自己寫的一個php是基於phpQuery的通用採集類

複製程式碼

還是小菜，第一次分享程式碼哈，這是自己以前寫的一個php的採集類，自己一直在用，自我感覺很簡單很強大，只要懂一點點選擇器的知識就可以採集任何頁面了，也支持https頁面，做簡單的採集就夠用了。

 /**
 *通用列表採集類別
 *版本V1.3
 *作者:JAE
 *部落格:http://blog.jaekj.com
*/
 require_once '../phpQuery/phpQuery/phpQuery.php';
 類QueryList{

 private $pageURL;
 private $regArr = array();
 public $jsonArr = array();
 private $regRange;
 private; > /************************************************
 *參數: 頁面位址選擇器陣列塊選擇器
 * 【選擇器陣列】說明：格式array("名稱"=>array("選擇器","類型"),.......)
 * 【類型】說明：值"text" ,"html" ,"屬性"
 *【塊選擇器】：指先按照規則選出幾個大塊，然後再分別再在塊裡面進行相關的選擇
 ********************************************** ***/
 函數QueryList($pageURL,$regArr=array(),$regRange='')
 {
 $this->pageURL = $pageURL;

 // 為了能夠取得https://
 $ch =curl_init();
curl_setopt($ch, CURLOPT_URL,$this->pageURL);
curl_setopt($ch, CURLOPT_pageURL);
curl_setopt($ch, CURLOPT_PESSL_IFYER， );
curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, false);
curl_setopt($ch,CURLOPT_RETURNTRANSFER,1);
 $this->html = curl_exec($ch); ) );

 if(!empty($regArr))
 {

 $this->regArr = $regArr;
 $this->regRange = $regRange;
 $this->getList();
 }

 }
 函數setQuery($regArr,$regRange='')
 {
 $this->jsonArr=array ( );
 $this->regArr = $regArr;
 $this->; regRange = $regRange;
 $this->getList();
 }
 中間函數getList( )
 {

 $hobj = phpQuery::newDocumentHTML($this-> html);
 if(!empty($this->regRange))
 {
 $empty($this->regRange))
 {
 $robjrob = pq($hobj)->find($this->regRange);

 $i=0;
 foreach($robj as $item)
 {

 while ( list($key,$reg_value)=each($this->regArr ))
 {
 $iobj = pq($item)->find($reg_value[0]);

 switch($reg_value[1])
 {
 case 'text':
 $this->jsonArr[$i][$key] = trim(pq($iobj)->text() ) ;
 中斷;
 case 'html' :
 $this->jsonArr[$i][$key] = trim(pq($iobj)->html());
 中斷;
預設值：
 $this->jsonArr[$i][$key] = pq($iobj)->attr($reg_value[1]);
break;

 }
 }
 // 重設導覽導覽
 reset($this->regArr);
 $i ;
 }
 }
 else
 {
 while(list($key,$reg_value)=each($this->regArr))
 {
 $lobj = pq($hobj)->find($reg_value[0]);


 $ i=0;
 foreach($lobj as $item)
 {
 switch($reg_value[1])
 {
 case 'text':
 $這個->; jsonArr[$i ][$key] = trim(pq($item)->text());
 中斷;
 case 'html':
 $this->jsonArr[$i ][ $key] = trim(pq($item)->html());
中斷;
 預設值:
 $this->jsonArr[$i ][$key] = pq ($item )->attr($reg_value[1]);
 規則;

 }


 }


 }
 } }
 function getJSON()
 {
 return json_encode($this->jsonArr);
 }

}

require 'Query/QueryList.class.php';


//採集OSC的程式碼分享列表，標題連結作者
$url = "http://www.oschina.net/code/list";
$reg = array("title"=>array(".code_title a:eq(0)","text "),"url"=>array(".code_title a:eq(0)","href"),"author"=>array("img","title"));
$rang = " .code_list li";
$hj = new QueryList($url,$reg,$rang);
$arr = $hj->jsonArr;
print_r($arr);
//如果還想採目前頁面右邊的TOP40活躍貢獻者影像，得到JSON資料,可以這樣寫
$reg = array("portrait"=>array(".hot_top img","src"));
$hj->setQuery($reg);
$json = $hj->getJSON();
echo $json . "";

