JarvisButlerBot

A modular telegram bot and group butler that aims to understand your text messages.

Setup

Download a release or build the source yourself, execute the JarvisButlerBot.exe and enter your Telegram Bot Token.

To install a module, copy the modules output dll file into the %APPDATA%\Crazypokemondev\JarvisButlerBot\modules folder and put all its dependency dll files into the %APPDATA%\Crazypokemondev\JarvisButlerBot\lib folder. You can skip any files that JarvisButlerBot itself depends on (those will already be loaded).

Developing a module

Create a .NET Framework Class Library project (currently, JARVIS runs on .NET Framework 4.7.2).

Install JarvisButlerBot.ModuleCore from NuGet.

In the library project, create a public class anywhere you like. Annotate it with the JarvisModuleAttribute and have it derive from JarvisModule like this:

using JarvisModuleCore.Classes;
using JarvisModuleCore.ML;
…
[JarvisModule]
public class ExampleModule : JarvisModule
{
  …
}

If your module depends on any other libraries (dll files that are copied into the execution directory), add them to the lib folder as mentioned above and specify their file names in the JarvisModuleAttribute like this:

[JarvisModule(new string[] { "ExampleDependency.dll" })]
public class ExampleModule : JarvisModule
…

Any task that JARVIS should be able to execute must be public, non-static and in a module class, annotated and typed like this (the async keyword can be left out, if you want):

[JarvisTask("example.taskid")]
public async void ExampleTask(Telegram.Bot.Types.Message message, Jarvis jarvis)
{
  …
}

This method will be called every time JARVIS either recognizes one of the commands you set in the attribute or the ML module predicts your Task ID for a message. The Message will not necessarily be a text message, but it will definitely contain text as either a caption, normal text or the question of a poll.

For your module class to compile, you will have to override several things:

• Id: a unique identifier string for your module, recommended to be used as a prefix for your task ids
• Name: a name for your module
• Version: a version number for your module. If you don't know what to use, go with Version.Parse("1.0.0")
• MLTrainingData: an array of TaskPredicitionInputs to train the model to recognize your tasks. I recommend about 50 elements per task. An example data set can be found here. For the prediction, the bots individual username will be replaced by @Username, usernames of other users by @User and inline mentions by @Mention.

Generating training data

You can ask Jarvis to start saving your messages, he will then start doing so while still responding as normal. Once you ask him to stop, or if you're not a global admin, once you've hit 50 messages, he'll send you all the data of the messages he received from you, as c# code, with the taskid %taskid%, so you can just go and replace that with whatever task you want to use the data for! Formerly, this generated JSON, however this seemed more convenient to me.

Official modules

Apart from the default modules that are included in the main repository and releases, there currently is one other module that is being developed by the creator of the JARVIS bot:

• JarvisGroupButlerModule

Publishing your modules

If you wrote your own module and want it to be added to @JarvisButlerBot on Telegram, message me at @Olfi01, I will happily add it if I think it fits!