Chatbot Lex Handler Lambda

NOTE: Rosa (www.rosa.bot) is the new name for ForgotPW

Respond to Amazon Lex chat requests for all intents

Intents

This lambda function will respond to intents determined by AWS Lex which will receive the text messages, determine the intent, and invoke this lambda function with the determined event.

Example Intent: Store Password

You> Store password for Amazon
FPW> Click here to enter your password for Amazon:
https://app.forgotpw.com/#/store?arid=jEH93Hlx9t1xWp

You> Store password for Amazon
FPW> I found an existing entry for Amazon Web Services, do you want to 1) overwrite it, or 2) create a new entry for "Amazon" ?
You> 2
FPW> Click here to enter your password for Amazon:
https://app.forgotpw.com/#/store?arid=jEH93Hlx9t1xWp
FPW> Your password for Amazon has been stored.

You> Store password for Amazon
FPW> I found an existing entry for Amazon, do you want to overwrite it?
You> Yes
FPW> Click here to enter your password for Amazon:
https://app.forgotpw.com/#/store?arid=jEH93Hlx9t1xWp
FPW> Your password for Amazon has been stored.

Authorized Requests

This service will generate an ARID (authorized request ID), which will encapsulate a request to store or retrieve a password for the specified application as well as the users phone number:

s3://forgotpw-authorized-requests-dev/jEH93Hlx9t1xWp

{
    "expireEpoch": "",
    "userToken": "",
    "rawApplication": "",
    "normalizedApplication": ""
}

The userToken can be queried usin the phone-token-service given the phone number, which is available in the userId field passed to the event data to the Lambda function from Twilio. https://docs.aws.amazon.com/lex/latest/dg/lambda-input-response-format.html

This service will then create and return the chat respons which will contain the link to the authorized request, e.g. https://app.forgotpw.com/#/store?arid=jEH93Hlx9t1xWp

When the web application receives this link it will query S3 for the specified authorized request, validate the expire time, and process the request with the provided user input for the secret.

The S3 bucket which contains these requests will have a TTL to expire items, however since we will want to expire these much faster than we can rely on S3 for expiring them, we'll also enforce our own expiry check.

Setup - Dev Environment

Install the Serverless CLI.

# install the serverless framework
npm install serverless -g

Deploy

The deploy environment will install production dependencies only to keep the package size within Lambda's 250MB limit. Be sure to re-build the docker container each time. Requires pip install iam-docker-run.

export AWS_ENV="dev" # | prod
export PROFILE="fpw$AWS_ENV"
# must re-build docker container each deploy!
docker build -f Dockerfile.deploy -t forgotpw-chatbot-lex-lambda:deploy .
iam-docker-run \
    --interactive \
    --profile $PROFILE \
    -e AWS_ENV \
    --image forgotpw-chatbot-lex-lambda:deploy

Troubleshooting

You may need to run the following command to grant Lex permissions to invole the Lambda.

export AWS_ENV="prod" && export PROFILE="fpw$AWS_ENV"
iam-starter --profile $PROFILE --command aws lambda add-permission --function-name fpw-chatbot-lex-handler --statement-id chatbot-fulfillment --action "lambda:InvokeFunction" --principal "lex.amazonaws.com"

License

GNU General Public License v3.0

See LICENSE to see the full text.