//採OSC內容頁內容
$url = "http://www.oschina.net/code/snippet_186288_23816";
$reg = array("title"=>array(".QTitle h1","text"), "con"=>array(".Content","html"));
$hj = new QueryList($url,$reg);
$arr = $hj->jsonArr;

複製程式碼

 /**
 *自己寫的百度和Google搜尋API
 *版本V2.0
 *作者:JAE
 *部落格:http://blog.jaekj.com
 **/
require_once 'QueryList_class.php' 🎜> private $searcher; 私人$num;
 私人$page;
 私人$regArr; > 公共 $jsonArr; if ($searcher=='baidu')
 {
 $this-> ;regArr = array("title"=>array("h3.t a,#thing_single_box a","text"),"tCon "=>array("div.c-abstract,font:slice(0,2 ),div#weibo,tabletree:eq(0),div.c-abstract-size p:eq(0),div.vd_sitcom_new_tinfo" ,"text"),"url"=>array("h3.t a, #thing_single_box a","href")); -op';
 $this->regZnum=array("zNum"=>array("span.nums","text")); 'google' )
 {
 $this->regArr = array("title"=>array("h3.r a","text"),"tCon"=>array("span.st", "text"),"url"=>array("h3.r a","href")); >regZnum=array("zNum"=>array("div#resultStates","text")) ;key = $key;
 $this->num = $num;
 $this->page = $page-1;
$this->getList();私有函數 getList()
 {
 $s = urlcode($this->key); >num*$this->page;
 if($this->searcher=='baidu')
 {
 $url = "http://www.baidu.com/s ?pn= $start&rn=$num&wd=$s"
 $reg_znum='/[d,] /';
 }
 else if($this->searcher=='google')
 {
 $url="https://www.google.com.hk/search?filter=0&lr=&newwindow=1&safe=images&hl=one&as_qdr=all&num=$num&start=$start&q=$s";
 $reg_znum= '/([d,] ) 結果？ {
 $searcherObj->jsonArr[$i]['url'] = $this-&getBaiduRealURL($searcherObj->jsonArr[$i]['url']);
 else if($this->searcher=='google')
{
$searcherObj ->jsonArr[$i]['url'] = $this->getGoogleRealURL($searcherObj ->jsonArr [$i]['url']);
}
}
 $ this->jsonArr = $searcherObj->jsonArr 預設值

 $searcherObj->setQuery(this$- >regZnum); $zNum=$arr; [0]:$zNum=0; /方向鍵
 $zPage = ceil($zNum/$this->num); 頁'=>((int)$this->第1頁),'zNum'=>$zNum,'zPage'=>$zPage,"s"=>"$this->key" ,'other'= >array('author'=>'JAE', 'QQ'=>'734708094','博客'=>'http://blog.jaekj.com'),'data'=>$this->jsonArr) ;


 }
 function getJSON()
 {
 return json_encode($this->jsonArr);
 }
 private function getBduReunction getBduRe > {
 //預設建立預設出口
 $header = get_headers($url,1);
 if (strpos($header[0],'301') || strpos( $ header [ 0 ] , '302' ) )
 {
 if ( is_array ( $ header [ ' Location ' ] ) )
 {
 // 回傳$ header [ ' Location ' ][ count ( $header[ '位置'])-1];
 return $header['位置'][0];
}
 else
 {
 return $header['Location'];
 }
 }
 else
 {
 return $url;
 }
 {
 return $url;
 }
 }
 }
 private function getGoogleRealURL($url)
 {
 $reg_url = '/q=(. )&/U';
 return preg_match($reg_url,$url,$arr)?urldecode ($arr[1]):$url;

 }
 }