The goal of this project is to provide a comprehensive yet easy to use Elasticsearch client for Swift. The Vapor server side framework has a large community around it so integrating with Vapor was a logical first step. That said, this library should be very easy to port to another framework (Perfect, Kitura) or even use by itself for command line utilities and other such purposes.
Main priorities are to provide index management (field mapping, settings, tokenizers and analyzers), CRUD support and search results with support for aggregations. Currently these goals are all being met on some level.
This project is under heavy development and the public API has been changing (with no backward compatability) every week. That said, the changes tend to be fairly minor as long as you're diligent with pulling the latest code every week.
All development and testing is being done using the Elasticsearch 6.x series. This has implications around document types as they have been depricated in Elasticsearch 6.0 and will be removed in 7.0. Given that multiple types per index is a thing of the past and this client is a thing of the future, supporting multiple types per index didn't seem like a good fit. If using an older version of Elasticsearch, keep this limitation in mind.
Add Elasticsearch
to the Package dependencies:
dependencies: [
...,
.package(url: "https://github.com/ryangrimm/VaporElasticsearch", .branch("master"))
]
as well as to your target (e.g. "App"):
targets: [
...
.target(
name: "App",
dependencies: [... "Elasticsearch" ...]
),
...
]
Make sure that you've imported Elasticsearch
everywhere needed:
import Elasticsearch
Add the ElasticsearchDatabase
in your configure.swift
file:
let esConfig = ElasticsearchClientConfig(hostname: "localhost", port: 9200)
let es = try ElasticsearchDatabase(config: esConfig)
var databases = DatabasesConfig()
databases.add(database: es, as: .elasticsearch)
services.register(databases)
var databases = DatabasesConfig()
databases.enableLogging(on: .elasticsearch)
services.register(databases)
struct Document: Codable {
var id: String
var title: String
}
func list(_ req: Request) throws -> Future<[Document]> {
let query = Query(
Match(field: "id", value: "42")
)
return req.withNewConnection(to: .elasticsearch) { conn in
return try conn.search(
decodeTo: Document.self,
index: "documents",
query: SearchContainer(query)
)
}.map(to: [Document].self ) { searchResponse in
guard let hits = searchResponse.hits else { return [Document]() }
let results = hits.hits.map { $0.source }
return results
}
}
//let client: ElasticsearchClient = ...
let synonymFilter = SynonymFilter(name: "synonym_filter",
synonyms: ["file, document", "nice, awesome, great"])
let synonymAnalyzer = CustomAnalyzer(
name: "synonym_analyzer",
tokenizer: StandardTokenizer(),
filter: [synonymFilter]))
let index = client.configureIndex(name: "documents")
.indexSettings(index: IndexSettings(shards: 5, replicas: 1))
.property(key: "id", type: MapKeyword())
.property(key: "title", type: MapText(analyzer: synonymAnalyzer))
try index.create()
//let client: ElasticsearchClient = ...
try client.deleteIndex(name: "documents")
bulk
to insert documents//let client: ElasticsearchClient = ...
let doc1 = Document(id: 1, title: "hello world")
let doc2 = Document(id: 5, title: "awesome place")
let bulk = client.bulkOperation()
bulk.defaultHeader.index = "documents"
try bulk.create(doc: doc1, id: String(doc1.id))
try bulk.create(doc: doc2, id: String(doc2.id))
// if you want to overwrite documents, use `bulk.index` instead
try bulk.send()