Bug Fixes and thiladhun scraper intergration

This commit is contained in:
2020-08-11 03:39:08 +05:00
parent 0cede5b708
commit 5df4011f13
14 changed files with 4478 additions and 137 deletions

View File

@@ -11,7 +11,7 @@ class MihaaruScraper
protected $title;
protected $content;
protected $image;
protected $tags = [];
protected $topics = [];
protected $author;
public function __construct()
@@ -25,13 +25,11 @@ class MihaaruScraper
$crawler = $this->client->request('GET', $url);
$crawler->filter('h1')->each(function ($node) {
$title = $node->text();
$this->title = $title;
$this->title = $node->text();
});
$crawler->filter('.container img')->eq(3)->each(function ($node) {
$image = $node->attr('src');
$this->image = $image;
$this->image = $node->attr('src');
});
$crawler->filter('.by-line address')->each(function ($node) {
@@ -49,7 +47,7 @@ class MihaaruScraper
$crawler->filter('.article-tags')->each(function ($node) {
$this->tags[] = [
$this->topics[] = [
"name" => $node->text(),
"slug" => str_replace("https://mihaaru.com/", "", $node->attr('href'))
];
@@ -57,7 +55,7 @@ class MihaaruScraper
//Remove all the alphabets from string
//preg_replace("/[a-zA-Z]/", "",$string);
$data = [
return [
'source' => 'Mihaaru',
'title' => $this->title,
'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'),
@@ -67,9 +65,7 @@ class MihaaruScraper
'date' => $date,
'guid' => $guid,
'author' => $this->author,
'topics' => $this->tags,
'topics' => $this->topics
];
return $data;
}
}