YUPANA-42 - transform bios UUID field to omit invalid value

yupana/processor/report_slice_processor.py

@@ -23,6 +23,7 @@
 import re
 import threading
 
+from uuid import UUID
 from aiokafka import AIOKafkaProducer
 from kafka.errors import KafkaConnectionError
 from processor.abstract_processor import (AbstractProcessor, FAILED_TO_VALIDATE)
@@ -247,6 +248,31 @@ def _remove_empty_mac_addresses(self, host: dict):
             ))
         return host
 
+    @staticmethod
+    def is_valid_uuid(uuid):
+        """Validate a UUID string."""
+        try:
+            uuid_obj = UUID(uuid)
+        except ValueError:
+            return False
+
+        return str(uuid_obj) == uuid
+
+    def _remove_invalid_bios_uuid(self, host):
+        """Remove invalid bios UUID."""
+        uuid = host.get('bios_uuid')
+        if not isinstance(uuid, str):
+            return host
+
+        if not self.is_valid_uuid(uuid):
+            LOG.error(format_message(
+                self.prefix,
+                "Invalid uuid: %s for host with FQDN '%s'"
+                % (uuid, host.get('fqdn', ''))))
+            del host['bios_uuid']
+
+        return host
+
     def _match_regex_and_find_os_details(self, os_release):
         """Match Regex with os_release and return os_details."""
         source_os_release = os_release.strip()
@@ -397,6 +423,7 @@ def _transform_single_host(self, host: dict):
         host = self._remove_empty_ip_addresses(host)
         host = self._remove_empty_mac_addresses(host)
         host = self._remove_display_name(host)
+        host = self._remove_invalid_bios_uuid(host)
         return host
 
     # pylint:disable=too-many-locals

yupana/processor/tests_report_slice_processor.py

@@ -268,7 +268,7 @@ def test_determine_retry_limit(self):
 
     async def async_test_transition_to_hosts_uploaded(self):
         """Test the transition to hosts being uploaded."""
-        hosts = [{str(self.uuid): {'bios_uuid': 'value', 'name': 'value',
+        hosts = [{str(self.uuid): {'bios_uuid': str(self.uuid), 'name': 'value',
                                    'system_platform_id': str(self.uuid)}},
                  {str(self.uuid2): {'insights_client_id': 'value', 'name': 'foo',
                                     'system_platform_id': str(self.uuid2)}},
@@ -303,7 +303,7 @@ def test_transition_to_hosts_uploaded(self):
 
     async def async_test_transition_to_hosts_uploaded_kafka_mode(self):
         """Test the transition to hosts being uploaded."""
-        hosts = [{str(self.uuid): {'bios_uuid': 'value', 'name': 'value',
+        hosts = [{str(self.uuid): {'bios_uuid': str(self.uuid), 'name': 'value',
                                    'system_platform_id': str(self.uuid)}},
                  {str(self.uuid2): {'insights_client_id': 'value', 'name': 'foo',
                                     'system_platform_id': str(self.uuid2)}},
@@ -373,7 +373,7 @@ def test_test_transition_to_hosts_uploaded_no_candidates(self):
 
     async def async_test_transition_to_hosts_uploaded_exception(self):
         """Test the transition to hosts being uploaded."""
-        hosts = {str(self.uuid): {'bios_uuid': 'value', 'name': 'value'},
+        hosts = {str(self.uuid): {'bios_uuid': str(self.uuid), 'name': 'value'},
                  str(self.uuid2): {'insights_client_id': 'value', 'name': 'foo'},
                  str(self.uuid3): {'ip_addresses': 'value', 'name': 'foo'},
                  str(self.uuid4): {'mac_addresses': 'value', 'name': 'foo'},
@@ -406,7 +406,7 @@ async def async_test_upload_to_host_inventory_via_kafka(self):
         """Test uploading to inventory via kafka."""
         self.processor.report_or_slice = self.report_slice
         hosts = {
-            str(self.uuid): {'bios_uuid': 'value', 'name': 'value'},
+            str(self.uuid): {'bios_uuid': str(self.uuid), 'name': 'value'},
             str(self.uuid2): {'insights_client_id': 'value', 'name': 'foo'},
             str(self.uuid3): {'ip_addresses': 'value', 'name': 'foo'},
             str(self.uuid4): {'mac_addresses': 'value', 'name': 'foo'},
@@ -442,7 +442,7 @@ def test_upload_to_host_inventory_via_kafka(self):
     async def async_test_upload_to_host_inventory_via_kafka_exception(self):
         """Test uploading to inventory via kafka."""
         self.processor.report_or_slice = self.report_slice
-        hosts = {str(self.uuid): {'bios_uuid': 'value', 'name': 'value'},
+        hosts = {str(self.uuid): {'bios_uuid': str(self.uuid), 'name': 'value'},
                  str(self.uuid2): {'insights_client_id': 'value', 'name': 'foo'},
                  str(self.uuid3): {'ip_addresses': 'value', 'name': 'foo'},
                  str(self.uuid4): {'mac_addresses': 'value', 'name': 'foo'},
@@ -479,7 +479,7 @@ def test_upload_to_host_inventory_via_kafka_exception(self):
     async def async_test_upload_to_host_inventory_via_kafka_send_exception(self):
         """Test uploading to inventory via kafka."""
         self.processor.report_or_slice = self.report_slice
-        hosts = {str(self.uuid): {'bios_uuid': 'value', 'name': 'value'},
+        hosts = {str(self.uuid): {'bios_uuid': str(self.uuid), 'name': 'value'},
                  str(self.uuid2): {'insights_client_id': 'value', 'name': 'foo'},
                  str(self.uuid3): {'ip_addresses': 'value', 'name': 'foo'},
                  str(self.uuid4): {'mac_addresses': 'value', 'name': 'foo'},
@@ -855,3 +855,30 @@ def test_remove_empty_strings_in_ipv6_addresses(self):
         filtered_nics = [nic for nic in nics if nic.get('name') == 'eth0']
         self.assertTrue(len(filtered_nics))
         self.assertEqual(len(filtered_nics[0]['ipv6_addresses']), 1)
+
+    def test_remove_invalid_bios_uuid(self):
+        """Test remove invalid bios UUID."""
+        host = {
+            'fqdn': 'virt-who.example.com',
+            'bios_uuid': '45AA7104-5CB0-4A75-945D-7173C8DC5744443'
+        }
+        host = self.processor._remove_invalid_bios_uuid(host)
+        self.assertEqual(host, {'fqdn': 'virt-who.example.com'})
+
+    def test_remove_empty_bios_uuid(self):
+        """Test remove empty bios UUID field."""
+        host = {
+            'fqdn': 'virt-who.example.com',
+            'bios_uuid': ''
+        }
+        host = self.processor._remove_invalid_bios_uuid(host)
+        self.assertEqual(host, {'fqdn': 'virt-who.example.com'})
+
+    def test_do_not_remove_valid_bios_uuid(self):
+        """Test do not remove valid bios UUID."""
+        host = {
+            'fqdn': 'virt-who.example.com',
+            'bios_uuid': '123e4567-e89b-12d3-a456-426614174000'
+        }
+        new_host = self.processor._remove_invalid_bios_uuid(host)
+        self.assertEqual(new_host, host)