Home

Awesome

SuperScript Bootstrap

Welcome to the wonderful world of building chatbots!

Getting the enviroment setup

We will assume you already have Node and NPM installed. To use SuperScript you need to make sure you have Node 6.x installed.

SuperScript also requires MongoDB - You can download and install Mongo from https://www.mongodb.org/downloads#production

Once installed you run it from the command line by sudo mongod

You can start by cloneing this repo locally using the command git clone https://github.com/silentrob/hello-superscript.git

You will then want to run npm install This will load in all the dependencies.

This example will show you how to create a basic trigger, topic and plugin, and use 2 sample clients. (Telnet and Slack)

Basic Workflow

Pre step

Compile conversation data

We can compile the data two ways, either by typing parse from the command line which generates a intermediate json representation that we can manually import, or by doing a parse and load combind step. Lets do the latter as it removes the need to manually import the JSON file ourselves.

Run ./node_modules/.bin/cleanup --mongo telnetbot

We pass in a --mongo telnetbot param to say we want to store (and flush) from the telnetbot database in mongo, this is also a handy way to seperate other bots.

The output should look something like:

users is clear
topics is clear
replies is clear
gambits is clear
Time to Process 1.009 seconds
Number of topics 2 parsed.
Number of gambits 1 parsed.
Number of replies 1 parsed.
Number of convos 0 parsed.
Number of Gambits: 0
Number of Replies: 0
Everything imported

It should be noted that this task will also cull the existing data before adding new data.

As we can see we have added one Gambit and One Reply as expected. We also have two topics, one called main, and one the system creates called random.

Lets streamline this step

You can lean on NPM to run commands directly, and I have added that script to the package.json file, so from now on, we can just type: npm run reload and it call the cleanup script for us.

Starting the client

The next step is to see the bot in action, for that we will fire up our first client and chat with our new bot. SuperScript will work with on any messaging platform. Slack, Telegram, Twilio, IRC, WebSockets or anything that exposes text output and a way to reply.

Some of the more advanced client logic is deliberately left upto you to build, this is because there are many differnet chat paradigms one on one, multi-party chat, direct messaging. Usually what you want to do is specific to your own app. For now we will cover a basic telnet example. You can find more clients here: https://github.com/silentrob/superscript/tree/master/clients

Lets fire up the client by running node telnet.js

And in a new window lets chat with it by running telnet 127.0.0.1 2000

Adding a plugin.

More information about Plugins can be found here - https://github.com/silentrob/superscript/wiki/Plugins-and-Functions

Plugins are designed to break out of the scripting interface and allow you to do whatever you want using JavaScript and NPM.

Some general ideas are :

Lets add the most basic example like the one found at the link above.

exports.getWeather = function(city, cb) {
  cb(null, "It is probably sunny in " + city);
}

I have already copied that plugin and saved/exported it from the ./plugins/example.js file.

This plugin has two arugments comming in city and cb. The system ALWAYS adds the cb for you and ALWAYS expects it to be called when you are finished. This allows the system to keep going and not hang waiting forever.

If we add a trigger like:

+ what is the weather in *1
- ^getWeather(<cap1>)

This will pass in *1 into the function as <cap1> becoming the city param inside the function.

NB

All functions exported inside the plugin folder automatically become available as plugins to the system.

NB2

The plugin callback above shows 2 return params, error and reply If there is a error, it will not display anything and exit the matching loop. If the reply is a string, it becomes the reply we send to the user. If the reply is a Object with a text or reply property, that will become the reply string and all other properties will become OOB (Out of Bound) data. There is also a third param which is a boolean (true/false) value that tells the system to stop matching or keep matching for more.