#!/bin/bash
set -euo pipefail

DATA_ROOT="${DATA_ROOT:-/home/v-menggao/code/data}"

# HF datasets
huggingface-cli download --repo-type dataset nlphuji/flickr30k --local-dir "${DATA_ROOT}/flickr30k"
# huggingface-cli download --repo-type dataset Lin-Chen/ShareGPT4V --local-dir "${DATA_ROOT}/ShareGPT4V"
# huggingface-cli download --repo-type dataset BeichenZhang/Urban1k --local-dir "${DATA_ROOT}/Urban1k"
# huggingface-cli download --repo-type dataset royokong/coco_test --local-dir "${DATA_ROOT}/coco_test"

# sugar-crepe repo
if [ ! -d "${DATA_ROOT}/sugar-crepe" ]; then
  git clone https://github.com/RAIVNLab/sugar-crepe.git "${DATA_ROOT}/sugar-crepe"
else
  echo "sugar-crepe already exists, skipping clone"
fi

# # COCO val2017 images
# mkdir -p "${DATA_ROOT}/MSCOCO-2017"
# wget -c -O "${DATA_ROOT}/MSCOCO-2017/val2017.zip" http://images.cocodataset.org/zips/val2017.zip
# unzip -n "${DATA_ROOT}/MSCOCO-2017/val2017.zip" -d "${DATA_ROOT}/MSCOCO-2017"