Flummi is a client library for Elastic Search. It has been successfully tested with Elastic Search versions 6 and 7. 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.
- 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
- 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!
Flummi versions starting with 7.x are intended for use with Elastic Search 6 and 7. Use older versions of the Library for older ES versions. Flummi versions starting with 8.x are intended for use with Elastic Search 8, because the field _type is not supported in ES8. https://www.elastic.co/guide/en/elasticsearch/reference/current/migrating-8.0.html#breaking-changes-8.0
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>8.0.0.0</version>
</dependency>
For gradle users:
compile "de.otto:flummi:8.0.0.0"
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");
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");
}
}
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();
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();
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") // type isn't supported in ES8
)
.add(
new IndexActionBuilder("products")
.setSource(bouncingBall2)
.setId("bblxlxb-4712")
.setType("product") // type isn't supported in ES8
)
.execute();
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));
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().
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"));
Flummi version 7.x is compatible with Elastic Search 6.x and 7.x Flummi version 8.x is compatible with Elastic Search 8.x
Flummi is build with JDK 8
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!