class DataValidationTest(unittest.TestCase):
validator = None
def setUp(self):
self.validator = DataFileValidator()
self.base_dir = os.path.dirname(os.path.realpath(__file__))
self.invalid_file_yaml = os.path.join(self.base_dir,
'test_data/invalid_file.yaml')
self.valid_file_yaml = os.path.join(self.base_dir,
'test_data/valid_file.yaml')
self.valid_file_json = os.path.join(self.base_dir,
'test_data/valid_file.json')
self.invalid_file_json = os.path.join(self.base_dir,
'test_data/invalid_file.json')
self.valid_file_error_percent_yaml = os.path.join(
self.base_dir, 'test_data/valid_data_with_error.yaml')
self.invalid_syntax_data_file = os.path.join(
self.base_dir, 'test_data/invalid_data_file.yaml')
self.invalid_parser_file = os.path.join(
self.base_dir, 'test_data/invalid_parser_file.yaml')
self.valid_custom_file = os.path.join(
self.base_dir, 'test_data/valid_file_custom.yaml')
def test_no_file_path_supplied(self):
try:
self.validator.validate(file_path=None)
except LookupError as le:
assert (le)
def test_valid_yaml_file(self):
print('___DATA_VALIDATION: Testing valid yaml submission___')
is_valid = self.validator.validate(file_path=self.valid_file_yaml)
self.validator.print_errors(self.valid_file_yaml)
self.assertEqual(is_valid, True)
def test_invalid_yaml_file(self):
print('___DATA_VALIDATION: Testing invalid yaml submission___')
self.assertEqual(
self.validator.validate(file_path=self.invalid_file_yaml), False)
self.validator.print_errors(self.invalid_file_yaml)
def test_valid_file_with_percent_errors(self):
print('___DATA_VALIDATION: Testing valid yaml percent error ___')
self.assertEqual(
self.validator.validate(
file_path=self.valid_file_error_percent_yaml), False)
self.validator.print_errors(self.valid_file_error_percent_yaml)
def test_valid_json_file(self):
print('___DATA_VALIDATION: Testing valid json submission___')
is_valid = self.validator.validate(file_path=self.valid_file_json)
self.validator.print_errors(self.valid_file_json)
self.assertEqual(is_valid, True)
self.validator.print_errors(self.valid_file_json)
def test_invalid_json_file(self):
print('___DATA_VALIDATION: Testing invalid json submission___')
self.assertEqual(
self.validator.validate(file_path=self.invalid_file_json), False)
self.validator.print_errors(self.invalid_file_json)
def test_load_data_with_custom_data_type(self):
self.validator = DataFileValidator()
custom_schema_path = os.path.join(self.base_dir,
'test_data/custom_data_schema.json')
self.validator.load_custom_schema('different', custom_schema_path)
self.assertTrue('different' in self.validator.custom_data_schemas)
self.assertTrue(
self.validator.validate(file_path=self.valid_custom_file))
def test_load_invalid_custom_schema(self):
self.validator.custom_data_schemas = {}
print('Loading invalid schema')
try:
self.validator.load_custom_schema('different')
except UnsupportedDataSchemaException as udse:
self.assertTrue(
udse.message ==
"There is no schema defined for the 'different' data type.")
self.assertTrue(udse.message == udse.__unicode__())
def test_load_invalid_data_file(self):
print('Loading invalid data file')
self.assertFalse(
self.validator.validate(file_path=self.invalid_syntax_data_file))
self.assertTrue(
self.validator.has_errors(self.invalid_syntax_data_file))
self.assertTrue(
len(self.validator.get_messages(self.invalid_syntax_data_file)) ==
1)
self.validator.print_errors(self.invalid_syntax_data_file)
for message in self.validator.get_messages(
self.invalid_syntax_data_file):
self.assertTrue(
message.message.index("There was a problem parsing the file.")
== 0)
def test_invalid_parser_yaml_file(self):
print('___DATA_VALIDATION: Testing invalid parser yaml submission___')
self.assertEqual(
self.validator.validate(file_path=self.invalid_parser_file), False)
self.validator.print_errors(self.invalid_parser_file)
def test_ioerror_yaml_file(self):
print('___DATA_VALIDATION: Testing ioerror yaml submission___')
self.assertEqual(
self.validator.validate(file_path=self.valid_file_yaml[:-1]),
False)
self.validator.print_errors(self.valid_file_yaml[:-1])
'data_file'] if directory else doc['data_file']
