def test_strip_newlines_local_custom_mini(self):
test_filename_uri = './tap_spreadsheets_anywhere/test/sample_with_bad_newlines.csv'
iterator = get_row_iterator(TEST_TABLE_SPEC['tables'][0],
test_filename_uri)
for row in iterator:
self.assertTrue(
row['id'].isnumeric(),
"Parsed ID is not a number for: {}".format(row['id']))
def test_handle_newlines_local_json(self):
test_filename_uri = './tap_spreadsheets_anywhere/test/sample.json'
iterator = get_row_iterator(TEST_TABLE_SPEC['tables'][3],
test_filename_uri)
for row in iterator:
self.assertTrue(
isinstance(row['id'], float) or isinstance(row['id'], int),
"Parsed ID is not a number for: {}".format(row['id']))
def test_renamed_https_object(self):
table_spec = TEST_TABLE_SPEC['tables'][6]
modified_since = dateutil.parser.parse(table_spec['start_date'])
target_files = file_utils.get_matching_objects(table_spec,
modified_since)
assert len(target_files) == 1
target_uri = table_spec['path'] + '/' + table_spec['pattern']
iterator = get_row_iterator(TEST_TABLE_SPEC['tables'][6], target_uri)
row = next(iterator)
self.assertTrue(len(row) > 1, "Not able to read a row.")
def test_https_bucket(self):
table_spec = TEST_TABLE_SPEC['tables'][4]
modified_since = dateutil.parser.parse(table_spec['start_date'])
target_files = file_utils.get_matching_objects(table_spec,
modified_since)
assert len(target_files) == 1
target_uri = table_spec['path'] + '/' + target_files[0]["key"]
iterator = get_row_iterator(TEST_TABLE_SPEC['tables'][4], target_uri)
row = next(iterator)
self.assertTrue(int(row['id']) > 0, row['id'] + " was not positive")
def test_indirect_https_bucket(self):
table_spec = TEST_TABLE_SPEC['tables'][5]
modified_since = dateutil.parser.parse(table_spec['start_date'])
target_files = file_utils.get_matching_objects(table_spec,
modified_since)
assert len(target_files) == 1
target_uri = table_spec['path'] + '/' + table_spec['pattern']
iterator = get_row_iterator(TEST_TABLE_SPEC['tables'][5], target_uri)
row = next(iterator)
self.assertTrue(row['1976'] == '1976',
"Row did not contain expected data")
