def test_update_progress(azure):
"""
Upload a 32 bytes file in chunks of 8 and test that the progress is incrementally
updated.
"""
calls = []
def recording_callback(progress, total):
calls.append((progress, total))
def transfer(adlfs,
src,
dst,
offset,
size,
blocksize,
buffersize,
shutdown_event=None):
return size, None
client = ADLTransferClient(azure,
transfer=transfer,
chunksize=8,
chunked=True,
progress_callback=recording_callback)
client.submit('foo', AzureDLPath('bar'), 32)
client.run()
assert calls == [(8, 32), (16, 32), (24, 32), (32, 32)]
def test_merge(azure):
calls = []
def merge(adlfs, outfile, files, shutdown_event=None, overwrite=False):
calls.append(files)
def transfer(adlfs,
src,
dst,
offset,
size,
blocksize,
buffersize,
shutdown_event=None):
return size, None
class XLoaderMock(object):
_overwrite = False
file_size = 32
chunk_size = 8
client = ADLTransferClient(azure,
parent=XLoaderMock(),
transfer=transfer,
merge=merge,
chunksize=chunk_size,
chunked=True)
client.submit('foo', AzureDLPath('bar'), file_size)
client.run()
assert len(calls[0]) == file_size / chunk_size
def mock_client(adl, nthreads):
def transfer(adlfs, src, dst, offset, size, buffersize, blocksize, shutdown_event=None):
pass
def merge(adlfs, outfile, files, shutdown_event=None):
pass
return ADLTransferClient(
adl,
'foo',
transfer=transfer,
merge=merge,
nthreads=nthreads)
def test_upload_one_empty_file(local_files, azure):
with azure_teardown(azure):
bigfile, littlefile, emptyfile, a, b, c = local_files
# transfer client w/ deterministic temporary directory
from azure.datalake.store.multithread import put_chunk
client = ADLTransferClient(azure, transfer=put_chunk,
unique_temporary=False)
# single chunk, empty file
up = ADLUploader(azure, test_dir / 'emptyfile', emptyfile, nthreads=1,
overwrite=True)
assert azure.info(test_dir / 'emptyfile')['length'] == 0
azure.rm(test_dir / 'emptyfile')
def test_upload_single_file_in_dir(tempdir, azure):
with azure_teardown(azure):
lpath_dir = tempdir
lfilename = os.path.join(lpath_dir, 'singlefile')
with open(lfilename, 'wb') as f:
f.write(b'0123456789')
# transfer client w/ deterministic temporary directory
from azure.datalake.store.multithread import put_chunk
client = ADLTransferClient(azure, transfer=put_chunk,
unique_temporary=False)
up = ADLUploader(azure, test_dir / 'singlefiledir', lpath_dir, nthreads=1,
overwrite=True)
assert azure.info(test_dir / 'singlefiledir' / 'singlefile')['length'] == 10
azure.rm(test_dir / 'singlefiledir' / 'singlefile')
def test_submit_and_run(azure):
def transfer(adlfs,
src,
dst,
offset,
size,
blocksize,
buffersize,
shutdown_event=None):
time.sleep(0.1)
return size, None
client = ADLTransferClient(azure,
transfer=transfer,
chunksize=8,
chunked=False)
client.submit('foo', 'bar', 16)
client.submit('abc', '123', 8)
nfiles = len(client.progress)
assert nfiles == 2
assert len([client.progress[i].chunks for i in range(nfiles)])
assert all([client.progress[i].state == 'pending' for i in range(nfiles)])
assert all([
chunk.state == 'pending' for f in client.progress for chunk in f.chunks
])
expected = {('bar', 0), ('bar', 8), ('123', 0)}
assert {(chunk.name, chunk.offset)
for f in client.progress for chunk in f.chunks} == expected
client.run()
assert all([client.progress[i].state == 'finished' for i in range(nfiles)])
assert all([
chunk.state == 'finished' for f in client.progress
for chunk in f.chunks
])
assert all([
chunk.expected == chunk.actual for f in client.progress
for chunk in f.chunks
])
def test_upload_empty_folder(tempdir, azure):
with azure_teardown(azure):
os.mkdir(os.path.join(tempdir, "dir1"))
os.mkdir(os.path.join(tempdir, "dir1", "b"))
with open(os.path.join(tempdir, "dir1", "file.txt"), 'wb') as f:
f.write(b'0123456789')
# transfer client w/ deterministic temporary directory
from azure.datalake.store.multithread import put_chunk
client = ADLTransferClient(azure, transfer=put_chunk,
unique_temporary=False)
# single chunk, empty file
up = ADLUploader(azure, test_dir / "dir1", os.path.join(tempdir, "dir1") , nthreads=1,
overwrite=True)
assert azure.info(test_dir / "dir1" /"b")['type'] == 'DIRECTORY'
azure.rm(test_dir / "dir1", recursive=True)
def test_shutdown(azure):
def transfer(adlfs,
src,
dst,
offset,
size,
blocksize,
buffersize,
retries=5,
shutdown_event=None):
while shutdown_event and not shutdown_event.is_set():
time.sleep(0.1)
return size, None
client = ADLTransferClient(azure,
transfer=transfer,
chunksize=1,
chunked=False)
client.submit('foo', 'bar', 16)
client.run(monitor=False)
client.shutdown()
assert client.progress[0].state == 'finished'
def test_upload_one(local_files, azure):
with azure_teardown(azure):
bigfile, littlefile, emptyfile, a, b, c = local_files
# transfer client w/ deterministic temporary directory
from azure.datalake.store.multithread import put_chunk
client = ADLTransferClient(azure, transfer=put_chunk,
unique_temporary=False)
# single chunk
up = ADLUploader(azure, test_dir / 'littlefile', littlefile, nthreads=1,
overwrite=True)
assert azure.info(test_dir / 'littlefile')['length'] == 10
# multiple chunks, one thread
size = 10000
up = ADLUploader(azure, test_dir / 'bigfile', bigfile, nthreads=1,
chunksize=size//5, client=client, run=False,
overwrite=True)
up.run()
assert azure.info(test_dir / 'bigfile')['length'] == size
azure.rm(test_dir / 'bigfile')
