Flummi Elastic Search HTTP REST Client
Java Shell
Latest commit f567f76 Jan 17, 2017 Martin Scheidweiler bump version to 5.0.25.0
Permalink
Failed to load latest commit information.
bin upgrade to gradle 3.2.1 Dec 15, 2016
src remove unnecessary nesting of sum aggregations Jan 17, 2017
.gitignore Remove unnecessary stuff from build.gradle Jul 5, 2016
.travis.yml
HISTORY.md bump version to 5.0.25.0 Jan 17, 2017
LICENSE dynp-poc|Rename everything to de.otto.flummi.* Jul 5, 2016
README.md update readme with new versioning Jan 11, 2017
build.gradle bump version to 5.0.25.0 Jan 17, 2017

README.md

Flummi Elastic Search HTTP Client

Flummi is a client library for Elastic Search. It has been successfully tested with Elastic Search versions 2.3, 2.4 and 5.1. It provides a comprehensive Java query DSL API and communicates with the Elastic Search Cluster via HTTP/JSON. It is licensed under the Apache 2 License.

Why should I use Flummi?

  • Flummi uses HTTP and JSON for communication with Elastic Search. Its only dependencies are Gson and AsyncHttpClient, so it is good for you if you don't want to have your application depend on the full ElasticSearch JAR.
  • Flummi's API is as close as possible to the original Elmastic Search transport client API. This makes it very easy to port existing client code to Flummi.
  • Flummi uses the Elastic Search Scroll API for downloading large result sets as a stream of smaller pages.
  • It supports parent-child relationships

Limitations

  • Flummi does not support cluster load balancing yet. You can use a hardware loadbalancer or HTTP Proxy such as nginx.
  • Although it supports the most common query and request types, it is not yet fully feature complete. When you need a request or query type that is not yet supported by Flummi, please feel free to add it and send us a Pull Request!

Which version should I use?

For Elasticsearch 2.x we recommend to use version 0.21.0.

For Elasticsearch 5.x you should use the latest version.

How to use Flummi

You can simply include Flummi in your Maven or Gradle build as follows.

For Maven users:

<dependency>
    <groupId>de.otto</groupId>
    <artifactId>flummi</artifactId>
    <version>5.0.23.0</version>
</dependency>

For gradle users:

compile "de.otto:flummi:5.0.23.0"

Flummi versions starting with 5.x are intended for use with Elastic Search 5.x, flummi versions starting with 2.x should be used with elastic 2.x.

Getting started

For using Flummi in a Java application, initialize it as follows.

AsyncHttpClient asyncHttpClient = new AsyncHttpClient();
Flummi flummi = new Flummi(asyncHttpClient, "http://elasticsearch.base.url:9200");

Using Flummi with Spring

For using Flummi in a Spring or Spring Boot application, you can add a simple @Configuration class for initialization and then autowire Flummi in your beans.

@Configuration
public class FlummiConfiguration {

   @Bean
   public AsyncHttpClient asyncHttpClient() {
     return new AsyncHttpClient();
   }

   @Bean
   public Flummi flummi() {
     return new Flummi(asyncHttpClient(), "http://elasticsearch.base.url:9200");
   }
}

Creating an index

The following example creates a products index with a customized analyzer for the name property.

JsonObject settings = GsonHelper.object(
    "analysis", GsonHelper.object(
        "analyzer", GsonHelper.object(
            "lowercase-analyzer", GsonHelper.object(
                "tokenizer", "keyword-tokenizer",
                "filter", "lowercase-filter"
            )
        ),
        "tokenizer", GsonHelper.object(
            "keyword-tokenizer", GsonHelper.object(
                "type", "keyword"
            )
        ),
        "filter", GsonHelper.object(
            "lowercase-filter", GsonHelper.object(
                "type", "lowercase"
            )
        )
    )
);
JsonObject mappings = GsonHelper.object(
    "products", GsonHelper.object(
        "properties", GsonHelper.object(
            "name", GsonHelper.object(
                "type", "string",
                "store", "yes",
                "analyzer", "lowercase-analyzer",
                "fields", GsonHelper.object(
                    "raw", GsonHelper.object(
                      "type", "string",
                      "index", "not_analyzed"
                    )
                )
            ),
            "color", GsonHelper.object(
                "type", "string",
                "store", "no",
                "index", "not_analyzed"
            )
        )
    )
);

