fixup! fixup! fixup! fixup! fixup! fixup! Check difference of archive…

… and bucket

migrations/02_add_archive_count.py

@@ -81,12 +81,20 @@
         )
 
     in_bucket = {summary.key: summary.size for summary in bucket.objects.all()}
-    print(f"{bucket.name}: {sum(in_bucket.values()) / 2**30:.1f} GB in {len(in_bucket)} items")
+    print(
+        f"{bucket.name}: {sum(in_bucket.values()) / 2**30:.1f} GB in {len(in_bucket)} items"
+    )
     in_archive = {
         get_key_for_url(bucket, row["uri"]): row["avro_archive_id"]
-        for row in connection.execute(sa.select([Archive.c.avro_archive_id, Archive.c.uri])).fetchall()
+        for row in connection.execute(
+            sa.select([Archive.c.avro_archive_id, Archive.c.uri]).where(
+                Archive.c.refcount > 0
+            )
+        ).fetchall()
     }
-    missing_in_bucket = sorted(in_archive[key] for key in set(in_archive.keys()).difference(in_bucket))
+    missing_in_bucket = sorted(
+        in_archive[key] for key in set(in_archive.keys()).difference(in_bucket)
+    )
     missing_in_archive = [key for key in set(in_bucket).difference(in_archive)]
     print(f"{len(missing_in_bucket)} missing in bucket: {missing_in_bucket}")
     print(f"{len(missing_in_archive)} missing in archive {missing_in_archive}")