INNER CODE UNIT · Python
main
FoundationVision/Liquid · data_process/convert_DCLM_data.py:19
def main(args):
input_path = args.input_path
if not os.path.exists(args.temp_path):
os.makedirs(args.temp_path)
files_list = []
for dirpath, dirnames, filenames in os.walk(input_path):
for file in filenames:
if file.endswith('zst'):
full_path = os.path.join(dirpath, file)
files_list.append(full_path)
files_list.sort()
# files_list = files_list[:10]
for idx in tqdm(range(len(files_list))):
filepath = files_list[idx]
alldata_in_one_path = list(read_jsonl_zst(filepath))
all_datas = []
for data in alldata_in_one_path: