def test_verify_file_contained_by_dataset_files():
"""Test is_consistent_for_file method.
Check that ``DatasetConsistency::is_consistent_for_file()`` succeeds
when dataset files contains the file
:returns: ``None``
"""
# Init metax client
configuration = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
client = Metax(
configuration.get('metax_url'),
configuration.get('metax_user'),
configuration.get('metax_password'),
verify=configuration.getboolean('metax_ssl_verification')
)
dataset = {
'identifier': 'dataset_identifier',
'research_dataset': {'files': [
{'identifier': 'file_identifier'}
], 'directories': []}
}
file_metadata = {
'identifier': 'file_identifier',
'file_path': "/path/to/file",
'parent_directory': {'identifier': 'parent_directory_identifier'}
}
try:
dirs = DatasetConsistency(client, dataset)
dirs.is_consistent_for_file(file_metadata)
except InvalidDatasetMetadataError as exc:
pytest.fail(
'_verify_file_contained_by_dataset raised exception: ' + str(exc)
)
def app(request, test_config):
"""Create web app and Mock Metax HTTP responses.
:returns: An instance of the REST API web app.
"""
# Create app and change the default config file path
app_ = create_app()
app_.config.update(
SIPTOOLS_RESEARCH_CONF=test_config
)
app_.config['TESTING'] = True
# Create temporary directories
conf = Configuration(test_config)
cache_dir = os.path.join(conf.get("packaging_root"), "file_cache")
os.mkdir(cache_dir)
tmp_dir = os.path.join(conf.get("packaging_root"), "tmp")
os.mkdir(tmp_dir)
# Mock Metax
def _fin():
httpretty.reset()
httpretty.disable()
httpretty.enable()
request.addfinalizer(_fin)
mock_metax()
return app_
def test_run_workflowtask(testpath):
"""Test WorkflowTask execution.
Executes a TestTask, checks that output file is created, checks that new
event field is created to mongo document.
:param testpath: temporary directory
:returns: ``None``
"""
# Run task like it would be run from command line
run_luigi_task('TestTask', testpath)
# Check that output file is created
with open(os.path.join(testpath, 'output_file')) as output:
assert output.read() == 'Hello world'
# Check that new event is added to workflow database
conf = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
mongoclient = pymongo.MongoClient(host=conf.get('mongodb_host'))
collection = (mongoclient[conf.get('mongodb_database')][conf.get(
'mongodb_collection')])
document = collection.find_one()
# Check 'messages' field
assert document['workflow_tasks']['TestTask']['messages'] ==\
'Test task was successfull'
# Check 'result' field
assert document['workflow_tasks']['TestTask']['result'] == 'success'
# Parse the 'timestamp' field to make sure it is correct format
datetime.datetime.strptime(
document['workflow_tasks']['TestTask']['timestamp'],
'%Y-%m-%dT%H:%M:%S.%f')
# Check that there is no extra documents in mongo collection
assert collection.count() == 1
def test_validate_file_metadata_invalid_metadata(requests_mock):
"""Test ``_validate_file_metadata``.
Function should raise exceptions with descriptive error messages.
:param requests_mock: Mocker object
:returns: ``None``
"""
file_metadata = copy.deepcopy(BASE_FILE)
file_metadata['file_characteristics'] = {
"file_created": "2014-01-17T08:19:31Z"
}
tests.utils.add_metax_dataset(requests_mock, files=[file_metadata])
# Init metax client
configuration = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
client = Metax(configuration.get('metax_url'),
configuration.get('metax_user'),
configuration.get('metax_password'),
verify=configuration.getboolean('metax_ssl_verification'))
expected_error = (
"Validation error in metadata of path/to/file: 'file_format' is"
" a required property\n\nFailed validating 'required' in schema")
with pytest.raises(InvalidFileMetadataError, match=expected_error):
# pylint: disable=protected-access
siptools_research.metadata_validator._validate_file_metadata(
{'identifier': 'dataset_identifier'}, client, configuration)
def test_validate_datacite(requests_mock):
"""Test _validate_datacite.
