diff --git a/app/Services/Scrapers/DhenScraper.php b/app/Services/Scrapers/DhenScraper.php index bff30e5..dcf11cf 100644 --- a/app/Services/Scrapers/DhenScraper.php +++ b/app/Services/Scrapers/DhenScraper.php @@ -1,4 +1,5 @@ client = new Client; @@ -23,25 +24,30 @@ class DhenScraper $title = $crawler->filter('h1')->first()->text(); - if($crawler->filter('article .article-visual img')->count() == 1) - { + if ($crawler->filter('article .article-visual img')->count() == 1) { $image = $crawler->filter('article .article-visual img')->first()->attr('src'); } - $crawler->filter('article .article-entry p')->each(function ($node) { - $this->content[] = preg_replace("/[a-zA-Z]/","",$node->text()); - }); + if ($crawler->filter('article .article-entry p')->count() > 0) { + $crawler->filter('article .article-entry p')->each(function ($node) { + $this->content[] = preg_replace("/[a-zA-Z]/", "", $node->text()); + }); + } else { + $crawler->filter('.im_message_body .im_message_text')->each(function ($node) { + $this->content[] = preg_replace("/[a-zA-Z]/", "", $node->text()); + }); + } - - if($crawler->filter(".story-item--meta a")->count() == 1) - { + + + if ($crawler->filter(".story-item--meta a")->count() == 1) { $this->author = $crawler->filter('.story-item--meta a')->first()->text(); } //Remove all the alphabets from string //preg_replace("/[a-zA-Z]/", "",$string); - return [ + return [ 'source' => 'Dhen', 'title' => $title, 'og_title' => $crawler->filter('meta[property*="og:title"]')->first()->attr('content'), @@ -49,7 +55,7 @@ class DhenScraper 'content' => $this->content, 'url' => $url, 'date' => Carbon::parse(str_replace("- ", "", $crawler->filter('.story-item--meta time')->first()->text()))->format("Y-m-d H:i:s"), - 'guid' => str_replace("https://dhen.mv/","",$url), + 'guid' => str_replace("https://dhen.mv/", "", $url), 'author' => $this->author, 'topics' => [ [ @@ -59,4 +65,4 @@ class DhenScraper ] ]; } -} \ No newline at end of file +}