def test_generator_one_hot_and_class_index(self):
from adlkit.data_provider.tests.mock_config import mock_sample_specification, \
mock_batches, mock_expected_malloc_requests, mock_file_index_list, mock_class_index_map
mock_class_index_map = copy.deepcopy(mock_class_index_map)
mock_sample_specification = copy.deepcopy(mock_sample_specification)
mock_batches = copy.deepcopy(mock_batches)
mock_expected_malloc_requests = copy.deepcopy(
mock_expected_malloc_requests)
mock_file_index_list = copy.deepcopy(mock_file_index_list)
batch_size = 500
tmp_data_provider = FileDataProvider(mock_sample_specification,
batch_size=batch_size,
read_multiplier=2,
make_one_hot=True,
make_class_index=True,
sleep_duration=sleep_duration)
tmp_data_provider.start_queues()
max_batches = len(mock_batches)
# inserting mock data, once again
for batch in mock_batches:
# try:
# tmp_data_provider.in_queue.put(batch, timeout=1)
success = tmp_data_provider.comm_driver.write('in',
batch,
block=False)
# except Queue.Full:
if not success:
max_batches -= 1
tmp_data_provider.malloc_requests = mock_expected_malloc_requests
tmp_data_provider.start_reader(FileReader, io_ctlr=IOController())
# lg.debug("I am expecting to write to {0}".format(hex(id(tmp_data_provider.shared_memory[0][0][1][0]))))
this = BaseGenerator(
# out_queue=tmp_data_provider.out_queue,
comm_driver=tmp_data_provider.comm_driver,
batch_size=tmp_data_provider.config.batch_size,
read_size=tmp_data_provider.config.read_size,
max_batches=max_batches,
shared_memory_pointer=tmp_data_provider.shared_memory,
file_index_list=mock_file_index_list,
class_index_map=mock_class_index_map)
count = 0
for tmp in this.generate():
count += 1
self.assertEqual(len(tmp), 4)
self.assertEqual(tmp[0].shape, (batch_size, 5))
self.assertEqual(tmp[1].shape, (batch_size, 5))
self.assertEqual(tmp[2].shape, (batch_size, ))
self.assertEqual(tmp[3].shape, (batch_size, 3))
self.assertEqual(count, max_batches)
def test_generator(self):
from adlkit.data_provider.tests.mock_config import mock_sample_specification, mock_batches, \
mock_expected_malloc_requests, mock_file_index_list, mock_class_index_map
mock_sample_specification = copy.deepcopy(mock_sample_specification)
mock_batches = copy.deepcopy(mock_batches)
mock_expected_malloc_requests = copy.deepcopy(
mock_expected_malloc_requests)
mock_file_index_list = copy.deepcopy(mock_file_index_list)
mock_class_index_map = copy.deepcopy(mock_class_index_map)
tmp_data_provider = FileDataProvider(mock_sample_specification,
batch_size=500,
read_multiplier=2,
n_readers=3,
sleep_duration=sleep_duration)
tmp_data_provider.start_queues()
max_batches = len(mock_batches)
# max_batches = 100
# inserting mock data into the queue (should probably be a function at this point)
for batch in mock_batches:
# try:
# tmp_data_provider.in_queue.put(batch, timeout=1)
success = tmp_data_provider.comm_driver.write('in',
batch,
block=False)
# except Queue.Full:
if not success:
max_batches -= 1
tmp_data_provider.malloc_requests = mock_expected_malloc_requests
tmp_data_provider.start_reader(FileReader, io_ctlr=IOController())
# lg.debug("I am expecting to write to {0}".format(hex(id(tmp_data_provider.shared_memory[0][0][1][0]))))
this = BaseGenerator(
comm_driver=tmp_data_provider.comm_driver,
batch_size=tmp_data_provider.config.batch_size,
read_size=tmp_data_provider.config.read_size,
max_batches=max_batches,
shared_memory_pointer=tmp_data_provider.shared_memory,
file_index_list=mock_file_index_list,
class_index_map=mock_class_index_map)
batch_size = tmp_data_provider.config.batch_size
self.assertGreater(max_batches, 0)
for _ in range(max_batches):
tmp = this.generate().next()
self.assertEqual(len(tmp), 2)
self.assertEqual(tmp[0].shape, (batch_size, 5))
self.assertEqual(tmp[1].shape, (batch_size, 5))
def test_start_reader(self):
from adlkit.data_provider.tests.mock_config import mock_sample_specification, \
mock_batches, mock_expected_malloc_requests
mock_sample_specification = copy.deepcopy(mock_sample_specification)
mock_batches = copy.deepcopy(mock_batches)
mock_expected_malloc_requests = copy.deepcopy(
mock_expected_malloc_requests)
tmp_data_provider = FileDataProvider(mock_sample_specification,
batch_size=500,
read_multiplier=2,
make_one_hot=True,
make_class_index=True,
sleep_duration=sleep_duration)
tmp_data_provider.start_queues()
# inserting mock data into the queue
for batch in mock_batches:
try:
tmp_data_provider.in_queue.put(batch, timeout=1)
except Queue.Full:
pass
tmp_data_provider.malloc_requests = mock_expected_malloc_requests
reader_id = tmp_data_provider.start_reader(H5Reader)
# lg.debug("I am expecting to write to {0}".format(hex(id(tmp_data_provider.shared_memory[0][0][1][0]))))
max_batches = 5
out = list()
while True:
if len(out) == max_batches:
break
batch = None
try:
batch = tmp_data_provider.out_queue.get(timeout=1)
except Queue.Empty:
pass
finally:
if batch is not None:
out.append(batch)
for item in out:
tmp_worker_id, tmp_bucket_index, tmp_data_sets, batch_id = item
# Check that each bucket was successfully updated
self.assertEqual(
tmp_data_provider.shared_memory[tmp_worker_id]
[tmp_bucket_index][0].value, 1)
# check for correct reader_id assignment
self.assertEqual(len(tmp_data_provider.readers), reader_id + 1)
# check that mock data successfully was processed
self.assertEquals(
len(out), max_batches,
"test consumed {0} of {1} expected batches from the out_queue".
