#!/usr/bin/env python3
import json

# 1) Load the LayoutSAM image list
with open('./images_to_download.json') as f:
    images = json.load(f)

# 2) Figure out which shards we need
shards = sorted({img.split('/')[0] + '.tar' for img in images})

# 3) Write out clean HF URLs—no headers, no query strings
base = "https://huggingface.co/datasets/Aber-r/SA-1B_backup/resolve/main/"
with open('layoutsam_shard_links.txt', 'w') as out:
    for shard in shards:
        out.write(f"{shard}\t{base}{shard}\n")

print(f"Wrote {len(shards)} entries to layoutsam_shard_links.txt")