Allow to exclude some fields from diff_against check (#680)

* [#674] Allow to exclude some fields from `diff_against` check

* Add command test

* Run make format command

* Edit CHANGES.rst

* Fix help message

Author: MaximZemskov

AUTHORS.rst

@@ -74,6 +74,7 @@ Authors
 - Marty Alchin
 - Matheus Cansian (`mscansian <https://github.com/mscansian>`_)
 - Mauricio de Abreu Antunes
+- Maxim Zemskov (`MaximZemskov <https://github.com/MaximZemskov>`_)
 - Micah Denbraver
 - Michael England
 - Miguel Vargas

CHANGES.rst

@@ -5,6 +5,7 @@ Unreleased
 ----------
 - Add default date to ``bulk_create_with_history`` and ``bulk_update_with_history`` (gh-687)
 - Exclude ManyToManyFields when using ``bulk_create_with_history`` (gh-699)
+- Added ``--excluded_fields`` argument to ``clean_duplicate_history`` command (gh-674)
 
 2.11.0 (2020-06-20)
 -------------------

docs/utils.rst

@@ -24,6 +24,12 @@ so you can schedule, for instance, an hourly cronjob such as
 
     $ python manage.py clean_duplicate_history -m 60 --auto
 
+You can also use ``--excluded_fields`` to provide a list of fields to be excluded
+from the duplicate check
+
+.. code-block:: bash
+
+    $ python manage.py clean_duplicate_history --auto --excluded_fields field1 field2
 
 clean_old_history
 -----------------------

simple_history/management/commands/clean_duplicate_history.py

@@ -31,9 +31,15 @@ def add_arguments(self, parser):
         parser.add_argument(
             "-m", "--minutes", type=int, help="Only search the last MINUTES of history"
         )
+        parser.add_argument(
+            "--excluded_fields",
+            nargs="+",
+            help="List of fields to be excluded from the diff_against check",
+        )
 
     def handle(self, *args, **options):
         self.verbosity = options["verbosity"]
+        self.excluded_fields = options.get("excluded_fields")
 
         to_process = set()
         model_strings = options.get("models", []) or args
@@ -109,7 +115,7 @@ def log(self, message, verbosity_level=1):
             self.stdout.write(message)
 
     def _check_and_delete(self, entry1, entry2, dry_run=True):
-        delta = entry1.diff_against(entry2)
+        delta = entry1.diff_against(entry2, excluded_fields=self.excluded_fields)
         if not delta.changed_fields:
             if not dry_run:
                 entry1.delete()

simple_history/models.py

@@ -586,19 +586,22 @@ def __get__(self, instance, owner):
 
 
 class HistoricalChanges(object):
-    def diff_against(self, old_history):
+    def diff_against(self, old_history, excluded_fields=None):
         if not isinstance(old_history, type(self)):
             raise TypeError(
                 ("unsupported type(s) for diffing: " "'{}' and '{}'").format(
                     type(self), type(old_history)
                 )
             )
-
+        if excluded_fields is None:
+            excluded_fields = []
         changes = []
         changed_fields = []
         old_values = model_to_dict(old_history.instance)
         current_values = model_to_dict(self.instance)
         for field, new_value in current_values.items():
+            if field in excluded_fields:
+                continue
             if field in old_values:
                 old_value = old_values[field]
                 if old_value != new_value:

simple_history/tests/tests/test_commands.py

@@ -386,6 +386,29 @@ def test_auto_cleanup_custom_history_field(self):
         )
         self.assertEqual(CustomManagerNameModel.log.all().count(), 2)
 
+    def test_auto_cleanup_with_excluded_fields(self):
+        p = Poll.objects.create(
+            question="Will this be deleted?", pub_date=datetime.now()
+        )
+        self.assertEqual(Poll.history.all().count(), 1)
+        p.pub_date = p.pub_date + timedelta(days=1)
+        p.save()
+        self.assertEqual(Poll.history.all().count(), 2)
+        out = StringIO()
+        management.call_command(
+            self.command_name,
+            auto=True,
+            excluded_fields=("pub_date",),
+            stdout=out,
+            stderr=StringIO(),
+        )
+        self.assertEqual(
+            out.getvalue(),
+            "Removed 1 historical records for "
+            "<class 'simple_history.tests.models.Poll'>\n",
+        )
+        self.assertEqual(Poll.history.all().count(), 1)
+
 
 class TestCleanOldHistory(TestCase):
     command_name = "clean_old_history"

simple_history/tests/tests/test_models.py

@@ -659,6 +659,15 @@ def test_history_diff_with_incorrect_type(self):
         with self.assertRaises(TypeError):
             new_record.diff_against("something")
 
+    def test_history_diff_with_excluded_fields(self):
+        p = Poll.objects.create(question="what's up?", pub_date=today)
+        p.question = "what's up, man?"
+        p.save()
+        new_record, old_record = p.history.all()
+        delta = new_record.diff_against(old_record, excluded_fields=("question",))
+        self.assertEqual(delta.changed_fields, [])
+        self.assertEqual(delta.changes, [])
+
 
 class GetPrevRecordAndNextRecordTestCase(TestCase):
     def assertRecordsMatch(self, record_a, record_b):