flummi.admin().indices()
   .prepareCreate("products")
   .setSettings(settings)
   .setMappings(mappings)
   .execute();

Indexing documents

A simple example that adds a product to the products index

JsonObject bouncingBall1 = GsonHelper.object(
   "name", "Bouncing Ball small",
   "color", "green"
   );

flummi.prepareIndex()
    .setId("bblsml-4710")
    .setSource(bouncingBall1)
    .setIndexName("products")
    .setDocumentType("product")
    .execute();

Bulk Requests

A bulk request is a single HTTP request that contains multiple actions. For indexing large amounts of data, this is much more efficient than sending one request for every document. The following simple example adds some products to the product index using a Bulk Request

JsonObject bouncingBall1 = GsonHelper.object(
   "name", "Bouncing Ball with smiley",
   "color", "yellow"
   );
JsonObject bouncingBall2 = GsonHelper.object(
   "name", "Bouncing Ball XL extra bouncy",
   "color", "transparent"
   );

flummi.prepareBulk()
    .add(
       new IndexActionBuilder("products")
           .setSource(bouncingBall1)
           .setId("bblsmly-4711")
           .setType("product")
           )
    .add(
       new IndexActionBuilder("products")
           .setSource(bouncingBall2)
           .setId("bblxlxb-4712")
           .setType("product")
           )
    .execute();

Executing Queries

A simple example that finds up to 10 yellow-colored products in the products index:

SearchRequestBuilder searchRequestBuilder = flummi
   .prepareSearch("products")
   .setTypes("product")
   .setSize(10)
   .setQuery(
      QueryBuilders.termQuery("color", "yellow")
        .build()
   )
   .setTimeoutMillis(150);

SearchResponse searchResponse = searchRequestBuilder.execute()

System.out.println("Found " + searchResponse.getHits().getTotalHits() + " products");
searchResponse.getHits()
   .stream().map(hit -> hit.getSource().get("name").getAsString())
   .forEach(name -> System.out.println("Name: " + name));

Streaming large result sets with the Scroll API

For streaming large result sets, Flummi uses the Elastic Search Scroll API to split the result set into smaller pages and thus reduce memory usage and network bandwidth. To use it, simply setScroll("1m") on your SearchRequestBuilder before calling execute().

Aggregation queries

The following example shows how to do simple terms bucket aggregations.

SearchRequestBuilder searchRequestBuilder = flummi
   .prepareSearch("products")
   .setTypes("product")
   .setSize(10)
   .setQuery(
      QueryBuilders.matchAll().build()
   )
   .addAggregation(
      new TermsBuilder("Colors").field("color").size(0)
   );

SearchResponse searchResponse = searchRequestBuilder.execute()

AggregationResult colors = searchResponse.getAggregations().get("Colors");
colors.getBuckets().forEach(bucket -> System.out.println(
   "Found " + bucket.getDocCount() + " " + bucket.getKey() + " products"));

Compatibility

Flummi is compatible with Elastic Search 2.3 and 2.4. It probably also works with 5.0 but we haven't tested this comprehensively yet.

Contribution Guide

You want to contribute new features to Flummi? Great!

Flummi is built using the gradle wrapper gradlew. After cloning the git repository, you can create an IntelliJ Idea project file with the following command

./bin/gradlew idea

Before you push, you might want to run all the unit tests with the following command

./bin/gradlew clean check

And don't forget to send us your pull request!