def test_import_since_regular_value_returned_unchanged(self):
'''Test that any value other than 'big_bang' or 'last_changed' for
`import_since` is returned unchanged.'''
FisbrokerPlugin().source_config = {'import_since': "2020-03-01"}
import_since = FisbrokerPlugin().get_import_since_date(None)
_assert_equal(import_since, "2020-03-01")
def test_timeout_config_returned_as_int(self):
'''Test that get_timeout() always returns an int, if the `timeout``
config is set.'''
FisbrokerPlugin().source_config = { 'timeout': '100' }
timeout = FisbrokerPlugin().get_timeout()
_assert_equal(timeout, 100)
def test_timedelta_config_returned_as_int(self):
'''Test that get_timedelta() always returns an int, if the `timedelta``
config is set.'''
FisbrokerPlugin().source_config = { 'timedelta': '1' }
timedelta = FisbrokerPlugin().get_timedelta()
_assert_equal(timedelta, 1)
def test_import_since_big_bang_means_none(self):
'''Test that 'big_bang' for the `import_since` config means
returns None.'''
FisbrokerPlugin().source_config = { 'import_since': "big_bang" }
import_since = FisbrokerPlugin().get_import_since_date(None)
_assert_equal(import_since, None)
def test_timedelta_must_be_int(self):
'''Test that the `timedelta` config must be an int.'''
config = '{ "timedelta": 2 }'
assert FisbrokerPlugin().validate_config(config)
# invalid timedelta:
config = '{ "timedelta": "two" }'
with assert_raises(ValueError):
assert FisbrokerPlugin().validate_config(config)
def test_timeout_must_be_int(self):
'''Test that the `timeout` config must be an int.'''
config = '{ "timeout": 30 }'
assert FisbrokerPlugin().validate_config(config)
# invalid timout:
config = '{ "timeout": "hurtz" }'
with assert_raises(ValueError):
assert FisbrokerPlugin().validate_config(config)
def test_import_since_must_be_valid_iso(self):
'''Test that the `import_since` config must be a valid ISO8601 date.'''
config = '{ "import_since": "2019-01-01" }'
assert FisbrokerPlugin().validate_config(config)
# invalid date:
config = '{ "import_since": "2019.01.01" }'
with assert_raises(ValueError):
assert FisbrokerPlugin().validate_config(config)
def test_import_since_date_is_none_if_no_jobs(self):
'''Test that, if the `import_since` setting is `last_error_free`, but
no jobs have run successfully (or at all), get_import_since_date()
returns None.'''
source, job = self._create_source_and_job()
FisbrokerPlugin().source_config['import_since'] = "last_error_free"
import_since = FisbrokerPlugin().get_import_since_date(job)
_assert_equal(import_since, None)
def test_skip_on_missing_release_date(self):
'''Test if get_package_dict() returns 'skip' for a service resource
without a release date.'''
data_dict = self._csw_resource_data_dict('wfs-no-release-date.xml')
# LOG.info("iso_valalala: %s", data_dict['iso_values'])
# assert False
_assert_equal(FisbrokerPlugin().get_package_dict(self.context, data_dict), 'skip')
def test_last_error_free_does_not_return_reimport_job(self):
'''Test that reimport jobs are ignored for determining
the last error-free job.'''
# do a successful job
source, job_a = self._create_source_and_job()
object_ids = gather_stage(FisbrokerPlugin(), job_a)
for object_id in object_ids:
harvest_object = HarvestObject.get(object_id)
fetch_and_import_stages(FisbrokerPlugin(), harvest_object)
job_a.status = u'Finished'
job_a.save()
LOG.debug("successful job done ...")
# do an unsuccessful job
# This harvest job should fail, because the mock FIS-broker will look for a different
# file on the second harvest run, will not find it and return a "no_record_found"
# error.
job_b = self._create_job(source.id)
object_ids = gather_stage(FisbrokerPlugin(), job_b)
for object_id in object_ids:
harvest_object = HarvestObject.get(object_id)
fetch_and_import_stages(FisbrokerPlugin(), harvest_object)
job_b.status = u'Finished'
job_b.save()
LOG.debug("unsuccessful job done ...")
