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')
def test_set_expiry(azure):
with azure_teardown(azure):
# this future time gives the milliseconds since the epoch that have occured as of 01/31/2030 at noon
epoch_time = datetime.datetime.utcfromtimestamp(0)
final_time = datetime.datetime(2030, 1, 31, 12)
time_in_milliseconds = (final_time - epoch_time).total_seconds() * 1000
# create the file
azure.touch(a)
# first get the existing expiry, which should be never
initial_expiry = azure.info(a,
invalidate_cache=True)['msExpirationTime']
azure.set_expiry(a, 'Absolute', time_in_milliseconds)
cur_expiry = azure.info(a, invalidate_cache=True)['msExpirationTime']
# this is a range of +- 100ms because the service does a best effort to set it precisely, but there is
# no guarantee that the expiry will be to the exact millisecond
assert time_in_milliseconds - 100 <= cur_expiry <= time_in_milliseconds + 100
assert initial_expiry != cur_expiry
# now set it back to never expire and validate it is the same
azure.set_expiry(a, 'NeverExpire')
cur_expiry = azure.info(a)['msExpirationTime']
assert initial_expiry == cur_expiry
# now validate the fail cases
# bad enum
with pytest.raises(ValueError):
azure.set_expiry(a, 'BadEnumValue')
# missing time
with pytest.raises(ValueError):
azure.set_expiry(a, 'Absolute')
def test_set_expiry(azure):
with azure_teardown(azure):
azure.touch(a)
# first get the existing expiry, which should be never
initial_expiry = azure.info(a)['msExpirationTime']
# this future time gives the milliseconds since the epoch that have occured as of 01/31/2030 at noon
epoch_time = datetime.datetime.utcfromtimestamp(0)
final_time = datetime.datetime(2030, 1, 31, 12)
time_in_milliseconds = (final_time - epoch_time).total_seconds() * 1000
azure.set_expiry(a, 'Absolute', time_in_milliseconds)
cur_expiry = azure.info(a)['msExpirationTime']
assert time_in_milliseconds == cur_expiry
assert initial_expiry != cur_expiry
# now set it back to never expire and validate it is the same
azure.set_expiry(a, 'NeverExpire')
cur_expiry = azure.info(a)['msExpirationTime']
assert initial_expiry == cur_expiry
# now validate the fail cases
# bad enum
with pytest.raises(ValueError):
azure.set_expiry(a, 'BadEnumValue')
# missing time
with pytest.raises(ValueError):
azure.set_expiry(a, 'Absolute')
def test_chmod(azure):
with azure_teardown(azure):
azure.touch(a)
assert azure.info(a)['permission'] == '770'
azure.chmod(a, '0555')
assert azure.info(a)['permission'] == '555'
with pytest.raises((OSError, IOError)):
with azure.open(a, 'ab') as f:
try:
f.write(b'data')
except Exception as e:
print(e)
raise e
azure.chmod(a, '0770')
azure.rm(a)
azure.mkdir(test_dir / 'deep')
azure.touch(test_dir / 'deep' / 'file')
azure.chmod(test_dir / 'deep', '660')
with pytest.raises((OSError, IOError)):
azure.ls(test_dir / 'deep')
azure.chmod(test_dir / 'deep', '770')
def test_info(azure):
with azure_teardown(azure):
with azure.open(a, 'wb') as f:
f.write(b'a' * 5)
info = azure.info(a, invalidate_cache=False)
assert info['length'] == 5
assert info['name'] == a
assert info['type'] == 'FILE'
assert azure.info(test_dir, invalidate_cache=True)['type'] == 'DIRECTORY'
def test_info(azure):
with azure_teardown(azure):
with azure.open(a, 'wb') as f:
f.write(b'a' * 5)
info = azure.info(a)
assert info['length'] == 5
assert info['name'] == a
assert info['type'] == 'FILE'
assert azure.info(test_dir)['type'] == 'DIRECTORY'
def test_info_invalidate_cache(azure, second_azure):
with azure_teardown(azure):
# construct initial cache and ensure the file doesn't already exist
assert not azure.exists(a, invalidate_cache=False)
assert not second_azure.exists(a, invalidate_cache=False)
with azure.open(a, 'wb') as f:
f.write(b'a' * 5)
# verify that it works in the fs that did the write and not on the other
info = azure.info(a, invalidate_cache=False)
with pytest.raises(FileNotFoundError):
second_azure.info(a, invalidate_cache=False)
# then invalidate
second_info = second_azure.info(a, invalidate_cache=True)
assert info['length'] == 5
assert info['name'] == a
assert info['type'] == 'FILE'
assert info['length'] == second_info['length']
assert info['name'] == second_info['name']
assert info['type'] == second_info['type']
# assure that the cache was properly repopulated on the info call
assert second_azure.info(test_dir,
invalidate_cache=False)['type'] == 'DIRECTORY'
def test_file_creation_open(azure):
with azure_teardown(azure):
if azure.exists(a):
azure.rm(a)
assert not azure.exists(a)
f = azure.open(a, "wb")
assert azure.exists(a)
f.close()
assert azure.info(a)['length'] == 0
def test_upload_many(local_files, azure):
with azure_teardown(azure):
bigfile, littlefile, emptyfile, a, b, c = local_files
root = os.path.dirname(bigfile)
# single thread
up = ADLUploader(azure, test_dir, root, nthreads=1, overwrite=True)
assert azure.info(test_dir / 'littlefile')['length'] == 10
assert azure.cat(test_dir / 'nested1/nested2/a') == b'0123456789'
assert len(azure.du(test_dir, deep=True)) == 6
assert azure.du(test_dir, deep=True, total=True) == 10000 + 40
def test_chown(azure):
with azure_teardown(azure):
azure.touch(a)
# fake IDs
user_id = '470c0ccf-c91a-4597-98cd-48507d2f1486'
group_id = '6b190b7a-0acf-43c8-ab14-965f5aea6243'
# Account doesn't have permission to change owner
owner = azure.info(a)['owner']
with pytest.raises(PermissionError):
azure.chown(a, owner=user_id)
assert owner == azure.info(a)['owner']
# Account doesn't have permission to change group
group = azure.info(a)['group']
with pytest.raises(PermissionError):
azure.chown(a, group=group_id)
assert group == azure.info(a)['group']
# Account doesn't have permission to change owner/group
with pytest.raises(PermissionError):
azure.chown(a, owner=owner, group=group)
assert owner == azure.info(a)['owner']
assert group == azure.info(a)['group']
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_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)