lolcat
/
4get
6
13
Fork 4
Code Issues 13 Pull Requests Packages Projects Releases Wiki Activity

more error handling for yep

This commit is contained in:
lolcat 2024-05-23 09:05:04 -04:00
parent 92d0102738
commit 2007907972
1 changed files with 51 additions and 44 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

95
scraper/yep.php
View File

@ -350,21 +350,7 @@ class yep{
throw new Exception("Failed to fetch JSON"); throw new Exception("Failed to fetch JSON");
} }
// detect cloudflare page $this->detect_cf($json);
$this->fuckhtml->load($json);
if(
count(
$this->fuckhtml
->getElementsByClassName(
"cf-wrapper",
"div"
)
) !== 0
){
throw new Exception("Blocked by Cloudflare");
}
$json = json_decode($json, true); $json = json_decode($json, true);
//$json = json_decode(file_get_contents("scraper/yep.json"), true); //$json = json_decode(file_get_contents("scraper/yep.json"), true);
@ -539,26 +525,27 @@ class yep{
try{ try{
$json = $json =
json_decode( $this->get(
$this->get( $this->backend->get_ip(), // no nextpage!
$this->backend->get_ip(), // no nextpage! "https://api.yep.com/fs/2/search",
"https://api.yep.com/fs/2/search", [
[ "client" => "web",
"client" => "web", "gl" => $country == "all" ? $country : strtoupper($country),
"gl" => $country == "all" ? $country : strtoupper($country), "no_correct" => "false",
"no_correct" => "false", "q" => $search,
"q" => $search, "safeSearch" => $nsfw,
"safeSearch" => $nsfw, "type" => "images"
"type" => "images" ]
]
),
true
); );
}catch(Exception $error){ }catch(Exception $error){
throw new Exception("Failed to fetch JSON"); throw new Exception("Failed to fetch JSON");
} }
$this->detect_cf($json);
$json = json_decode($json, true);
if($json === null){ if($json === null){
throw new Exception("Failed to decode JSON"); throw new Exception("Failed to decode JSON");
@ -636,27 +623,27 @@ class yep{
// https://api.yep.com/fs/2/search?client=web&gl=CA&no_correct=false&q=undefined+variable+javascript&safeSearch=off&type=web // https://api.yep.com/fs/2/search?client=web&gl=CA&no_correct=false&q=undefined+variable+javascript&safeSearch=off&type=web
$json = $json =
json_decode( $this->get(
$this->get( $this->backend->get_ip(),
$this->backend->get_ip(), "https://api.yep.com/fs/2/search",
"https://api.yep.com/fs/2/search", [
[ "client" => "web",
"client" => "web", "gl" => $country == "all" ? $country : strtoupper($country),
"gl" => $country == "all" ? $country : strtoupper($country), "limit" => "99999",
"limit" => "99999", "no_correct" => "false",
"no_correct" => "false", "q" => $search,
"q" => $search, "safeSearch" => $nsfw,
"safeSearch" => $nsfw, "type" => "news"
"type" => "news" ]
]
),
true
); );
}catch(Exception $error){ }catch(Exception $error){
throw new Exception("Failed to fetch JSON"); throw new Exception("Failed to fetch JSON");
} }
$this->detect_cf($json);
$json = json_decode($json, true);
//$json = json_decode(file_get_contents("scraper/yep.json"), true); //$json = json_decode(file_get_contents("scraper/yep.json"), true);
if($json === null){ if($json === null){
@ -698,6 +685,26 @@ class yep{
} }
private function detect_cf($payload){
// detect cloudflare page
$this->fuckhtml->load($payload);
if(
count(
$this->fuckhtml
->getElementsByClassName(
"cf-wrapper",
"div"
)
) !== 0
){
throw new Exception("Blocked by Cloudflare");
}
}
private function titledots($title){ private function titledots($title){
$substr = substr($title, -4); $substr = substr($title, -4);