src/visualizr/dataset_util.py

from os import path
from shutil import copytree

from visualizr import logger
from visualizr.dist_utils import barrier, get_rank


def use_cached_dataset_path(source_path, cache_path):
    if get_rank() == 0:
        if not path.exists(cache_path):
            logger.info(f"copying the data: {source_path} to {cache_path}")
            copytree(source_path, cache_path)
    barrier()
    return cache_path