def _stream2tempfiles(
stream, jobs=1, chunksize=CHUNKSIZE, writer=_PICKLE_IO, **kwargs):
"""
Sort a stream of data into temporary files. Caller is responsible for
deleting files. Tempfile paths are generated with `tempfile.mkstemp()`.
Parameters
----------
stream : iter
Input stream to sort.
jobs : int, optional
Sort data with a pool of N workers.
chunksize : int, optional
Process this many objects from the input stream in each job. Also
the maximum amount of objects per tempfile.
writer : None or tinysort.io.BaseSerializer, optional
Instance of the serializer for writing the stream to disk.
kwargs : **kwargs, optional
Keyword arguments for `sorted()`.
Returns
-------
list
Temporary file paths.
"""
tasks = ({
'data': data,
'writer': writer,
'sort_args': kwargs
} for data in tools.slicer(stream, chunksize))
with tools.runner(_mp_sort_into_tempfile, tasks, jobs) as run:
return list(run)
def test_slicer_odd():
it = tools.slicer(range(5), 2)
assert next(it) == (0, 1)
assert next(it) == (2, 3)
assert next(it) == (4, )
with pytest.raises(StopIteration):
next(it)
def test_slicer_even():
it = tools.slicer(six.moves.xrange(100), 10)
for idx, actual in enumerate(it):
assert isinstance(actual, tuple)
assert len(actual) == 10
# Verify that the values are correct
assert actual == tuple((10 * idx) + i for i in range(len(actual)))
assert idx == 9
