curl采集完整实例仅供参考(完整版)

<?php

header('Content-type:text/html;charset=utf-8');
error_reporting(E_ERROR | E_WARNING | E_PARSE);
set_time_limit(0);//d等待时间.不限制
ini_set('memory_limit','200m');//设置内存


$url = $_GET['url'];
if(!isset($url)){
  die("请输入地址");
}
//获取网页
$str = get_str($url);


// $block_rule ='//si'; s空格换行字符  i 不区分大小写
$block_rule ='/<li class="clearfix">(.*?)<p class="info">/si'; 
preg_match_all($block_rule,$str,$fenlei);
if(!empty($fenlei)){
	// print_r($fenlei[1]);
  $count = count($fenlei[1]);
  for($i=0;$i<$count;$i++){
    $cat_rule='/<a href="(.*?)" target="_blank">(.*?)<\/a>/si';
    preg_match_all($cat_rule,$fenlei[1][$i],$cats);
    if(!empty($cats)){
      // echo "<pre>";
      // print_r($cats[1]);
      // print_r($cats[2]);
      if(!empty($cats[1][1])){
          $url_r = $cats[1][1];
          $name_r = $cats[2][1];
          $content_str = get_str($url_r);
          $content_rule ='/<div class="newfujian">(.*?)rel="nofollow" target="_blank"><\/a>.*?<\/div>.*?<\/div>/si'; 
          // echo $content_str;
          @preg_match_all($content_rule,$content_str,$cont);
          if(!empty($cont)){
             echo $name_r.$cont[1][0].'<br> <hr>';
          }else{
             // echo $name_r.'--<a href="'.$url_r.'" target="_blank">'.$url_r.'</a><br> ';
             echo $name_r.'</a><br> ';


          }

      }
      

    }
  }
}
  

// echo $str;




//curl获取网页内容
function get_str($url){
	$ch = curl_init($url);

	curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
	curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);

	$str = curl_exec($ch);
	curl_close($ch);

	return $str;

}