php爬虫抓取百度贴吧图片

 1 <?php 2 //运行时间 3 @set_time_limit(60); 4 //贴吧名称 5 $tbname = "%CD%BC%C6%AC"; 6 //抓取类型 0-按照帖子顺序 1-按照贴图顺序 7 $type = 0; 8 //列表页url 9 $listurltpl = "http://tieba.baidu.com/f?kw=%s".($type?"&tp=1":"&pn=");10 //图册页url11 $galleryurltpl = "http://tieba.baidu.com/photo/bw/picture/guide?kw=%s&tid=%s&next=9999";12 //图片url13 $imageurltpl = "http://imgsrc.baidu.com/forum/pic/item/%s.jpg";14 //本地的目录15 $savepath = "h:/images/";16 //帖子子文件夹17 $filedirtpl = $savepath."%s/";18 //图片文件19 $filenametpl = $savepath."%s/%s.jpg";20 21 $listurl = sprintf($listurltpl,$tbname);22 //抓取起始点23 $pn = 0;24 while(1)25 {26     if (!$type) $listurl .= $pn;27     //得到列表页源代码28     $listhtml = file_get_contents($listurl);29     //匹配出帖子id30     if($type)31         preg_match_all('/<div class=\"aep_wrapper\" id=\"pic_item_(\d+)\" tid=\"\d+\">/',$listhtml,$m1);32     else33         preg_match_all('/<ul class=\"threadlist_media j_threadlist_media\" id=\"fm(\d+)\"/',$listhtml,$m1);34     //得到帖子id列表35     $tidlist = $m1[1];36     echo "Fetching ... <br /> \r\n";37     foreach($tidlist as $tid)38     {39         echo "--Gallery $tid <br /> \r\n";40         $galleryurl = sprintf($galleryurltpl,$tbname,$tid);41         //得到帖子图册的源代码42         $galleryhtml = file_get_contents($galleryurl);43         //匹配出图片id44         preg_match_all('/\{\"original\":\{\"id\":\"(\w+)\"/',$galleryhtml,$m2);45         //得到图片id列表46         $pidlist = $m2[1];47         foreach($pidlist as $pid)48         {49             echo "----Picture {$tid}/{$pid}.jpg ";50             $filedir = sprintf($filedirtpl,$tid);51             $filename = sprintf($filenametpl,$tid,$pid);52             //文件是否存在53             if(!is_file($filename))54             {55                 $imageurl = sprintf($imageurltpl,$pid);56                 //下载图片57                 $imagebin = file_get_contents($imageurl);58                 //目录是否存在59                 if(!is_dir($filedir))60                     mkdir($filedir);61                 //保存图片62                 file_put_contents($filename,$imagebin);63                 $rnd = rand(2000,5000);64                 echo "Downloaded! ";65                 //延时休息66                 sleep(1.0*$rnd/1000);67                 echo "Sleep $rnd us <br />\r\n";68             }69             else70                 echo "Existed! <br />\r\n";71         }72     }73     //翻到下一页74     if (!$type) $pn += 50;75 }