get_readable_fileobj#, encoding=None, cache=False, show_progress=True, remote_timeout=None, sources=None, http_headers=None, *, use_fsspec=None, fsspec_kwargs=None, close_files=True)[source]#

Yield a readable, seekable file-like object from a file or URL.

This supports passing filenames, URLs, and readable file-like objects, any of which can be compressed in gzip, bzip2 or lzma (xz) if the appropriate compression libraries are provided by the Python installation.

name_or_objstr or file-like object

The filename of the file to access (if given as a string), or the file-like object to access.

If a file-like object, it must be opened in binary mode.

encodingstr, optional

When None (default), returns a file-like object with a read method that returns str (unicode) objects, using locale.getpreferredencoding as an encoding. This matches the default behavior of the built-in open when no mode argument is provided.

When 'binary', returns a file-like object where its read method returns bytes objects.

When another string, it is the name of an encoding, and the file-like object’s read method will return str (unicode) objects, decoded from binary using the given encoding.

cachebool or “update”, optional

Whether to cache the contents of remote URLs. If “update”, check the remote URL for a new version but store the result in the cache.

show_progressbool, optional

Whether to display a progress bar if the file is downloaded from a remote server. Default is True.


Timeout for remote requests in seconds (default is the configurable

sourceslist of str, optional

If provided, a list of URLs to try to obtain the file from. The result will be stored under the original URL. The original URL will not be tried unless it is in this list; this is to prevent long waits for a primary server that is known to be inaccessible at the moment.

http_headersdict or None

HTTP request headers to pass into urlopen if needed. (These headers are ignored if the protocol for the name_or_obj/sources entry is not a remote HTTP URL.) In the default case (None), the headers are User-Agent: some_value and Accept: */*, where some_value is set by

use_fsspecbool, optional

Use to open the file? Defaults to False unless name_or_obj starts with the Amazon S3 storage prefix s3:// or the Google Cloud Storage prefix gs://. Can also be used for paths with other prefixes (e.g. http://) but in this case you must explicitly pass use_fsspec=True. Use of this feature requires the optional fsspec package. A ModuleNotFoundError will be raised if the dependency is missing.

Added in version 5.2.

fsspec_kwargsdict, optional

Keyword arguments passed on to This can be used to configure cloud storage credentials and caching behavior. For example, pass fsspec_kwargs={"anon": True} to enable anonymous access to Amazon S3 open data buckets. See fsspec’s documentation for available parameters.

Added in version 5.2.

close_filesbool, optional

Close the file object when exiting the context manager. Default is True.

Added in version 5.2.

filefile-like (readable)


This function is a context manager, and should be used for example as:

with get_readable_fileobj('file.dat') as f:
    contents =

If a URL is provided and the cache is in use, the provided URL will be the name used in the cache. The contents may already be stored in the cache under this URL provided, they may be downloaded from this URL, or they may be downloaded from one of the locations listed in sources. See download_file for details.