download_assets.py

from huggingface_hub import hf_hub_download
import os

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

    repo_id   = "MoizK/mindmedic-assets"
    repo_type = "dataset"
    # allow either env var name
    token     = os.getenv("HUGGINGFACE_HUB_TOKEN") or os.getenv("HUGGINGFACE_API_TOKEN")

    # list of PDF filenames 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,
                local_dir="data",
                token=token,
            )
            print(f"✅ Downloaded {fname} → {path}")
        except Exception as e:
            print(f"⚠️ Failed to download {fname}: {e}")

    # Now fetch the FAISS index files from the nested path
    index_files = ["index.faiss", "index.pkl"]
    for fname in index_files:
        remote_path = f"vectorstore/db_faiss/{fname}"
        try:
            path = hf_hub_download(
                repo_id=repo_id,
                repo_type=repo_type,
                filename=remote_path,  # this is where they actually live in the repo
                local_dir=".",         # so that ./vectorstore/db_faiss/... is created
                token=token,
            )
            print(f"✅ Downloaded {remote_path} → {path}")
        except Exception as e:
            print(f"⚠️ Failed to download {remote_path}: {e}")

if __name__ == "__main__":
    download_assets()