def test_substream_for_single_thread_upload_large_block(self):
FILE_PATH = 'largest_blob_from_path.temp.{}.dat'.format(
str(uuid.uuid4()))
with open(FILE_PATH, 'wb') as stream:
largeStream = LargeStream(LARGE_BLOCK_SIZE, 4 * 1024 * 1024)
chunk = largeStream.read()
while chunk:
stream.write(chunk)
chunk = largeStream.read()
with open(FILE_PATH, 'rb') as stream:
substream = SubStream(stream, 0, 2 * 1024 * 1024, None)
# this is to mimic stage large block: SubStream.read() is getting called by http client
data1 = substream.read(2 * 1024 * 1024)
substream.read(2 * 1024 * 1024)
substream.read(2 * 1024 * 1024)
# this is to mimic rewinding request body after connection error
substream.seek(0)
# this is to mimic retry: stage that large block from beginning
data2 = substream.read(2 * 1024 * 1024)
self.assertEqual(data1, data2)
self._teardown(FILE_PATH)
def test_sub_stream_with_length_equal_to_buffer(self, resource_group, location, storage_account, storage_account_key):
data = os.urandom(6 * 1024 * 1024)
# assuming the max size of the buffer is 4MB, this test needs to be updated if that has changed
# the block size is 2MB for this test
expected_data = data[0: 2 * 1024 * 1024]
wrapped_stream = BytesIO(expected_data) # simulate stream given by user
lockObj = Lock() # simulate multi-threaded environment
substream = SubStream(wrapped_stream, stream_begin_index=0, length=2 * 1024 * 1024, lockObj=lockObj)
try:
# substream should start with position at 0
self.assertEqual(substream.tell(), 0)
# reading a chunk that is smaller than the buffer
data_chunk_1 = substream.read(1 * 1024 * 1024)
self.assertEqual(len(data_chunk_1), 1 * 1024 * 1024)
# reading a chunk that is bigger than the buffer, should not read anything beyond
data_chunk_2 = substream.read(4 * 1024 * 1024)
self.assertEqual(len(data_chunk_2), 1 * 1024 * 1024)
# assert data is consistent
self.assertEqual(data_chunk_1 + data_chunk_2, expected_data)
# test seek
substream.seek(1 * 1024 * 1024, SEEK_SET)
data_chunk_2 = substream.read(1 * 1024 * 1024)
# assert data is consistent
self.assertEqual(data_chunk_1 + data_chunk_2, expected_data)
finally:
wrapped_stream.close()
substream.close()
def test_sub_stream_with_length_larger_than_buffer(self, resource_group,
location,
storage_account,
storage_account_key):
data = os.urandom(12 * 1024 * 1024)
# assuming the max size of the buffer is 4MB, this test needs to be updated if that has changed
# the block size is 6MB for this test
expected_data = data[0:6 * 1024 * 1024]
wrapped_stream = BytesIO(data) # simulate stream given by user
lockObj = Lock() # simulate multi-threaded environment
substream = SubStream(wrapped_stream,
stream_begin_index=0,
length=6 * 1024 * 1024,
lockObj=lockObj)
try:
# substream should start with position at 0
self.assertEqual(substream.tell(), 0)
# reading a chunk that is smaller than the buffer
data_chunk_1 = substream.read(2 * 1024 * 1024)
self.assertEqual(len(data_chunk_1), 2 * 1024 * 1024)
# reading a chunk that is bigger than the data remaining in buffer, force a buffer swap
data_chunk_2 = substream.read(4 * 1024 * 1024)
self.assertEqual(len(data_chunk_2), 4 * 1024 * 1024)
# assert data is consistent
self.assertEqual(data_chunk_1 + data_chunk_2, expected_data)
self.assertEqual(6 * 1024 * 1024, substream.tell())
# attempt to read more than what the sub stream contains should return nothing
empty_data = substream.read(1 * 1024 * 1024)
self.assertEqual(0, len(empty_data))
self.assertEqual(6 * 1024 * 1024, substream.tell())
# test seek outside of current buffer, which is at the moment the last 2MB of data
substream.seek(0, SEEK_SET)
data_chunk_1 = substream.read(4 * 1024 * 1024)
data_chunk_2 = substream.read(2 * 1024 * 1024)
# assert data is consistent
self.assertEqual(data_chunk_1 + data_chunk_2, expected_data)
# test seek inside of buffer, which is at the moment the last 2MB of data
substream.seek(4 * 1024 * 1024, SEEK_SET)
data_chunk_2 = substream.read(2 * 1024 * 1024)
# assert data is consistent
self.assertEqual(data_chunk_1 + data_chunk_2, expected_data)
finally:
wrapped_stream.close()
substream.close()