ontonotes.py 文件源码

def dataset_path_iterator(file_path: str) -> Iterator[str]:
        """
        An iterator returning file_paths in a directory
        containing CONLL-formatted files.
        """
        logger.info("Reading CONLL sentences from dataset files at: %s", file_path)
        for root, _, files in tqdm.tqdm(list(os.walk(file_path))):
            for data_file in files:
                # These are a relic of the dataset pre-processing. Every
                # file will be duplicated - one file called filename.gold_skel
                # and one generated from the preprocessing called filename.gold_conll.
                if not data_file.endswith("gold_conll"):
                    continue

                yield os.path.join(root, data_file)