# reset the mock server's counter
reset_mock_server(1)
# do a reimport job
package_id = "3d-gebaudemodelle-im-level-of-detail-2-lod-2-wms-f2a8a483"
self._get_test_app().get(
url="/api/harvest/reimport?id={}".format(package_id),
headers={'Accept': 'application/json'},
extra_environ={'REMOTE_USER': self.context['user'].encode('ascii')}
)
LOG.debug("reimport job done ...")
new_job = self._create_job(source.id)
last_error_free_job = FisbrokerPlugin().last_error_free_job(new_job)
# job_a should be the last error free job:
_assert_equal(last_error_free_job.id, job_a.id)
def test_last_error_free_returns_correct_job(self):
'''Test that, after a successful job A, last_error_free() returns A.'''
source, job = self._create_source_and_job()
object_ids = gather_stage(FisbrokerPlugin(), job)
for object_id in object_ids:
harvest_object = HarvestObject.get(object_id)
fetch_and_import_stages(FisbrokerPlugin(), harvest_object)
job.status = u'Finished'
job.save()
new_job = self._create_job(source.id)
last_error_free_job = FisbrokerPlugin().last_error_free_job(new_job)
_assert_equal(last_error_free_job, job)
# the import_since date should be the time job_a finished:
FisbrokerPlugin().source_config['import_since'] = "last_error_free"
import_since = FisbrokerPlugin().get_import_since_date(new_job)
import_since_expected = (job.gather_started +
timedelta(hours=FisbrokerPlugin().get_timedelta()))
_assert_equal(import_since, import_since_expected.strftime("%Y-%m-%dT%H:%M:%S%z"))
# the query constraints should reflect the import_since date:
constraint = FisbrokerPlugin().get_constraints(new_job)[0]
_assert_equal(constraint.literal, PropertyIsGreaterThanOrEqualTo(
'modified', import_since).literal)
_assert_equal(constraint.propertyname, PropertyIsGreaterThanOrEqualTo(
'modified', import_since).propertyname)
def _run_job_for_single_document(self, harvest_job, object_id):
harvester = FisbrokerPlugin()
# we circumvent gather_stage() and fetch_stage() and just load the
# content with a known object_id and create the harvest object:
url = harvest_job.source.url
# _get_content() returns XML
content = harvester._get_content(url)
obj = HarvestObject(guid=object_id,
job=harvest_job,
content=content,
extras=[HarvestObjectExtra(key='status',value='new')])
obj.save()
assert obj, obj.content
harvester.import_stage(obj)
Session.refresh(obj)
harvest_job.status = u'Finished'
harvest_job.save()
return obj
def test_last_error_free_does_not_return_unsuccessful_job(self):
'''Test that, after a successful job A, followed by an unsuccessful
job B, last_error_free() returns A.'''
source, job_a = self._create_source_and_job()
object_ids = gather_stage(FisbrokerPlugin(), job_a)
for object_id in object_ids:
harvest_object = HarvestObject.get(object_id)
fetch_and_import_stages(FisbrokerPlugin(), harvest_object)
job_a.status = u'Finished'
job_a.save()
