def test__match_headers(self):
self.kwargs['field_dict'] = OrderedDict([('Sno', 'BIGINT'),
('Some,Text', 'STRING')])
assert S3ToHiveOperator(**self.kwargs)._match_headers(
['Sno', 'Some,Text']), "Header row doesnt match expected value"
# Testing with different column order
assert not S3ToHiveOperator(**self.kwargs)._match_headers(
['Some,Text', 'Sno']), "Header row doesnt match expected value"
# Testing with extra column in header
assert not S3ToHiveOperator(**self.kwargs)._match_headers([
'Sno', 'Some,Text', 'ExtraColumn'
]), "Header row doesnt match expected value"
def test__get_top_row_as_list(self):
self.kwargs['delimiter'] = '\t'
fn_txt = self._get_fn('.txt', True)
header_list = S3ToHiveOperator(
**self.kwargs)._get_top_row_as_list(fn_txt)
self.assertEqual(header_list, ['Sno', 'Some,Text'],
msg="Top row from file doesnt matched expected value")
self.kwargs['delimiter'] = ','
header_list = S3ToHiveOperator(
**self.kwargs)._get_top_row_as_list(fn_txt)
self.assertEqual(header_list, ['Sno\tSome', 'Text'],
msg="Top row from file doesnt matched expected value")
def test_execute(self, mock_hiveclihook):
conn = boto3.client('s3')
conn.create_bucket(Bucket='bucket')
# Testing txt, zip, bz2 files with and without header row
for (ext, has_header) in product(['.txt', '.gz', '.bz2', '.GZ'],
[True, False]):
self.kwargs['headers'] = has_header
self.kwargs['check_headers'] = has_header
logging.info("Testing %s format %s header", ext,
'with' if has_header else 'without')
self.kwargs['input_compressed'] = ext.lower() != '.txt'
self.kwargs['s3_key'] = 's3://bucket/' + self.s3_key + ext
ip_fn = self._get_fn(ext, self.kwargs['headers'])
op_fn = self._get_fn(ext, False)
# Upload the file into the Mocked S3 bucket
conn.upload_file(ip_fn, 'bucket', self.s3_key + ext)
# file parameter to HiveCliHook.load_file is compared
# against expected file output
mock_hiveclihook(
).load_file.side_effect = lambda *args, **kwargs: self.assertTrue(
self._check_file_equality(args[0], op_fn, ext),
f'{ext} output file not as expected',
)
# Execute S3ToHiveTransfer
s32hive = S3ToHiveOperator(**self.kwargs)
s32hive.execute(None)
def test__delete_top_row_and_compress(self):
s32hive = S3ToHiveOperator(**self.kwargs)
# Testing gz file type
fn_txt = self._get_fn('.txt', True)
gz_txt_nh = s32hive._delete_top_row_and_compress(
fn_txt, '.gz', self.tmp_dir)
fn_gz = self._get_fn('.gz', False)
assert self._check_file_equality(
gz_txt_nh, fn_gz, '.gz'), "gz Compressed file not as expected"
# Testing bz2 file type
bz2_txt_nh = s32hive._delete_top_row_and_compress(
fn_txt, '.bz2', self.tmp_dir)
fn_bz2 = self._get_fn('.bz2', False)
assert self._check_file_equality(
bz2_txt_nh, fn_bz2, '.bz2'), "bz2 Compressed file not as expected"
def test_execute_with_select_expression(self, mock_hiveclihook):
conn = boto3.client('s3')
conn.create_bucket(Bucket='bucket')
select_expression = "SELECT * FROM S3Object s"
bucket = 'bucket'
# Only testing S3ToHiveTransfer calls S3Hook.select_key with
# the right parameters and its execute method succeeds here,
# since Moto doesn't support select_object_content as of 1.3.2.
for (ext, has_header) in product(['.txt', '.gz', '.GZ'],
[True, False]):
input_compressed = ext.lower() != '.txt'
key = self.s3_key + ext
self.kwargs['check_headers'] = False
self.kwargs['headers'] = has_header
self.kwargs['input_compressed'] = input_compressed
self.kwargs['select_expression'] = select_expression
self.kwargs['s3_key'] = f's3://{bucket}/{key}'
ip_fn = self._get_fn(ext, has_header)
# Upload the file into the Mocked S3 bucket
conn.upload_file(ip_fn, bucket, key)
input_serialization = {'CSV': {'FieldDelimiter': self.delimiter}}
if input_compressed:
input_serialization['CompressionType'] = 'GZIP'
if has_header:
input_serialization['CSV']['FileHeaderInfo'] = 'USE'
# Confirm that select_key was called with the right params
with mock.patch(
'airflow.providers.amazon.aws.hooks.s3.S3Hook.select_key',
return_value="") as mock_select_key:
# Execute S3ToHiveTransfer
s32hive = S3ToHiveOperator(**self.kwargs)
s32hive.execute(None)
mock_select_key.assert_called_once_with(
bucket_name=bucket,
key=key,
expression=select_expression,
input_serialization=input_serialization,
)
def test_bad_parameters(self):
self.kwargs['check_headers'] = True
self.kwargs['headers'] = False
with pytest.raises(AirflowException, match="To check_headers.*"):
S3ToHiveOperator(**self.kwargs)