def test_get_participant_data(self, mock_get):
mock_get.return_value.status_code = 200
mock_get.return_value.json.return_value = self.json_response_entry
expected_response = psr.get_participant_data(self.fake_url,
self.fake_headers)
self.assertEqual(expected_response, self.participant_data)
def test_fake_website(self, mock_get, mock_token):
mock_token.return_value = self.fake_token
status_code = 500
error_msg = 'Error: API request failed because <Response [{status_code}]>'
mock_get.return_value = FakeHTTPResponse(status_code=status_code)
with self.assertRaises(RuntimeError) as e:
_ = psr.get_participant_data(self.fake_url, self.fake_headers)
self.assertEqual(str(e.exception),
error_msg.format(status_code=status_code))
self.assertEqual(mock_get.call_count, 1)
status_code = 404
mock_get.return_value = FakeHTTPResponse(status_code=status_code)
with self.assertRaises(RuntimeError) as e:
_ = psr.get_participant_data(self.fake_url, self.fake_headers)
self.assertEqual(str(e.exception),
error_msg.format(status_code=status_code))
self.assertEqual(mock_get.call_count, 2)
def test_get_participant_data(self, mock_get_session, mock_token):
mock_token.return_value = self.fake_token
mock_session = MagicMock()
mock_get_session.return_value = mock_session
mock_session.get.return_value.status_code = 200
mock_session.get.return_value.json.return_value = self.json_response_entry
actual_response = psr.get_participant_data(self.fake_url,
self.fake_headers)
self.assertEqual(actual_response, self.participant_data)
def test_get_participant_data(self, mock_get):
"""
Mocks calling the participant summary api.
"""
# pre conditions
mock_get.return_value.status_code = 200
mock_get.return_value.json.return_value = self.json_response_entry
# test
expected_response = psr.get_participant_data(self.url, self.headers)
# post conditions
self.assertEqual(expected_response, self.participant_data)
def get_participants(api_project_id, existing_pids):
"""
Method to hit the participant summary API based on cutoff dates and max age. Filters out participants that already
exist in the pipeline_tables._deid_map table.
Loops through the results to double check the API returned the correct participants.
:param api_project_id: project_id to send to API call
:param existing_pids: list of pids that already exist in mapping table
:return: dataframe with single column person_id from participant summary API, which needs RIDS created for
"""
# create datetimes from cutoff dates and datetimes to bin API call requests
bin_1_gt_datetime = datetime.strptime('2019-08-31T23:59:59',
'%Y-%m-%dT%H:%M:%S')
bin_1_lt_datetime = datetime.strptime('2020-08-02T00:00:00',
'%Y-%m-%dT%H:%M:%S')
bin_2_datetime = datetime.strptime('2019-01-01T00:00:00',
'%Y-%m-%dT%H:%M:%S')
bin_3_gt_datetime = datetime.strptime('2018-12-31T23:59:59',
'%Y-%m-%dT%H:%M:%S')
bin_3_lt_datetime = datetime.strptime('2019-09-01T00:00:00',
'%Y-%m-%dT%H:%M:%S')
# Make request to get API version. This is the current RDR version for reference
# See https://github.com/all-of-us/raw-data-repository/blob/master/opsdataAPI.md for documentation of this api.
request_url_cutoff_participants = "https://{0}.appspot.com/rdr/v1/ParticipantSummary?_sort=" \
"consentForStudyEnrollmentAuthored&withdrawalStatus={1}" \
"&consentForStudyEnrollmentAuthored=gt{2}&consentForStudyEnrollmentAuthored=" \
"lt{3}".format(api_project_id, 'NOT_WITHDRAWN', bin_1_gt_datetime,
bin_1_lt_datetime)
request_url_max_age_participants_1 = "https://{0}.appspot.com/rdr/v1/ParticipantSummary?_sort=" \
"consentForStudyEnrollmentAuthored&withdrawalStatus={1}" \
"&consentForStudyEnrollmentAuthored=lt{2}".format(api_project_id,
'NOT_WITHDRAWN',
bin_2_datetime)
request_url_max_age_participants_2 = "https://{0}.appspot.com/rdr/v1/ParticipantSummary?_sort=" \
"consentForStudyEnrollmentAuthored&withdrawalStatus={1}" \
"&consentForStudyEnrollmentAuthored=gt{2}&consentForStudyEnrollmentAuthored=" \
"lt{3}".format(api_project_id, 'NOT_WITHDRAWN', bin_3_gt_datetime,
bin_3_lt_datetime)
list_url_requests = [
request_url_cutoff_participants, request_url_max_age_participants_1,
request_url_max_age_participants_2
]
participant_data = []
# loop through urls and create new tokens each request to avoid API timing out
for url in list_url_requests:
token = get_access_token()
headers = {
'content-type': 'application/json',
'Authorization': 'Bearer {0}'.format(token)
}
participant_data = participant_data + get_participant_data(
url, headers)
participants = pd.DataFrame(columns=['person_id'])
# Loop through participant_data to retrieve only person_id
for participant in participant_data:
participant = participant['resource']
participant_id = int(participant['participantId'].replace('P', ''))
# remove pids that already exist in mapping table
if participant_id not in existing_pids:
# turn string into datetime to compare
participants = participants.append({'person_id': participant_id},
ignore_index=True)
return participants.drop_duplicates()