curl的一个采集某小说站的实例

最新推荐文章于 2022-07-23 03:46:31 发布

u011886490

最新推荐文章于 2022-07-23 03:46:31 发布

阅读量802

点赞数

分类专栏：系统运维

本文链接：https://blog.csdn.net/lampzdrj/article/details/10599927

版权

298 篇文章 0 订阅

订阅专栏

<?php
function insert($href='',$pattern='',$novel='',$author='')
{
$href = "http://www.kenshuge.com/Html/Book/4/4410/Index.shtm";
$hrefb = strrev($href);
$hrefb = strstr($hrefb,'/');
$hrefb = strrev($hrefb);
$pattern = '/\<a\s+href="([^"]+)"\s*title="([^"]+)"\s*\>(.+?)\<\/a\>/i';
$url = curl_init($href);
curl_setopt($url,CURLOPT_RETURNTRANSFER,true);
try{
$pdo = New PDO('mysql:location=localhost;dbname=test','testuser','mytest');
$pdo->query('set names utf8');
}catch( PDOException $e ){
echo $e->getMessage();
}
$string = curl_exec($url);
$string = iconv('gbk', 'utf-8', $string);
preg_match_all( $pattern, $string, $array );
$stmt = $pdo->prepare("insert into novel (`novelName`,`author`,`chapterName`,`chapterinfo`,`chapterLink`) values(?,?,?,?,?) ;");
$j = count($array[1]);
echo '<table border="1" width="1250">';
$novel = '异界全职业大师';
$author = '庄毕凡';
for ($i = 0; $i < $j; $i++) {
$link = $hrefb.$array[1][$i];
$info = $array[2][$i];
$title = $array[3][$i];
$stmt->bindParam(1, $novel) ;
$stmt->bindParam(2, $author);
$stmt->bindParam(3, $title);
$stmt->bindParam(4, $info);
$stmt->bindParam(5, $link);
$rst = $stmt->execute();
$color = $rst ? 'green' : 'red';
echo '<tr><td>';
echo '<font color="'.$color.'">';
echo $link;
echo '</font>';
echo '</td><td>';
echo '<font color="'.$color.'">';
echo $info;
echo '</font>';
echo '</td><td>';
echo '<font color="'.$color.'">';
echo $title;
echo '</font>';
echo '</td></tr>';
if (strstr($title,'大结局')!==false) {
break;
}
}
echo '</table>';
curl_close($url);
}
insert();