fcallejon/fes

F# Elasticsearch

FEs

FEs tries to be an F# client for Elasticsearch.

Motivation

Although both Elasticsearch.Net and NEST are great to use from C#, using them from F# means doing weird stuff around lambdas and such.

Also using Fleece to decode/encode JSON gives better control over documents going back and forth.

About Fleece

I'm in the process of updating it to the future version. For this reason there are a few warnings that will be output by the build.

How to use it

There is no nuget yet sadly, meaning to use this library it will have to be clone/downloaded, build and use the output.

1. Add the following opens:

   open Fes
   open Fes.DSL
   open Fes.Builders
   open Fes.Builders.Indices

2. Create the mapping for an Index:

   let req =
        createIndexRequest {
            name "indexName"
            mappings
                [|
                    mappingDefinition {
                        name "field1"
                        fieldType (FieldTypes.DateTypes.Date |> Fields.Date)
                    }
                    mappingDefinition {
                        name "field2"
                        fieldType (FieldTypes.Keyword |> Fields.Keywords)
                    }
                    mappingDefinition {
                        name "nestedField1"
                        fieldType Fields.Nested
                        mappings
                            [| ([| ("innerField1", FieldTypes.Integer |> Fields.Numeric)
                                   ("innerField2", FieldTypes.Keyword |> Fields.Keywords ) |]
                                |> Mapping.FieldMapping.Properties) |]
                    }
                |]
            settings (
                indexSettings {
                    shards 3us
                    replicas 0us
                })
            aliases [| indexAlias { name "Test" } |]
        }

3. Make the call: At the moment any Http Client will work. The following is just an example using some Http functions from Fes that I'm planning to move outside of the "core" library. Both Http.toRequest and Http.Response.toResult would map from and to HttpRequestMessage and HttpResponseMessage respectively.

    [<RequireQualifiedAccess>]
    module ElasticsearchClient =
    open System.Net.Http
   
    let inline execute (client: HttpClient) req =
        (Http.toRequest >> Async.retn) req
        |> AsyncResult.bind (client.SendAsync >> AsyncResult.waitTask)
        |> AsyncResult.bind Http.Response.toResult
   
    let client =
        let client = new HttpClient()
        client.BaseAddress <- getEsServer()
        client
    let inline executeElasticsearchCall (req: 'a) =
        ElasticsearchClient.execute client req
    let createResult : Result<IndexCreateResponse, exn> =
        executeElasticsearchCall req |> Async.RunSynchronously

Road map (WIP)

• Update Fleece (WIP)
• DSL for Indices
  • DSL to Create Indices
  • DSL to Update Indices
  • DSL to Manage Aliases
  • Add CE constructs
• DSL to Index Documents
  • Initial Implementation to index Documents
  • Add CE constructs for initial implementations
• DSL for Search
  • Initial Implementation for Queries
  • Add CE constructs for initial implementations
• Http Client
  • Initial Exceptions/Errors handling
  • Rethink Http Client
    • Http Client independent
    • Improve Exceptions/Errors handling
  • Add retries

Contributing

I'm doing this as a weekend-only project, if you are interested in contributing please do. Adding tests would be a great place to help, but ES Rest API it's really big so help there will also be appreciated.

Also moving and rethinking the Http Client or anything in Roadmap is welcome 😁.