DialogFlow Chatbot with NewBot Framework

The goal: to create a chatbot quickly in Javascript with the understanding of the natural language (DialogFlow).

Prerequisites

• Have a Google Cloud and DialogFlow account
• Have NodeJS and NPM installed on your machine
• Knowledge of Javascript

Why NewBot Framework?

• Universal, a single code to work everywhere
• Creation of conversational website
• Modular and structured
• Using the NLP (Natural Language Processing)
• Conversational script syntax to write user-chatbot conversations easily and quickly
• Internationalization
• Unit tests
• Emulator with NewBot CLI

Compatibility

All platforms, i.e.

• Facebook Messenger
• Google Assistant
• Amazon Alexa
• Slack
• Telegram
• Viber
• Skype
• etc.

Above all

Install NewBot CLI to take advantage of command lines

npm install -g newbot-cli

Create a project

newbot new <your directory name>

Go to the generated folder

You can test the chatbot with the command newbot serve. See below

Retrieve the credentials of DialogFlow

Have a Google Cloud and DialogFlow account

Click on the service in Google Cloud

Create new Key

Download the private key and add it to the root of the project

Add DialogFlow in our project

Install the newbot-dialogflow module :

npm install newbot-dialogflow

In the main.js file, add the DialogFlow skill :

import dialogflowSkill from 'newbot-dialogflow'
import code from './main.converse'

export default {
    code,
    skills: {
        dialogflow: dialogflowSkill({
            projectId: 'newbot-fttkoh',
            credentials: 'newbot-fttkoh-69d17227a8b7.json'
        })
    }
}

• projectId : The project ID (in Chatbot Settings)
• credentials: Path to credentials file

Input unknown

Add this code to the main.converse file :

@Intent('input.unknown')
unknown() {
    > { :intent.response }
}

• An input.unknow intention is already present in DialogFlow, it is triggered when no other intention is triggered
• We then display the response received from DialogFlow

:intent.date.value is a magic variable that retrieves the date entered by the user

The language used is specific to NewBot. It is called ConverseScript and allows you to create conversational scripts with a syntax close to Javascript
More : https://newbot.io/docs/syntax/variables.html

Add our own intention: we buy a pizza!

1. Create a new intent in DialogFlow
2. Put a name ("Buy Pizza" by example)
3. Add trainings phrases
4. Add action name. Here, this name is input.buy
5. You can add parameters. Here, we recover the date in the sentence
6. Add response(s)

main.converse content contains a new intention!

@Intent('input.buy')
buy() {
    date = :intent.date.value
    > { :intent.response }
    callApi(date)
}


@Intent('input.unknown')
unknown() {
    > { :intent.response }
}

We retrieve the date entity and call a function that can trigger an API

callApi() is a function created by us, present in Javascript :

main.js

import dialogflowSkill from 'newbot-dialogflow'
import code from './main.converse'

export default {
    code,
    skills: {
        dialogflow: dialogflowSkill({
            projectId: 'newbot-fttkoh',
            credentials: 'newbot-fttkoh-69d17227a8b7.json'
        })
    },
    functions: {
        callApi(date) {
            console.log(`Call your own API. date is ${date}`)
        }
    }
}

Test in Emulator

Type the following command at the root of your project

newbot serve

Go to localhost:3000 and test the project in the emulator

Documentation

Here : https://newbot.io/en/docs