fix: exclude 'restPath' field from being collected

collector_module/opmon_collector/collector_worker.py

@@ -70,7 +70,9 @@ def work(self):
         for _ in range(self.settings['collector']['repeat-limit']):
             try:
                 response = self._request_opmon_data()
-                self.records = self._parse_attachment(response)
+                records = self._parse_attachment(response)
+                sanitized_records = self._sanitize_records(records)
+                self.records = sanitized_records
                 if self.settings['collector'].get('documents-log-directory', ''):
                     self._store_records_to_file()
                 self._store_records_to_database()
@@ -196,6 +198,19 @@ def _parse_attachment(self, opmon_response):
             self.log_exception('Cannot parse response attachment.', str(e))
             raise e
 
+    @staticmethod
+    def _sanitize_records(records):
+        """
+        Temporary solution to address a privacy concern - remove 'restPath' field from records.
+        To be removed after X-Road version 7.6.2 release.
+        """
+        sanitized_records = []
+        for record in records:
+            sanitized_record = record.copy()
+            sanitized_record.pop('restPath', None)
+            sanitized_records.append(sanitized_record)
+        return sanitized_records
+
     def _get_records_logger(self) -> logging.Logger:
         host_name = re.sub('[^0-9a-zA-Z.-]+', '.', self.server_data['server'])
         records_logger = logging.getLogger(host_name)

collector_module/opmon_collector/tests/test_collector_worker.py

@@ -195,6 +195,25 @@ def test_worker_status(mock_server_manager, basic_data):
     assert worker.status == CollectorWorker.Status.ALL_COLLECTED
 
 
+def test_sanitize_records(mock_server_manager, basic_data):
+    worker = CollectorWorker(basic_data)
+
+    records = [
+        {"id": 1, "foo": "valid_data_foo", "restPath": "some_path/*"},
+        {"id": 2, "bar": "valid_data_bar", "restPath": "some_path/*"},
+        {"id": 3, "data3": "valid_data", "restPath": "some_path/*"},
+        {"restPath": "some_path/*", "id": 4, "data4": "valid_data4"},
+        {"data5": "valid_data5", "restPath": "some_path/*", "id": 5}
+    ]
+
+    sanitized_records = worker._sanitize_records(records)
+
+    assert len(sanitized_records) == 5
+    for record in sanitized_records:
+        assert 'restPath' not in record
+        assert 'id' in record
+
+
 @responses.activate
 @pytest.mark.parametrize(
     'mock_response_contents', [('metrics_client_proxy_ssl_auth_failed.dat',)], indirect=True