php采集百度下面的相关搜索关键词

 1 <?php
 2 
 3 $html = getArray(‘狂怒‘);
 4 $res = getHtmlByContent($html);
 5 print_r($res);
 6 
 7 
 8 
 9 /**
10  * getHtmlByContent 
11  * 
12  * @param mixed $html 
13  * @access public
14  * @return void
15  */
16 function getHtmlByContent($html){
17     preg_match_all(‘/<table cellpadding="0">.*<\/table>/‘,$html,$res);
18 //  print_r($res[0][0]);
19     preg_match_all(‘/<th>.*?<\/th>/‘,$res[0][0],$result);
20     $result = $result[0];
21     if(!empty($result) && is_array($result)){
22         foreach($result as $k=>$v){
23             $result[$k] = strip_tags($v);
24         }   
25     }   
26     //print_r($result);
27     return $result;
28 }
29 
30 
31 /**
32  * getArray 
33  * 
34  * @param mixed $kw 
35  * @access public
36  * @return void
37  */
38 function getArray($kw){
39     $url="http://www.baidu.com/s?wd=".$kw;
40     $curl=curl_init();
41     curl_setopt($curl,CURLOPT_URL,$url);
42     curl_setopt($curl,CURLOPT_RETURNTRANSFER,1);
43     //curl_setopt($curl, CURLOPT_POSTFIELDS, $curlPost);
44     $data = curl_exec($curl);
45     $data = preg_replace("/[\r\n\t]+/","",$data);
46     return $data;
47 
48 }