def _copy_single_object(
self,
gcs_hook: GoogleCloudStorageHook,
sftp_hook: SFTPHook,
source_path: str,
destination_object: str,
) -> None:
"""
Helper function to copy single object.
"""
self.log.info(
"Executing copy of %s to gs://%s/%s",
source_path,
self.destination_bucket,
destination_object,
)
with NamedTemporaryFile("w") as tmp:
sftp_hook.retrieve_file(source_path, tmp.name)
gcs_hook.upload(
bucket_name=self.destination_bucket,
object_name=destination_object,
filename=tmp.name,
mime_type=self.mime_type,
)
if self.move_object:
self.log.info("Executing delete of %s", source_path)
sftp_hook.delete_file(source_path)
def execute(self, context):
sftp_hook = SFTPHook(ftp_conn_id=self.sftp_conn_id)
s3_hook = S3Hook(self.s3_conn_id)
sftp_files = sftp_hook.list_directory(self.sftp_path)
filtered_files_by_extensions = [
key for key in sftp_files
if key.lower().endswith(self.file_extensions)
]
for sftp_file in filtered_files_by_extensions:
with NamedTemporaryFile("w") as f:
sftp_hook.retrieve_file(f'{self.sftp_path}/{sftp_file}',
f.name)
s3_key = self.get_s3_key(f'{self.s3_prefix}/{sftp_file}')
s3_hook.load_file(filename=f.name,
key=s3_key,
bucket_name=self.s3_bucket,
replace=True)
# Add the empty _SUCCESS file to indicate the task is done successfully
s3_key = self.get_s3_key(f'{self.s3_prefix}/_SUCCESS')
s3_hook.load_string('',
key=s3_key,
bucket_name=self.s3_bucket,
replace=True)
def execute(self, context):
source_hook = SFTPHook(ftp_conn_id=self.source_conn_id)
source_files = source_hook.list_directory(self.source_path)
for target in self.target_full_path:
target_connection = target[0]
target_path = target[1]
self.log.info(
f"Beginning transfer to SFTP site {target_connection} and directory {target_path}"
)
target_hook = SFTPHook(ftp_conn_id=target_connection)
target_files = target_hook.list_directory(target_path)
for file in source_files:
if self.filter_function is None or self.filter_function(file):
if self.overwrite_target is True or file not in target_files:
source_hook.retrieve_file(
op.join(self.source_path, file),
op.join(self.work_path, file))
self.log.info(
"Downloaded the file %s from the source SFTP",
file)
try:
target_hook.store_file(
op.join(target_path, file),
op.join(self.work_path, file))
self.log.info(
"Uploaded the file %s to the destination SFTP",
file)
finally:
os.remove(os.path.join(self.work_path, file))
def moveFromSourceToLocal(**kwargs):
"""
Use information from the dag_run passed in by the filefinder DAG to start pulling down a ready file.
"""
# Variablelize (my word) the dag_run config needed for this step.
# This might be a good candidate for externalizing
sftpConn = kwargs['dag_run'].conf['SFTP_Connection_Name']
sourceFullPath = kwargs['dag_run'].conf['File_Name']
# Strip the ".ready" from the filename as we get the basename of the file
fileName = os.path.basename(kwargs['dag_run'].conf['File_Name']).replace(
'.ready', '')
destFullPath = os.path.join(LOCAL_LANDING_PATH, fileName)
sftpHook = SFTPHook(ftp_conn_id=sftpConn)
conn = sftpHook.get_conn()
initialMD5sum = getMD5sumRemote(conn, sourceFullPath)
logging.info('Initial MD5Sum: {}'.format(initialMD5sum))
sftpHook.retrieve_file(sourceFullPath, destFullPath)
currentMD5sum = getMD5sumLocal(destFullPath)
logging.info('currentMD5Sum: {}'.format(currentMD5sum))
if initialMD5sum != currentMD5sum:
logging.error(
'MD5Sum mismatch. Initial: {} Post-Transfer: {}'.format(
initialMD5sum, currentMD5sum))
raise Exception(
'MD5Sum values before and after transfer do not match. Possible transfer issue. Initial: {} Post-Transfer: {}'
.format(initialMD5sum, currentMD5sum))
def sftp_to_pg(**kwargs):
today = datetime.date.today().strftime('%y%m%d')
conn = SFTPHook('sftp_cityftp')
files = conn.describe_directory('/Home/IET/PNC')
file_name = [
fn for fn in files.keys()
if fn.startswith(f"tls.cityofdetroit.out.{today}")
][0]
conn.retrieve_file(f"/Home/IET/PNC/{file_name}", f"/tmp/{file_name}")
pg_conn = PostgresHook('etl_postgres')
pg_conn.run("truncate table escrow.escrow")
pg_conn.run(
f"copy escrow.escrow from '/tmp/{file_name}' (FORMAT CSV, HEADER FALSE) "
)
def moveFileFromSourceToTemp(**context):
sourceHook = SFTPHook(ftp_conn_id=SOURCE_SFTP_CONN_ID)
fileList = context['task_instance'].xcom_pull(key = 'file_name', task_ids = TASKNAME)
for file in fileList:
sourceFullPath = os.path.join(SOURCE_FILEPATH, file)
transferFullPath = os.path.join(TRANSFER_FILEPATH, file)
try:
sourceHook.retrieve_file(sourceFullPath, transferFullPath)
except:
raise Exception('Unable to move {} from remote {} path to local {} path'.format(file, sourceFullPath, transferFullPath))
try:
sourceHook.delete_file(sourceFullPath)
except:
raise Exception('Unable to delete {} from the source machine'.format(sourceFullPath))
logging.info('Transferred {} to local machine {}'.format(file, TRANSFER_FILEPATH))
class SFTPHookTest(unittest.TestCase):
def setUp(self):
configuration.load_test_config()
self.hook = SFTPHook()
os.makedirs(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
with open(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS), 'a') as f:
f.write('Test file')
def test_get_conn(self):
output = self.hook.get_conn()
self.assertEqual(type(output), pysftp.Connection)
def test_close_conn(self):
self.hook.conn = self.hook.get_conn()
self.assertTrue(self.hook.conn is not None)
self.hook.close_conn()
self.assertTrue(self.hook.conn is None)
def test_describe_directory(self):
