def test_validate_metadata_datacite_bad_request(requests_mock):
"""Test validate_metadata.
Function should raise exception with correct error message if Metax fails
to generate datacite for dataset, for example when `publisher` attribute is
missing.
:param requests_mock: Mocker object
:returns: ``None``
"""
dataset_file = copy.deepcopy(TXT_FILE)
tests.utils.add_metax_dataset(requests_mock, files=[dataset_file])
# Mock datacite request response. Mocked response has status code 400, and
# response body contains error information.
requests_mock.get(tests.conftest.METAX_URL +
'/datasets/dataset_identifier?'
'dataset_format=datacite',
json={"detail": "Bad request"},
status_code=400)
# Try to validate invalid dataset
expected_error = "Datacite generation failed: Bad request"
with pytest.raises(InvalidDatasetMetadataError, match=expected_error):
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata_invalid_datacite(requests_mock):
"""Test validate_metadata.
Function should raise exception with correct error message for invalid
datacite where required attribute identifier is missing.
:param requests_mock: Mocker object
:returns: ``None``
"""
dataset_file = copy.deepcopy(TXT_FILE)
tests.utils.add_metax_dataset(requests_mock, files=[dataset_file])
requests_mock.get(
"https://metaksi/rest/v1/datasets/dataset_identifier?"
"dataset_format=datacite",
content=get_invalid_datacite())
# Try to validate invalid dataset
with pytest.raises(InvalidDatasetMetadataError) as exception_info:
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
# Check error message
assert str(exception_info.value).startswith(
"Datacite metadata is invalid: Element "
"'{http://datacite.org/schema/kernel-4}resource': Missing child "
"element(s).")
def validate_dataset_metadata(dataset_id):
"""Validate dataset metadata.
:returns: HTTP Response
"""
# Validate dataset metadata
try:
validate_metadata(
dataset_id,
app.config.get('SIPTOOLS_RESEARCH_CONF'),
dummy_doi="true"
)
except InvalidDatasetError as exc:
is_valid = False
error = "Metadata did not pass validation"
detailed_error = str(exc)
else:
is_valid = True
error = ''
detailed_error = ''
response = jsonify({'dataset_id': dataset_id,
'is_valid': is_valid,
'error': error,
'detailed_error': detailed_error})
response.status_code = 200
return response
def test_validate_metadata_missing_file(requests_mock):
"""Test validate_metadata with an empty dataset.
Function should raise InvalidDatasetMetadataError for datasets, which do
not contain any files.
:param requests_mock: Mocker object
:returns: ``None``
"""
tests.utils.add_metax_dataset(requests_mock)
expected_error = "Dataset must contain at least one file"
with pytest.raises(InvalidDatasetMetadataError, match=expected_error):
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata_http_error_raised(requests_mock):
"""Test validate_metadata.
Function should raise HTTPError if Metax fails.
:param requests_mock: Mocker object
:returns: ``None``
"""
tests.utils.add_metax_dataset(requests_mock)
requests_mock.get(
'https://metaksi/rest/v1/datasets/dataset_identifier/files',
status_code=500,
reason='Something not to be shown to user')
expected_error = '500 Server Error: Something not to be shown to user'
with pytest.raises(HTTPError, match=expected_error):
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata_invalid_contract_metadata(requests_mock):
"""Test validate_metadata.
Function raises exception with correct error message for invalid dataset.
:param requests_mock: Mocker object
:returns: ``None``
"""
invalid_contract = copy.deepcopy(BASE_CONTRACT)
del invalid_contract['contract_json']['organization']['name']
tests.utils.add_metax_dataset(requests_mock, contract=invalid_contract)
# Try to validate invalid dataset
expected_error = ("'name' is a required property\n\n"
"Failed validating 'required' in schema")
with pytest.raises(InvalidContractMetadataError, match=expected_error):
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata_invalid(requests_mock):
"""Test validate_metadata.
Function should raise exception with correct error message for invalid
dataset.
:returns: ``None``
"""
dataset = copy.deepcopy(BASE_DATASET)
del dataset['contract']
requests_mock.get("https://metaksi/rest/v1/datasets/dataset_identifier",
json=dataset)
# Try to validate invalid dataset
expected_error = "'contract' is a required property"
with pytest.raises(InvalidDatasetMetadataError, match=expected_error):
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata_invalid_file_path(requests_mock):
"""Test validate_metadata.
Function should raise exception if some of the
file paths point outside SIP.
