60 lines
1.6 KiB
PHP
60 lines
1.6 KiB
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 = $contentMarker;
|
||
|
|
}
|
||
|
|
|
||
|
|
public function getHTML()
|
||
|
|
{
|
||
|
|
// echo($this->url);
|
||
|
|
// echo($this->contentMarker);
|
||
|
|
$page = file_get_contents($this->url);
|
||
|
|
$strForPregMatch = "/" . "{$this->contentMarker}" . "(.*)<\/div>/isU";
|
||
|
|
$arr = [];
|
||
|
|
$rez = preg_match_all($strForPregMatch, $page, $arr);
|
||
|
|
//echo '<prE>'.print_r($arr[1]).'</pre>';
|
||
|
|
|
||
|
|
|
||
|
|
//$tmp = explode("<br>" , $page);
|
||
|
|
//$content = $tmp[4].$tmp[5].$tmp[6];
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
return $content = $arr[1][0].'</div>';
|
||
|
|
|
||
|
|
}
|
||
|
|
|
||
|
|
public function normalizeURLFile($content)
|
||
|
|
{
|
||
|
|
|
||
|
|
|
||
|
|
$rez = preg_match_all('/href="(.*)"/isU',$content,$arr);
|
||
|
|
//echo '<prE>'.print_r($arr,1).'</pre>';
|
||
|
|
foreach ($arr[1] as $el) {
|
||
|
|
//if (strpos($el, '/abitur') !== false) {
|
||
|
|
if (strpos($el, 'http') !== 0) {
|
||
|
|
//str_replace('/abitur', 'https://mkgtu.ru/abitur',$el);
|
||
|
|
$content = str_replace($el,'https://mkgtu.ru'.$el,$content);
|
||
|
|
}
|
||
|
|
}
|
||
|
|
//$content = $arr[1][0].'</div>';
|
||
|
|
//$content = preg_replace('/href="(?!http|#)/is', 'https:\/\/mkgtu.ru', $content, -1, $count);
|
||
|
|
|
||
|
|
//$content = str_replace('href="','href="https://mkgtu.ru',$content);
|
||
|
|
//$pos = strripos($content,'/abitur');
|
||
|
|
//echo substr_replace($content, 'https://mkgtu.ru',$pos,0)
|
||
|
|
|
||
|
|
return $content;
|
||
|
|
}
|
||
|
|
}
|