output = self.hook.describe_directory(TMP_PATH)
self.assertTrue(TMP_DIR_FOR_TESTS in output)
def test_list_directory(self):
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [])
def test_create_and_delete_directory(self):
new_dir_name = 'new_dir'
self.hook.create_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name in output)
self.hook.delete_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name not in output)
def test_store_retrieve_and_delete_file(self):
self.hook.store_file(
remote_full_path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS,
TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [TMP_FILE_FOR_TESTS])
retrieved_file_name = 'retrieved.txt'
self.hook.retrieve_file(
remote_full_path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS,
TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, retrieved_file_name))
self.assertTrue(retrieved_file_name in os.listdir(TMP_PATH))
os.remove(os.path.join(TMP_PATH, retrieved_file_name))
self.hook.delete_file(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [])
def test_get_mod_time(self):
self.hook.store_file(
remote_full_path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS,
TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
output = self.hook.get_mod_time(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
self.assertEqual(len(output), 14)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_default(self, get_connection):
connection = Connection(login='******', host='host')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_enabled(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"no_host_key_check": true}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, True)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_disabled(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"no_host_key_check": false}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_disabled_for_all_but_true(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"no_host_key_check": "foo"}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_ignore(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"ignore_hostkey_verification": true}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, True)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_no_ignore(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"ignore_hostkey_verification": false}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
def tearDown(self):
shutil.rmtree(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
os.remove(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
class TestSFTPHook(unittest.TestCase):
@provide_session
def update_connection(self, login, session=None):
connection = (session.query(Connection).filter(
Connection.conn_id == "sftp_default").first())
old_login = connection.login
connection.login = login
session.commit()
return old_login
def setUp(self):
self.old_login = self.update_connection(SFTP_CONNECTION_USER)
self.hook = SFTPHook()
os.makedirs(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, SUB_DIR))
with open(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS), 'a') as file:
file.write('Test file')
with open(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, SUB_DIR,
TMP_FILE_FOR_TESTS), 'a') as file:
file.write('Test file')
def test_get_conn(self):
output = self.hook.get_conn()
self.assertEqual(type(output), pysftp.Connection)
def test_close_conn(self):
self.hook.conn = self.hook.get_conn()
self.assertTrue(self.hook.conn is not None)
self.hook.close_conn()
self.assertTrue(self.hook.conn is None)
def test_describe_directory(self):
output = self.hook.describe_directory(TMP_PATH)
self.assertTrue(TMP_DIR_FOR_TESTS in output)
def test_list_directory(self):
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [SUB_DIR])
def test_create_and_delete_directory(self):
new_dir_name = 'new_dir'
self.hook.create_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name in output)
self.hook.delete_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name not in output)
def test_create_and_delete_directories(self):
base_dir = "base_dir"
sub_dir = "sub_dir"
new_dir_path = os.path.join(base_dir, sub_dir)
self.hook.create_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_path))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(base_dir in output)
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, base_dir))
self.assertTrue(sub_dir in output)
self.hook.delete_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_path))
self.hook.delete_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, base_dir))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_path not in output)
self.assertTrue(base_dir not in output)
def test_store_retrieve_and_delete_file(self):
self.hook.store_file(
remote_full_path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS,
TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [SUB_DIR, TMP_FILE_FOR_TESTS])
retrieved_file_name = 'retrieved.txt'
self.hook.retrieve_file(
remote_full_path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS,
TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, retrieved_file_name))
self.assertTrue(retrieved_file_name in os.listdir(TMP_PATH))
os.remove(os.path.join(TMP_PATH, retrieved_file_name))
self.hook.delete_file(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [SUB_DIR])
def test_get_mod_time(self):
self.hook.store_file(
remote_full_path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS,
TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
output = self.hook.get_mod_time(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
self.assertEqual(len(output), 14)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_default(self, get_connection):
connection = Connection(login='******', host='host')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_enabled(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"no_host_key_check": true}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, True)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_disabled(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"no_host_key_check": false}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_disabled_for_all_but_true(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"no_host_key_check": "foo"}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_ignore(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"ignore_hostkey_verification": true}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, True)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_no_ignore(self, get_connection):
connection = Connection(login='******',
host='host',
extra='{"ignore_hostkey_verification": false}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@parameterized.expand([
(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS), True),
(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS), True),
(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS + "abc"), False),
(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, "abc"), False),
])
def test_path_exists(self, path, exists):
result = self.hook.path_exists(path)
self.assertEqual(result, exists)
@parameterized.expand([
("test/path/file.bin", None, None, True),
("test/path/file.bin", "test", None, True),
("test/path/file.bin", "test/", None, True),
("test/path/file.bin", None, "bin", True),
("test/path/file.bin", "test", "bin", True),
("test/path/file.bin", "test/", "file.bin", True),
("test/path/file.bin", None, "file.bin", True),
("test/path/file.bin", "diff", None, False),
("test/path/file.bin", "test//", None, False),
("test/path/file.bin", None, ".txt", False),
("test/path/file.bin", "diff", ".txt", False),
])
def test_path_match(self, path, prefix, delimiter, match):
result = self.hook._is_path_match(path=path,
prefix=prefix,
delimiter=delimiter)
self.assertEqual(result, match)
def test_get_tree_map(self):
tree_map = self.hook.get_tree_map(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
files, dirs, unknowns = tree_map
self.assertEqual(files, [
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, SUB_DIR,
TMP_FILE_FOR_TESTS)
])
self.assertEqual(dirs,
[os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, SUB_DIR)])
self.assertEqual(unknowns, [])
def tearDown(self):
shutil.rmtree(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
os.remove(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
self.update_connection(self.old_login)
class SFTPHookTest(unittest.TestCase):
def setUp(self):
configuration.load_test_config()
self.hook = SFTPHook()
os.makedirs(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
with open(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS), 'a') as f:
f.write('Test file')
def test_get_conn(self):
output = self.hook.get_conn()
self.assertEqual(type(output), pysftp.Connection)
def test_close_conn(self):
self.hook.conn = self.hook.get_conn()
self.assertTrue(self.hook.conn is not None)
self.hook.close_conn()
self.assertTrue(self.hook.conn is None)
def test_describe_directory(self):
output = self.hook.describe_directory(TMP_PATH)
self.assertTrue(TMP_DIR_FOR_TESTS in output)
def test_list_directory(self):
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [])
def test_create_and_delete_directory(self):
new_dir_name = 'new_dir'
self.hook.create_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name in output)
self.hook.delete_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name not in output)
def test_store_retrieve_and_delete_file(self):
self.hook.store_file(
remote_full_path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS,
TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [TMP_FILE_FOR_TESTS])
retrieved_file_name = 'retrieved.txt'
self.hook.retrieve_file(
remote_full_path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS,
TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, retrieved_file_name))
self.assertTrue(retrieved_file_name in os.listdir(TMP_PATH))
os.remove(os.path.join(TMP_PATH, retrieved_file_name))
self.hook.delete_file(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [])
def test_get_mod_time(self):
self.hook.store_file(
remote_full_path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS,
TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
output = self.hook.get_mod_time(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
self.assertEqual(len(output), 14)
def tearDown(self):
shutil.rmtree(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
os.remove(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
class SFTPHookTest(unittest.TestCase):
def setUp(self):
configuration.load_test_config()
self.hook = SFTPHook()
os.makedirs(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
with open(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS), 'a') as f:
f.write('Test file')
def test_get_conn(self):
output = self.hook.get_conn()
self.assertEqual(type(output), pysftp.Connection)
