remove{suffix,prefix} replaces {l,r}strip in pytorch/data/torchdata/d…

…atapipes/iter/util/bz2fileloader.py +1

Summary:
`x.lstrip("string")` is equivalent to `re.sub(r"[string]", "", x)`. This means that `"fbcode/fbcode_file".lstrip("fbcode/")` returns `_file`. This can easily be an unintended behaviour!

Instead, as of Python 3.10, we should use `string.removeprefix` or `string.removesuffix`.

This diff makes one or more such fixes that we believe are likely to be bugs. Please double-check before approving.

Reviewed By: zsol

Differential Revision: D55386281

fbshipit-source-id: 16a082f06f2820e87bbe9df69877a8d07ee97218

torchdata/datapipes/iter/util/bz2fileloader.py

@@ -53,7 +53,7 @@ def __iter__(self) -> Iterator[Tuple[str, BufferedIOBase]]:
             pathname, data_stream = data
             try:
                 extracted_fobj = bz2.open(data_stream, mode="rb")  # type: ignore[call-overload]
-                new_pathname = pathname.rstrip(".bz2")
+                new_pathname = pathname.removesuffix(".bz2")
                 yield new_pathname, StreamWrapper(extracted_fobj, data_stream, name=new_pathname)  # type: ignore[misc]
             except Exception as e:
                 warnings.warn(f"Unable to extract files from corrupted bzip2 stream {pathname} due to: {e}, abort!")