# This harvest job should fail, because the mock FIS-broker will look for a different
# file on the second harvest run, will not find it and return a "no_record_found"
# error.
job_b = self._create_job(source.id)
object_ids = gather_stage(FisbrokerPlugin(), job_b)
for object_id in object_ids:
harvest_object = HarvestObject.get(object_id)
fetch_and_import_stages(FisbrokerPlugin(), harvest_object)
job_b.status = u'Finished'
job_b.save()
new_job = self._create_job(source.id)
last_error_free_job = FisbrokerPlugin().last_error_free_job(new_job)
# job_a should be the last error free job:
_assert_equal(last_error_free_job, job_a)
# the import_since date should be the time job_a finished:
FisbrokerPlugin().source_config['import_since'] = "last_error_free"
import_since = FisbrokerPlugin().get_import_since_date(new_job)
import_since_expected = (job_a.gather_started +
timedelta(hours=FisbrokerPlugin().get_timedelta()))
_assert_equal(import_since, import_since_expected.strftime("%Y-%m-%dT%H:%M:%S%z"))
# the query constraints should reflect the import_since date:
constraint = FisbrokerPlugin().get_constraints(new_job)[0]
_assert_equal(constraint.literal, PropertyIsGreaterThanOrEqualTo('modified', import_since).literal)
_assert_equal(constraint.propertyname, PropertyIsGreaterThanOrEqualTo(
'modified', import_since).propertyname)
def command(self):
'''Implementation of the paster command
'''
self._load_config()
if not self.args:
self.parser.print_usage()
sys.exit(1)
cmd = self.args[0]
if cmd == 'list_sources':
LOG.debug("listing all instances of FisbrokerPlugin ...")
sources = self.list_sources()
self.print_harvest_sources(sources)
elif cmd == 'list_datasets':
LOG.debug("listing datasets harvested by FisbrokerPlugin ...")
sources = [source.get('id') for source in self.list_sources()]
if len(self.args) >= 2:
sources = [unicode(self.args[1])]
for source in sources:
start = time.time()
packages = self.list_packages(source)
self.print_datasets(packages)
LOG.debug("there were %i results ...", len(packages))
end = time.time()
LOG.debug("This took %f seconds", end - start)
elif cmd == 'reimport_dataset':
LOG.debug("reimporting datasets ...")
package_ids = []
if self.options.dataset_id:
LOG.debug("reimporting a single dataset ...")
package_ids = [unicode(self.options.dataset_id)]
else:
sources = []
if self.options.source_id:
LOG.debug(
"reimporting all dataset from a single source: %s ...",
self.options.source_id)
sources = [unicode(self.options.source_id)]
else:
LOG.debug("reimporting all dataset from all sources ...")
sources = [
source.get('id') for source in self.list_sources()
]
for source in sources:
package_ids += [
package['name']
for package in self.list_packages(source)
]
start = time.time()
self.reimport_dataset(package_ids)
end = time.time()
LOG.debug("This took %f seconds", end - start)
elif cmd == 'last_successful_job':
class MockHarvestJob:
pass
sources = []
if self.options.source_id:
LOG.debug(
"finding last successful job from a single source: %s ...",
self.options.source_id)
sources = [unicode(self.options.source_id)]
else:
LOG.debug("finding last successful job from all sources ...")
sources = [source.get('id') for source in self.list_sources()]
for source in sources:
harvest_job = MockHarvestJob()
harvest_job.source = HarvestSource.get(source)
harvest_job.id = 'fakeid'
last_successful_job = FisbrokerPlugin.last_error_free_job(
harvest_job)
LOG.debug(last_successful_job)
else:
print 'Command %s not recognized' % cmd
def reimport_batch(self, package_ids, context):
'''Batch-reimport all packages in `package_ids` from their original
harvest source.'''
