d' these descriptors
  are read from disk.

* If a **directory** argument is provided and the file is already present
  these descriptors are read from disk.

* If a **directory** argument is provided and the file is not present the
  file is downloaded this location then read.

* If the file has neither been downloaded and no **directory** argument
  is provided then the file is downloaded to a temporary directory that's
  deleted after it is read.

:param str directory: destination to download into
:param str descriptor_type: `descriptor type
  <https://metrics.torproject.org/collector.html#data-formats>`_, this is
  guessed if not provided
:param datetime.datetime start: publication time to begin with
:param datetime.datetime end: publication time to end with
:param stem.descriptor.__init__.DocumentHandler document_handler: method in
  which to parse a :class:`~stem.descriptor.networkstatus.NetworkStatusDocument`
:param int timeout: timeout when connection becomes idle, no timeout
  applied if **None**
:param int retries: maximum attempts to impose

:returns: iterator for :class:`~stem.descriptor.__init__.Descriptor`
  instances in the file

:raises:
  * **ValueError** if unable to determine the descirptor type
  * **TypeError** if we cannot parse this descriptor type
  * :class:`~stem.DownloadFailed` if the download fails
NÚ