fix(backup): Apply prod-UserOption fix to comparison (#69460)

The fixes in #68943 and
#69027 worked, but now fail on
comparison. This PR makes comparison clever enough to detect these
"duplicate" UserOption entries and avoid them.

This is a temporary fix. A more robust solution will prevent ever
exporting such UserOptions, but that will take a while to percolate
through the support window, so this should hold us over for the time
being.

Author: azaslavsky

src/sentry/backup/validate.py

@@ -91,6 +91,10 @@ def build_model_map(
         need_ordering: dict[NormalizedModelName, dict[tuple, JSONData]] = defaultdict(dict)
         pks_to_usernames: dict[int, str] = dict()
 
+        # TODO(azaslavsky): This is a temporary fix. Remove it once we have a more
+        # sustainable export-side solution.
+        user_options_seen_ids: dict[tuple[int, str], InstanceID] = dict()
+
         for model in models:
             pk = model["pk"]
             model_name = NormalizedModelName(model["model"])
@@ -106,6 +110,20 @@ def build_model_map(
                 id, found = ordinal_counters[model_name].assign(model, pk, side)
                 findings.extend(found)
                 model_map[model_name][id] = model
+
+                # TODO(azaslavsky): This is a temporary fix. Remove it once we have a more
+                # sustainable export-side solution.
+                if (
+                    str(model_name) == "sentry.useroption"
+                    and not model["fields"].get("project_id", None)
+                    and not model["fields"].get("organization_id", None)
+                ):
+                    duplicate_id = user_options_seen_ids.pop(
+                        (model["fields"]["user"], model["fields"]["key"]), None
+                    )
+                    user_options_seen_ids[(model["fields"]["user"], model["fields"]["key"])] = id
+                    if duplicate_id is not None:
+                        del model_map[model_name][duplicate_id]
                 continue
 
             custom_ordinal_parts = []
@@ -139,6 +157,18 @@ def build_model_map(
                 findings.extend(found)
                 model_map[model_name][id] = model
 
+        # TODO(azaslavsky): This is a temporary fix. Remove it once we have a more sustainable
+        # export-side solution.
+        user_option_model_name = NormalizedModelName("sentry.useroption")
+        user_option_model_map = model_map.get(user_option_model_name, None)
+        if user_option_model_map is not None:
+            counter = 1
+            model_map[user_option_model_name] = ordereddict()
+            for model in user_option_model_map.values():
+                id = InstanceID(str(user_option_model_name), counter)
+                model_map[user_option_model_name][id] = model
+                counter += 1
+
         return (model_map, ordinal_counters)
 
     def json_lines(obj: JSONData) -> list[str]:

tests/sentry/backup/test_validate.py

@@ -792,3 +792,101 @@ def test_good_user_custom_ordinal():
     findings = out.findings
 
     assert len(findings) == 0
+
+
+# TODO(azaslavsky): This is a temporary fix. Remove it once we have a more sustainable export-side
+# solution.
+def test_good_user_option_temp_fix():
+    left = json.loads(
+        """
+            [
+                {
+                    "model": "sentry.user",
+                    "pk": 1,
+                    "fields": {
+                        "password": "abc123",
+                        "last_login": null,
+                        "username": "testing@example.com",
+                        "name": "",
+                        "email": "testing@example.com"
+                    }
+                },
+                {
+                    "model": "sentry.useroption",
+                    "pk": 3,
+                    "fields": {
+                        "user": 1,
+                        "key": "foo",
+                        "value": "1"
+                    }
+                },
+                {
+                    "model": "sentry.useroption",
+                    "pk": 5,
+                    "fields": {
+                        "user": 1,
+                        "key": "foo",
+                        "value": "2"
+                    }
+                },
+                {
+                    "model": "sentry.useroption",
+                    "pk": 8,
+                    "fields": {
+                        "user": 1,
+                        "key": "bar",
+                        "value": "4"
+                    }
+                }
+            ]
+        """
+    )
+
+    right = json.loads(
+        """
+            [
+                {
+                    "model": "sentry.user",
+                    "pk": 1,
+                    "fields": {
+                        "password": "abc123",
+                        "last_login": null,
+                        "username": "testing@example.com",
+                        "name": "",
+                        "email": "testing@example.com"
+                    }
+                },
+                {
+                    "model": "sentry.useroption",
+                    "pk": 11,
+                    "fields": {
+                        "user": 1,
+                        "key": "foo",
+                        "value": "2"
+                    }
+                },
+                {
+                    "model": "sentry.useroption",
+                    "pk": 21,
+                    "fields": {
+                        "user": 1,
+                        "key": "bar",
+                        "value": "3"
+                    }
+                },
+                {
+                    "model": "sentry.useroption",
+                    "pk": 22,
+                    "fields": {
+                        "user": 1,
+                        "key": "bar",
+                        "value": "4"
+                    }
+                }
+            ]
+        """
+    )
+    out = validate(left, right)
+    findings = out.findings
+
+    assert len(findings) == 0