ckan_fb_mapping = {}
# first, do checks that can be done without connection to FIS-Broker
for package_id in package_ids:
package = Package.get(package_id)
if not package:
raise PackageIdDoesNotExistError(package_id)
if not dataset_was_harvested(package):
raise PackageNotHarvestedError(package_id)
harvester = harvester_for_package(package)
harvester_url = harvester.url
harvester_type = harvester.type
if not harvester_type == HARVESTER_ID:
raise PackageNotHarvestedInFisbrokerError(package_id)
fb_guid = fisbroker_guid(package)
if not fb_guid:
raise NoFisbrokerIdError(package_id)
ckan_fb_mapping[package.id] = fb_guid
# get the harvest source for FIS-Broker datasets
fb_source = get_fisbroker_source()
if not fb_source:
raise NoFBHarvesterDefined()
source_id = fb_source.get('id', None)
# Create and start a new harvest job
job_dict = toolkit.get_action('harvest_job_create')(context, {'source_id': source_id})
harvest_job = HarvestJob.get(job_dict['id'])
harvest_job.gather_started = datetime.datetime.utcnow()
assert harvest_job
# instatiate the CSW connector (on the reasonable assumption that harvester_url is
# the same for all package_ids)
package_id = None
reimported_packages = []
try:
csw = CatalogueServiceWeb(harvester_url)
for package_id, fb_guid in ckan_fb_mapping.items():
# query connector to get resource document
csw.getrecordbyid([fb_guid], outputschema=namespaces['gmd'])
# show resource document
record = csw.records.get(fb_guid, None)
if record:
obj = HarvestObject(guid=fb_guid,
job=harvest_job,
content=record.xml,
package_id=package_id,
extras=[
HarvestObjectExtra(key='status',value='change'),
HarvestObjectExtra(key='type',value='reimport'),
])
obj.save()
assert obj, obj.content
harvester = FisbrokerPlugin()
harvester.force_import = True
harvester.import_stage(obj)
rejection_reason = self._dataset_rejected(obj)
if rejection_reason:
raise FBImportError(package_id, rejection_reason)
harvester.force_import = False
Session.refresh(obj)
reimported_packages.append(record)
else:
raise NotFoundInFisbrokerError(package_id, fb_guid)
except RequestException as error:
raise NoConnectionError(package_id, harvester_url, str(error.__class__.__name__))
# successfully finish harvest job
harvest_job.status = u'Finished'
harvest_job.finished = datetime.datetime.utcnow()
harvest_job.save()
return reimported_packages
def test_skip_on_dataset_resource(self):
'''Test if get_package_dict() returns 'skip' for a dataset
CSW resource (as opposed to a service resource).'''
data_dict = self._csw_resource_data_dict('dataset-open-data.xml')
_assert_equal(FisbrokerPlugin().get_package_dict(self.context, data_dict), 'skip')
def test_skip_on_missing_email(self):
'''Test if get_package_dict() returns 'skip' for a service resource
without an email in the responsible party information.'''
data_dict = self._csw_resource_data_dict('wfs-no-email.xml')
_assert_equal(FisbrokerPlugin().get_package_dict(self.context, data_dict), 'skip')
def test_skip_on_missing_license_info(self):
'''Test if get_package_dict() returns 'skip' for a service resource
without parseable license information.'''
data_dict = self._csw_resource_data_dict('wfs-no-license.xml')
_assert_equal(FisbrokerPlugin().get_package_dict(self.context, data_dict), 'skip')
def test_empty_config(self):
'''Test that an empty config just returns unchanged.'''
_assert_equal(FisbrokerPlugin().validate_config(None), None)
_assert_equal(FisbrokerPlugin().validate_config({}), {})
def test_undefined_import_since_is_none(self):
'''Test that an undefined `import_since` config returns None.'''
FisbrokerPlugin().source_config = {}
import_since = FisbrokerPlugin().get_import_since_date(None)
_assert_equal(import_since, None)
def test_undefined_time_delta_gives_default(self):
'''Test that an undefined `timedelta` config returns the default.'''
FisbrokerPlugin().source_config = {}
timedelta = FisbrokerPlugin().get_timedelta()
_assert_equal(timedelta, TIMEDELTA_DEFAULT)
def test_undefined_timeout_gives_default(self):
'''Test that an undefined `timeout` config returns the default.'''
FisbrokerPlugin().source_config = {}
timeout = FisbrokerPlugin().get_timeout()
_assert_equal(timeout, TIMEOUT_DEFAULT)
def test_skip_on_closed_data_resource(self):
'''Test if get_package_dict() returns 'skip' for a closed data
CSW resource.'''
data_dict = self._csw_resource_data_dict('wfs-closed-data.xml')
_assert_equal(FisbrokerPlugin().get_package_dict(self.context, data_dict), 'skip')
