From 3bd714ea9d11baa4fa847ba9b75018dce7d65a58 Mon Sep 17 00:00:00 2001
From: Ilya Kreymer <ikreymer@users.noreply.github.com>
Date: Tue, 25 Jun 2024 08:54:42 -0700
Subject: [PATCH] QA stats aggregation: exclude isFile / isError pages from
 stats (#1879)

Follow-up to: #1868, exclude pages that have isFile or isError set to
true from the stats aggregation.
---
 backend/btrixcloud/pages.py | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/backend/btrixcloud/pages.py b/backend/btrixcloud/pages.py
index 414a898a..dadb0e22 100644
--- a/backend/btrixcloud/pages.py
+++ b/backend/btrixcloud/pages.py
@@ -577,7 +577,13 @@ class PageOps:
             boundaries.append(1.1)
 
         aggregate = [
-            {"$match": {"crawl_id": crawl_id}},
+            {
+                "$match": {
+                    "crawl_id": crawl_id,
+                    "isFile": {"$ne": True},
+                    "isError": {"$ne": True},
+                }
+            },
             {
                 "$bucket": {
                     "groupBy": f"$qa.{qa_run_id}.{key}",