qwant gibberish check

This commit is contained in:
lolcat 2024-04-27 14:25:39 -04:00
parent 81dc93802c
commit f2eb164d40
1 changed files with 23 additions and 0 deletions

View File

@ -453,6 +453,8 @@ class qwant{
switch($item["type"]){ // ignores ads switch($item["type"]){ // ignores ads
case "web": case "web":
$first_iteration = true;
foreach($item["items"] as $result){ foreach($item["items"] as $result){
if(isset($result["thumbnailUrl"])){ 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"][] = [ $out["web"][] = [
"title" => $this->trimdots($result["title"]), "title" => $this->trimdots($result["title"]),
"description" => $this->trimdots($result["desc"]), "description" => $this->trimdots($result["desc"]),
@ -493,6 +514,8 @@ class qwant{
"sublink" => $sublinks, "sublink" => $sublinks,
"table" => [] "table" => []
]; ];
$first_iteration = false;
} }
break; break;