def test_download_file_sliced(self):
buckets = Buckets(os.getenv('KBC_TEST_API_URL'),
os.getenv('KBC_TEST_TOKEN'))
try:
buckets.delete('in.c-py-test-files', force=True)
except exceptions.HTTPError as e:
if e.response.status_code != 404:
raise
buckets.create(name='py-test-files', stage='in')
tables = Tables(os.getenv('KBC_TEST_API_URL'),
os.getenv('KBC_TEST_TOKEN'))
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
os.close(file)
table_id = tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-files')
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'foo', 'col2': 'bar'})
os.close(file)
tables.load(table_id=table_id, file_path=path, is_incremental=True)
file_id = tables.export(table_id=table_id)
temp_path = tempfile.TemporaryDirectory()
local_path = self.files.download(file_id, temp_path.name)
with open(local_path, mode='rt') as file:
lines = file.readlines()
self.assertEqual(['"foo","bar"\n', '"ping","pong"\n'], sorted(lines))
class TestTables(unittest.TestCase):
def setUp(self):
self.tables = Tables(os.getenv('KBC_TEST_API_URL'),
os.getenv('KBC_TEST_TOKEN'))
self.buckets = Buckets(os.getenv('KBC_TEST_API_URL'),
os.getenv('KBC_TEST_TOKEN'))
try:
self.buckets.delete('in.c-py-test-tables', force=True)
except exceptions.HTTPError as e:
if e.response.status_code != 404:
raise
self.buckets.create(name='py-test-tables', stage='in')
# https://github.com/boto/boto3/issues/454
warnings.simplefilter("ignore", ResourceWarning)
def tearDown(self):
try:
self.buckets.delete('in.c-py-test-tables', force=True)
except exceptions.HTTPError as e:
if e.response.status_code != 404:
raise
def test_create_table_minimal(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
table_info = self.tables.detail(table_id)
with self.subTest():
self.assertEqual(table_id, table_info['id'])
with self.subTest():
self.assertEqual('in.c-py-test-tables', table_info['bucket']['id'])
def test_create_table_primary_key(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
writer.writerow({'col1': 'pong', 'col2': 'ping'})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables',
primary_key=['col1', 'col2'])
table_info = self.tables.detail(table_id)
with self.subTest():
self.assertEqual(table_id, table_info['id'])
with self.subTest():
self.assertEqual('in.c-py-test-tables', table_info['bucket']['id'])
with self.subTest():
self.assertEqual(['col1', 'col2'], table_info['primaryKey'])
def test_table_detail(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
table_info = self.tables.detail(table_id)
with self.subTest():
self.assertEqual(table_id, table_info['id'])
with self.subTest():
self.assertEqual('some-table', table_info['name'])
with self.subTest():
self.assertTrue(
'in.c-py-test-tables.some-table' in table_info['uri'])
with self.subTest():
self.assertEqual([], table_info['primaryKey'])
with self.subTest():
self.assertEqual(['col1', 'col2'], table_info['columns'])
with self.subTest():
self.assertTrue('created' in table_info)
with self.subTest():
self.assertTrue('lastImportDate' in table_info)
with self.subTest():
self.assertTrue('lastChangeDate' in table_info)
with self.subTest():
self.assertTrue('rowsCount' in table_info)
with self.subTest():
self.assertTrue('metadata' in table_info)
with self.subTest():
self.assertTrue('bucket' in table_info)
with self.subTest():
self.assertTrue('columnMetadata' in table_info)
def test_delete_table(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
table_info = self.tables.detail(table_id)
self.assertEqual(table_id, table_info['id'])
self.tables.delete(table_id)
try:
self.tables.detail('some-totally-non-existent-table')
except exceptions.HTTPError as e:
if e.response.status_code != 404:
raise
def test_invalid_create(self):
try:
self.tables.detail('some-totally-non-existent-table')
except exceptions.HTTPError as e:
if e.response.status_code != 404:
raise
def test_import_table_incremental(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
table_info = self.tables.detail(table_id)
with self.subTest():
self.assertEqual(table_id, table_info['id'])
with self.subTest():
self.assertEqual(1, table_info['rowsCount'])
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'foo', 'col2': 'bar'})
os.close(file)
self.tables.load(table_id=table_id,
file_path=path,
is_incremental=True)
table_info = self.tables.detail(table_id)
with self.subTest():
self.assertEqual(table_id, table_info['id'])
with self.subTest():
self.assertEqual(2, table_info['rowsCount'])
def test_import_table_no_incremental(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