Function should raises exception with readable error message when datacite
XML contains multiple errors.
:param requests_mock: Mocker object
:returns: ``None``
"""
# Init metax client
configuration = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
metax_client = Metax(
configuration.get('metax_url'),
configuration.get('metax_user'),
configuration.get('metax_password'),
verify=configuration.getboolean('metax_ssl_verification'))
requests_mock.get(
"https://metaksi/rest/v1/datasets/dataset_identifier?"
"dataset_format=datacite",
content=get_very_invalid_datacite())
# Try to validate datacite
expected_error = "Datacite metadata is invalid:"
# pylint: disable=protected-access
with pytest.raises(InvalidDatasetMetadataError, match=expected_error):
metadata_validator._validate_datacite('dataset_identifier',
metax_client)
def run(self):
"""Send SIP file to DP service using sftp.
:returns: ``None``
"""
# Read host/user/ssh_key_path from onfiguration file
conf = Configuration(self.config)
# Init SFTP connection
with paramiko.SSHClient() as ssh:
ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
ssh.connect(conf.get('dp_host'),
username=conf.get('dp_user'),
key_filename=conf.get('dp_ssh_key'))
with ssh.open_sftp() as sftp:
# Copy tar to remote host. Validation workflow starts
# when ".incomplete" suffix is removed from target file
# path.
tar_file = os.path.basename(self.workspace) + '.tar'
sftp.put(os.path.join(self.workspace, tar_file),
os.path.join('transfer', tar_file + '.incomplete'),
confirm=False)
sftp.rename(os.path.join('transfer', tar_file + '.incomplete'),
os.path.join('transfer', tar_file))
with self.output().open('w') as log:
log.write('Dataset id=' + self.dataset_id)
def test_run_failing_task(testpath, ):
"""Test running task that fails.
Executes FailingTestTask and checks that report of failed event is
added to mongo document.
:param testpath: temporary directory
:returns: ``None``
"""
# Run task like it would be run from command line
run_luigi_task('FailingTestTask', testpath)
# Check that new event is added to workflow database
conf = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
mongoclient = pymongo.MongoClient(host=conf.get('mongodb_host'))
collection = (mongoclient[conf.get('mongodb_database')][conf.get(
'mongodb_collection')])
document = collection.find_one()
# Check 'messages' field
assert document['workflow_tasks']['FailingTestTask']['messages'] ==\
'An error occurred while running a test task: Shit hit the fan'
# Check 'result' field
assert document['workflow_tasks']['FailingTestTask']['result'] ==\
'failure'
# Parse the 'timestamp' field to make sure it is correct format
datetime.datetime.strptime(
document['workflow_tasks']['FailingTestTask']['timestamp'],
'%Y-%m-%dT%H:%M:%S.%f')
# Check that there is no extra documents in mongo collection
assert collection.count() == 1
def test_validate_file_metadata(requests_mock):
"""Test _validate_file_metadata.
Check that dataset directory caching is working correctly in
DatasetConsistency when the files have common root directory in
dataset.directories property.
:param requests_mock: Mocker object
:returns: ``None``
"""
dataset = copy.deepcopy(BASE_DATASET)
dataset['research_dataset']['directories'] = [{'identifier': 'root_dir'}]
file_1 = copy.deepcopy(TXT_FILE)
file_1['identifier'] = 'file_identifier1'
file_2 = copy.deepcopy(TXT_FILE)
file_2['identifier'] = 'file_identifier2'
requests_mock.get(tests.conftest.METAX_URL +
'/directories/pid:urn:dir:wf1',
json={
'identifier': 'first_par_dir',
'directory_path': '',
'parent_directory': {
'identifier': 'second_par_dir'
}
},
status_code=200)
requests_mock.get(tests.conftest.METAX_URL + '/directories/second_par_dir',
json={
'identifier': 'second_par_dir',
'directory_path': '',
'parent_directory': {
'identifier': 'root_dir'
}
},
status_code=200)
requests_mock.get(tests.conftest.METAX_URL + '/directories/root_dir',
json={
'identifier': 'second_par_dir',
'directory_path': '/'
},
status_code=200)
files_adapter = requests_mock.get(tests.conftest.METAX_URL +
'/datasets/dataset_identifier/files',
json=[file_1, file_2],
status_code=200)
# Init metax client
configuration = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
client = Metax(configuration.get('metax_url'),
configuration.get('metax_user'),
configuration.get('metax_password'),
verify=configuration.getboolean('metax_ssl_verification'))
# pylint: disable=protected-access
siptools_research.metadata_validator._validate_file_metadata(
dataset, client, configuration)
assert files_adapter.call_count == 1
def test_directory_validation_caching_works(requests_mock):
"""Test directory validation caching.
