README.md

Query Processing Service

This service is responsible for the actual conversion of NLP to SQL and generating output data

It utilized two LLMs namely : Gemini and Worqhat

How to start

Step 1: Configure the environment variables

Create a .env file and input the data accordingly

WORQHAT_API_KEY=<worqhat-api-key>
GEMINI_API_KEY=<gemini-api-key>
GEMINI_MODEL_NAME=gemini-1.5-flash

Step 2: Configure firebase credentials

Goto /app/services/firebase_services/data/firebase-config.json file and input the given information

{
    "type": "",
    "project_id": "",
    "private_key_id": "",
    "private_key": "",
    "client_email": "",
    "client_id": "",
    "auth_uri": "",
    "token_uri": "",
    "auth_provider_x509_cert_url": "",
    "client_x509_cert_url": "",
    "universe_domain": ""
}

Step 3: Create a virtual environment and activate it

pip install virtualenv

virtualenv my_env

my_env/Scripts/activate

Step 4: Install the required modules

pip install -r requirements.txt

Step 5: Run the fast api server

fastapi dev .\app\main.py

API Routes

1. /api/metadata/generate
   Method: GET

This route is responsible for the Generating the meta-data from the given database

2. /api/query
   Method: POST

This route is used for actual conversion and execution of the query

Expected payload

{
  "query": "string",
  "model": "string",
  "dbms": "string",
  "credentials": {
    "host": "string",
    "user": "string",
    "password": "string",
    "database": "string",
    "port": "string"
  }
}

3. /api/data-visualization Method: POST

This method is responsible for the creation of the graphs, charts and insights

Expected payload is the results of the /api/query API Endpoint

{
  "results": [
    "string"
  ]
}