def test_read_range_at_beginning(self):
fbs = LineSource('dummy_pattern', validate=False)
file_name, expected_data = write_data(10)
assert len(expected_data) == 10
source = SingleFileSource(fbs, file_name, 0, 10 * 6)
range_tracker = source.get_range_tracker(0, 20)
read_data = [value for value in source.read(range_tracker)]
self.assertCountEqual(expected_data[:4], read_data)
def test_read_range_at_middle(self):
fbs = LineSource('dummy_pattern', validate=False)
file_name, expected_data = write_data(10)
assert len(expected_data) == 10
source = SingleFileSource(fbs, file_name, 0, 10 * 6)
range_tracker = source.get_range_tracker(20, 40)
read_data = [value for value in source.read(range_tracker)]
self.assertItemsEqual(expected_data[4:7], read_data)
def test_produces_splits_desiredsize_smaller_than_size(self):
fbs = LineSource('dummy_pattern', validate=False)
file_name, expected_data = write_data(10)
assert len(expected_data) == 10
source = SingleFileSource(fbs, file_name, 0, 10 * 6)
splits = [split for split in source.split(desired_bundle_size=25)]
self.assertEquals(3, len(splits))
read_data = []
for split in splits:
source = split.source
range_tracker = source.get_range_tracker(split.start_position,
split.stop_position)
data_from_split = [data for data in source.read(range_tracker)]
read_data.extend(data_from_split)
self.assertItemsEqual(expected_data, read_data)
def test_produce_split_with_start_and_end_positions(self):
fbs = LineSource('dummy_pattern', validate=False)
file_name, expected_data = write_data(10)
assert len(expected_data) == 10
source = SingleFileSource(fbs, file_name, 0, 10 * 6)
splits = [split for split in
source.split(desired_bundle_size=15, start_offset=10,
stop_offset=50)]
self.assertEquals(3, len(splits))
read_data = []
for split in splits:
source = split.source
range_tracker = source.get_range_tracker(split.start_position,
split.stop_position)
data_from_split = [data for data in source.read(range_tracker)]
read_data.extend(data_from_split)
self.assertCountEqual(expected_data[2:9], read_data)