table_info = self.tables.detail(table_id)
with self.subTest():
self.assertEqual(table_id, table_info['id'])
with self.subTest():
self.assertEqual(1, table_info['rowsCount'])
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'foo', 'col2': 'bar'})
os.close(file)
self.tables.load(table_id=table_id,
file_path=path,
is_incremental=False)
table_info = self.tables.detail(table_id)
with self.subTest():
self.assertEqual(table_id, table_info['id'])
with self.subTest():
self.assertEqual(1, table_info['rowsCount'])
def test_table_preview(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
writer.writerow({'col1': 'foo', 'col2': 'bar'})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
contents = self.tables.preview(table_id=table_id)
lines = contents.split('\n')
self.assertEqual(['', '"col1","col2"', '"foo","bar"', '"ping","pong"'],
sorted(lines))
def test_table_export(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
writer.writerow({'col1': 'foo', 'col2': 'bar'})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
result = self.tables.export(table_id=table_id)
self.assertIsNotNone(result)
def test_table_export_file_plain(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
writer.writerow({'col1': 'foo', 'col2': 'bar'})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
temp_path = tempfile.TemporaryDirectory()
local_path = self.tables.export_to_file(table_id=table_id,
path_name=temp_path.name,
is_gzip=False)
with open(local_path, mode='rt') as file:
lines = file.readlines()
self.assertEqual(
['"col1","col2"\n', '"foo","bar"\n', '"ping","pong"\n'],
sorted(lines))
def test_table_export_file_gzip(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
writer.writerow({'col1': 'foo', 'col2': 'bar'})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
temp_path = tempfile.TemporaryDirectory()
local_path = self.tables.export_to_file(table_id=table_id,
path_name=temp_path.name,
is_gzip=True)
with open(local_path, mode='rt') as file:
lines = file.readlines()
self.assertEqual(
['"col1","col2"\n', '"foo","bar"\n', '"ping","pong"\n'],
sorted(lines))
def test_table_export_sliced(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'ping', 'col2': 'pong'})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
table_info = self.tables.detail(table_id)
with self.subTest():
self.assertEqual(table_id, table_info['id'])
with self.subTest():
self.assertEqual(1, table_info['rowsCount'])
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({'col1': 'foo', 'col2': 'bar'})
os.close(file)
self.tables.load(table_id=table_id,
file_path=path,
is_incremental=True)
temp_path = tempfile.TemporaryDirectory()
local_path = self.tables.export_to_file(table_id=table_id,
path_name=temp_path.name)
with open(local_path, mode='rt') as file:
lines = file.readlines()
self.assertEqual(
['"col1","col2"\n', '"foo","bar"\n', '"ping","pong"\n'],
sorted(lines))
def test_table_columns(self):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(
csv_file,
fieldnames=['col1', 'col2', 'col3', 'col4'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow({
'col1': 'ping',
'col2': 'pong',
'col3': 'king',
'col4': 'kong'
})
os.close(file)
table_id = self.tables.create(name='some-table',
file_path=path,
bucket_id='in.c-py-test-tables')
temp_path = tempfile.TemporaryDirectory()
local_path = self.tables.export_to_file(table_id=table_id,
path_name=temp_path.name,
is_gzip=False,
columns=['col3', 'col2'])
with open(local_path, mode='rt') as file:
lines = file.readlines()
self.assertEqual(['"col3","col2"\n', '"king","pong"\n'], sorted(lines))
class TestWorkspaces(unittest.TestCase):
def setUp(self):
self.workspaces = Workspaces(os.getenv('KBC_TEST_API_URL'),
os.getenv('KBC_TEST_TOKEN'))
self.buckets = Buckets(os.getenv('KBC_TEST_API_URL'),
os.getenv('KBC_TEST_TOKEN'))
self.jobs = Jobs(os.getenv('KBC_TEST_API_URL'),
os.getenv('KBC_TEST_TOKEN'))
self.tables = Tables(os.getenv('KBC_TEST_API_URL'),
os.getenv('KBC_TEST_TOKEN'))
self.files = Files(os.getenv('KBC_TEST_API_URL'),
os.getenv('KBC_TEST_TOKEN'))
try:
file_list = self.files.list(tags=['sapi-client-python-tests'])
for file in file_list:
self.files.delete(file['id'])
except exceptions.HTTPError as e:
if e.response.status_code != 404:
raise
try:
self.buckets.delete('in.c-py-test-buckets', force=True)
except exceptions.HTTPError as e:
if e.response.status_code != 404:
raise
# https://github.com/boto/boto3/issues/454
warnings.simplefilter("ignore", ResourceWarning)
def tearDown(self):
try:
if hasattr(self, 'workspace_id'):
self.workspaces.delete(self.workspace_id)
except exceptions.HTTPError as e:
