Crate opensearch

source ·
Expand description

Official Rust client for OpenSearch

OpenSearch is an official Rust client for OpenSearch, providing an efficient asynchronous client for all stable OpenSearch APIs that’s easy to use.

Versions and Compatibility

Rust clientOpenSearch
1.x1.x
2.x2.x, 1.x^
  • ^: With the exception of some previously deprecated APIs

A major version of the client is compatible with the same major version of OpenSearch. Since OpenSearch is developed following Semantic Versioning principles, Any minor/patch version of the client can be used against any minor/patch version of OpenSearch within the same major version lineage. For example,

  • A 1.5.0 client can be used against 1.0.0 OpenSearch
  • A 1.4.0 client can be used against 1.5.1 OpenSearch

In the former case, a 1.5.0 client may contain additional API functions that are not available in 1.0.0 OpenSearch. In this case, these APIs cannot be used, but for any APIs available in OpenSearch, the respective API functions on the client will be compatible.

In the latter case, a 1.4.0 client won’t contain API functions for APIs that are introduced in OpenSearch 1.5.0+, but for all other APIs available in OpenSearch, the respective API functions on the client will be compatible.

In some instances, a new major version of OpenSearch may remain compatible with an older major version of the client, which may not warrant a need to update the client. Please consult COMPATIBILITY.md for more details.

No compatibility assurances are given between different major versions of the client and OpenSearch. Major differences likely exist between major versions of OpenSearch, particularly around request and response object formats, but also around API urls and behaviour.

Features

The following are a list of Cargo features that can be enabled or disabled:

  • native-tls (enabled by default): Enables TLS functionality provided by native-tls.
  • rustls-tls: Enables TLS functionality provided by rustls.
  • beta-apis: Enables beta APIs. Beta APIs are on track to become stable and permanent features. Use them with caution because it is possible that breaking changes are made to these APIs in a minor version.
  • experimental-apis: Enables experimental APIs. Experimental APIs are just that - an experiment. An experimental API might have breaking changes in any future version, or it might even be removed entirely. This feature also enables beta-apis.
  • aws-auth: Enables authentication with Amazon OpenSearch and OpenSearch Serverless. Performs AWS SigV4 signing using credential types from aws-types.

Getting started

Add the opensearch crate and version to Cargo.toml. Choose the version that is compatible with the version of OpenSearch you’re using

[dependencies]
opensearch = "1.0.0"

The following optional dependencies may also be useful to create requests and read responses

serde = "~1"
serde_json = "~1"

Async support with tokio

The client uses reqwest to make HTTP calls, which internally uses the tokio runtime for async support. As such, you may require to take a dependency on tokio in order to use the client. For example, in Cargo.toml, you may need the following dependency

tokio = { version = "*", features = ["full"] }

and to attribute async main function with #[tokio::main]

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // your code ...
    Ok(())
}

and attribute test functions with #[tokio::test]

#[tokio::test]
async fn my_test() -> Result<(), Box<dyn std::error::Error>> {
    // your code ...
    Ok(())
}

Create a client

To create a client to make API calls to OpenSearch running on http://localhost:9200

let client = OpenSearch::default();

Alternatively, you can create a client to make API calls against OpenSearch running on a specific url::Url

let transport = Transport::single_node("https://example.com")?;
let client = OpenSearch::new(transport);

More control over how a Transport is built can be achieved using TransportBuilder to build a transport, and passing it to OpenSearch::new create a new instance of OpenSearch

let url = Url::parse("https://example.com")?;
let conn_pool = SingleNodeConnectionPool::new(url);
let transport = TransportBuilder::new(conn_pool).disable_proxy().build()?;
let client = OpenSearch::new(transport);

Making API calls

The client exposes all stable OpenSearch APIs, either on the root OpenSearch client, or on a namespace client that groups related APIs, such as Cat, which groups the Cat related APIs. All API functions are async and can be awaited.

