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

1. Add ClickHouseVapor as a dependency to your Package.swift

dependencies:[.package(url:"https://github.com/patrick-zippenfenig/ClickHouseVapor.git", from:"1.0.0")],targets:[.target(name:"MyApp", dependencies:["ClickHouseVapor"])]

2. Build your project:

swift build

1. 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 letapp=Application(.testing)defer{ app.shutdown()} app.clickHouse.configuration =tryClickHousePoolConfiguration( hostname:"localhost", port:9000, user:"default", password:"admin", database:"default", maxConnectionsPerEventLoop:2, requestTimeout:.seconds(10))

2. Define a table with fields and an engine.

   publicclassTestModel:ClickHouseModel{@Field(key:"timestamp", isPrimary:true, isOrderBy:true)vartimestamp:[Int64]@Field(key:"stationID", isPrimary:true, isOrderBy:true)varid:[String]@Field(key:"fixed", fixedStringLen:10)varfixed:[String]@Field(key:"temperature")vartemperature:[Float]requiredpublicinit(){}publicstaticvarengine:ClickHouseEngine{returnClickHouseEngineReplacingMergeTree( table:"test", database:nil, cluster:nil, partitionBy:"toYYYYMM(toDateTime(timestamp))")}}

3. Create a table. For simplicity this example is calling wait(). It is discouraged to use wait() in production.

   tryTestModel.createTable(on: app.clickHouse).wait()

4. Insert data

   letmodel=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()

5. Query all data again

   letresult=tryTestModel.select(on: app.clickHouse).wait()print(result.temperature) // [ 11.1, 10.4, 8.9 ] // Filter data in more detail letresult2=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 letsql="SELECT timestamp, stationID FROM default.test"letresult2=try!TestModel.select(on: app.clickHouse, sql: sql).wait()

6. If you have several models that follow a common base scheme, you can also use inheritance to keep your code tidy:

   openclassTestParentClass{@Field(key:"timestamp", isPrimary:true, isOrderBy:true)vartimestamp:[Int64]@Field(key:"stationID", isPrimary:true, isOrderBy:true, isLowCardinality:true)varid:[String]}publicfinalclassInheritedTestModel:TestParentClass,ClickHouseModel{@Field(key:"temperature")vartemperature:[Float]overridepublicinit(){}publicstaticvarengine:ClickHouseEngine{returnClickHouseEngineReplacingMergeTree( 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.

MIT