22 lines
867 B
Python
Raw Normal View History

2021-09-09 01:14:21 +09:00
import argparse
2022-06-03 16:36:32 +09:00
from sensai_dataset.generator.commands import generate_dataset, generate_reduced_dataset
from sensai_dataset.generator.constants import SENSAI_COMPLETE_DIR, SENSAI_DIR, DATASET_SOURCE_DIR
2021-09-09 01:14:21 +09:00
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='dataset generator')
2022-06-03 16:36:32 +09:00
parser.add_argument('-m', '--matcher', type=str, default='chats_*.parquet')
2021-09-09 01:14:21 +09:00
args = parser.parse_args()
print('source: ' + DATASET_SOURCE_DIR)
2022-06-03 16:36:32 +09:00
print('SENSAI_COMPLETE_DIR: ' + SENSAI_COMPLETE_DIR)
print('SENSAI_DIR: ' + SENSAI_DIR)
2021-09-09 01:14:21 +09:00
generate_dataset(source_dir=DATASET_SOURCE_DIR,
2022-06-03 16:36:32 +09:00
target_dir=SENSAI_COMPLETE_DIR,
2021-09-09 01:14:21 +09:00
matcher=args.matcher)
2022-06-03 16:36:32 +09:00
generate_reduced_dataset(source_dir=DATASET_SOURCE_DIR,
target_dir=SENSAI_DIR,
matcher=args.matcher)