noury/mirror-4get
1
0
Fork 0
mirror of https://git.lolcat.ca/lolcat/4get.git synced 2024-11-14 03:49:48 +01:00
Code Issues Projects Releases Packages Wiki Activity

fixed brave news

Browse source
This commit is contained in:
lolcat 2023-11-09 08:50:53 -05:00
parent 165d80f80b
commit 9fd993b47b
1 changed files with 38 additions and 115 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

153
scraper/brave.php
View file

@ -1139,131 +1139,54 @@ class brave{
$proxy
);
$news =
$this->fuckhtml
->getElementsByClassName(
"snippet inline gap-standard",
"div"
);
preg_match(
'/const data ?= ?(\[{.*}]);/',
$html,
$json
);
foreach($news as $article){
if(!isset($json[1])){
$data = [
"title" => null,
"author" => null,
"description" => null,
"date" => null,
"thumb" =>
[
"url" => null,
"ratio" => null
],
"url" => null
];
throw new Exception("Failed to grep javascript object");
}
$json = $this->fuckhtml->parseJsObject($json[1], true);
if($json === null){
$this->fuckhtml->load($article);
$elems =
$this->fuckhtml
->getElementsByTagName("*");
// get title
$data["title"] =
$this->fuckhtml
->getTextContent(
$this->fuckhtml
->getElementsByClassName(
"snippet-title",
$elems
)
[0]
["innerHTML"]
);
// get description
$data["description"] =
$this->titledots(
$this->fuckhtml
->getTextContent(
$this->fuckhtml
->getElementsByClassName(
"snippet-description",
$elems
)
[0]
["innerHTML"]
)
);
// get date
$date =
explode(
"•",
$this->fuckhtml
->getTextContent(
$this->fuckhtml
->getElementsByClassName(
"snippet-url",
$elems
)[0]
)
);
throw new Exception("Failed to parse javascript object");
}
foreach(
$json[1]["data"]["body"]["response"]["news"]["results"]
as $news
){
if(
count($date) !== 1 &&
trim($date[1]) != ""
!isset($news["thumbnail"]["src"]) ||
$news["thumbnail"]["src"] == "void 0"
){
$data["date"] =
strtotime(
$date[1]
);
}
// get URL
$data["url"] =
$this->fuckhtml->getTextContent(
$this->unshiturl(
$this->fuckhtml
->getElementsByClassName(
"result-header",
$elems
)
[0]
["attributes"]
["href"]
)
);
// get thumbnail
$thumb =
$this->fuckhtml
->getElementsByTagName(
"img"
);
if(
count($thumb) === 2 &&
trim(
$thumb[1]
["attributes"]
["src"]
) != ""
){
$data["thumb"] = [
"url" =>
$this->fuckhtml->getTextContent(
$this->unshiturl(
$thumb[1]
["attributes"]
["src"]
)
),
$thumb = [
"url" => null,
"ratio" => null
];
}else{
$thumb = [
"url" => $this->unshiturl($news["thumbnail"]["src"]),
"ratio" => "16:9"
];
}
$out["news"][] = $data;
$out["news"][] = [
"title" => $news["title"],
"author" => null,
"description" => $news["description"],
"date" => !isset($news["age"]) || $news["age"] == "void 0" ? null : strtotime($news["age"]),
"thumb" => $thumb,
"url" => $news["url"]
];
}
return $out;