From 93baed4b9048cb4f686f356381725382db56003b Mon Sep 17 00:00:00 2001
From: ghost <localhost>
Date: Wed, 20 Dec 2023 08:44:35 +0200
Subject: [PATCH] delete deprecated documents with HTTP code not 200 on second
 scan

---
 src/cli/document/crawl.php | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/src/cli/document/crawl.php b/src/cli/document/crawl.php
index 7f3cd6d..7a49873 100644
--- a/src/cli/document/crawl.php
+++ b/src/cli/document/crawl.php
@@ -178,6 +178,16 @@ foreach($index->search('')
         // Update HTTP code or skip on empty
         if ($code = curl_getinfo($request, CURLINFO_HTTP_CODE))
         {
+            // Delete deprecated document from index as HTTP code still not 200
+            if ($code != 200 && !empty($data['code']) && $data['code'] != 200)
+            {
+                $index->deleteDocument(
+                    $document->getId()
+                );
+
+                continue;
+            }
+
             $data['code'] = $code;
 
         } else continue;