julicq/PDF-RAG-Query

RAG model for PDF database

PDF Document Search with LangChain

This project implements a Streamlit app that allows users to upload PDF files, add them to a Chroma database, and query the database to retrieve information from the uploaded PDFs using LangChain.

Features

• Upload PDF files through the Streamlit interface.
• Add uploaded PDFs to a Chroma vector store database.
• Reset the database to clear all stored documents.
• Query the database to retrieve information from the PDFs using LangChain.

Requirements

• Python 3.7 or later
• The following Python packages:
  • streamlit
  • langchain
  • langchain_community
  • pypdf2
  • hashlib

Installation

1. Clone the repository:

   git clone https://github.com/your-username/pdf-document-search-langchain.git
   cd pdf-document-search-langchain
   

2. Create a virtual environment and activate it:

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`
   

3. Install the required packages:

   pip install -r requirements.txt
   

Usage

1. Run the Streamlit app:

   streamlit run streamlit_app.py
   

2. Upload PDFs:

   • Use the "Upload PDFs" section to upload multiple PDF files. The files will be saved in the data directory with their original names.

3. Add PDFs to the Database:

   • Click the "Add PDFs to Database" button to process and add the uploaded PDFs to the Chroma vector store database.

4. Reset Database:

   • Click the "Reset Database" button to clear all stored documents in the Chroma database (only works when database is not in use currently)

5. Query the Database:

   • Enter a query in the "Enter your query" text input and click the "Search" button to retrieve information from the PDFs in the database.

Project Structure

• app.py: The main Streamlit app script.
• main_script.py: Contains functions to load documents, split them into chunks, and add them to the Chroma database.
• query_script.py: Contains functions to query the Chroma database using LangChain.
• get_embedding_function.py: Contains the function to get the embedding function for Chroma.
• data/: Directory where uploaded PDF files are stored.
• chroma/: Directory where the Chroma database is stored.

Acknoledgements

• Langchain
• Streamlit