def test_load_run(self, mock_harvest_time, mock_check_if_identifier_exists,
mock_update_science_metadata,
mock_load_science_metadata):
"""
SCENARIO: One document is to be loaded for the first time, not
updated.
EXPECTED RESULT: The call counts must reflect that the load routine
is called and not the update routine. Verify that the sid is a UUID and
that the pid is the MD5SUM of the metadata document.
"""
mock_harvest_time.return_value = '1900-01-01T00:00:00Z'
mock_check_if_identifier_exists.return_value = {'outcome': 'no'}
mock_update_science_metadata.return_value = True
mock_load_science_metadata.return_value = True
harvester = NKNHarvester(host=self.host, port=self.port)
# External calls to read the:
#
# 1) sitemap (raw HTML directory listing)
# 2) Remote HTML document for record 1 (another directory listing)
# 3) Remote XML document for record 1
#
uuid = '0a42d2bc-700a-4cf2-a7ac-ad6b892da7f0'
contents = [
ir.read_binary('tests.data.nkn', 'index.html'),
ir.read_binary(f'tests.data.nkn.{uuid}', 'index.html'),
ir.read_binary(f'tests.data.nkn.{uuid}', 'metadata.xml'),
]
status_codes = [200, 200, 200]
headers = [
{
'Content-Type': 'text/html;charset=UTF-8'
},
{
'Content-Type': 'text/html;charset=UTF-8'
},
{
'Content-Type': 'application/xml'
},
]
regex = [
re.compile('https://www.northwestknowledge.net/data/'),
re.compile('https://www.northwestknowledge.net/data/'),
re.compile('https://www.northwestknowledge.net/data/'),
]
with aioresponses() as m:
z = zip(regex, contents, status_codes, headers)
for regex, content, status_code, headers in z:
m.get(regex, body=content, status=status_code, headers=headers)
with self.assertLogs(logger=harvester.logger, level='DEBUG'):
asyncio.run(harvester.run())
self.assertEqual(mock_load_science_metadata.call_count, 1),
self.assertEqual(mock_update_science_metadata.call_count, 0),
# Verify the PID and SID
args, kwargs = mock_load_science_metadata.call_args_list[0]
actual = kwargs['system_metadata'].identifier.value()
expected = '679742d8c458378928ed21b2868db95b'
self.assertEqual(actual, expected)
actual = kwargs['system_metadata'].seriesId.value()
expected = uuid
self.assertEqual(actual, expected)
def test_update_run(self, mock_harvest_time,
mock_check_if_identifier_exists,
mock_update_science_metadata,
mock_load_science_metadata):
"""
SCENARIO: One document is to be updated.
EXPECTED RESULT: The document is updated, not loaded for the first
time. Verify that the sid is a UUID and that the pid is the MD5SUM of
the metadata document.
"""
record_date = dt.datetime(2017,
4,
28,
10,
44,
0,
tzinfo=dt.timezone.utc)
mock_harvest_time.return_value = record_date.strftime(DATETIME_FORMAT)
mock_check_if_identifier_exists.return_value = {
'outcome': 'yes',
'record_date': record_date - dt.timedelta(days=1),
'current_version_id': 1,
}
mock_update_science_metadata.return_value = True
mock_load_science_metadata.return_value = True
harvester = NKNHarvester(host=self.host, port=self.port)
# External calls to read the:
#
# 1) sitemap (raw HTML directory listing)
# 2) Remote HTML document for record 1 (another directory listing)
# 3) Remote XML document for record 1
# 4) Existing XML document for record 1 (retrieved from the member
# node)
#
uuid = '0a42d2bc-700a-4cf2-a7ac-ad6b892da7f0'
contents = [
ir.read_binary('tests.data.nkn', 'index.html'),
ir.read_binary(f'tests.data.nkn.{uuid}', 'index.html'),
ir.read_binary(f'tests.data.nkn.{uuid}', 'metadata.xml'),
ir.read_binary(f'tests.data.nkn.{uuid}', 'metadata.prior.xml')
]
status_codes = [200, 200, 200, 200]
headers = [
{
'Content-Type': 'text/html;charset=UTF-8'
},
{
'Content-Type': 'text/html;charset=UTF-8'
},
{
'Content-Type': 'application/xml'
},
{
'Content-Type': 'application/xml'
},
]
regex = [
re.compile('https://www.northwestknowledge.net/data/'),
re.compile('https://www.northwestknowledge.net/data/'),
re.compile('https://www.northwestknowledge.net/data/'),
re.compile('https://nkn.mn.org:443/mn/v2/'),
]
with aioresponses() as m:
z = zip(regex, contents, status_codes, headers)
for regex, content, status_code, headers in z:
m.get(regex, body=content, status=status_code, headers=headers)
with self.assertLogs(logger=harvester.logger, level='DEBUG'):
asyncio.run(harvester.run())
self.assertEqual(mock_load_science_metadata.call_count, 0),
self.assertEqual(mock_update_science_metadata.call_count, 1),
# Verify the PID and SID
args, kwargs = mock_update_science_metadata.call_args_list[0]
actual = kwargs['system_metadata'].identifier.value()
expected = '679742d8c458378928ed21b2868db95b'
self.assertEqual(actual, expected)
actual = kwargs['system_metadata'].seriesId.value()
expected = uuid
self.assertEqual(actual, expected)