A simple column-oriented ORM for the ClickHouse database in Swift.
Features:
- Fixed datatype ORM
- Connection pool
- Fully asynchronous based on the adapter ClickHouseNio
- Integrated with Vapor 4
- Add
ClickHouseVapor
as a dependency to yourPackage.swift
dependencies: [
.package(url: "https://github.com/patrick-zippenfenig/ClickHouseVapor.git", from: "1.0.0")
],
targets: [
.target(name: "MyApp", dependencies: ["ClickHouseVapor"])
]
- Build your project:
swift build
-
Configure the connection credentials with a Vapor 4 application. Usually this is done in
config.swift
.Note:
maxConnectionsPerEventLoop
controls the number of connections per thread. If you have 4 CPU cores and Vapor is using 4 eventLoops, 8 connections will be used.requestTimeout
is the timeout to establish a connection. It does not limit query runtime.import ClickHouseVapor let app = Application(.testing) defer { app.shutdown() } app.clickHouse.configuration = try ClickHousePoolConfiguration( hostname: "localhost", port: 9000, user: "default", password: "admin", database: "default", maxConnectionsPerEventLoop: 2, requestTimeout: .seconds(10) )
-
Define a table with fields and an engine.
public class TestModel : ClickHouseModel { @Field(key: "timestamp", isPrimary: true, isOrderBy: true) var timestamp: [Int64] @Field(key: "stationID", isPrimary: true, isOrderBy: true) var id: [String] @Field(key: "fixed", fixedStringLen: 10) var fixed: [ String ] @Field(key: "temperature") var temperature: [Float] required public init() { } public static var engine: ClickHouseEngine { return ClickHouseEngineReplacingMergeTree( table: "test", database: nil, cluster: nil, partitionBy: "toYYYYMM(toDateTime(timestamp))" ) } }
-
Create a table. For simplicity this example is calling
wait()
. It is discouraged to usewait()
in production.try TestModel.createTable(on: app.clickHouse).wait()
-
Insert data
let model = TestModel() model.id = [ "x010", "ax51", "cd22" ] model.fixed = [ "", "123456", "12345678901234" ] model.timestamp = [ 100, 200, 300 ] model.temperature = [ 11.1, 10.4, 8.9 ] try model.insert(on: app.clickHouse).wait()
-
Query all data again
let result = try TestModel.select(on: app.clickHouse).wait() print(result.temperature) // [ 11.1, 10.4, 8.9 ] // Filter data in more detail let result2 = try! TestModel.select( on: app.clickHouse, fields: ["timestamp", "stationID"], where: "temperature > 10", order: "timestamp DESC", limit: 10, offset: 0 ).wait() print(result2.id) // ["ax51", "x010"] print(result2.timestamp) // [200, 100] // Perform raw queries, but assign the result to TestModel let sql = "SELECT timestamp, stationID FROM default.test" let result2 = try! TestModel.select(on: app.clickHouse, sql: sql).wait()
-
If you have several models that follow a common base scheme, you can also use inheritance to keep your code tidy:
open class TestParentClass { @Field(key: "timestamp", isPrimary: true, isOrderBy: true) var timestamp: [Int64] @Field(key: "stationID", isPrimary: true, isOrderBy: true, isLowCardinality: true) var id: [String] } public final class InheritedTestModel: TestParentClass, ClickHouseModel { @Field(key: "temperature") var temperature: [Float] override public init() {} public static var engine: ClickHouseEngine { return ClickHouseEngineReplacingMergeTree( table: "testInherited", database: nil, cluster: nil, partitionBy: "toYYYYMM(toDateTime(timestamp))" ) } }
- Query timeouts
- Implement more engines
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.