Two files are contained by same directory. Metax is called only once for
each directory in tree and hence the directory validation as well.
:returns: ``None``
"""
# Init metax client
configuration = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
client = Metax(configuration.get('metax_url'),
configuration.get('metax_user'),
configuration.get('metax_password'),
verify=configuration.getboolean('metax_ssl_verification'))
first_par_dir_adapter = requests_mock.get(
tests.conftest.METAX_URL + '/directories/first_par',
json={
'identifier': 'first_par',
'directory_path': '/second_par/first_par',
'parent_directory': {
'identifier': 'second_par'
}
},
status_code=200)
second_par_dir_adapter = requests_mock.get(
tests.conftest.METAX_URL + '/directories/second_par',
json={
'identifier': 'second_par',
'directory_path': '/second_par',
'parent_directory': {
'identifier': 'root'
}
},
status_code=200)
root_dir_adapter = requests_mock.get(tests.conftest.METAX_URL +
'/directories/root',
json={
'identifier': 'root',
'directory_path': '/'
},
status_code=200)
file2_metadata = copy.deepcopy(FILE_METADATA)
file2_metadata['file_path'] = ["/path/to/file2"]
try:
validator = DirectoryValidation(client)
validator.is_valid_for_file(FILE_METADATA)
validator.is_valid_for_file(file2_metadata)
except InvalidDatasetMetadataError as exc:
pytest.fail('test_successful_directory_validation fails: ' + str(exc))
# verify that metax is called only once for directories
assert first_par_dir_adapter.call_count == 1
assert second_par_dir_adapter.call_count == 1
assert root_dir_adapter.call_count == 1
def find_file_categories(self):
"""Create logical structure map of dataset files.
Returns dictionary with filecategories as keys and filepaths as
values.
:returns: logical structure map dictionary
"""
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'))
dataset_files = metax_client.get_dataset_files(self.dataset_id)
dataset_metadata = metax_client.get_dataset(self.dataset_id)
languages = get_dataset_languages(dataset_metadata)
dirpath2usecategory = get_dirpath_dict(metax_client, dataset_metadata)
logical_struct = dict()
for dataset_file in dataset_files:
file_id = dataset_file['identifier']
# Get the use category of file. The path to the file in
# logical structmap is stored in 'use_category' in metax.
filecategory = find_file_use_category(file_id, dataset_metadata)
# If file listed in datasets/<id>/files is not listed in
# 'files' section of dataset metadata, look for
# parent_directory of the file from 'directories' section.
# The "use_category" of file is the "use_category" of the
# parent directory.
if filecategory is None:
name_len = len(dataset_file["file_name"])
filecategory = find_dir_use_category(
dataset_file["file_path"][:-name_len], dirpath2usecategory,
languages)
# If file category was not found even for the parent
# directory, raise error
if filecategory is None:
raise InvalidDatasetMetadataError(
"File category for file {} was not found".format(file_id))
# Append path to logical_struct[filecategory] list. Create
# list if it does not exist already
if filecategory not in logical_struct.keys():
logical_struct[filecategory] = []
logical_struct[filecategory].append(dataset_file['file_path'])
return logical_struct
def run(self):
"""Get dataset 1 from Metax.
