Serverless Bartender Bot

Serverless Bartender Bot is a chat-bot powered by Amazon Lex and built on top of Serverless Architecture.

Build Project

You can build the project and create a fat jar.

./gradlew clean build

Installation & Deployment

Deploy Infrastructure

Serverless Framework is used to describe the entire infrastructure as code. Under the hood, it is converting our serverless.yml file to the CloudFormation template. Checkout serverless.yml for infrastructure definitions.

Once you have setup your local environment to use serverless and AWS, you can deploy the entire infrastructure.

sls deploy

Deploy Lex Bot

Use following instructions to deploy the bot.

1 . Give permission to Lex to invoke the lambda function.

aws lambda add-permission --region eu-west-1 --function-name serverless-bartenderbot-dev-lexevent --statement-id serverless-bartender-bot --action "lambda:InvokeFunction" --principal "lex.amazonaws.com"

2 . Update previously exported bot json file with your own aws account id.

sed "s/<aws-account-id>/$(aws sts get-caller-identity | jq -r .Account)/g" BartenderBot_Export.json.template > BartenderBot_Export.json

3 . Zip bot json file.

zip bot.zip BartenderBot_Export.json

4 . Start the bot import job.

aws lex-models start-import --region eu-west-1 --payload fileb://bot.zip --resource-type BOT --merge-strategy OVERWRITE_LATEST

5 . Import process may take a while. You can check the status of the job.

aws lex-models get-import --import-id import-id-from-previous-command --region eu-west-1 

6 . Once the import operation succeed, you can build the bot.

I admit that this part is a bit hacky, but there is no other way to put bot into ready state.

aws lex-models get-bot --name Bartender_Bot --version-or-alias "\$LATEST" --region eu-west-1 | jq 'with_entries(select([.key] | inside(["name", "description", "intents", "clarificationPrompt", "abortStatement", "idleSessionTTLInSeconds", "voiceId", "checksum", "processBehavior", "locale", "childDirected", "createVersion"])))' | jq -r > BartenderBot.json
aws lex-models put-bot --region eu-west-1 --name Bartender_Bot --cli-input-json file://BartenderBot.json

7 . Check whether the status of the bot is READY.

aws lex-models create-bot-version --region eu-west-1 --name Bartender_Bot

8 . Now, you can test the bot.

aws lex-runtime post-text --region eu-west-1 --bot-alias "\$LATEST" --bot-name Bartender_Bot --user-id "$(uuidgen)" --input-text "Hi"

You will get the following response.

{
    "sessionAttributes": {},
    "message": "Hi. How can i help you ?",
    "messageFormat": "PlainText",
    "dialogState": "ElicitIntent"
}

There is also an user-interface exists on Amazon Lex Console where you can test your bot.

Seed DynamoDB

Tho bot is designed to fetch cocktails from DynamoDB, so we need to seed DynamoDB with cocktails. Previously, I've developed a scrapper Cocktail Fetcher to help you to have a Cocktail DB.

First, run the following command to get service information.

sls info

You will most likely get the following response.

Service Information
service: serverless-bartenderbot
stage: dev
region: eu-west-1
stack: serverless-bartenderbot-dev
resources: 28
api keys:
  None
endpoints:
  POST - https://xxxx.execute-api.eu-west-1.amazonaws.com/dev/cocktail/feed
  GET - https://xxxx.execute-api.eu-west-1.amazonaws.com/dev/cocktail/all
  POST - https://xxxx.execute-api.eu-west-1.amazonaws.com/dev/cocktail/filter
functions:
  lexevent: serverless-bartenderbot-dev-lexevent
  createcocktail: serverless-bartenderbot-dev-createcocktail
  getcocktails: serverless-bartenderbot-dev-getcocktails
  filtercocktail: serverless-bartenderbot-dev-filtercocktail
layers:
  None

Now, you need to seed all the cocktails within repository.json to the feed endpoint. Run the following command to start seeding. This process may take a while.

./cocktail-seeder.sh https://xxxx.execute-api.eu-west-1.amazonaws.com/dev/cocktail/feed

Let's try to filter the cocktails to see if the seed is succeed.

curl -X POST https://xxxx.execute-api.eu-west-1.amazonaws.com/dev/cocktail/filter -d '{"alcoholType" : "NONALCOHOLIC", "cocktailCategory" : "Ordinary Drink"}'

Response

{
  "items": [
    {
      "cocktailId": "fa522ede-9e82-3bba-a7b3-656fa5209cdf",
      "cocktailName": "Pysch Vitamin Light",
      "alcoholType": "NONALCOHOLIC",
      "category": "Ordinary Drink",
      "glass": "Collins Glass",
      "imageUrl": "https://www.thecocktaildb.com/images/media/drink/xsqsxw1441553580.jpg",
      "instructions": "Shake with ice."
    }
  ],
  "status": "OK"
}

Tests

You can run all tests.

./gradlew test