scripts/get_assets.pyFiles

#!/usr/bin/env python
"""
Lightweight helper to download the COCO FAISS index + caption array
into ./scripts/ the first time the Space boots.
"""
import pathlib, subprocess, sys

FILES = {
    "coco_caption_clip.index":
        "https://huggingface.co/datasets/stephenebert/coco-faiss-assets/resolve/main/coco_caption_clip.index",
    "coco_caption_texts.npy":
        "https://huggingface.co/datasets/stephenebert/coco-faiss-assets/resolve/main/coco_caption_texts.npy",
}

DEST = pathlib.Path(__file__).resolve().parent

def fetch(url, out):
    cmd = ["curl", "-L", "--progress-bar", "-o", str(out), url]
    print("⤵️ ", " ".join(cmd), flush=True)
    subprocess.check_call(cmd)

for fname, url in FILES.items():
    path = DEST / fname
    if path.exists():
        print(f"✔️ {fname} already present")
        continue
    print(f"⬇️ Downloading {fname} …")
    fetch(url, path)

print("✅ All assets ready")