39 lines
1.2 KiB
PHP
39 lines
1.2 KiB
PHP
<?php
|
|
|
|
namespace Scraping;
|
|
|
|
use \Goutte\Client;
|
|
|
|
class Scraper
|
|
{
|
|
public $links = [];
|
|
|
|
public static function scrape($method, $url, $filters, $level, $download_dir){
|
|
$client = new Client();
|
|
|
|
$crawler = $client->request($method, $url);
|
|
|
|
if (strstr($client->getResponse()->getHeaders()['content-type'][0], 'text/html')) {
|
|
print str_pad('['.$level.']', $level*2, ' ', STR_PAD_LEFT).' '.$url."\n";
|
|
|
|
$crawler->filterXPath($filters[$level])->each(function ($node) use ($method, $filters, $level) {
|
|
$new_url = $node->link()->getUri();
|
|
$new_level = $level+1;
|
|
|
|
if ($new_level < count($filters)) {
|
|
self::scrape($method, $new_url, $filters, $new_level, $download_dir);
|
|
}
|
|
});
|
|
} else {
|
|
$content_disposition = $client->getResponse()->getHeaders()['content-disposition'][0];
|
|
|
|
$filename = time();
|
|
|
|
if (preg_match('/filename="([^"]+)"/', $content_disposition, $m)) {
|
|
$filename = $m[1];
|
|
}
|
|
|
|
file_put_contents( $download_dir.'/'.$filename, $client->getResponse()->getContent() );
|
|
}
|
|
}
|
|
} |