:returns: ``None``
"""
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'))
metax_client.get_dataset('1')
def __init__(self, config_file):
"""Initialize new pymongo client if it does not exist already
(singleton design pattern)
:param config_file: path to configurtion file
"""
if self._collection is None:
conf = Configuration(config_file)
self._client = pymongo.MongoClient(host=conf.get("mongodb_host"),
port=27017)
self._collection = (self._client[conf.get("mongodb_database")][
conf.get("mongodb_collection")])
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 test_successful_directory_validation_fails(requests_mock):
"""Test validation of invalid directory tree.
The root directory is missing the `directory_path` attribute
:returns: ``None``
"""
# Init metax client
configuration = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
client = Metax(configuration.get('metax_url'),
configuration.get('metax_user'),
configuration.get('metax_password'),
verify=configuration.getboolean('metax_ssl_verification'))
first_par_dir_adapter = requests_mock.get(
tests.conftest.METAX_URL + '/directories/first_par',
json={
'identifier': 'first_par',
'directory_path': '/second_par/first_par',
'parent_directory': {
'identifier': 'second_par'
}
},
status_code=200)
second_par_dir_adapter = requests_mock.get(
tests.conftest.METAX_URL + '/directories/second_par',
json={
'identifier': 'second_par',
'directory_path': '/second_par',
'parent_directory': {
'identifier': 'root'
}
},
status_code=200)
root_dir_adapter = requests_mock.get(tests.conftest.METAX_URL +
'/directories/root',
json={'identifier': 'root'},
status_code=200)
with pytest.raises(InvalidDatasetMetadataError) as exc_info:
validator = DirectoryValidation(client)
validator.is_valid_for_file(FILE_METADATA)
assert str(exc_info.value).startswith(
"Validation error in metadata of root: "
"'directory_path' is a required property")
assert first_par_dir_adapter.call_count == 1
assert second_par_dir_adapter.call_count == 1
assert root_dir_adapter.call_count == 1
def run(self):
"""Create premis events.
Reads dataset metadata from Metax and creates premis event XML
files. Premis event XML files are written to SIP creation
directory and premis event reference file is written to
workspace directory.
:returns: ``None``
"""
config_object = Configuration(self.config)
tmp = os.path.join(config_object.get('packaging_root'), 'tmp/')
with TemporaryDirectory(prefix=tmp) as temporary_workspace:
_create_premis_events(self.dataset_id, temporary_workspace,
self.config)
# Move PREMIS event files to SIP creation path when all of
# them are succesfully created to avoid atomicity problems.
# PREMIS event reference file is moved to ouput target path
# after other files are moved to SIP creation directory.
with self.output().temporary_path() as target_path:
shutil.move(
os.path.join(temporary_workspace,
'premis-event-md-references.jsonl'),
target_path)
for file_ in os.listdir(temporary_workspace):
shutil.move(os.path.join(temporary_workspace, file_),
self.sip_creation_path)
def download_file(file_metadata,
linkpath="",
config_file="/etc/siptools_research.conf",
upload_database=None):
"""Get file from IDA or upload-rest-api and create a hard
link to linkpath.
:param file_metadata: File metadata from Metax
:param linkpath: Path where the hard link is created
:param config_file: Configuration file
:param upload_database: upload_rest_api.database.Database object
:returns: ``None``
"""
conf = Configuration(config_file)
pas_storage_id = conf.get("pas_storage_id")
file_storage = file_metadata["file_storage"]["identifier"]
if file_storage == pas_storage_id:
if upload_database is None:
raise ValueError("upload_database parameter required")
filepath = _get_local_file(file_metadata, upload_database, conf)
else:
filepath = _get_ida_file(file_metadata, conf)
if linkpath:
os.link(filepath, linkpath)
def test_successful_directory_validation(requests_mock):
"""Directory validation of valid directory tree.
