From f5b3480f442f6538057f20cbc853a7009bc55bd7 Mon Sep 17 00:00:00 2001
From: abram axel booth <boothaa@gmail.com>
Date: Fri, 31 Jul 2026 17:22:09 -0400
Subject: [PATCH] wip: fix ENG-11883

---
 .../reporters/osfstorage_file_count.py        | 54 +++++++++++++++----
 1 file changed, 43 insertions(+), 11 deletions(-)

diff --git a/osf/metrics/reporters/osfstorage_file_count.py b/osf/metrics/reporters/osfstorage_file_count.py
index 5db9ad1cff6..20b598b45fd 100644
--- a/osf/metrics/reporters/osfstorage_file_count.py
+++ b/osf/metrics/reporters/osfstorage_file_count.py
@@ -2,7 +2,11 @@
 from django.db.models import Q
 import logging
 
-from osf.models import AbstractNode, Preprint
+from osf.models import (
+    AbstractNode,
+    Preprint,
+    SpamStatus,
+)
 from osf.metrics.daily_reports import DailyOsfstorageFileCountReport
 from osf.metrics.utils import cycle_coverage_date
 from ._base import DailyReporter
@@ -13,32 +17,60 @@
 
 class OsfstorageFileCountReporter(DailyReporter):
 
-    def report(self, date):
+    def report(self, report_date):
         from addons.osfstorage.models import OsfStorageFile
 
-        file_qs = OsfStorageFile.objects
+        file_qs = (
+            OsfStorageFile.objects
+            .filter(created__date__lte=report_date)
+            .exclude(deleted__date__lte=report_date)
+        )
         abstract_node_content_type = ContentType.objects.get_for_model(AbstractNode)
         preprint_content_type = ContentType.objects.get_for_model(Preprint)
+        node_ids = (
+            AbstractNode.objects
+            .exclude(spam_status=SpamStatus.SPAM)
+            .values('id')
+        )
+        preprint_ids = (
+            Preprint.objects
+            .exclude(spam_status=SpamStatus.SPAM)
+            .values('id')
+        )
+
+        total_query = Q(
+            target_object_id__in=node_ids,
+            target_content_type=abstract_node_content_type,
+        ) | Q(
+            target_object_id__in=preprint_ids,
+            target_content_type=preprint_content_type,
+        )
 
         public_query = Q(
-            target_object_id__in=AbstractNode.objects.filter(is_public=True).values('id'),
-            target_content_type__in=[abstract_node_content_type, preprint_content_type],
+            target_object_id__in=node_ids.filter(is_public=True),
+            target_content_type=abstract_node_content_type,
+        ) | Q(
+            target_object_id__in=preprint_ids.filter(is_public=True),
+            target_content_type=preprint_content_type,
         )
 
         private_query = Q(
-            target_object_id__in=AbstractNode.objects.filter(is_public=False).values('id'),
-            target_content_type__in=[abstract_node_content_type, preprint_content_type],
+            target_object_id__in=node_ids.filter(is_public=False),
+            target_content_type=abstract_node_content_type,
+        ) | Q(
+            target_object_id__in=preprint_ids.filter(is_public=False),
+            target_content_type=preprint_content_type,
         )
 
-        daily_query = Q(created__date=date)
+        daily_query = Q(created__date=report_date)
 
         yield DailyOsfstorageFileCountReport(
-            cycle_coverage=cycle_coverage_date(date),
+            cycle_coverage=cycle_coverage_date(report_date),
             files=dict(
-                total=file_qs.count(),
+                total=file_qs.filter(total_query).count(),
                 public=file_qs.filter(public_query).count(),
                 private=file_qs.filter(private_query).count(),
-                total_daily=file_qs.filter(daily_query).count(),
+                total_daily=file_qs.filter(total_query & daily_query).count(),
                 public_daily=file_qs.filter(public_query & daily_query).count(),
                 private_daily=file_qs.filter(private_query & daily_query).count(),
             ),