def test_close_conn(self):
self.hook.conn = self.hook.get_conn()
self.assertTrue(self.hook.conn is not None)
self.hook.close_conn()
self.assertTrue(self.hook.conn is None)
def test_describe_directory(self):
output = self.hook.describe_directory(TMP_PATH)
self.assertTrue(TMP_DIR_FOR_TESTS in output)
def test_list_directory(self):
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [])
def test_create_and_delete_directory(self):
new_dir_name = 'new_dir'
self.hook.create_directory(os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name in output)
self.hook.delete_directory(os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name not in output)
def test_store_retrieve_and_delete_file(self):
self.hook.store_file(
remote_full_path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS)
)
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [TMP_FILE_FOR_TESTS])
retrieved_file_name = 'retrieved.txt'
self.hook.retrieve_file(
remote_full_path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, retrieved_file_name)
)
self.assertTrue(retrieved_file_name in os.listdir(TMP_PATH))
os.remove(os.path.join(TMP_PATH, retrieved_file_name))
self.hook.delete_file(path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [])
def test_get_mod_time(self):
self.hook.store_file(
remote_full_path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS)
)
output = self.hook.get_mod_time(path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
self.assertEqual(len(output), 14)
def tearDown(self):
shutil.rmtree(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
os.remove(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
class SFTPHookTest(unittest.TestCase):
def setUp(self):
configuration.load_test_config()
self.hook = SFTPHook()
os.makedirs(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
with open(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS), 'a') as f:
f.write('Test file')
def test_get_conn(self):
output = self.hook.get_conn()
self.assertEqual(type(output), pysftp.Connection)
def test_close_conn(self):
self.hook.conn = self.hook.get_conn()
self.assertTrue(self.hook.conn is not None)
self.hook.close_conn()
self.assertTrue(self.hook.conn is None)
def test_describe_directory(self):
output = self.hook.describe_directory(TMP_PATH)
self.assertTrue(TMP_DIR_FOR_TESTS in output)
def test_list_directory(self):
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [])
def test_create_and_delete_directory(self):
new_dir_name = 'new_dir'
self.hook.create_directory(os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name in output)
self.hook.delete_directory(os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, new_dir_name))
output = self.hook.describe_directory(
os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertTrue(new_dir_name not in output)
def test_store_retrieve_and_delete_file(self):
self.hook.store_file(
remote_full_path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS)
)
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [TMP_FILE_FOR_TESTS])
retrieved_file_name = 'retrieved.txt'
self.hook.retrieve_file(
remote_full_path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, retrieved_file_name)
)
self.assertTrue(retrieved_file_name in os.listdir(TMP_PATH))
os.remove(os.path.join(TMP_PATH, retrieved_file_name))
self.hook.delete_file(path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
output = self.hook.list_directory(
path=os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
self.assertEqual(output, [])
def test_get_mod_time(self):
self.hook.store_file(
remote_full_path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS),
local_full_path=os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS)
)
output = self.hook.get_mod_time(path=os.path.join(
TMP_PATH, TMP_DIR_FOR_TESTS, TMP_FILE_FOR_TESTS))
self.assertEqual(len(output), 14)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_default(self, get_connection):
connection = Connection(login='******', host='host')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_enabled(self, get_connection):
connection = Connection(
login='******', host='host',
extra='{"no_host_key_check": true}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, True)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_disabled(self, get_connection):
connection = Connection(
login='******', host='host',
extra='{"no_host_key_check": false}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_disabled_for_all_but_true(self, get_connection):
connection = Connection(
login='******', host='host',
extra='{"no_host_key_check": "foo"}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_ignore(self, get_connection):
connection = Connection(
login='******', host='host',
extra='{"ignore_hostkey_verification": true}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, True)
@mock.patch('airflow.contrib.hooks.sftp_hook.SFTPHook.get_connection')
def test_no_host_key_check_no_ignore(self, get_connection):
connection = Connection(
login='******', host='host',
extra='{"ignore_hostkey_verification": false}')
get_connection.return_value = connection
hook = SFTPHook()
self.assertEqual(hook.no_host_key_check, False)
def tearDown(self):
shutil.rmtree(os.path.join(TMP_PATH, TMP_DIR_FOR_TESTS))
os.remove(os.path.join(TMP_PATH, TMP_FILE_FOR_TESTS))
def get_file(**kwargs):
conn = SFTPHook('sftp_novatus')
d = conn.describe_directory('/outgoing/')
conn.retrieve_file(path, f"/tmp/{file_name}")