:returns: ``None``
"""
# Init metax client
configuration = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
client = Metax(configuration.get('metax_url'),
configuration.get('metax_user'),
configuration.get('metax_password'),
verify=configuration.getboolean('metax_ssl_verification'))
first_par_dir_adapter = requests_mock.get(
tests.conftest.METAX_URL + '/directories/first_par',
json={
'identifier': 'first_par',
'directory_path': '/second_par/first_par',
'parent_directory': {
'identifier': 'second_par'
}
},
status_code=200)
second_par_dir_adapter = requests_mock.get(
tests.conftest.METAX_URL + '/directories/second_par',
json={
'identifier': 'second_par',
'directory_path': '/second_par',
'parent_directory': {
'identifier': 'root'
}
},
status_code=200)
root_dir_adapter = requests_mock.get(tests.conftest.METAX_URL +
'/directories/root',
json={
'identifier': 'root',
'directory_path': '/'
},
status_code=200)
try:
validator = DirectoryValidation(client)
validator.is_valid_for_file(FILE_METADATA)
except InvalidDatasetMetadataError as exc:
pytest.fail('test_successful_directory_validation fails: ' + str(exc))
assert first_par_dir_adapter.call_count == 1
assert second_par_dir_adapter.call_count == 1
assert root_dir_adapter.call_count == 1
def test_workflow(testpath, module_name, task, requests_mock):
"""Test workflow dependency tree.
Run a task (and all tasks it requires) and check that report of
successful task is added to mongodb.
:param testpath: temporary directory
:param module_name: submodule of siptools_research.workflow that
contains Task to be tested
:param task: Task class name
:param requests_mock: Mocker object
:returns: ``None``
"""
tests.utils.add_metax_dataset(requests_mock,
tests.metax_data.datasets.BASE_DATASET,
files=[tests.metax_data.files.TXT_FILE])
requests_mock.get('https://ida.test/files/pid:urn:identifier/download',
text='foo')
# Init pymongo client
conf = Configuration(tests.conftest.TEST_CONFIG_FILE)
mongoclient = pymongo.MongoClient(host=conf.get('mongodb_host'))
with mock.patch.object(RemoteAnyTarget, '_exists', _mock_exists):
workspace = os.path.join(testpath, 'workspaces', 'workspace')
module = importlib.import_module('siptools_research.workflow.'
+ module_name)
task_class = getattr(module, task)
luigi.build(
[task_class(
workspace=workspace,
dataset_id='dataset_identifier',
config=tests.conftest.UNIT_TEST_CONFIG_FILE
)],
local_scheduler=True
)
collection = (mongoclient[conf.get('mongodb_database')]
[conf.get('mongodb_collection')])
document = collection.find_one()
# Check 'result' field
assert document['workflow_tasks'][task]['result'] == 'success'
if module_name == "cleanup":
assert document["completed"]
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 run(self):
"""Read list of required files from Metax and download them.
Files are written to path based on ``file_path`` in Metax.
:returns: ``None``
"""
upload_database = upload_rest_api.database.Database()
# Find file identifiers from Metax dataset metadata.
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'))
dataset_files = metax_client.get_dataset_files(self.dataset_id)
# Download files to temporary directory which will be moved to
# output target path when all files have been downloaded
with self.output().temporary_path() as temporary_directory:
os.mkdir(temporary_directory)
for dataset_file in dataset_files:
identifier = dataset_file["identifier"]
# Full path to file
target_path = os.path.normpath(
os.path.join(temporary_directory,
dataset_file["file_path"].strip('/')))
if not target_path.startswith(temporary_directory):
raise InvalidFileMetadataError(
'The file path of file %s is invalid: %s' %
(identifier, dataset_file["file_path"]))
# Create the download directory for file if it does not
# exist already
if not os.path.isdir(os.path.dirname(target_path)):
# TODO: Use exist_ok -parameter when moving to
# python3
os.makedirs(os.path.dirname(target_path))
download_file(dataset_file, target_path, self.config,
upload_database)
def test_send_sip(testpath):
"""Test the SendSipToDP task.
