PDF Query System

A PDF query system that allows users to upload PDF files and ask questions about their content using Ollama.

Quick Start

Step 1: Install Ollama

macOS (with Homebrew):

brew install ollama

Other platforms: Download from https://ollama.ai

Step 2: Pull the llama2 Model

ollama pull llama2

Step 3: Start Ollama Server

Open a terminal and run:

ollama serve

Keep this terminal open. Ollama will run on http://127.0.0.1:11434.

Step 4: Start FastAPI Backend

Open a new terminal and run:

cd /path/to/pdf_query
source venv/bin/activate
python3 -m uvicorn main:app --reload --host 127.0.0.1 --port 8000

(Replace /path/to/pdf_query with your actual project directory path)

Step 5: Open Browser

Navigate to: http://127.0.0.1:8000

Upload a PDF and ask questions!

Setup (First Time Only)

Backend Dependencies

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Frontend Dependencies

npm install
npm run build

Project Structure

• main.py - FastAPI application entry point
• llm_client.py - Ollama LLM client wrapper
• pdf_utils.py - PDF text extraction utilities
• src/ - React frontend source code
• dist/ - Built React app (generated after npm run build)
• requirements.txt - Python dependencies
• package.json - Node.js dependencies

API Endpoint

POST /query-pdf

Accepts:

• file: PDF file (multipart/form-data)
• question: User's question (form field)

Returns:

{
  "answer": "Answer from Ollama",
  "context_truncated": false,
  "context_length": 1234
}