FbIN/fossq
FbIN/fossq
Watch
1
0
Fork
You've already forked fossq
 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Synced

Browse source 
Baidu fix (46411a6c83)
DDG fix (8328d93b17)
This commit is contained in:
FbIN Support 2026-07-05 16:22:57 +05:30
commit d6159f66fb
2 changed files with 31 additions and 123 deletions
Download patch file Download diff file Expand all files Collapse all files

44
scraper/baidu.php
View file

@ -666,14 +666,24 @@ class baidu{
//
// Don't parse as a search result if it's a card
//
$card =
$tpl_probe =
$this->fuckhtml
->getElementsByClassName(
"cosc-card",
$div
->getElementsByAttributeValue(
"tpl",
"www_index",
[$datafield]
);
if(count($card) !== 0){
if(
count($tpl_probe) === 0 &&
count(
$this->fuckhtml
->getElementsByClassName(
"cosc-card",
$div
)
) !== 0
){
//
// Parse chinese youtube shorts
@ -870,10 +880,13 @@ class baidu{
// class:FYB_RD -> News garbage, IGNORE
$result =
$this->fuckhtml
->getElementsByClassName(
"result",
[$datafield]
array_merge(
$tpl_probe,
$this->fuckhtml
->getElementsByClassName(
"result",
[$datafield]
)
);
if(count($result) !== 0){
@ -885,7 +898,7 @@ class baidu{
$title =
$this->fuckhtml
->getElementsByClassName(
"sc-link",
"cosc-title-a",
"a"
);
@ -905,8 +918,9 @@ class baidu{
$description =
$this->fuckhtml
->getElementsByClassName(
"c-color",
->getElementsByAttributeValue(
"data-sanssr-cmpt",
"card/www-summary-1",
$div
);
@ -2051,7 +2065,7 @@ class baidu{
$parts = explode("src=", $url);
if(count($parts) !== 2){
// stuff
// stuff is broken
return $url;
}
@ -2062,11 +2076,11 @@ class baidu{
if(count($q) !== 2){
// stuff
// stuff is broken, again
return $url;
}
// baidu devs are fucking retarded and dont follow spec:
// baidu devs are not following specs:
// &fmt=auto?s=BB32F3A050471AEC72886934030090C4&sec=1753203600&t=0fb2194775d3bd3d1bb114b818479e0a
parse_str(str_replace("?", "&", $q[1]), $query);

110
scraper/ddg.php
View file

@ -54,8 +54,8 @@ class ddg{
$headers =
["User-Agent: " . config::USER_AGENT,
"Accept: */*",
"Accept-Language: en-US,en;q=0.5",
"Accept-Encoding: gzip",
"Accept-Language: en-US,en;q=0.9",
"Accept-Encoding: gzip, deflate, br, zstd",
"Referer: https://duckduckgo.com/",
"DNT: 1",
"Sec-GPC: 1",
@ -920,112 +920,6 @@ class ddg{
){
throw new Exception("DuckDuckGo returned a JSA challenge");
// get JSA initial token
if(
!preg_match(
'/let jsa *= *([0-9]+)/',
$js,
$jsa
)
){
$jsa = 0;
}else{
$jsa = (int)$jsa[1];
}
// get function bodies
preg_match_all(
'/let *([A-Za-z0-9]+) *= *function\(.*\) *{(.*)};/sU',
$js,
$functions
);
$parsed_functions = [];
for($i=0; $i<count($functions[0]); $i++){
$functions[2][$i] = trim($functions[2][$i]);
if(
preg_match(
'/return num *\* *([0-9]+)/i',
$functions[2][$i],
$num
)
){
$parsed_functions[$functions[1][$i]] = [
"type" => "multiplication",
"num" => (int)$num[1]
];
continue;
}
if(
preg_match(
'/innerHTML *= *`([^`]+)`/i',
$functions[2][$i],
$challenge
)
){
$challenge[1] =
preg_replace(
'/<\/(br)>/',
'<$1>',
$challenge[1]
);
$parsed_functions[$functions[1][$i]] = [
"type" => "challenge",
"text" => $challenge[1]
];
}
}
// get function call order
preg_match_all(
'/jsa *= *([A-Za-z0-9]+)\(jsa\)/i',
$js,
$call_order
);
foreach($call_order[1] as $order){
if(!isset($parsed_functions[$order])){
throw new Exception("JS challenge solve failure: DuckDuckGo called an unknown function");
}
if($parsed_functions[$order]["type"] == "multiplication"){
$jsa = $jsa * $parsed_functions[$order]["num"];
continue;
}
if($parsed_functions[$order]["type"] == "challenge"){
// @TODO get parsed length
//$parsed_functions[$order]["text"]
$jsa = $jsa + strlen($parsed_functions[$order]["text"]);
}
}
try{
$js = $this->get(
$proxy,
"https://links.duckduckgo.com" . $challenge_url[1] . $jsa,
[],
ddg::req_xhr
);
}catch(Exception $error){
throw new Exception("Failed to get challenged d.js");
}
}
//