Ratpack Workflow

A workflow engine for Ratpack, built on the Ratpack Execution Model. Requires Ratpack 1.3.3.

Try it

Add this to your build.gradle:

repositories {
  maven { url 'http://oss.jfrog.org/artifactory/repo' }
}

dependencies {
  compile 'com.danveloper.ratpack.workflow:ratpack-workflow:0.3.3'
}

Quick Start

The project ships with an extension to the regular Ratpack server specification. Nothing about your application needs to change except the class from which you import the RatpackServer. The below example shows importing the Groovy DSL from the GroovyRatpackWorkflow class, instead of Ratpack's Groovy class.

@Grab('com.danveloper.ratpack.workflow:ratpack-workflow-groovy:0.3.3')

import com.danveloper.ratpack.workflow.server.RatpackWorkflow
import static com.danveloper.ratpack.workflow.server.GroovyRatpackWorkflow.ratpack

ratpack {
  workflow {
    work("type", "version") {
      // do one thing
      ...
      // trigger completion
      complete()
    }
    flow("name", "version") {
      all {
        // always run this
        ...
        // move on to the next
        next()
      }
      all {
        ...
        complete()
      }
    }
  }
  handlers {
    prefix("flows") {
      get(":id", RatpackWorkflow.flowStatusGetHandler())
      get(RatpackWorkflow.flowListHandler())
      post(RatpackWorkflow.flowSubmissionHandler())
    }
    prefix("works") {
      get(":id", RatpackWorkflow.workStatusGetHandler())
      get(RatpackWorkflow.workListHandler())
      post(RatpackWorkflow.workSubmissionHandler())
    }
    // ...
  }
}

RESTful HTTP handlers are provided via static methods on the RatpackWorkflow class. You can use these endpoints to list jobs, lookup specific jobs, and submit jobs.

Submitting Work

A properly formed JSON payload can be POSTed to the FlowSubmissionHandler to start a flow.

{
  "name": "AWESOME-FLOW-001"
  "description": "My awesome flow thingy",
  "tags": {
    "key": "val"
  },
  "works": [
    {
      "type": "workSomething",
      "version": "1.0",
      "data": {
        "key": "val",
        "key2": "val2"
      }
    },
    {
      "type": "workSomethingElse",
      "version": "1.0",
      "data": {
        "somekey": "someval",
        "somekey2": "someval2"
      }
    }
  ]
}

Given this JSON, a workflow definition like the following will be matched:

class WorkSomethingConfig {
  String key
  String key2
}

class WorkSomethingElseConfig {
  String somekey
  String somekey2
}

ratpack {
  workflow {
    work("workSomething", "1.0") {
      def workConfig = config.mapData(WorkSomethingConfig)
      assert workConfig.key == "val"
      assert workConfig.key2 == "val2"
      complete()
    }
    work("workSomethingElse", "1.0") {
      def workConfig = config.mapData(WorkSomethingElseConfig)
      assert workConfig.somekey == "someval"
      assert workConfig.somekey2 == "someval2"
      complete()
    }
  }
  handlers {
    // ...
  }
}

Backing Persistence

By default, work and flow status will use the InMemoryWorkStatusRepository and InMemoryFlowStatusRepository repositories respectively. By including the ratpack-workflow-redis module in your project, you can elect to bind the RedisWorkStatusRepository and RedisFlowStatusRepository types via a registry definition.

Redis support, by default, will connect to localhost:6379. Binding a custom JedisPool will allow you to configure connection properties.

import com.danveloper.ratpack.workflow.redis.*
import static com.danveloper.ratpack.workflow.server.GroovyRatpackWorkflow.ratpack

ratpack {
  bindings {
    bind(FlowStatusRepository, RedisFlowStatusRepository)
    bind(WorkStatusRepository, RedisWorkStatusRepository)
  }
  workflow {
    // ...
  }
  handlers {
    // ...
  }
}

Work vs. Flows

There are two types of operations that can be submitted to the workflow processor: work and flow. A work type represents an atomic operation, whereas a flow type encapsulates many work types that can be chained together.

Contributors

Dan Woods