37 lines
924 B
PHP
37 lines
924 B
PHP
|
<?php
|
||
|
|
|
||
|
|
namespace App\Services;
|
||
|
|
|
||
|
|
class PageScrapper
|
||
|
|
{
|
||
|
|
private string $url;
|
||
|
|
private string $contentMarker;
|
||
|
|
|
||
|
|
public function __construct($url, $contentMarker = '<div class=["\']content_info["\']>')
|
||
|
|
{
|
||
|
|
$this->url = $url;
|
||
|
|
$this->contentMarker = $contentMarker;
|
||
|
|
}
|
||
|
|
|
||
|
|
public function getHTML()
|
||
|
|
{
|
||
|
|
$page = file_get_contents($this->url);
|
||
|
|
$strForPregMatch = "/" . "{$this->contentMarker}" . "(.*)<\/div>/isU";
|
||
|
|
$arr = [];
|
||
|
|
$rez = preg_match_all($strForPregMatch, $page, $arr);
|
||
|
|
return $content = $arr[1][0].'</div>';
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
public function normalizeURLFile($content)
|
||
|
|
{
|
||
|
|
$rez = preg_match_all('/href="(.*)"/isU',$content,$arr);
|
||
|
|
foreach ($arr[1] as $el) {
|
||
|
|
if (strpos($el, 'http') !== 0) {
|
||
|
|
$content = str_replace($el,'https://mkgtu.ru'.$el,$content);
|
||
|
|
}
|
||
|
|
}
|
||
|
|
return $content;
|
||
|
|
}
|
||
|
|
}
|