ingest.py

import os
import shutil
from src.processor import build_index

DATA_DIR = "./data"
DB_DIR = "./chroma_db"

def main():
    print(f"Checking for data in {DATA_DIR}...")
    if not os.path.exists(DATA_DIR):
        print(f"Create a '{DATA_DIR}' directory and put your PDFs there.")
        return

    if not any(f.endswith(".pdf") for f in os.listdir(DATA_DIR)):
        print("No PDF files found in data directory.")
        return

    print("Building Vector Database... (This may take a while for large docs)")
    
    # Optional: Clear old DB if you want a fresh start every time
    # if os.path.exists(DB_DIR):
    #     shutil.rmtree(DB_DIR)

    vectorstore = build_index(DATA_DIR, DB_DIR)
    print(f"Success! Database built at {DB_DIR}")

if __name__ == "__main__":
    main()