🐠 Babelfish.db

Your universal translator for human-to-database communication - making data speak your language!

---

Overview

Babelfish.db is an intelligent database assistant designed to facilitate natural language interactions with both SQL and NoSQL databases. It enables users to query their databases using plain English and generates responses in a conversational manner. Additionally, it can generate detailed PDF reports of the interactions.

Try it out for pubically hosted databases: https://babelfish-db.streamlit.app/

Features

• Natural Language Querying: Allows users to ask questions in natural language.
• SQL Database Support: Connects to MySQL databases and automatically generates SQL queries based on user questions.
• NoSQL Database Support: Connects to MongoDB databases and constructs MongoDB aggregation pipelines based on user queries.
• PDF Report Generation: Creates detailed reports of conversations in PDF format.

Technology Stack

• Streamlit: For building the web interface.
• LangChain: For processing natural language queries.
• Groq: For leveraging high-speed & accurate text generation capabilities with LLMs (large language models) to assist with query generation and responses.
• MySQL: For SQL database connectivity.
• MongoDB: For NoSQL database connectivity.
• FPDF: For generating PDF reports.

Installation

1. Clone the repository:

   git clone https://github.com/whonehuljain/Babelfish.db
   cd Babelfish.db

2. Create a virtual environment and activate it (Optional):

   python -m venv env
   source env/Scripts/activate

3. Install the dependencies:

   pip install -r requirements.txt

Usage

1. Launch the Application:

   streamlit run main.py

   Remember to replace your groq api key in main.py before launching the app.

2. Connect to a Database:

   • Use the sidebar to select the database type (MySQL or MongoDB).
   • Enter the necessary connection details and click "Connect".

3. Interact with Your Database:

   • Type your questions in the chat input box.
   • The assistant will generate and run the appropriate queries, then respond in natural language.

4. Generate PDF Reports:

   • After interacting with the assistant, click the "Generate Report" button to create a PDF report of the conversation.
   • Download the PDF using the provided download button.

Technical Details

1. SQL Integration

• Schema Retrieval: Connect to the SQL database and retrieve the table schema.
• Query Generation: Pass the schema and user query to the LLM to generate the SQL query.
• Execution and Response: Execute the SQL query and pass the results along with the user query to the LLM for a natural language response.

2. MongoDB Integration

• Document Retrieval: Query the MongoDB collection to retrieve two representative documents.
• Schema Inference: Pass these documents to the LLM to infer the database schema.
• Aggregation Pipeline: Use the inferred schema to generate an aggregation pipeline query.
• Execution and Response: Execute the aggregation pipeline and pass the results and user query to the LLM for a natural language response.

3. PDF Report Generation

• Capture the Conversation: Capture the conversation history and format it for PDF output when the user presses the “Generate Report” button.
• Download Option: Provide a “Download” option for users to save the conversation history as a PDF file.

Examples

MySQL Query Example

• User: How many employees are there in the company?
• Assistant: There are 150 employees in the company.

MongoDB Query Example

• User: List all the products in the 'electronics' category.
• Assistant: Here are the products in the 'electronics' category: ...

Troubleshooting

• Connection Issues: Ensure that the database credentials are correct and that the database server is accessible.
• Query Errors: Check the schema and the examples generated to ensure that the queries are being formed correctly.

License

This project is licensed under the MIT License.