多线程 QQ 号码爬虫

php代码

<?php
/*
Homepage: http://www.php.cn
*/
if(!extension_loaded(&#39;pthreads&#39;)) die (&#39;Please install pthreads&#39;);

include_once(&#39;Snoopy.class.php&#39;);

class CrawlerWorker extends Worker {

	protected  static $dbh;
	public function __construct() {

	}
	public function run(){
	/*
		$dbhost = &#39;db.example.com&#39;;			// 数据库服务器
	    $dbuser = &#39;example.com&#39;;        	// 数据库用户名
        $dbpw = &#39;password&#39;;             	// 数据库密码
		$dbname = &#39;example&#39;;				// 数据库名

		self::$dbh  = new PDO("mysql:host=$dbhost;port=3306;dbname=$dbname", $dbuser, $dbpw, array(
			PDO::MYSQL_ATTR_INIT_COMMAND => &#39;SET NAMES \&#39;UTF8\&#39;&#39;,
			PDO::MYSQL_ATTR_COMPRESS => true,
			PDO::ATTR_PERSISTENT => true
			)
		);
	*/
	}
	protected function getInstance(){
        return self::$dbh;
    }

}

/* the collectable class implements machinery for Pool::collect */
class Crawler extends Stackable {
	public $depth = 3;
	private static $level = 0;
	public function __construct($qq) {
		$this->qq = $qq;
	}
	public function run() {

		try {
			$dbh  = $this->worker->getInstance();
			$this->recursion(array($this->qq));
		}
		catch(PDOException $e) {
			$error = sprintf("%s,%s\n", $mobile, $id );
			file_put_contents("mobile_error.log", $error, FILE_APPEND);
		}
		//printf("runtime: %s, %s\n", date(&#39;Y-m-d H:i:s&#39;), $this->worker->getThreadId());
		//$lst = $this->qzone($this->qq);
		//print_r($lst);
	}
	public function recursion($qqs){
		
		if( self::$level <= $this->depth){
			self::$level++;
		}else if(self::$level > 0){
			self::$level--;
		}
		printf("Level: %s\n", self::$level);
		//sleep(1);
		usleep(mt_rand(10000,1000000));
		if(self::$level >= $this->depth){
			return;
		}
		
		foreach($qqs as $uin) {
			$lst = $this->qzone($uin);
			print_r($lst);
			$this->recursion($lst);
		}
	}

	public function qzone($qq){
		$url = &#39;http://m.qzone.com/mqz_get_visitor?g_tk=1191852101&res_mode=0&res_uin=&#39;.$qq.&#39;&offset=0&count=100&page=1&format=json&t=1401762986882&sid=dODKVcYv6azjN87cxXQ5mao1xgakYjHg18c8aa5e0201%3D%3D&#39;;
		$snoopy = new Snoopy;
		 
		// need an proxy?
		//$snoopy->proxy_host = "my.proxy.host";
		//$snoopy->proxy_port = "8080";
		 
		// set browser and referer:
		$snoopy->agent = "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)";
		$snoopy->referer = "http://m.qzone.com/";
		 
		// set some cookies:
		//$snoopy->cookies["SessionID"] = &#39;238472834723489&#39;;
		//$snoopy->cookies["favoriteColor"] = "blue";
		 
		// set an raw-header:
		$snoopy->rawheaders["Pragma"] = "no-cache";
		 
		// set some internal variables:
		$snoopy->maxredirs = 2;
		$snoopy->offsiteok = false;
		$snoopy->expandlinks = false;
		 
		// set username and password (optional)
		//$snoopy->user = "joe";
		//$snoopy->pass = "bloe";
		 
		// fetch the text of the website www.google.com:
		if($snoopy->fetchtext($url)){ 
			// other methods: fetch, fetchform, fetchlinks, submittext and submitlinks

			// response code:
			//print "response code: ".$snoopy->response_code."<br/>\n";
		 
			// print the headers:
			//print "<b>Headers:</b><br/>";
			//while(list($key,$val) = each($snoopy->headers)){
			//	print $key.": ".$val."<br/>\n";
			//}

			// print the texts of the website:
			//print_r( json_decode($snoopy->results) );
			
			$results = array();
			$tmp = json_decode($snoopy->results);
			
			if($tmp){
				if(property_exists($tmp, &#39;data&#39;)){
					foreach( $tmp->data->list as $lst ){
						$results[] = $lst->uin;
					}
				}
			}
			return ($results);
			
		}
		else {
			print "Snoopy: error while fetching document: ".$snoopy->error."\n";
		}		
	}
}

$pool = new Pool(100, \CrawlerWorker::class, []);

#foreach (range(1000, 100000) as $number) {
#	$pool->submit(new Crawler($number));
#}

$pool->submit(new Crawler(&#39;13721218&#39;));
$pool->submit(new Crawler(&#39;291379&#39;));
//$pool->submit(new Crawler(&#39;xxx&#39;));
//$pool->submit(new Crawler(&#39;xxx&#39;));
//$pool->submit(new Crawler(&#39;xxx&#39;));
// 以此类推
//$pool->submit(new Crawler(&#39;nnn&#39;));

$pool->shutdown();
?>