format(len(out), max_batches))
def test_start_reader(self):
from adlkit.data_provider.tests.mock_config import mock_sample_specification, \
mock_batches, mock_expected_malloc_requests
mock_sample_specification = copy.deepcopy(mock_sample_specification)
mock_batches = copy.deepcopy(mock_batches)
mock_expected_malloc_requests = copy.deepcopy(
mock_expected_malloc_requests)
tmp_data_provider = FileDataProvider(mock_sample_specification,
batch_size=500,
read_multiplier=2,
make_one_hot=True,
make_class_index=True,
sleep_duration=sleep_duration)
tmp_data_provider.start_queues()
# inserting mock data into the queue
for batch in mock_batches:
success = tmp_data_provider.comm_driver.write('in', batch)
self.assertTrue(success)
tmp_data_provider.malloc_requests = mock_expected_malloc_requests
reader_id = tmp_data_provider.start_reader(FileReader,
io_ctlr=IOController())
# lg.debug("I am expecting to write to {0}".format(hex(id(tmp_data_provider.shared_memory[0][0][1][0]))))
max_batches = 5
out = list()
start_time = datetime.datetime.utcnow()
end_time = datetime.timedelta(seconds=100) + start_time
while datetime.datetime.utcnow() < end_time and len(
out) != max_batches:
# try:
batch = tmp_data_provider.comm_driver.read('out', block=False)
# except Queue.Empty:
# pass
# finally:
if batch is not None:
out.append(batch)
for item in out:
# tmp_worker_id, tmp_bucket_index, tmp_data_sets, batch_id = item
tmp_bucket_index, tmp_data_sets, batch_id = item
# Check that each bucket was successfully updated
self.assertEqual(
tmp_data_provider.shared_memory[tmp_bucket_index][0].value, 1)
# check for correct reader_id assignment
self.assertEqual(len(tmp_data_provider.readers), reader_id + 1)
# check that mock data successfully was processed
self.assertEquals(
len(out), max_batches,
"test consumed {0} of {1} expected batches from the out_queue".
format(len(out), max_batches))
def test_end_to_end(self):
from adlkit.data_provider.tests.mock_config import mock_sample_specification
mock_sample_specification = copy.deepcopy(mock_sample_specification)
batch_size = 5
max_batches = 5
tmp_data_provider = FileDataProvider(mock_sample_specification,
batch_size=batch_size,
read_multiplier=2,
make_one_hot=True,
make_class_index=True,
make_file_index=True,
n_readers=4,
sleep_duration=sleep_duration,
max_batches=max_batches)
tmp_data_provider.start_queues()
tmp_data_provider.start_filler(H5Filler)
tmp_data_provider.process_malloc_requests()
tmp_data_provider.start_reader(H5Reader)
# lg.debug("I am expecting to write to {0}".format(hex(id(tmp_data_provider.shared_memory[0][0][1][0]))))
generator_id = tmp_data_provider.start_generator(BaseGenerator)
count = 0
for tmp in tmp_data_provider.generators[generator_id].generate():
count += 1
self.assertEqual(len(tmp), 5)
self.assertEqual(tmp[0].shape, (batch_size, 5))
self.assertEqual(tmp[1].shape, (batch_size, 5))
self.assertEqual(tmp[2].shape, (batch_size, ))
self.assertEqual(tmp[3].shape, (batch_size, 3))
self.assertEqual(len(tmp[4][0]), 2)
self.assertEqual(
count, max_batches,
"we have a leaky generator here {0} != {1}".format(
count, max_batches))
