download_assets.py

from huggingface_hub import hf_hub_download
import os

def download_assets():
    """Download necessary assets from Hugging Face Hub."""
    # Prepare your local folders
    os.makedirs("data", exist_ok=True)
    os.makedirs("vectorstore/db_faiss", exist_ok=True)

    repo_id   = "MoizK/mindmedic-assets"
    repo_type = "dataset"
    token     = os.getenv("HUGGINGFACE_HUB_TOKEN") or os.getenv("HUGGINGFACE_API_TOKEN")

    # -- PDFs (these live at the repo root) --
    pdf_files = [
        "71763-gale-encyclopedia-of-medicine.-vol.-1.-2nd-ed.pdf",
        "Depression-NIM-2024.pdf",
        "Depression-and-Other-Common-Mental-Disorders-Global-Health-Estimates.pdf",
        "Doing-What-Matters-in-Times-of-Stress.pdf",
        "Generalized-Anxiety-Disorder-When-Worry-Gets-Out-of-Control.pdf",
        "WHO-mhGAP-Intervention-Guide-v2.pdf",
        "social-anxiety-disorder-more-than-just-shyness.pdf",
    ]

    for fname in pdf_files:
        try:
            path = hf_hub_download(
                repo_id=repo_id,
                repo_type=repo_type,
                filename=fname,     # root-level file
                local_dir="data",   # save under ./data/
                use_auth_token=token,
            )
            print(f"✅ Downloaded PDF {fname} → {path}")
        except Exception as e:
            print(f"⚠️ Failed to download PDF {fname}: {e}")

    # -- FAISS index files (nested under vectorstore/db_faiss in the repo) --
    index_files = ["index.faiss", "index.pkl"]
    for fname in index_files:
        try:
            path = hf_hub_download(
                repo_id=repo_id,
                repo_type=repo_type,
                subfolder="vectorstore/db_faiss",  # look in this folder in the repo
                filename=fname,                   # just the filename here
                local_dir="vectorstore/db_faiss", # save under ./vectorstore/db_faiss/
                use_auth_token=token,
            )
            print(f"✅ Downloaded index file {fname} → {path}")
        except Exception as e:
            print(f"⚠️ Failed to download index file {fname}: {e}")

if __name__ == "__main__":
    download_assets()