contribute to the must clause of the top-level boolean query.
• Each item in the aggregations list generates either a date_range aggregation if the value matches a configured value of date_aggregations[].agg_name, or a terms aggregation on the specified field otherwise. The value is interpreted as either a field alias or a literal Elasticsearch field.
• Each entry in the autocomplete map generates a specialized terms aggregation using the key as the field/alias and the value to produce a regular expression in the include parameter of the aggregation, such that only terms containing words prefixed by the value are included in the entry list for that aggregation (note the regular expression produced is case-insensitive). This parameter can be used to produce "autocomplete" style suggestions for a given text input and field, which can be useful for faceting on a field with many distinct values, for instance.
• Each item in the sort_clauses list generates an item in the Elasticsearch sort request parameter using path as the sort field/alias and direction as the sort direction.

As Field but with the following additional property.

- field: _generic_all_std
  boost: 0.0
- field: _all
  boost: 0.0

- field: @admin.id
  boost: 0.0
- field: admin.id
  boost: 0.0

name: my-glyph
type: elastic-generic-search
properties:
  search_config:
    search_fields:
      - field: my_search_field
        boost: 0.0
    id_fields:
      - field: my_id_field
        boost: 0.0
    fields:
      - field: my_field
        boost: 0.0
    exists_fields:
      - field: my_exists_field
        boost: 0.0
        requirement: none
  path_config:
    paths:
      alias_1: field_1
      alias_2: field_2
      alias_3: field_3
    require_alias: false
  aggregation_configs:
    - name: alias_1
      size: 20
      auto_complete_size: 100
  default_aggregation_size: 10
  date_aggregations:
    - agg_name: date_field_alias
      path: date_elasticsearch_field_1
      additional_path: date_elasticsearch_field_2
      format: dd/MM/yyyy
      date_ranges:
        - from: 1850
          to: 1900
        - from: 1900
          to: 1950
        - from: 1950
          to: 2000
  date_range_format: '{from} - {to}'
  global_filter: my_field:(value_1 OR value_2 OR value_3)
  match_operator: and
  default_operator: and
  request_timeout: 3000