def _create_premis_events(dataset_id, workspace, config):
"""Create premis events from provenance metadata.
Reads dataset provenance metadata from Metax. For each provenance
object a METS document that contains a PREMIS event element is
created.
:param dataset_id: dataset identifier
:param workspace: SIP creation directory
:param config: path to configuration file
:returns: ``None``
"""
config_object = Configuration(config)
metadata = Metax(config_object.get('metax_url'),
config_object.get('metax_user'),
config_object.get('metax_password'),
verify=config_object.getboolean(
'metax_ssl_verification')).get_dataset(dataset_id)
dataset_languages = get_dataset_languages(metadata)
provenances = metadata["research_dataset"]["provenance"]
for provenance in provenances:
event_type = get_localized_value(
provenance["preservation_event"]["pref_label"],
languages=dataset_languages)
try:
event_datetime = provenance["temporal"]["start_date"]
except KeyError:
event_datetime = 'OPEN'
event_detail = get_localized_value(provenance["description"],
languages=dataset_languages)
event_outcome = get_localized_value(
provenance["event_outcome"]["pref_label"],
languages=dataset_languages)
event_outcome_detail = get_localized_value(
provenance["outcome_description"], languages=dataset_languages)
premis_event.premis_event(workspace=workspace,
event_type=event_type,
event_datetime=event_datetime,
event_detail=event_detail,
event_outcome=event_outcome,
event_outcome_detail=event_outcome_detail)
def find_dir_use_category(parent_path, dirpath2usecategory, languages):
"""Find use category of path.
Find use_category of the closest parent directory listed in the
research_dataset. This is done by checking how well the directory
paths in the research_dataset match with the parent directory path.
:param parent_path: path to the parent directory of the file
:param dirpath2usecategory: Dictionary, which maps research_dataset
directory paths to the corresponding
use_categories.
:param languages: A list of ISO 639-1 formatted language codes of
the dataset
:returns: `use_category` attribute of directory
"""
max_matches = 0
use_category = None
for dirpath in dirpath2usecategory:
matches = _match_paths(parent_path, dirpath)
if matches > max_matches:
max_matches = matches
use_category = dirpath2usecategory[dirpath]
if use_category:
return get_localized_value(use_category["pref_label"],
languages=languages)
return None
def test_get_localized_value():
"""Test that correct localized value is selected depending on priority,
or 'und' or 'zxx' is used as a fallback if nothing else matches
:returns: ``None``
"""
assert get_localized_value(DATASET["testa"],
languages=["en", "fi"]) == "Test in English"
assert get_localized_value(DATASET["testa"],
languages=["fi", "en"]) == "Testi suomeksi"
# Use 'und' or 'xzz' if no localized value exists for 'fi' or 'en'
assert get_localized_value(DATASET["testb"],
languages=["en", "fi"]) == "#1,234,567"
with pytest.raises(KeyError):
get_localized_value(DATASET["testc"], languages=["en", "fi"])
def get_provenance_ids(self):
"""List identifiers of provenance events.
Gets list of dataset provenance events from Metax, and reads
provenance IDs of the events from event.xml files found in the
workspace directory.
:returns: list of provenance IDs
"""
config_object = Configuration(self.config)
metax_client = Metax(
config_object.get('metax_url'),
config_object.get('metax_user'),
config_object.get('metax_password'),
verify=config_object.getboolean('metax_ssl_verification'))
metadata = metax_client.get_dataset(self.dataset_id)
languages = get_dataset_languages(metadata)
# Get the reference file path from Luigi task input
# It already contains the workspace path.
event_ids = get_md_references(
read_md_references(
self.workspace,
os.path.basename(
self.input()['create_provenance_information'].path)))
event_type_ids = {}
for event_id in event_ids:
event_file = event_id[1:] + "-PREMIS%3AEVENT-amd.xml"
event_file_path = os.path.join(self.sip_creation_path, event_file)
if not os.path.exists(event_file_path):
continue
root = ET.parse(encode_path(event_file_path)).getroot()
event_type = root.xpath("//premis:eventType",
namespaces=NAMESPACES)[0].text
event_type_ids[event_type] = event_id
provenance_ids = []
for provenance in metadata["research_dataset"]["provenance"]:
event_type = get_localized_value(
provenance["preservation_event"]["pref_label"],
languages=languages)
provenance_ids += [event_type_ids[event_type]]
return provenance_ids
def find_file_use_category(identifier, dataset_metadata):
"""Look for file with identifier from dataset metadata.
Returns the `use_category` of file if it is found. If file is not
found from list, return None.
:param identifier: file identifier
:param dataset_metadata: dataset metadata dictionary
:returns: `use_category` attribute of file
"""
languages = get_dataset_languages(dataset_metadata)
if 'files' in dataset_metadata['research_dataset']:
for file_ in dataset_metadata['research_dataset']['files']:
if file_['identifier'] == identifier:
return get_localized_value(file_['use_category']['pref_label'],
languages=languages)
# Nothing found
return None