Run task and check that .tar is copied to digital preservation
server. This test uses real DPS test instance. To run this test,
a valid ssh key for `tpas` user on `customer-test-tpas-1` must be
copied to current working directory and renamed as `ssh_key`.
:param testpath: Temporary directory fixture
:returns: ``None``
"""
# Create workspace with directories and files required by the task
workspace = testpath
os.makedirs(os.path.join(workspace, 'sip-in-progress'))
tar_file_name = os.path.basename(workspace) + ".tar"
shutil.copy('tests/data/testsips/simple_sip.tar',
os.path.join(workspace, tar_file_name))
# Init sftp connection to digital preservation server for instant
# verification later on
conf = Configuration(tests.conftest.TEST_CONFIG_FILE)
ssh = paramiko.SSHClient()
ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
ssh.connect(conf.get('dp_host'),
username=conf.get('dp_user'),
key_filename=conf.get('dp_ssh_key'))
sftp = ssh.open_sftp()
# Init and run task
task = SendSIPToDP(workspace=workspace,
dataset_id='1',
config=tests.conftest.TEST_CONFIG_FILE)
task.run()
assert task.complete()
# Check that tar-file is created on remote host.
# NOTE: Tar is copied to ~/transfer/. From there it is automatically
# moved to /var/spool/preservation/ and after validation it is moved
# to ~/rejected/<datedir>/<workspace>.tar/. There is a risk that
# file is moved from ~/transfer before this test is finished.
target_file_path = "transfer/" + tar_file_name
logging.debug('Looking for file: %s on server: %s', target_file_path,
conf.get('dp_host'))
sftp.stat(target_file_path)
def test_verify_file_contained_by_dataset_missing_from_dataset(requests_mock):
"""Test is_consistent_for_file method.
Check that ``DatasetConsistency::is_consistent_for_file()`` raises
exception with descriptive error messages when dataset files nor
directories do not contain the file.
:returns: ``None``
"""
# Init metax client
configuration = Configuration(tests.conftest.UNIT_TEST_CONFIG_FILE)
client = Metax(
configuration.get('metax_url'),
configuration.get('metax_user'),
configuration.get('metax_password'),
verify=configuration.getboolean('metax_ssl_verification')
)
dataset = {
'identifier': 'dataset_identifier',
'research_dataset': {
'files': [],
'directories': []
}
}
file_metadata = {
'identifier': 'file_identifier',
'file_path': "/path/to/file",
'parent_directory': {
'identifier': 'parent_directory_identifier'
}
}
requests_mock.get(
tests.conftest.METAX_URL + '/directories/parent_directory_identifier',
json={'identifier': 'parent_directory_identifier'},
status_code=200
)
with pytest.raises(InvalidDatasetMetadataError) as exc_info:
dirs = DatasetConsistency(client, dataset)
dirs.is_consistent_for_file(file_metadata)
assert str(exc_info.value) == ("File not found from dataset files nor "
"directories: /path/to/file")
def run(self):
"""Report preservation status to Metax.
Checks the path of ingest report file in digital preservation
service. If the ingest report is in ~/accepted/.../ directory,
the dataset has passed validation.If the report is found in
~/rejected/.../ directory, or somewhere else, an exception is
risen. The event handlers will deal with the exceptions.
:returns: ``None``
"""
# List of all matching paths ValidateSIP found
ingest_report_paths = self.input()[0].existing_paths()
# Only one ingest report should be found
assert len(ingest_report_paths) == 1
# 'accepted' or 'rejected'?
directory = ingest_report_paths[0].split('/')[0]
if directory == 'accepted':
# Init metax
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'))
# Set Metax preservation state of this dataset to 6 ("in
# longterm preservation")
metax_client.set_preservation_state(
self.dataset_id,
state=DS_STATE_IN_DIGITAL_PRESERVATION,
system_description='Accepted to preservation')
with self.output().open('w') as output:
output.write('Dataset id=' + self.dataset_id)
elif directory == 'rejected':
# Raise exception that informs event handler that dataset
# did not pass validation
raise InvalidSIPError("SIP was rejected")
else:
raise ValueError('Report was found in incorrect '
'path: %s' % ingest_report_paths[0])
def output(self):
"""Return the output target of this Task.