if e.response.status_code != 404:
raise
try:
self.buckets.delete('in.c-py-test-tables', force=True)
except exceptions.HTTPError as e:
if e.response.status_code != 404:
raise
def test_create_workspace(self):
workspace = self.workspaces.create()
self.workspace_id = workspace['id']
with self.subTest():
self.assertTrue('id' in workspace)
with self.subTest():
self.assertTrue('type' in workspace)
self.assertTrue(workspace['type'] in ['table', 'file'])
with self.subTest():
self.assertTrue('name' in workspace)
with self.subTest():
self.assertTrue('component' in workspace)
with self.subTest():
self.assertTrue('configurationId' in workspace)
with self.subTest():
self.assertTrue('created' in workspace)
with self.subTest():
self.assertTrue('connection' in workspace)
with self.subTest():
self.assertTrue('backend' in workspace['connection'])
with self.subTest():
self.assertTrue('creatorToken' in workspace)
def test_load_tables_to_workspace(self):
bucket_id = self.buckets.create('py-test-tables')['id']
table1_id = self.__create_table(bucket_id, 'test-table-1', {
'col1': 'ping',
'col2': 'pong'
})
table2_id = self.__create_table(bucket_id, 'test-table-2', {
'col1': 'king',
'col2': 'kong'
})
workspace = self.workspaces.create()
self.workspace_id = workspace['id']
job = self.workspaces.load_tables(workspace['id'], {
table1_id: 'destination_1',
table2_id: 'destination_2'
})
self.jobs.block_until_completed(job['id'])
job = self.tables.create_raw(bucket_id,
'back-and-forth-table',
data_workspace_id=workspace['id'],
data_table_name='destination_1')
self.jobs.block_until_completed(job['id'])
new_table = self.tables.detail(bucket_id + '.back-and-forth-table')
self.assertEqual('back-and-forth-table', new_table['name'])
# test load files into an abs workspace
def test_load_files_to_workspace(self):
if (os.getenv('SKIP_ABS_TEST')):
self.skipTest(
'Skipping ABS test because env var SKIP_ABS_TESTS was set')
# put a test file to storage
file, path = tempfile.mkstemp(prefix='sapi-test')
os.write(file, bytes('fooBar', 'utf-8'))
os.close(file)
# We'll put 2 files with the same tag to test multiple results
file1_id = self.files.upload_file(
path, tags=['sapi-client-python-tests', 'file1'])
file2_id = self.files.upload_file(
path, tags=['sapi-client-python-tests', 'file2'])
file1 = self.files.detail(file1_id)
file2 = self.files.detail(file2_id)
# create a workspace and load the file to it
workspace = self.workspaces.create('abs')
self.workspace_id = workspace['id']
self.workspaces.load_files(workspace['id'], {
'tags': ['sapi-client-python-tests'],
'destination': 'data/in/files'
})
# assert that the file was loaded to the workspace
blob_service_client = BlobServiceClient.from_connection_string(
workspace['connection']['connectionString'])
blob_client_1 = blob_service_client.get_blob_client(
container=workspace['connection']['container'],
blob='data/in/files/%s/%s' % (file1['name'], str(file1['id'])))
self.assertEqual(
'fooBar',
blob_client_1.download_blob().readall().decode('utf-8'))
blob_client_2 = blob_service_client.get_blob_client(
container=workspace['connection']['container'],
blob='data/in/files/%s/%s' % (file2['name'], str(file2['id'])))
self.assertEqual(
'fooBar',
blob_client_2.download_blob().readall().decode('utf-8'))
# now let's test that we can use the 'and' operator. in this case file2 should not get loaded
self.workspaces.load_files(
workspace['id'], {
'tags': ['sapi-client-python-tests', 'file1'],
'operator': 'and',
'destination': 'data/in/and_files'
})
# file 1 should be there
blob_client_1 = blob_service_client.get_blob_client(
container=workspace['connection']['container'],
blob='data/in/and_files/%s/%s' % (file1['name'], str(file1['id'])))
self.assertEqual(
'fooBar',
blob_client_1.download_blob().readall().decode('utf-8'))
# file 2 should not
blob_client_2 = blob_service_client.get_blob_client(
container=workspace['connection']['container'],
blob='data/in/and_files/%s/%s' % (file2['name'], str(file2['id'])))
with self.assertRaises(ResourceNotFoundError) as context:
blob_client_2.download_blob().readall().decode('utf-8')
self.assertTrue(
'The specified blob does not exist' in str(context.exception))
def __create_table(self, bucket_id, table_name, row):
file, path = tempfile.mkstemp(prefix='sapi-test')
with open(path, 'w') as csv_file:
writer = csv.DictWriter(csv_file,
fieldnames=['col1', 'col2'],
lineterminator='\n',
delimiter=',',
quotechar='"')
writer.writeheader()
writer.writerow(row)
return self.tables.create(name=table_name,
file_path=path,
bucket_id=bucket_id)