php curl通过连接获取标题
- WBOYoriginal
- 2016-06-07 11:34:261319parcourir
php curl通过连接获取标题/****/<br>
//Gary xu<br>
//1122557724@qq.com<br>
/****/<br>
namespace Xuyaoxiang;<br>
<br>
class Snoopy {<br>
<br>
public $pattern_array=array(<br>
'title'=>'/<title>(\s*.*)/i',<br>
'description'=>'/<meta>/',<br>
'charset'=>'/charset=\"?([\w-]+)\"?/i',<br>
);<br>
<br>
public $user_agent='Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/49.0.2623.110 Safari/537.36'; //模拟浏览器头部数据<br>
<br>
public $target_code="utf-8"; //目标编码<br>
<br>
public $url;<br>
<br>
public $data;<br>
<br>
public $pattern_key;<br>
<br>
function __construct($url)<br>
{<br>
$this->url=$url;<br>
}<br>
<br>
public function set_pattern($key,$val)<br>
{<br>
$this->pattern_array[$key]=$val;<br>
}<br>
<br>
<br>
<br>
function get_content($pattern_key)<br>
{<br>
$this->pattern_key=$pattern_key;<br>
<br>
if($this->pattern_key==''){return false;}<br>
<br>
$this->curl_get_data();<br>
<br>
if($this->data==false){return false;}<br>
<br>
$charset=$this->get_charset();<br>
<br>
$this->check_charset($charset);<br>
<br>
$content=$this->get_key_content();<br>
<br>
return trim($content[1]);<br>
}<br>
<br>
<br>
function curl_get_data()<br>
{<br>
$curl=curl_init();<br>
// 设置你需要抓取的URL<br>
<br>
curl_setopt($curl, CURLOPT_URL, $this->url);<br>
<br>
// 设置header<br>
curl_setopt($curl, CURLOPT_HEADER, 0);<br>
<br>
// 设置cURL 参数，要求结果保存到字符串中还是输出到屏幕上。<br>
curl_setopt($curl, CURLOPT_RETURNTRANSFER, 1);<br>
<br>
curl_setopt($curl, CURLOPT_USERAGENT, $this->user_agent);<br>
<br>
// 运行cURL，请求网页<br>
<br>
$this->data = curl_exec($curl); <br>
<br>
curl_close($curl);<br>
}<br>
<br>
function check_charset($page_charset)<br>
{<br>
if($page_charset!=$this->target_code)<br>
{<br>
$this->data=mb_convert_encoding($this->data,$this->target_code,$page_charset);<br>
}<br>
}<br>
<br>
function get_key_content()<br>
{<br>
preg_match($this->pattern_array[$this->pattern_key],$this->data,$content);<br>
return $content; <br>
}<br>
<br>
function get_charset()<br>
{<br>
preg_match($this->pattern_array['charset'],$this->data,$reg_charset);<br>
return $page_charset=strtolower($reg_charset[1]); <br>
}<br>
}<br>
<br>
<br>
header("Content-type:text/html;charset=utf-8");<br>
$snoopy=new snoopy("http://www.qq.com");<br>
<br>
$title=$snoopy->get_content('title');<br>
<br>
print_r($title);</title>
AD：真正免费，域名+虚机+企业邮箱=0元