TL;DR

Elastisch is a battle tested, small but feature rich and well documented Clojure client for ElasticSearch. It supports virtually every Elastic Search feature and has solid documentation.

1.1.0-beta1 is a development release that has several new features, most notably a native client with the same API as the REST one Elastisch already has.

Changes in 1.1.0-beta1

Native Client

Elastisch 1.1.0 includes a major new feature: native ElasticSearch client. The client uses ElasticSearch’s Java API, and can be used with both transport and node clients.

Rationale

Native client is more bandwidth efficient. It also can use SMILE (binary JSON format) to be more efficient on the wire.

Namespace Layout

Native client API in Elastisch is nearly identical to that of the REST API client and resides in clojurewerkz.elastisch.native and clojurewerkz.elastisch.native.* namespaces (similarly to how clojurewerkz.elastisch.rest clojurewerkz.elastisch.rest.* namespaces are organized).

Connections

Transport client (used for TCP/remote connections) connections are set up using clojurewerkz.elastisch.native/connect!. Note that you need to provide node configuration that at least has cluster name in it:

1
2
3
4
5
6

(require '[clojurewerkz.elastisch.native :as es])

;; note that transport client uses port 9300 by default.
;; it also can connect to multiple cluster nodes
(es/connect! [["127.0.0.1" 9300]]
             {"cluster.name" "elasticsearch_antares" })

Cluster name and transport node addresses can be retrieved via HTTP API, for example:

1
2

curl http://localhost:9200/_cluster/nodes
{"ok":true,"cluster_name":"elasticsearch_antares","nodes":...}}

Performing Operations

The Native client tries to be as close as possible to the existing REST client API. For example, document operation functions in clojurewerkz.elastisch.native.document, such as clojurewerkz.elastisch.native.document/create, follow clojurewerkz.elastisch.rest.document function signatures as closely as possible:

1
2
3
4
5
6
7
8
9

;; in the REPL
(require '[clojurewerkz.elastisch.native :as es])
(require '[clojurewerkz.elastisch.native.document :as doc])

(es/connect! [["127.0.0.1" 9300]]
             {"cluster.name" "elasticsearch_antares" })

(doc/put index-name index-type id document)
(doc/get index-name index-type id)

The same with returned results. Note, however, that ES transport client does have (very) minor differences with the REST API and it is not always possible for Elastisch to completely cover such differences.

Async Operations

Native client offers a choice of synchronous (blocking calling thread until a response is received) and asynchronous (returns a future) versions of multiple API operations:

1
2
3
4
5
6
7
8
9
10
11
12
13
14

;; in the REPL
(require '[clojurewerkz.elastisch.native :as es])
(require '[clojurewerkz.elastisch.native.document :as doc])

(es/connect! [["127.0.0.1" 9300]]
             {"cluster.name" "elasticsearch_antares" })

(doc/put index-name index-type id document)

;; returns a response
(doc/get index-name index-type id)
;; returns a future that will eventually
;; contain a response
(doc/async-get index-name index-type id)

One notable exception to this is administrative operations (such as opening or closing an index). The rationale for this is that they are rarely executed on the hot code path (e.g. in tight loops), so convenience and better error visibility is more important for them.

GH issues: #17, #18, #20.

Note that native ElasticSearch client currently relies on ElasticSearch 0.90.0.Beta1 client libraries and some operations will only work with that version.

Bulk Request Support

Bulk requests are now supported. All the relevant code is in the clojurewerkz.elastisch.rest.bulk namespace. Here is a small example of bulk document indexing using this new API:

1
2
3

(require '[clojurewerkz.elastisch.rest.bulk :as eb])

(eb/bulk (eb/bulk-index doc1 doc2 doc3) :refresh true)

Contributed by Davie Moston.

Scroll Queries Support

Scroll queries are now easier to perform thanks to the new clojurewerkz.elastisch.rest.document/scroll function that takes a scroll id and amount of time retrieved documents and related information will be kept in memory for future retrieval. They are analogous to database cursors.

A short code example:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

(require '[clojurewerkz.elastisch.rest.document :as doc])
(require '[clojurewerkz.elastisch.query :as q])
(require '[clojurewerkz.elastisch.rest.response :refer [hits-from]])

(let [index-name   "articles"
      mapping-type "article"
      response     (doc/search index-name mapping-type
                               :query (q/query-string :query "*")
                               :search_type "scan"
                               :scroll "1m"
                               :size 1)
      scroll-id     (:_scroll_id response)
      scan-response (doc/scroll scroll-id :scroll "1m")
      scan-hits     (hits-from scan-response)]
  (println scan-hits))

Contributed by Davie Moston.

Cheshire Update

Cheshire dependency has been upgraded to version 5.0.2.

clj-http Update

clj-http dependency has been upgraded to version 0.6.4.

Count API No Longer Ignores Mapping Types

clojurewerkz.elastisch.rest.document/count no longer ignores mapping types.

GH issue: #6.

Count API now uses GET requests

clojurewerkz.elastisch.rest.document/count now correctly uses GET for requests without the query part andPOST for request that have it.

GH issue: #5.

Change Log

Elastisch change log is available on GitHub.

We recommend all users to give 1.1.0-beta1 a try.

Thank You, Contributors

Kudos to Davie Moston and Andrew Jones who contributed to this release.

Elastisch is a ClojureWerkz Project

Elastisch is part of the group of libraries known as ClojureWerkz, together with

• Welle, a Riak client with batteries included
• Monger, a Clojure MongoDB client for a more civilized age
• Neocons, a client for the Neo4J REST API
• Langohr, a Clojure client for RabbitMQ that embraces the AMQP 0.9.1 model
• Quartzite, a powerful scheduling library

and several others. If you like Elastisch, you may also like our other projects.

Let us know what you think on Twitter or on the Clojure mailing list.

Michael on behalf of the ClojureWerkz Team