:param requests_mock: Mocker object
:returns: ``None``
"""
invalid_file = copy.deepcopy(TXT_FILE)
invalid_file['file_path'] = "../../file_in_invalid_path"
tests.utils.add_metax_dataset(requests_mock, files=[invalid_file])
# Try to validate invalid dataset
expected_error = ("The file path of file pid:urn:identifier is invalid: "
"../../file_in_invalid_path")
with pytest.raises(InvalidFileMetadataError, match=expected_error):
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata_invalid_directory_metadata(requests_mock):
"""Test validate_metadata wihth invalid directory metadata.
Function should raise exception if directory metadata is not valid
(directory_path attribute is missing).
:param requests_mock: Mocker object
:returns: ``None``
"""
file_metadata = copy.deepcopy(TXT_FILE)
tests.utils.add_metax_dataset(requests_mock, files=[file_metadata])
requests_mock.get("https://metaksi/rest/v1/directories/pid:urn:dir:wf1",
json={"identifier": "pid:urn:dir:wf1"})
# Try to validate invalid dataset
expected_error = ("Validation error in metadata of pid:urn:dir:wf1: "
"'directory_path' is a required property")
with pytest.raises(InvalidDatasetMetadataError, match=expected_error):
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata_corrupted_datacite(requests_mock):
"""Test validate_metadata.
Function should raise exception if datacite XML is corrupted.
:param requests_mock: Mocker object
:returns: ``None``
"""
dataset_file = copy.deepcopy(TXT_FILE)
tests.utils.add_metax_dataset(requests_mock, files=[dataset_file])
requests_mock.get(
"https://metaksi/rest/v1/datasets/dataset_identifier?"
"dataset_format=datacite",
text="<resource\n")
# Try to validate invalid dataset
expected_error \
= "Couldn't find end of Start Tag resource line 1, line 2, column 1"
with pytest.raises(Exception, match=expected_error):
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_invalid_file_type(file_characteristics, version_info,
requests_mock):
"""Test validate_metadata.
Function should raise exception with correct error message for unsupported
file type.
:param file_characteristics: file characteristics dict in file metadata
:param version_info: expected version information in exception message
:param requests_mock: Mocker object
:returns: ``None``
"""
unsupported_file = copy.deepcopy(BASE_FILE)
unsupported_file['file_characteristics'] = file_characteristics
tests.utils.add_metax_dataset(requests_mock, files=[unsupported_file])
# Try to validate dataset with a file that has an unsupported file_format
expected_error = (
"Validation error in file path/to/file: Incorrect file format: "
"application/unsupported{}".format(version_info))
with pytest.raises(InvalidFileMetadataError, match=expected_error):
validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata(requests_mock):
"""Test validate_metadata.
Function should return ``True`` for a valid dataset.
:param requests_mock: Mocker object
:returns: ``None``
"""
files = [copy.deepcopy(TXT_FILE), copy.deepcopy(TXT_FILE)]
files[0]['identifier'] = "pid:urn:1"
files[1]['identifier'] = "pid:urn:2"
tests.utils.add_metax_dataset(requests_mock, files=files)
assert validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata_languages(translations, expectation, requests_mock):
"""Test validate_metadata.
Function should raise exception when one of the localized
fields has invalid values.
:returns: ``None``
"""
dataset = copy.deepcopy(BASE_DATASET)
dataset_file = copy.deepcopy(TXT_FILE)
dataset['research_dataset']['provenance'][0]['description'] = translations
tests.utils.add_metax_dataset(requests_mock,
dataset=dataset,
files=[dataset_file])
with expectation:
assert validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)
def test_validate_metadata_multiple_formats(requests_mock, file_format,
stream_type, xml):
"""Test validate_metadata.
Function validates different types of technical metadata.
:param requests_mock: Mocker object
:param file_format: file mimetype
:param stream_type: stream type contained within the file
:param xml: techincal metada as xml object
:returns: ``None``
"""
file_metadata = copy.deepcopy(BASE_FILE)
file_metadata.update({
"file_characteristics": {
"file_format": file_format
},
"file_characteristics_extension": {
"streams": {
0: {
"mimetype": file_format,
"stream_type": stream_type
}
}
}
})
tests.utils.add_metax_dataset(requests_mock, files=[file_metadata])
requests_mock.get("https://metaksi/rest/v1/files/{}/xml".format(
file_metadata['identifier']),
json=[METS_NS])
requests_mock.get("https://metaksi/rest/v1/files/{}/xml?"
"namespace={}".format(file_metadata['identifier'],
METS_NS),
content=six.binary_type(lxml.etree.tostring(xml)))
assert validate_metadata('dataset_identifier',
tests.conftest.UNIT_TEST_CONFIG_FILE)