INSTALL.md

How to install

Requirements

• Node.Js >= 6.0
• MongoDB (URL connection string in environment variable MONGO, example: export MONGO="mongodb://username:password@host1:27017/ovhchatbot")
• Facebook developer account with Facebook developer app created (more explanations below)
• Slack developer account with Slack developer app created (more explanations below)
• Api.ai account
• OVH account
• Public ip with domain linked in HTTPS

For each environment variable, if you want to use docker and docker-compose you can put the value in the Dockerfile of this project.

Create a api.ai account

• Create an account on api.ai website and create a new agent
• Go in settings of your agent in tab Export and Import
• Click on restore from zip and import the zip that are in the archives directory of this project.
• Go in settings, copy/paste Client access token and save it into the env variable named APIAI_TOKEN

Deployment

• You have to host this project on a public instance with a public ip and link this IP to a domain name
• To get a free SSL on your domain linked to the public ip of your chatbot API you can use the "SSL Gateway" service by OVH or simply use Let's encrypt. You just have to indicate the domain and ip linked to this domain.
• Save your domain name into env variable APP_URL (for example: export APP_URL=https://myCustomDomain.com)

Run the API on your server or with docker-compose

Without docker

$ npm i
$ APP_PORT=80 npm start

With docker and docker-compose

$ sudo docker-compose up

Create Facebook Developer app for chatbot

• On your server export an env variable named FB_VALIDATION_TOKEN with the value ovh_chatbot_dev for example. This value is the same that you fill in step 2 of this Facebook tutorial and the webhook url must be https://yourCustomDomain/api/v1.0/webhook

• Follow the four first steps described here, but don't take care about the sample app that Facebook give.

• Save the correspondant informations in env variable (FB_APP_SECRET --> in section dashboard, FB_APP_ACCESS_TOKEN --> in section messenger)

• (Optional) you can add a welcome text:

$ curl -X POST -H "Content-Type: application/json" -d '{
    "setting_type":"greeting",
    "greeting":{
      "text":"Your welcome text here"
    }
  }' "https://graph.facebook.com/v2.6/me/thread_settings?access_token=$FB_APP_ACCESS_TOKEN"   

Create Slack Developer app for chatbot

• Go on this page and create a new app
• Enable Interactive Messages and add request URL https://yourCustomDomain/api/v1.0/slack/actions
• Enable Event Subscriptions and add request URL https://yourCustomDomain/api/v1.0/slack
• In the same section subscribe to event message.im and save changes
• Add a bot user in section Bot Users
• In section OAuth & Permissions add permissions named bot and chat:write:bot
• In the same section click on Install App to Team in order to get the Bot User Oauth Access Token which are saved in our env variable as SLACK_BOT_ACCESS_TOKEN
• Then add a redirect URLs like this one https://yourCustomDomain/api/v1.0/slack/authorize
• In section Basic Information you can fetch the Client ID saved in our env variable as SLACK_ID, the Client Secret saved in SLACK_SECRET and the Verification Token in SLACK_TOKEN
• Go in Manage Distribution section and access to the Sharable URL to install slack bot into your team.

Create your api.ovh.com application

• Go on this page and click on First steps with the OVH API to create an application and save the application key and secret in env variable OVH_KEY and OVH_SECRET

Translations

• For the api.ai agent translation: Have a look at tools
• For the bot translation: Have a look at translations

List of all the Enviroment Variables used:

APIAI_TOKEN
APP_PORT (optional, default is 8080)
APP_URL
FB_APP_ACCESS_TOKEN
FB_APP_SECRET
FB_VALIDATION_TOKEN
OVH_BASE_PATH
OVH_KEY
OVH_SECRET
MONGO (optional, default is `mongodb://mongo:27017`)
SLACK_BOT_ACCESS_TOKEN
SLACK_ID
SLACK_SECRET
SLACK_TOKEN

If you need more informations you can join us on Gitter