:returns: remote target that may exist on digital preservation
server in any path formatted::
~/accepted/<datepath>/<document_id>.tar/
~/rejected/<datepath>/<document_id>.tar/
where datepath is any date between the date the SIP
was sent to the server and the current date.
:rtype: RemoteAnyTarget
"""
conf = Configuration(self.config)
database = Database(self.config)
# Get SendSIPToDP completion datetime or use the current UTC
# time. This is necessary since ValidateSip output is checked
# first time before any of the dependencies are ran.
# Dependencies are ran only if ValidateSip task is not
# completed.
try:
send_timestamp = database.get_event_timestamp(
self.document_id, "SendSIPToDP")
sip_to_dp_date = dateutil.parser.parse(send_timestamp).date()
except (ValueError, KeyError):
sip_to_dp_date = datetime.utcnow().date()
lim_date = datetime.today().date()
path = []
while sip_to_dp_date <= lim_date:
path.append('accepted/%s/%s.tar' %
(sip_to_dp_date, self.document_id))
path.append('rejected/%s/%s.tar' %
(sip_to_dp_date, self.document_id))
sip_to_dp_date += timedelta(days=1)
return RemoteAnyTarget(path, conf.get('dp_host'), conf.get('dp_user'),
conf.get('dp_ssh_key'))
def run(self):
"""Copy datacite.xml metadatafile from Metax.
Creates a METS document that contains dmdSec element with
datacite metadata.
:returns: ``None``
"""
# Get datacite.xml from Metax
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'))
dataset = metax_client.get_dataset(self.dataset_id)
datacite = metax_client.get_datacite(dataset['identifier'])
# Write datacite.xml to file
datacite_path = os.path.join(self.workspace, 'datacite.xml')
datacite.write(datacite_path)
tmp = os.path.join(config_object.get('packaging_root'), 'tmp/')
with TemporaryDirectory(prefix=tmp) as temporary_workspace:
# Create output files with siptools
import_description.import_description(
dmdsec_location=datacite_path,
workspace=temporary_workspace,
without_uuid=True)
# Move created files to SIP creation directory. PREMIS event
# reference file is moved to output target path after
# everything else is done.
with self.output().temporary_path() as target_path:
shutil.move(
os.path.join(temporary_workspace,
'premis-event-md-references.jsonl'),
target_path)
for file_ in os.listdir(temporary_workspace):
shutil.move(os.path.join(temporary_workspace, file_),
self.sip_creation_path)
def report_task_failure(task, exception):
"""Report task failure.
This function is triggered when a WorkflowTask fails. Adds report of
failed event to workflow database.
If task failed because dataset is invalid, the preservation status of
dataset is updated in Metax, and the workflow is disabled.
:param task: WorkflowTask object
:param exception: Exception that caused failure
:returns: ``None``
"""
database = Database(task.config)
database.add_event(task.document_id, task.task_name, 'failure',
"%s: %s" % (task.failure_message, str(exception)))
config_object = Configuration(task.config)
metax_client = metax_access.Metax(
config_object.get('metax_url'),
config_object.get('metax_user'),
config_object.get('metax_password'),
verify=config_object.getboolean('metax_ssl_verification'))
if isinstance(exception, siptools_research.exceptions.InvalidDatasetError):
# Disable workflow
database.set_disabled(task.document_id)
if isinstance(exception, siptools_research.exceptions.InvalidSIPError):
# Set preservation status for dataset in Metax
metax_client.set_preservation_state(
task.dataset_id,
state=(metax_access.
