- class torchdata.datapipes.iter.TarArchiveLoader(datapipe: Iterable[Tuple[str, BufferedIOBase]], mode: str = 'r:*', length: int = - 1)¶
Opens/decompresses tar binary streams from an Iterable DataPipe which contains tuples of path name and tar binary stream, and yields a tuple of path name and extracted binary stream (functional name:
datapipe – Iterable DataPipe that provides tuples of path name and tar binary stream
mode – File mode used by tarfile.open to read file object. Mode has to be a string of the form ‘filemode[:compression]’
length – a nominal length of the DataPipe
The opened file handles will be closed automatically if the default
DecoderDataPipeis attached. Otherwise, user should be responsible to close file handles explicitly or let Python’s GC close them periodically.
>>> from torchdata.datapipes.iter import FileLister, FileOpener >>> datapipe1 = FileLister(".", "*.tar") >>> datapipe2 = FileOpener(datapipe1, mode="b") >>> tar_loader_dp = datapipe2.load_from_tar() >>> for _, stream in tar_loader_dp: >>> print(stream.read()) b'0123456789abcdef'