# Just try to load YAML data file without validating schema.
# Script will terminate with an exception if there is a problem.
contents = yaml.load(open(data_file_path, 'r'), Loader=Loader)
# Validate the YAML data file if validator imported.
if not validator_imported:
print('%s is valid YAML.' % data_file_path)
else:
data_file_validator = DataFileValidator()
is_valid_data_file = data_file_validator.validate(
file_path=data_file_path, data=contents)
if not is_valid_data_file:
print('%s is invalid HEPData YAML.' % data_file_path)
data_file_validator.print_errors(data_file_path)
else:
# Check that the length of the 'values' list is consistent for
# each of the independent_variables and dependent_variables.
indep_count = [
len(indep['values'])
for indep in contents['independent_variables']
]
dep_count = [
len(dep['values'])
for dep in contents['dependent_variables']
]
if len(set(indep_count + dep_count)
) > 1: # if more than one unique count
print(
"%s has inconsistent length of 'values' list: " %
from hepdata_validator.data_file_validator import DataFileValidator
import argparse
parser = argparse.ArgumentParser(description='Validate yaml files.')
parser.add_argument('-filename',dest='filename', type=str, help='file to check')
args = parser.parse_args()
data_file_validator = DataFileValidator()
# the validate method takes a string representing the file path.
data_file_validator.validate(file_path=args.filename)
# if there are any error messages, they are retrievable through this call
data_file_validator.get_messages()
# the error messages can be printed
data_file_validator.print_errors('data.yaml')
class DataValidationTest(unittest.TestCase):
validator = None
def setUp(self):
self.validator = DataFileValidator()
self.base_dir = os.path.dirname(os.path.realpath(__file__))
self.invalid_file_yaml = os.path.join(
self.base_dir,
'test_data/invalid_file.yaml'
)
self.valid_file_yaml = os.path.join(
self.base_dir,
'test_data/valid_file.yaml'
)
self.valid_file_json = os.path.join(
self.base_dir,
'test_data/valid_file.json'
)
self.invalid_file_json = os.path.join(
self.base_dir,
'test_data/invalid_file.json')
self.valid_file_error_percent_yaml = os.path.join(
self.base_dir,
'test_data/valid_data_with_error.yaml'
)
self.invalid_syntax_data_file = os.path.join(
self.base_dir,
'test_data/invalid_data_file.yaml'
)
self.valid_custom_file = os.path.join(
self.base_dir,
'test_data/valid_file_custom.yaml')
def test_valid_yaml_file(self):
print '___DATA_VALIDATION: Testing valid yaml submission___'
is_valid = self.validator.validate(file_path=self.valid_file_yaml)
self.validator.print_errors(self.valid_file_yaml)
self.assertEqual(is_valid, True)
def test_invalid_yaml_file(self):
print '___DATA_VALIDATION: Testing invalid yaml submission___'
self.assertEqual(self.validator.validate(file_path=self.invalid_file_yaml),
False)
self.validator.print_errors(self.invalid_file_yaml)
def test_valid_file_with_percent_errors(self):
print '___DATA_VALIDATION: Testing valid yaml percent error ___'
self.assertEqual(self.validator.validate(file_path=self.valid_file_error_percent_yaml),
False)
self.validator.print_errors(self.valid_file_error_percent_yaml)
def test_valid_json_file(self):
print '___DATA_VALIDATION: Testing valid json submission___'
is_valid = self.validator.validate(file_path=self.valid_file_json)
self.validator.print_errors(self.valid_file_json)
self.assertEqual(is_valid, True)
self.validator.print_errors(self.valid_file_json)
def test_invalid_json_file(self):
print '___DATA_VALIDATION: Testing invalid json submission___'
self.assertEqual(self.validator.validate(file_path=self.invalid_file_json),
False)
self.validator.print_errors(self.invalid_file_json)
def test_load_data_with_custom_data_type(self):
self.validator = DataFileValidator()
custom_schema_path = os.path.join(self.base_dir, 'test_data/custom_data_schema.json')
self.validator.load_custom_schema('different', custom_schema_path)
self.assertTrue('different' in self.validator.custom_data_schemas)
self.assertTrue(self.validator.validate(file_path=self.valid_custom_file))
def test_load_invalid_custom_schema(self):
self.validator.custom_data_schemas = {}
print('Loading invalid schema')
try:
self.validator.load_custom_schema('different')
except UnsupportedDataSchemaException as udse:
self.assertTrue(udse.message == "There is no schema defined for the 'different' data type.")
self.assertTrue(udse.message == udse.__unicode__())
def test_load_invalid_data_file(self):
print('Loading invalid data file')
self.assertFalse(self.validator.validate(file_path=self.invalid_syntax_data_file))
self.assertTrue(self.validator.has_errors(self.invalid_syntax_data_file))
self.assertTrue(len(self.validator.get_messages(self.invalid_syntax_data_file)) == 1)
self.validator.print_errors(self.invalid_syntax_data_file)
for message in self.validator.get_messages(self.invalid_syntax_data_file):
self.assertTrue(message.message.index("There was a problem parsing the file.") == 0)
