From f2eb164d40340cf221cb7ad457ab35492da4d308 Mon Sep 17 00:00:00 2001
From: lolcat <will@lolcat.ca>
Date: Sat, 27 Apr 2024 14:25:39 -0400
Subject: [PATCH] qwant gibberish check

---
 scraper/qwant.php | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)

diff --git a/scraper/qwant.php b/scraper/qwant.php
index a8b69fe..7f441e5 100644
--- a/scraper/qwant.php
+++ b/scraper/qwant.php
@@ -453,6 +453,8 @@ class qwant{
 			switch($item["type"]){ // ignores ads
 				
 				case "web":
+					
+					$first_iteration = true;
 					foreach($item["items"] as $result){
 						
 						if(isset($result["thumbnailUrl"])){
@@ -483,6 +485,25 @@ class qwant{
 							}
 						}
 						
+						// detect gibberish results
+						if(
+							$first_iteration &&
+							preg_match(
+								"/^" .
+								preg_quote(
+									$this->trimdots(
+										$result["source"]
+									),
+									"/"
+								) .
+								"/",
+								$result["url"]
+							) !== 1
+						){
+							
+							throw new Exception("Qwant returned gibberish results");
+						}
+						
 						$out["web"][] = [
 							"title" => $this->trimdots($result["title"]),
 							"description" => $this->trimdots($result["desc"]),
@@ -493,6 +514,8 @@ class qwant{
 							"sublink" => $sublinks,
 							"table" => []
 						];
+						
+						$first_iteration = false;
 					}
 					break;