The following makes an API call to the cat indices API

let response = client
    .cat()
    .indices(CatIndicesParts::Index(&["*"]))
    .send()
    .await?;

let response_body = response.json::<Value>().await?;
for record in response_body.as_array().unwrap() {
    // print the name of each index
    println!("{}", record["index"].as_str().unwrap());
}

For APIs that contain parts of the Url path to be provided by the consumer, the Url path variants are modelled as an enum, such as CatIndicesParts in the above example, which models the variants of the CatIndices API.

Indexing

Indexing a single document can be achieved with the index API

let response = client
    .index(IndexParts::IndexId("tweets", "1"))
    .body(json!({
        "id": 1,
        "user": "kimchy",
        "post_date": "2009-11-15T00:00:00Z",
        "message": "Trying out OpenSearch, so far so good?"
    }))
    .send()
    .await?;

let successful = response.status_code().is_success();

For indexing multiple documents, the bulk API is a better option, allowing multiple operations to be sent in one API call

let mut body: Vec<JsonBody<_>> = Vec::with_capacity(4);

// add the first operation and document
body.push(json!({"index": {"_id": "1"}}).into());
body.push(json!({
    "id": 1,
    "user": "kimchy",
    "post_date": "2009-11-15T00:00:00Z",
    "message": "Trying out OpenSearch, so far so good?"
}).into());

// add the second operation and document
body.push(json!({"index": {"_id": "2"}}).into());
body.push(json!({
    "id": 2,
    "user": "forloop",
    "post_date": "2020-01-08T00:00:00Z",
    "message": "Bulk indexing with the rust client, yeah!"
}).into());

let response = client
    .bulk(BulkParts::Index("tweets"))
    .body(body)
    .send()
    .await?;

let response_body = response.json::<Value>().await?;
let successful = response_body["errors"].as_bool().unwrap() == false;

Searching

The following makes an API call to tweets/_search with the json body {"query":{"match":{"message":"OpenSearch"}}}

let response = client
    .search(SearchParts::Index(&["tweets"]))
    .from(0)
    .size(10)
    .body(json!({
        "query": {
            "match": {
                "message": "OpenSearch rust"
            }
        }
    }))
    .send()
    .await?;

let response_body = response.json::<Value>().await?;
let took = response_body["took"].as_i64().unwrap();
for hit in response_body["hits"]["hits"].as_array().unwrap() {
    // print the source document
    println!("{:?}", hit["_source"]);
}

Request bodies

For APIs that expect JSON, the body associated function of the API constrains the input to a type that implements serde::Serialize trait. An example of this was the indexing a single document example above.

Some APIs expect newline delimited JSON (NDJSON) however, so the body associated for these APIs constrain the input to a vector of types that implement Body trait. An example of this was the bulk indexing multiple documents above.

The Body trait represents the body of an API call, allowing for different body implementations. As well as those to represent JSON and NDJSON, a few other types also have implementations for Body, such as byte slice. Whilst these can’t be passed to the API functions directly, OpenSearch::send can be used

let body = b"{\"query\":{\"match_all\":{}}}";

let response = client
    .send(Method::Post,
        SearchParts::Index(&["tweets"]).url().as_ref(),
        HeaderMap::new(),
        Option::<&Value>::None,
        Some(body.as_ref()),
        None,
    )
    .await?;

Amazon OpenSearch and OpenSearch Serverless

For authenticating against an Amazon OpenSearch or OpenSearch Serverless endpoint using AWS SigV4 request signing, you must enable the aws-auth feature, then pass the AWS credentials to the TransportBuilder. The easiest way to retrieve AWS credentials in the required format is to use aws-config.

[dependencies]
opensearch = { version = "1", features = ["aws-auth"] }
aws-config = "0.10"

let creds = aws_config::load_from_env().await;
let url = Url::parse("https://...")?;
let region_provider = RegionProviderChain::default_provider().or_else("us-east-1");
let aws_config = aws_config::from_env().region(region_provider).load().await.clone();
let conn_pool = SingleNodeConnectionPool::new(url);
let transport = TransportBuilder::new(conn_pool)
    .auth(aws_config.clone().try_into()?)
    .service_name("es") // use "aoss" for OpenSearch Serverless
    .build()?;
let client = OpenSearch::new(transport);

Re-exports

Modules

Structs

Enums