<?php
namespace WebBot\Core\WebBot;
use WebBot\Core\Bot;
// set unlimited execution time
set_time_limit(0);
// set the default timeout to 1 seconds
Bot::$conf_default_timeout = 1;
// set delay between consecutive fetches (in milli seconds)
Bot::$conf_delay_between_fetches = 1000;
// do not use HTTPS protocol
Bot::$conf_force_https = false;
// don't include document field raw values
Bot::$conf_include_document_field_raw_values = false;
// set the directory for storing information
// $dir = your/custom/path;
if (defined('LIB_PATH')) {
$dir = LIB_PATH . '/WebBot/tmp/';
} else {
$dir = '../tmp/';
}
Bot::$logging = false;
Bot::$conf_store_dir = $dir;
public static function fetchCampaign($url)
{
$data = [];
try {
$bot = new Bot(['cloud' => $url], ['logging' => false]);
$bot->execute();
$documents = $bot->getDocuments();
// because only variables can be passed as reference
$doc = array_shift($documents);
$type = $doc->type;
if (preg_match("/image/i", $type)) {
$data["image"] = $data["url"] = $url;
$data["description"] = $data["title"] = "";
return $data;
}
$data["title"] = $doc->query("/html/head/title")->item(0)->nodeValue;
$data["url"] = $url;
$data["description"] = "";
$data["image"] = "";
$metas = $doc->query("/html/head/meta");
for ($i = 0; $i < $metas->length; $i++) {
$meta = $metas->item($i);
if ($meta->getAttribute('name') == 'description') {
$data["description"] = $meta->getAttribute('content');
}
if ($meta->getAttribute('property') == 'og:image') {
$data["image"] = $meta->getAttribute('content');
}
}
} catch (\Exception $e) {
$data["url"] = $url;
$data["image"] = $data["description"] = $data["title"] = "";
}
return $data;
}
