php 爬虫爱奇艺 视频、内容

<?php

function getdata( $i, $url) {

   $data = array();

// 把整个文件读入到字符串中

   $str = file_get_contents( $url);

   $str = substr( $str, strpos( $str, 'album-head-info clearfix'));

// print_r($str);

   $str = substr( $str, 0, strpos( $str, 'class="album-auto"'));

   $preg= '/<a .* ?id="j-album-title".* ?>(.* ?)< \/ a>/is'; 

   preg_match_all( $preg, $str, $match);

   //echo $str;exit;

   $title = $match[ 1][ 0];

   $data[ 'title'] = $title;

   $preg= '/<span .* ?class="info-intro-title-s".* ?>(.* ?)< \/ span>/is'; 

   preg_match_all( $preg, $str, $match);

   $other_title = $match[ 1][ 0];

   $data[ 'other_title'] = $other_title;

   //地区

   $preg= '/<p .* ?class="episodeIntro-area".* ?>.* ?<em>(.* ?)< \/ em>.* ?<a.* ?>(.* ?)< \/ a>.* ?< \/ p>/is'; 

   preg_match_all( $preg, $str, $match);

   $data[ 'area'] = trim( $match[ 2][ 0]);

   //语言

   $preg= '/<p .* ?class="episodeIntro-lang".* ?>.* ?<em>(.* ?)< \/ em>.* ?<span.* ?>(.* ?)< \/ span>.* ?< \/ p>/is'; 

   preg_match_all( $preg, $str, $match);

   $data[ 'lang'] = trim( $match[ 2][ 0]);

   //类型-悬疑/历史/剧情

   $preg= '/<a .* ?qwys_leixing.* ?>(.* ?)< \/ a>/is'; 

   preg_match_all( $preg, $str, $match);

   $data[ 'type'] = implode( '/', $match[ 1]);

  

// 时间

   $preg= '/<p .* ?class="episodeIntro-time".* ?>.* ?<em>(.* ?)< \/ em>.* ?<span.* ?>(.* ?)< \/ span>.* ?< \/ p>/is'; 

   preg_match_all( $preg, $str, $match);

   $data[ 'time'] = $match[ 2][ 0];

// 导演

   $preg= '/<p .* ?class="episodeIntro-director".* ?>.* ?<em>(.* ?)< \/ em>.* ?<a.* ?>(.* ?)< \/ a>.* ?< \/ p>/is'; 

   preg_match_all( $preg, $str, $match);

   $data[ 'daoyan'] = $match[ 2][ 0];

// 简介

   $preg= '/<span .* ?class="briefIntroTxt".* ?>(.* ?)< \/ span>/is';

   preg_match_all( $preg, $str, $match);

   //$data['summary'] = $match[1][0];

  

   if(! empty( $match[ 1][ 1])) {

     $data[ 'summary_all'] = $match[ 1][ 1];

  } elseif (! empty( $match[ 1][ 0])) {

     $data[ 'summary_all'] = $match[ 1][ 0];

  }

   $preg= '/<img .* ?src="(.* ?)".* ?id="j-album-img".* ?>/is'; 

   preg_match_all( $preg, $str, $match);

   $img = $match[ 1][ 0];

   $file = '/data/' . $i. '.jpg';

   if(! file_exists( $file)) {

     $f = file_get_contents( $img);

     if( $f) {

       file_put_contents( $file, $f);

    }

  }

  

   return $data;

}

// explode 将字符串打散

$data = file( 'dianshiju02.txt');

$ret = array();

$i = 5000;

foreach( $data as $v) {

   $i++;

   // if(strpos($v, 'mp4') !== false) {

   //  continue;

   // }

  

   $tmp = explode( " \t ", $v);

   // print_r($tmp);

   // exit;

   $num = ( int) $tmp[ 0];

   $_names = explode( "/", $tmp[ 1]);

   $_names = explode( "-", $_names[ 0]);

   $mp4 = $i . "/01.mp4";

   $infos = array();

   if(! empty( $tmp[ 2])) {

     print_r( $i);

     print_r( $tmp[ 2]);

     // exit;

     $infos = getdata( $i, $tmp[ 2]);

  } else {

     continue;

  }

   $ret[ $i] = array(

     'title' => $infos[ 'title'],

     'num' => $num,

     'img' => '//static0.qianqian.com/movies/' . $i . '.jpg',

     'mp4' => 'http://qukufile2.qianqian.com/data2/film_tv/tv/' . $mp4,

     'id' => $i,

     'infos' => $infos

);

}

// echo count($ret);

echo var_export( $ret, true);

40 雪山飞狐-01.mpg http://www.iqiyi.com/lib/m_204754714.html?src=search 33 大捕房-01.mp4 http://www.iqiyi.com/lib/m_202787314.html?src=search 40 嫁入豪门=01.mp4 http://www.iqiyi.com/lib/m_200881014.html?src=search 32 劝和小姐-01.mp4 http://www.iqiyi.com/lib/m_200840914.html?src=search 30 血色恋情-01.mp4 http://www.iqiyi.com/lib/m_202498214.html?src=search 32 锁定美军特使-01.mp4 http://www.iqiyi.com/lib/m_218730014.html?src=search 32 红狐-01.mp4 http://www.iqiyi.com/lib/m_202587214.html?src=search 34 大浴堂-01.mp4 http://www.iqiyi.com/lib/m_200880514.html?src=search 30 女婿难当-01.mp4 http://www.iqiyi.com/lib/m_206378814.html?src=search 27 风云1911-01.mp4 http://www.iqiyi.com/lib/m_202904114.html?src=search 28 醉红尘-01.mp4 http://www.iqiyi.com/lib/m_202964014.html?src=search 23 栗裕大将-01.mp4 http://www.iqiyi.com/lib/m_215180614.html?src=search 24 将军日记-01.mp4 http://www.iqiyi.com/lib/m_202547514.html?src=search