Method process_file

utils/dataset.py:635–639

(file)

            control_file_stems = {path.stem: path for path in self.control_path.glob('*') if path.is_file()} if self.control_path is not None else {}

            def process_file(file):
                if file.suffix != '.tar':
                    return [(None, str(file))]
                with tarfile.TarFile(file) as tar_f:
                    return [(str(file), name) for name in tar_f.getnames()]

            captions_json = self.path / CAPTIONS_JSON_FILE
            has_captions_json = captions_json.exists()