def setUp(self):
index_seq = [0, 2, 1, 2, 0, 1, 1]
data_seqs = [[0.0, 0.1, 0.2, 0.3],
[1.0, 1.1, 1.2, 1.3],
[2.0, 2.1, 2.2, 2.3]]
self.expected_result = [0.0, 2.0, 1.0, 2.1, 0.1, 1.1, 1.2]
self.iterator = MultiplexIterator(NativeCheckpointableIterator(index_seq), [NativeCheckpointableIterator(ds) for ds in data_seqs])
def setUp(self):
data = list(range(5))
batch_size = 3
self.expected_result = [data[0:3], data[3:]]
self.iterator = FixedBatchIterator(NativeCheckpointableIterator(data), batch_size=batch_size)
def test_different_line_endings(self):
# write data in binary mode with LF line endings
lf_dir = tempfile.mkdtemp()
lf_file = os.path.join(lf_dir, 'test.gz')
with gzip.open(lf_file, 'w') as f:
f.write('\n'.join(self.flattened_test_data).encode('utf-8'))
# write data in binary mode with CRLF line endings
crlf_dir = tempfile.mkdtemp()
crlf_file = os.path.join(crlf_dir, 'test.gz')
with gzip.open(crlf_file, 'w') as f:
f.write('\r\n'.join(self.flattened_test_data).encode('utf-8'))
lf_data = list(self._select_many_from_chunks(NativeCheckpointableIterator([lf_file])))
crlf_dat = list(self._select_many_from_chunks(NativeCheckpointableIterator([crlf_file])))
self.assertListEqual(lf_data, crlf_dat)
shutil.rmtree(lf_dir)
shutil.rmtree(crlf_dir)
def setUp(self):
data = list(range(53))
def transform(random: Random, item: int):
return item + random.random()
seed = 1
random = Random()
random.seed(seed)
self.expected_result = [n + random.random() for n in data]
self.iterator = SamplingRandomMapIterator(NativeCheckpointableIterator(data), transform=transform, seed=seed)
def setUp(self):
data = list(range(53))
self.expected_result = [0]
for i in data[1:]:
self.expected_result.append(self.expected_result[-1] + i)
def step_function(prev_state, item):
output = item + prev_state
new_state = output
return new_state, output
self.iterator = RecurrentIterator(NativeCheckpointableIterator(data), step_function, initial_state = 0)
def test(self):
for n in [0, 2, 3, 8, 9, 10, 11, 12]: # cover various boundary conditions
seq = list(range(n))
it = WindowedIterator(NativeCheckpointableIterator(seq), 3)
actual0 = list(itertools.islice(it, n * 3 // 10))
checkpoint = it.getstate()
actual1a = list(it)
it.setstate(checkpoint)
actual1b = list(it)
actual = actual0 + actual1a
expected = list(zip(seq, itertools.islice(seq, 1, None), itertools.islice(seq, 2, None)))
self.assertListEqual(actual, expected) # basic operation
self.assertListEqual(actual1a, actual1b) # checkpointing
def setUp(self):
self.expected_result = list(range(53))
source_iterator = NativeCheckpointableIterator(self.expected_result)
self.iterator = PrefetchIterator(source_iterator, buffer_size=13)
def setUp(self):
data1 = list(range(53))
data2 = [n * n for n in data1]
self.expected_result = list(zip(data1, data2))
self.iterator = ZipIterator(NativeCheckpointableIterator(data1), NativeCheckpointableIterator(data2))
def setUp(self):
data = list(range(53))
self.expected_result = [map_fun(n) for n in data]
self.iterator = ParallelMapIterator(NativeCheckpointableIterator(data), map_fun, 5, 7)
def test_shuffle_buffer_size_one(self):
# work on copy of data in case data is modified by class
items = list(BlockwiseShuffleIterator(NativeCheckpointableIterator(self.flattened_test_data.copy()), 1, 42))
self.assertListEqual(items, self.flattened_test_data)
def test_shuffle(self):
# work on copy of data in case data is modified by class
items = list(BufferedShuffleIterator(NativeCheckpointableIterator(self.flattened_test_data.copy()), 971, 42))
self.assertMultisetEqual(items, self.flattened_test_data)
def test_no_selector(self):
data = list(range(100))
sublists = [data[:10], data[10:42], data[42: 87], data[87:]]
result = list(SelectManyIterator(NativeCheckpointableIterator(sublists)))
self.assertListEqual(result, data)
def test(self):
items = list(self._select_many_from_chunks(NativeCheckpointableIterator(self.chunk_file_paths)))
self.assertListEqual(items, self.flattened_test_data)
def setUp(self):
self.expected_result = list(range(53))
self.iterator = NativeCheckpointableIterator(self.expected_result)
