Fix: tag/custom-field document_count scales badly with tag count (#13203)

* Fix (beta): tag/custom-field document_count scales badly with tag count

TagViewSet and CustomFieldViewSet's document_count annotation used a
per-row correlated subquery (annotate_document_count_for_related_queryset),
executed once per tag/custom-field row. Fine at a few dozen rows, but at
~1,000 tags it degrades to a full per-tag GroupAggregate over the tags M2M
table -- 6s+ in production reports, confirmed via EXPLAIN (loops=1000).

Replaced with a single, independent GROUP BY over the through table
(permission filter expressed as a plain WHERE, not an aggregate FILTER),
then injected via Case/When. Also tried a more "obvious" fix -- a plain
Count(filter=Q(id__in=permitted_ids), distinct=True) directly on the M2M
relation -- but that's worse: Postgres fails to plan the id__in check as a
semi-join once a large M2M bridge table is involved, and instead re-checks
subquery membership once per joined row.

Benchmarked against a synthetic corpus (400k documents, 1,000 tags @ ~5/doc,
matching real-world reports in discussion #13161): tag list wall-clock drops
from ~180s to ~8s for a permission-restricted user, ~21s to ~8s for a
superuser. No measurable change at typical home-instance scale (tens of
tags).

Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>

* Address review feedback: restore validation, scope aggregation to queryset

- Restore the document_count_source_field validation helper dropped during
  the refactor -- misconfiguring a viewset (document_count_through set
  without document_count_source_field) now fails fast with a clear error
  again instead of an obscure runtime failure.
- Restrict annotate_document_count_for_related_queryset()'s through-table
  aggregation to rows whose related_object_field is one of the annotated
  queryset's pks. No-op for the main tag-list call site (queryset is all
  tags), but avoids unnecessary work for narrower callers like the tag
  descendants branch in TagViewSet.list().

Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>

src/documents/views.py

@@ -482,7 +482,15 @@ class BulkPermissionMixin:
 class PermissionsAwareDocumentCountMixin(BulkPermissionMixin, PassUserMixin):
     """Mixin to add document count to queryset, permissions-aware if needed"""
 
-    # Default is simple relation path, override for through-table/count specialization.
+    # Direct FK/M2M relation name from this model to Document, used for the
+    # cheap Count(filter=...) path (Correspondent, DocumentType, StoragePath).
+    document_count_related_name: str = "documents"
+
+    # Set both of these instead, for models that only reach Document through
+    # an M2M/through-model table (Tag, CustomField). A plain Count(filter=...)
+    # over such a relation is fine for a direct FK, but forces a much more
+    # expensive plan once an M2M bridge table is involved -- see
+    # annotate_document_count_for_related_queryset() for why.
     document_count_through: type[Model] | None = None
     document_count_source_field: str | None = None
 
@@ -498,12 +506,14 @@ class PermissionsAwareDocumentCountMixin(BulkPermissionMixin, PassUserMixin):
     def get_document_count_filter(self):
         request = getattr(self, "request", None)
         user = getattr(request, "user", None) if request else None
-        return get_document_count_filter_for_user(user)
+        return get_document_count_filter_for_user(
+            user,
+            related_name=self.document_count_related_name,
+        )
 
     def get_queryset(self):
         base_qs = super().get_queryset()
 
-        # Use optimized through-table counting when configured.
         if self.document_count_through:
             user = getattr(getattr(self, "request", None), "user", None)
             return annotate_document_count_for_related_queryset(
@@ -513,10 +523,13 @@ class PermissionsAwareDocumentCountMixin(BulkPermissionMixin, PassUserMixin):
                 user=user,
             )
 
-        # Fallback: simple Count on relation with permission filter.
         filter = self.get_document_count_filter()
         return base_qs.annotate(
-            document_count=Count("documents", filter=filter),
+            document_count=Count(
+                self.document_count_related_name,
+                filter=filter,
+                distinct=True,
+            ),
         )