Note this tutorial assumes you are using Swift 4 on a unix-like machine.

You use PetriKit with the Swift Package Manager. Create an empty directory for your application, open a terminal inside and create a new Swift command line tool:

mkdir /some/path/to/your/app
cd /some/path/to/your/app
swift package init --type executable

You'll have to add PetriKit as a dependency to your application. Open the generated file Package.swift and edit its content as follows:

// swift-tools-version:4.2

import PackageDescription

let package = Package(
  name: "MyApp",
  dependencies: [
    .package(url: "https://github.com/kyouko-taiga/PetriKit.git", from: "2.0.0")
  ],
  targets: [
    .target(
      name: "MyApp",
      dependencies: ["PetriKit"]),
  ])

This will let you import PetriKit in your sources. Add the following statement at the beginning of Sources/MyApp/main.swift:

import PetriKit

And you can now use PetriKit.

Petri nets in PetriKit are described by the means of two protocols: Transition and PetriNet. Those protocols only describe the minimum requirements for a type to represent a transition (resp. a petri net). They can be seen as an abstract description of the structure of a Petri net. Hence, they can't be used as is, but rather should be implemented for a particular type of Petri net.

PetriKit comes with such an implementation for Place/Transition nets (or P/T-nets). Those are one of the most simple variant of Petri nets. You can create a P/T-net as follows:

enum Place {
  case p0, p1
}

let t0 = PTTransition<Place>(
  named         : "t0",
  preconditions : [PTArc(place: .p0)],
  postconditions: [PTArc(place: .p1)])
let t1 = PTTransition<Place>(
  named         : "t1",
  preconditions : [PTArc(place: .p1)],
  postconditions: [PTArc(place: .p0)])

let pn = PTNet(transitions: [t0, t1])

The above code create a P/T-net pn composed of two places and two transitions.

You can simulate the firing of a transition as follows:

let m = t0.fire(from: [.p0: 1, .p1: 1])

The above code assigns to m the marking obtained after firing the transition t0 from the marking [.p0: 1, .p1: 1] (i.e. one token in each place). Note that the method fire(from:) of the TransitionProtocol protocols returns an optional. That's because if the transition is not fireable from the given marking, the method should return nil.

You can also simulate the execution of a P/T-net from a given initial marking:

let m = pn.simulate(steps: 4, from: [.p0: 1, .p1: 0])

The above code simulate 4 steps of execution, randomly choosing one fireable transition at each step, or stops prematurely if the simulation reaches a deadlock.

The type PTNet comes with a special method saveAsDot that outputs the given P/T-net as a Graphviz file. Note that this will require you to import the Foundation library as well.

import Foundation

// ...

try pn.saveAsDot(to: URL(fileURLWithPath: "pn.dot"), withMarking: [.p0: 1, .p1: 2])

The above code will output the P/T-net pn with marking [.p0: 1, .p1: 2] to a file named pn.dot.