Skip to content
An example use case for play-akka-telegrambot4s
Scala Java HTML Shell Other
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Telegram Bot with Play Framework, Akka FSM, Scala.js, Binding.scala

This is work in progress.

This project is based on:

  1. Telegram Bot Demo with Scala/ Play
  2. Websockets with Play Framework, Scala.js, Binding.scala

In this example I want to combine everything from the 2. project above and extend it with:

Business Case

We want to have an overview of incidents that are reported by caretakers.

We implemented 2 Conversations:

1. Create an incident

They will report an incident to a Telegram Bot with their mobile phones:

  1. select type of incident.
  2. select level of incident (urgency).
  3. add a textual description.
  4. add optional photos.
  5. send the incident.


2. Edit an incident

  1. send an ident.
  2. select an edit action.
  3. do the change.


Control Panel

A web-page shows all incidents - the newest on top. To see the attached images you open a detail view. You can filter and sort the incidents (see images above).

Let's start with the simple parts:

Shared model

The great thing about a full-stack Scala app is that we only have to define our domain model once for the server and the client.

Client-Server Communication

Next to the model all that is needed is the JSON-un-/-marshalling. If we use ADTs (Algebraic Data Types - sealed traits in Scala) in combination with case classes this requires only a few lines of code.

Here is an example how that looks: SharedMessages

Handling dates

Dates are handled differently on JVM and JS. scalajs-java-time provides an implementation of java.time. This allows to have to work with Instants on both sides (Be aware not everything is supported).

case class Audit(user: String
                 , dateTime: Instant =

The JSON marshalling is now not too hard:

implicit val localInstantReads: Reads[Instant] =
    (json: JsValue) => {
        .map { epochSecond =>

implicit val localInstantWrites: Writes[Instant] =
    (instant: Instant) => JsNumber(instant.getEpochSecond)


Has its own README


The server part can be split into the following sub-chapters:

User management

When you go to http://localhost:9000 a web-socket is opened to show you incoming incidents.

The web-sockets are managed with Akka Actors. The implementation was taken from the Lightbend's Websocket example and adjusted to fit my needs.

See HomeController

Chat management

This is now handled by my small framework. See the documentation there.

Incident Conversation

Let's have a look now on the interesting part. Always start with a description of your conversation;)

  * report an incident with an IncidentType, a description and optional images.
  *     [Idle]  <-------------
  *       v                  |
  *   [SelectIncidentType]   |
  *       v                  |
  *   [SelectIncidentLevel]  |
  *       v                  |
  *    [AddDescription]      |
  *       v                  |
  *   [AddAdditionalInfo] <--|
  *       v                  |
  *       --------------------

Let's go through all states.


  // if no Conversation is active - the Conversation is in the Idle state
  when(Idle) {
    case Event(Command(msg, _), _) =>
      // the message contains only the command '/incidents' - so msg is only needed for the response.
      bot.sendMessage(msg, "Please select incident type!"
        // create the buttons for all IncidentTypes
        , Some(incidentSelector))
      // tell where to go next - we don't have any state
    // always handle all possible requests
    case other => notExpectedData(other)


  // first step after selecting IncidentType.
  when(SelectIncidentType) {
    case Event(Command(msg, callbackData: Option[String]), _) =>
      // now we check the callback data
      callbackData match {
        case Some(data) =>
          // ask the user for a description, as it is a text input no markup is needed. 
          bot.sendMessage(msg, "Add a description:")
          // when we go to the next step we add the IncidentType to the FSM.
          goto(SelectIncidentType) using IncidentTypeData(IncidentType.from(data))
        case None =>
          // when the user does not press a button - remind the user what we need
          bot.sendMessage(msg, "First you have to select the incident type!"
            , Some(incidentSelector))
          // and stay where we are


Analog SelectIncidentType


  when(AddDescription) {
    // now we always work with the state of the previous step
    case Event(Command(msg, _), IncidentTypeData(incidentType)) =>
      // all from the text input is in msg.text
      msg.text match {
        // check if the description has at least 5 characters
        case Some(descr) if descr.length >= 5 =>
          // ask for photos and provide a button to finish the report
          bot.sendMessage(msg, "You can now add a Photo or finish the report!"
            , bot.createDefaultButtons(finishReportTag)
          // now the state contains the IncidentType and the description
          goto(AddAdditionalInfo) using IncidentData(incidentType, descr)
        case _ =>
          // in any other case try to bring the user back on track
          bot.sendMessage(msg, "The description needs to have at least 5 characters!")


  when(AddAdditionalInfo) {
    case Event(Command(msg, callbackData: Option[String]), incidentData: IncidentData) =>
      callbackData match {
        // first check if the user hit the 'finish' button
        case Some(data) if data == finishReportTag =>
          // give a hint that the process is finished
          bot.sendMessage(msg, "Thanks for the Report.\n" +
            "\nIf you have another incident, click here: /incident")
          // send the Incident to the IncidentActor that informs the web-clients
          incidentActor ! incidentData.toIncident
          // go to the start step
        case _ =>
          // the process is asynchronous so a special step is needed
          bot.getFilePath(msg).map {
            case Some((fileId, path)) =>
              // if the user added a photo - she can add more photos
              bot.sendMessage(msg, "Ok, just add another Photo or finish the Report.", bot.createDefaultButtons(finishReportTag))
              // async: the result is send to itself (ChatConversation) - the uploaded photo is added to the state.
              self ! ExecutionResult(AddAdditionalInfo, incidentData.copy(assets = Asset(fileId, path) :: incidentData.assets))
            case _ =>
              // in any other case try to bring the user back on track
              bot.sendMessage(msg, "You can only add a Photo or finish the Report.", bot.createDefaultButtons(finishReportTag))
              // async: the result is send to itself (ChatConversation) - no state change.
              self ! ExecutionResult(AddAdditionalInfo, incidentData)
          // async: go to the special step (ChatConversation) - which waits until it gets the ExecutionResult

Here the whole class: IncidentConversation

The 2. conversation you find here: EditIncidentConversation

Run the application

$ sbt
> run

open http://localhost:9000 in a browser.

You can’t perform that action at this time.