def test_checks_records_error():
setup_main_database()
with tempfile.TemporaryDirectory() as tmpdir:
metadata_db = MetadataDB(tmpdir)
metadata_db.create_session_metadata("Test", True,
"http://www.test.com",
"2018-01-01-10-00-00")
metadata_db.add_filestatus({
'filename': 'test1.json',
'url': 'http://www.test.com',
'data_type': 'record_package'
})
# store details
source_session_id = database.start_store("test_source",
"2018-01-01-10-00-00", True,
metadata_db)
for data in metadata_db.list_filestatus():
with database.add_file(source_session_id, data) as database_file:
database_file.insert_record({'record': 'totally'}, {
'version': '0.1-does-not-exist',
'extensions': []
})
database.end_store(source_session_id)
record_id = 1
# Don't like hard coding ID in. Relies on DB assigning 1 to this new row. But I think we can assume that.
# Test
assert not database.is_record_check_done(record_id)
assert not database.is_check_done(source_session_id)
# check!
for data in metadata_db.list_filestatus():
checks.check_file(source_session_id, data)
# Test
assert database.is_record_check_done(record_id)
assert database.is_check_done(source_session_id)
with database.engine.begin() as connection:
s = sa.sql.select([database.record_check_error_table])
result = connection.execute(s)
data = result.fetchone()
assert 'The schema version in your data is not valid. Accepted values:' in data[
'error']
def test_checks_releases():
setup_main_database()
with tempfile.TemporaryDirectory() as tmpdir:
metadata_db = MetadataDB(tmpdir)
metadata_db.create_session_metadata("Test", True,
"http://www.test.com",
"2018-01-01-10-00-00")
metadata_db.add_filestatus({
'filename': 'test1.json',
'url': 'http://www.test.com',
'data_type': 'release_package'
})
# store details
source_session_id = database.start_store("test_source",
"2018-01-01-10-00-00", True,
metadata_db)
for data in metadata_db.list_filestatus():
with database.add_file(source_session_id, data) as database_file:
database_file.insert_release({'release': 'totally'},
{'extensions': []})
database.end_store(source_session_id)
release_id = 1
# Don't like hard coding ID in. Relies on DB assigning 1 to this new row. But I think we can assume that.
# Test
assert not database.is_release_check_done(release_id)
# check!
for data in metadata_db.list_filestatus():
checks.check_file(source_session_id, data)
# Test
assert database.is_release_check_done(release_id)
with database.engine.begin() as connection:
s = sa.sql.select([database.release_check_table])
result = connection.execute(s)
data = result.fetchone()
assert data['cove_output']['file_type'] == 'json'
assert len(data['cove_output']['validation_errors']) > 0
def run_store(self):
metadata = self.metadata_db.get_session()
if not metadata['fetch_success']:
raise Exception('Can not run store without a successful fetch')
if database.is_store_done(self.source_id, self.data_version,
self.sample):
return
source_session_id = database.start_store(self.source_id,
self.data_version,
self.sample, self.metadata_db)
for data in self.metadata_db.list_filestatus():
if data['data_type'].startswith('meta'):
continue
if database.is_store_file_done(source_session_id, data):
continue
with database.add_file(source_session_id, data) as database_file:
try:
with open(os.path.join(self.full_directory,
data['filename']),
encoding=data['encoding']) as f:
file_json_data = json.load(f)
except Exception as e:
# TODO better way of dealing with this?
raise e
return
objects_list = []
if data['data_type'] == 'record_package_list_in_results':
objects_list.extend(file_json_data['results'])
elif data['data_type'] == 'release_package_list_in_results':
objects_list.extend(file_json_data['results'])
elif data['data_type'] == 'record_package_list' or data[
'data_type'] == 'release_package_list':
objects_list.extend(file_json_data)
else:
objects_list.append(file_json_data)
del file_json_data
for json_data in objects_list:
error_msg = ''
if not isinstance(json_data, dict):
error_msg = "Can not process data in file {} as JSON is not an object".format(
data['filename'])
if data['data_type'] == 'release_package' or \
data['data_type'] == 'release_package_list_in_results' or \
data['data_type'] == 'release_package_list':
if 'releases' not in json_data:
error_msg = "Release list not found in file {}".format(
data['filename'])
elif not isinstance(json_data['releases'], list):
error_msg = "Release list which is not a list found in file {}".format(
data['filename'])
data_list = json_data['releases']
elif data['data_type'] == 'record_package' or \
data['data_type'] == 'record_package_list_in_results' or \
data['data_type'] == 'record_package_list':
if 'records' not in json_data:
error_msg = "Record list not found in file {}".format(
data['filename'])
elif not isinstance(json_data['records'], list):
error_msg = "Record list which is not a list found in file {}".format(
data['filename'])
data_list = json_data['records']
else:
error_msg = "data_type not a known type"
if error_msg:
raise Exception(error_msg)
package_data = {}
for key, value in json_data.items():
if key not in ('releases', 'records'):
package_data[key] = value
for row in data_list:
if not isinstance(row, dict):
error_msg = "Row in data is not a object {}".format(
data['filename'])
raise Exception(error_msg)
if data['data_type'] == 'record_package' or \
data['data_type'] == 'record_package_list_in_results' or \
data['data_type'] == 'record_package_list':
database_file.insert_record(row, package_data)
else:
database_file.insert_release(row, package_data)
database.end_store(source_session_id)