Add import history cleanup task and max_import_history setting fixes #13776

Author: Tejas Saubhage

dojo/db_migrations/0262_system_settings_max_import_history.py

@@ -0,0 +1,22 @@
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("dojo", "0261_remove_url_insert_insert_remove_url_update_update_and_more"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="system_settings",
+            name="max_import_history",
+            field=models.IntegerField(
+                blank=True,
+                null=True,
+                default=None,
+                verbose_name="Max Import History",
+                help_text="When set, the oldest import history records will be deleted when a test exceeds this number of imports. Leave empty to keep all history.",
+            ),
+        ),
+    ]

dojo/models.py

@@ -325,6 +325,9 @@ class System_Settings(models.Model):
                                               "issue reaches the maximum "
                                               "number of duplicates, the "
                                               "oldest will be deleted. Duplicate will not be deleted when left empty. A value of 0 will remove all duplicates."))
+    max_import_history = models.IntegerField(blank=True, null=True, default=None,
+                                    verbose_name=_("Max Import History"),
+                                    help_text=_("When set, the oldest import history records will be deleted when a test exceeds this number of imports. Leave empty to keep all history."))
 
     email_from = models.CharField(max_length=200, default="no-reply@example.com", blank=True)
 

dojo/settings/settings.dist.py

@@ -280,6 +280,7 @@
     # we limit the amount of duplicates that can be deleted in a single run of that job
     # to prevent overlapping runs of that job from occurrring
     DD_DUPE_DELETE_MAX_PER_RUN=(int, 200),
+    DD_IMPORT_HISTORY_MAX_PER_OBJECT=(int, 200),
     # when enabled 'mitigated date' and 'mitigated by' of a finding become editable
     DD_EDITABLE_MITIGATED_DATA=(bool, False),
     # new feature that tracks history across multiple reimports for the same test
@@ -1773,6 +1774,7 @@ def saml2_attrib_map_format(din):
             DEDUPLICATION_ALGORITHM_PER_PARSER[key] = value
 
 DUPE_DELETE_MAX_PER_RUN = env("DD_DUPE_DELETE_MAX_PER_RUN")
+IMPORT_HISTORY_MAX_PER_OBJECT = env("DD_IMPORT_HISTORY_MAX_PER_OBJECT")
 
 DISABLE_FINDING_MERGE = env("DD_DISABLE_FINDING_MERGE")
 

dojo/tasks.py

@@ -207,7 +207,44 @@ def jira_status_reconciliation_task(*args, **kwargs):
         return jira_status_reconciliation(*args, **kwargs)
 
 
-@app.task
+
+@app.task(bind=True)
+def async_import_history_cleanup(*args, **kwargs):
+    with pghistory.context(source="import_history_cleanup_task"):
+        _async_import_history_cleanup_impl()
+
+
+def _async_import_history_cleanup_impl():
+    """Delete oldest Test_Import records when a test exceeds max_import_history."""
+    try:
+        system_settings = System_Settings.objects.get()
+        max_history = system_settings.max_import_history
+        max_per_run = settings.IMPORT_HISTORY_MAX_PER_OBJECT
+    except System_Settings.DoesNotExist:
+        return
+
+    if max_history is None:
+        logger.info("skipping import history cleanup: max_import_history not configured")
+        return
+
+    logger.info("cleaning up import history (max per test: %s, max deletes per run: %s)", max_history, max_per_run)
+
+    tests_with_excess = Test_Import.objects \
+        .values("test") \
+        .annotate(import_count=Count("id")) \
+        .filter(import_count__gt=max_history)[:max_per_run]
+
+    total_deleted_count = 0
+    for entry in tests_with_excess:
+        test_id = entry["test"]
+        imports = Test_Import.objects.filter(test_id=test_id).order_by("created")
+        excess_count = entry["import_count"] - max_history
+        for test_import in imports[:excess_count]:
+            logger.debug("deleting Test_Import id %s for test %s", test_import.id, test_id)
+            test_import.delete()
+            total_deleted_count += 1
+
+    logger.info("total import history records deleted: %s", total_deleted_count)
 def fix_loop_duplicates_task(*args, **kwargs):
     # Wrap with pghistory context for audit trail
     with pghistory.context(source="fix_loop_duplicates"):