if ($newxml) {
$search_results = $newxml->xpath($scrubber_xml->xpath->domain);
if (!empty($search_results)) {
foreach ($search_results as $search_result) {
$domain = $search_result->div[0]->span->__toString();
$rank = explode(" ", $search_result->div[1]->span->attributes()->class->__toString())[1];
if ($rank == "r1" || $rank == "r2") {
file_put_contents("domains.txt", $domain . PHP_EOL, FILE_APPEND);
}
}
}
}
AngryCurl::add_debug_msg("->\t" . $request->options[CURLOPT_PROXY] . "\tOKAY\t" . $info['http_code'] . "\t" . $info['total_time'] . "\t" . $info['url']);
return;
}
echo "nothing happens!\n";
}
$AC = new AngryCurl('nothing');
$AC->__set('window_size', 20);
$AC->load_useragent_list('./lib/useragent_list.txt');
$AC->__set('use_proxy_list', false);
$AC->__set('use_useragent_list', true);
$doc = new DOMDocument();
$doc->strictErrorChecking = FALSE;
$doc->load("config.xml");
$scrubber_xml = simplexml_import_dom($doc);
$xml = new SimpleXMLElement('<xml/>');
for ($i = 199; $i < 999; $i++) {
$AC->get("https://www.mywot.com/en/scorecard?page=" . $i);
}
$AC->execute();
# Setting time and memory limits
ini_set('max_execution_time', 0);
ini_set('memory_limit', '128M');
define('AC_DIR', dirname(__FILE__) . DIRECTORY_SEPARATOR . '..');
# Including classes
require_once AC_DIR . DIRECTORY_SEPARATOR . 'classes' . DIRECTORY_SEPARATOR . 'RollingCurl.class.php';
require_once AC_DIR . DIRECTORY_SEPARATOR . 'classes' . DIRECTORY_SEPARATOR . 'AngryCurl.class.php';
# Initializing AngryCurl instance with callback function named 'callback_function'
$AC = new AngryCurl('callback_function');
# Initializing so called 'web-console mode' with direct cosnole-like output
$AC->init_console();
# Importing proxy and useragent lists, setting regexp, proxy type and target url for proxy check
# You may import proxy from an array as simple as $AC->load_proxy_list($proxy array);
$AC->load_proxy_list(AC_DIR . DIRECTORY_SEPARATOR . 'import' . DIRECTORY_SEPARATOR . 'proxy_list.txt', 200, 'http', 'http://google.com', 'title>G[o]{2}gle');
$AC->load_useragent_list(AC_DIR . DIRECTORY_SEPARATOR . 'import' . DIRECTORY_SEPARATOR . 'useragent_list.txt');
/* NOTE: IF USING request() - "on the fly" proxy server changing WILL apply
*
* You may use request(URL, METHOD, POST_DATA, HEADERS, CURL OPTIONS) to create add new requests
* METHOD may be GET or POST
* POST_DATA may be an array of POST-params
* HEADERS may be any HTTP headers
* CURL OPTIONS may be any of supported by CURL
*/
$AC->request('http://ya.ru');
/* NOTE: IF USING get() - "on the fly" proxy server changing WILL apply
*
* You may use shorcut get(URL, HEADERS, CURL OPTIONS) to create add new GET requests
* HEADERS may be any HTTP headers
* CURL OPTIONS may be any of supported by CURL
*/
