Rasa Haystack DB Demo

This repo is a bare-bones chat bot project using Rasa in combination with Haystack along with SqliteDB. While Rasa is used for the whole flow of the dialogue and intent management, Haystack is used to answer the long tail of "knowledge queries" that can be answered by searching an answer in a document corpus. This example repo sketches how to use the fallback intent or an dedicated knowledge base intent to offload such queries to haystack.

Installing and checking dependencies.

To check the docker version

docker -v && docker-compose -v

If this returns an error do the following steps else jump to Getting Stared

Docker :

curl -fsSL https://get.docker.com -o get-docker.sh
sudo sh get-docker.sh

Docker Compose :

sudo curl -L "https://github.com/docker/compose/releases/download/1.29.2/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose 
sudo chmod +x /usr/local/bin/docker-compose

Get Started

Start the Haystack REST API and a demo DocumentStore along with RASA connected to custome tracker store using SQLiteDB via Docker:

git clone https://github.com/athenasaurav/Rasa_Haystack_DB.git
cd Rasa_Haystack_DB
docker-compose build
docker-compose up -d

This will automatically train a model in this repository with rasa train

Interacting with RASA

To Interact with rasa we can use either of the following endpoints.

Using Default REST

Send POST request to http://ip:5005/webhooks/rest/webhook
Request format should be like :

{
    "message" : "hi",
    "sender": "sender"
}

You will receive response like this:

{
    "message" : "Hi! How can i help you?",
    "sender": "sender"
}

Using custom REST channel named senderid

Send POST request to http://ip:5005/webhooks/senderid/webhook
Request format should be like :

{
    "message" : "hi",
    "sender": "sender"
}

You will receive response like this:

{
    "message" : "Hi! How can i help you?",
    "sender": "sender"
}

Customizing Custom REST Url

To customize the rasa custom url we need to make following changes in Three files.

Step 1) Customize the rest url name in MyIO.py in actions/connector

nano actions/connector/MyIO.py

Step 2) Make these following changes in the MyIO.py file. For Ex: If you want the custom REST url should be /webhooks/example/webhook instead of /webhooks/senderid/webhook

Change Line no. 12, Class name from class senderidInput(InputChannel): to class exampleInput(InputChannel):
Change Line no. 15, Return name from return "senderid" to return "example"
Change Line no. 104, collector form collector = senderidOutput() to collector = exampleOutput()
Change Line no. 130, Class name from class senderidOutput(CollectingOutputChannel): to class exampleOutput(CollectingOutputChannel):
Change Line no. 133, Return name from return "senderid" to return "example"

Save this changes.

NOTE : Please Check your doing it all from Root Directory.

Step 4) Copy the Newly edited MyIO.py from actions/connector to backend/connector

rm backend/connector/MyIO.py

cp actions/connector/MyIO.py backend/connector/MyIO.py

Step 5) Update the name in the backend/credentials.yml

nano backend/connector/credentials.yml

Change the Following : Line no 9 : connector.MyIO.senderidInput: to connector.MyIO.exampleInput:

Step 6) Save and Exit. Your Custom REST API Endpoint is ready to be deployed.

Step 7) Rebuild and Restart the docker

docker-compose down
docker-compose build
docker-compose up -d

Step 8) Now you can send POST request to http://ip:5005/webhooks/example/webhook

Customizing Custom Tracker Store

To Customize Tracker Store we can set it in backend/endpoints.yml

tracker_store:
    type: SQL
    dialect: "sqlite"  # the dialect used to interact with the db
    url: "sqlite:///./rasa.db"  # (optional) host of the sql db, e.g. "localhost"
    db: "rasa.db"  # path to your db
    username:  # username used for authentication
    password:  # password used for authentication
    query: # optional dictionary to be added as a query string to the connection URL
      driver: my-driver

Set your Databse url, DB name and DB path accordingly. Incase if we dont change anything it will create a default rasa.db in docker which will only be accessiable inside docker.

RUN RASA WITH EXTERNAL HAYSTACK

To run RASA Only with SQLite DB Tracker Store Please visit RASA_without_Haystack

HayStack Official Documentation

To Read More about official implementation of HayStack and its tutorial. Please Visit HayStack Documentation & Tutorials