DS_STATE_REJECTED_IN_DIGITAL_PRESERVATION_SERVICE),
system_description=_get_description(task, exception))
elif isinstance(exception,
siptools_research.exceptions.InvalidDatasetMetadataError):
# Set preservation status for dataset in Metax
metax_client.set_preservation_state(
task.dataset_id,
state=metax_access.DS_STATE_METADATA_VALIDATION_FAILED,
system_description=_get_description(task, exception))
def get_identifiers(self):
"""Get file identifiers.
Return a list of all the file identifiers and the path to the
downloaded files.
:returns: Tuple (list of identifiers, cache_path)
"""
config_object = Configuration(self.config)
packaging_root = config_object.get("packaging_root")
cache_path = os.path.join(packaging_root, "file_cache")
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'))
try:
dataset_files = metax_client.get_dataset_files(self.dataset_id)
return [_file["identifier"] for _file in dataset_files], cache_path
except DatasetNotAvailableError:
return [], cache_path
def validate_metadata(dataset_id,
config="/etc/siptools_research.conf",
dummy_doi="false"):
"""Validate dataset.
Reads dataset metadata, file metadata, and additional techMD XML from
Metax and validates them against schemas. Raises error if dataset is not
valid. Raises InvalidDatasetError if dataset is invalid.
:param dataset_id: dataset identifier
:param config: configuration file path
:param: dummy_doi: 'true' if dummy preservation identifier is to be used
:returns: ``True``, if dataset metadata is valid.
"""
conf = Configuration(config)
metax_client = Metax(conf.get('metax_url'),
conf.get('metax_user'),
conf.get('metax_password'),
verify=conf.getboolean('metax_ssl_verification'))
# Get dataset metadata from Metax
dataset_metadata = metax_client.get_dataset(dataset_id)
# Validate dataset metadata
_validate_dataset_metadata(dataset_metadata, dummy_doi=dummy_doi)
# Validate dataset localization
_validate_dataset_localization(dataset_metadata)
# Validate contract metadata
_validate_contract_metadata(dataset_metadata['contract']['identifier'],
metax_client)
# Validate file metadata for each file in dataset files
_validate_file_metadata(dataset_metadata, metax_client, conf)
# Validate datacite provided by Metax
_validate_datacite(dataset_id, metax_client, dummy_doi=dummy_doi)
return True
def validate_files(dataset_id, config_file="/etc/siptools_research.conf"):
"""Validate all files in a dataset.
Raises InvalidFileError or MissingFileError if files are invalid.
:param dataset_id: dataset identifier
:param config: configuration file path
:returns: ``True`` if all files are well-formed.
"""
invalid_files = []
missing_files = []
conf = Configuration(config_file)
metax_client = Metax(
conf.get('metax_url'),
conf.get('metax_user'),
conf.get('metax_password'),
verify=conf.getboolean('metax_ssl_verification')
)
cache_path = os.path.join(conf.get("packaging_root"), "file_cache")
dataset_files = _download_files(
metax_client, dataset_id, config_file, missing_files
)
if missing_files:
raise MissingFileError(
"{} files are missing".format(len(missing_files)), missing_files
)
for dataset_file in dataset_files:
_validate_file(dataset_file, cache_path, invalid_files)
if invalid_files:
raise InvalidFileError(
"{} files are not well-formed".format(len(invalid_files)),
invalid_files
)
return True
def run(self):
"""Compile all metadata files into METS document.
:returns: ``None``
"""
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)
# Get preservation_identifier from Metax
preservation_id = metadata["preservation_identifier"]
# Get contract data from Metax
contract_id = metadata["contract"]["identifier"]
contract_metadata = metax_client.get_contract(contract_id)
contract_identifier = contract_metadata["contract_json"]["identifier"]
contract_org_name \
= contract_metadata["contract_json"]["organization"]["name"]
# Compile METS
mets = compile_mets.create_mets(workspace=self.sip_creation_path,
mets_profile='tpas',
contractid=contract_identifier,
objid=preservation_id,
organization_name=contract_org_name,
packagingservice='Packaging Service')
with self.output().open('wb') as outputfile:
mets.write(outputfile,
pretty_print=True,
xml_declaration=True,
encoding='UTF-8')
