`
NealCai
  • 浏览: 143280 次
  • 性别: Icon_minigender_1
  • 来自: 上海
社区版块
存档分类
最新评论

php 爬虫

    博客分类:
  • php
阅读更多
<?php

header("Content-Type:text/html; charset=gb2312");
$url1 = "http://i8i8.cc/";
$ch1 = curl_init();
curl_setopt($ch1, CURLOPT_URL, $url1 );
curl_setopt($ch1, CURLOPT_RETURNTRANSFER, true);
$output1 = curl_exec($ch1);
curl_close($ch1);
$regx1='/(<a href=")(\/cclass\/((9|10|11|12|13|14|15)_1\.html))(">)(.*)(<\/a>)/';
preg_match_all ($regx1,$output1,$result, PREG_SET_ORDER);

//print_r($result);
$conn=mysql_connect(localhost,test,111111);
mysql_select_db("test",$conn);
mysql_query('SET NAMES gb2312;');
foreach($result as $key =>$value)
{

$url2 = "http://i8i8.cc".$result[$key][2];
//print_r($url2);
$ch2 = curl_init();
curl_setopt($ch2, CURLOPT_URL, $url2 );
curl_setopt($ch2, CURLOPT_RETURNTRANSFER, true);
$output2 = curl_exec($ch2);
//print_r($output2);
curl_close($ch2);
$regx2='/(<li>)(<a href=")(.*\.html)(.*)(<img src=")(.*)(" alt=)(.*)(<span>)(.*)(<\/span>)/';
preg_match_all ($regx2,$output2,$movie, PREG_SET_ORDER);

//print_r($movie);

foreach($movie as$k =>$v )
{
$movie[$k][3]=substr($movie[$k][3],2);
$movie[$k][3]="http://i8i8.cc".$movie[$k][3];
//print_r($movie[$k][3]);

$url3 = $movie[$k][3];
$ch3 = curl_init();
curl_setopt($ch3, CURLOPT_URL, $url3 );
curl_setopt($ch3, CURLOPT_RETURNTRANSFER, true);
$output3 = curl_exec($ch3);
//var_dump($output3);
curl_close($ch3);

$regx_name='/(<div class=")(row_right)("><strong>)(.*)<\/strong>/';
preg_match_all ($regx_name,$output3,$movie_name, PREG_SET_ORDER);

$regx_actor='/(<div class="img">)(.*)(artlist&keyword=)(.*)(>)(.*)(<\/a>)/';
preg_match_all ($regx_actor,$output3,$movie_actor, PREG_SET_ORDER);

$regx_catalog='/(<div class="row_right"><A href="\/cclass\/(.*).html">)(.*)(<\/A> )/';
preg_match_all ($regx_catalog,$output3,$movie_catalog, PREG_SET_ORDER);

$regx_year='/<a href="\/year\/(.*).html">(.*)<\/a>/';
preg_match_all ($regx_year,$output3,$movie_year, PREG_SET_ORDER);
//print_r($movie_year);//[2]

$regx_update='/(更新时间:<\/div><div class="row_right">)(.*)<\/div>/';
preg_match_all ($regx_update,$output3,$movie_update, PREG_SET_ORDER);
//print_r($movie_update);//[2]

$regx_desc='/(主演。)(.*)<\/div>/';
preg_match_all ($regx_desc,$output3,$movie_desc, PREG_SET_ORDER);
//var_dump($movie_desc);//[2]
}



//mysql_query("insert into t_catalog values('{$key}','{$result[$key][6]}');");
}
mysql_close($conn);
?>
0
0
分享到:
评论

相关推荐

Global site tag (gtag.js) - Google Analytics