PHP网页抓取之抓取百度贴吧邮箱数据代码分享

<?php
$url2="";
$page="";
if($_GET['url2']==""){
$url2="http://tieba.baidu.com/p/2314539885?pn=1";
}else{
$url2=$_GET['url2'];
}

if($_GET['page']==""){
$page="1";
}else{
$page=$_GET['page'];
}
?>




帖子链接：



总页数：












帖子链接：







<?php
if($_GET['type']!=""){
$counts=0;
if($_GET['type']=="getAll"){
$pages=$_GET['page'];
$url = $_GET['url'];
for($i=0;$i<$pages;$i++){
$ch2 = curl_init();
curl_setopt($ch2, CURLOPT_URL, $url);
curl_setopt($ch2, CURLOPT_FOLLOWLOCATION, TRUE);
curl_setopt($ch2, CURLOPT_SSL_VERIFYHOST, FALSE);
curl_setopt($ch2, CURLOPT_SSL_VERIFYPEER, false);
curl_setopt($ch2, CURLOPT_RETURNTRANSFER, TRUE);
$texts = curl_exec($ch2);
curl_close($ch2);
$dat=getEmail($texts);
for($j=0;$j";
$counts++;
}
}
}else if($_GET['type']=="getNow"){
$url = $_GET['url2'];
$ch2 = curl_init();
curl_setopt($ch2, CURLOPT_URL, $url);
curl_setopt($ch2, CURLOPT_FOLLOWLOCATION, TRUE);
curl_setopt($ch2, CURLOPT_SSL_VERIFYHOST, FALSE);
curl_setopt($ch2, CURLOPT_SSL_VERIFYPEER, false);
curl_setopt($ch2, CURLOPT_RETURNTRANSFER, TRUE);
$texts = curl_exec($ch2);
curl_close($ch2);
$dat=getEmail($texts);
for($i=0;$i";
$counts++;
}
}
echo '
共采集到数据：'.$counts.'条
';
}
function getEmail($str){
$pattern = "/([a-z0-9\-_\.]+@[a-z0-9]+\.[a-z0-9\-_\.]+)/"; 
preg_match_all($pattern,$str,$emailArr); 
return $emailArr[0]; 
}
?>