README.md

🗳️ Hindi Voter PDF Processor with LLM API (OCR + OpenRouter + Gradio)

This app extracts voter information from scanned PDFs using OCR and formats it into a structured CSV using an LLM API (via OpenRouter).

---

📦 Features

• Extracts text from Hindi/English PDFs using EasyOCR
• Splits content to avoid LLM token limits
• Sends chunked JSON to LLM for conversion to clean CSV
• Uses OpenRouter LLM API (e.g., Gemma-3b)
• Interactive UI with Gradio
• Supports download of extracted JSON and final CSV

---

🌐 Get Your OpenRouter API Key

• Go to https://openrouter.ai

• Click Login (use Google/GitHub/Email)

• Navigate to the Models page

• Click on a model like gemma-3b, mistral, etc.

• On the model page, click "Create API Key"

• Copy the API key

🧪 How to Use the Gradio App

🔹 Tab 1: PDF Processing

• Upload a Hindi/English scanned PDF

• Click "Process PDF"

• View extracted text in JSON format

• Download JSON file if needed

🔹 Tab 2: LLM API Processing

• Paste your OpenRouter API key

• (Optional) Customize the prompt or add instructions

• Click "Call LLM API"

• View structured voter data in CSV format

• Download the CSV file

• Enable Debug Mode to see raw API responses for troubleshooting.

📁 Output Files

Extracted JSON and CSV files are saved in the processed_json/ folder.