PHP开发通用采集入库程序三

最新推荐文章于 2024-07-15 23:09:31 发布

astarblog

最新推荐文章于 2024-07-15 23:09:31 发布

阅读量37

点赞数

文章标签： php mysql laravel thinkphp 软件开发

本文链接：https://blog.csdn.net/u010711304/article/details/107915216

版权

进入开发阶段 – 单条入库获取下一条
实现真正大批量的入库功能

--
-- 数据库: `caiji`
--

-- --------------------------------------------------------

--
-- 表的结构 `news`
--

CREATE TABLE `news` (
`id` int(11) NOT NULL auto_increment,
`title` varchar(50) collate gb2312_bin NOT NULL,
`content` text collate gb2312_bin NOT NULL,
PRIMARY KEY (`id`)
) ENGINE=MyISAM DEFAULT CHARSET=gb2312 COLLATE=gb2312_bin AUTO_INCREMENT=58 ;

-- --------------------------------------------------------

--
-- 表的结构 `tmp_url`
--

CREATE TABLE `tmp_url` (
`id` int(11) NOT NULL auto_increment,
`title` varchar(100) NOT NULL,
`url` varchar(150) NOT NULL,
PRIMARY KEY (`id`)
) ENGINE=MyISAM DEFAULT CHARSET=gb2312 AUTO_INCREMENT=166 ;

<?php

mysql_connect("localhost","root","");
mysql_select_db("caiji");
mysql_query("set names 'GBK'");

<?php
include_once("conn.php");

if($_GET[id]<=60 && $_GET[id]){
$con=file_get_contents("http://it.sohu.com/7/1002/17/column203661721_32".$_GET[id].".shtml");

$preg = "#<h1>·<a href='(.*)' target='_blank'>(.*)</a><span>#iUs";

preg_match_all($preg,$con,$arr);

foreach($arr[1] as $id=>$v){

$sql="INSERT INTO `tmp_url` (`id`, `title`, `url`) VALUES (NULL, '".$arr[2][$id]."', '".$v."');";
mysql_query($sql);

     }
$_GET[id]++;
     echo "正在采集列表....".$_GET[id];
echo "<script>location.href='ls.php?id=".$_GET[id]."'</script>";
}else{

     echo "采集结束";
}
?>

<?php
include_once("conn.php");

$gid =(int)$_GET[id];

   $sql="select * from tmp_url where id='$gid'";
   $q=mysql_query($sql);
   $row = mysql_fetch_array($q);

   $con=file_get_contents($row[url]);
    echo $titles="<b>".zz("#<title>(.*)</title>#iUs",$con)."</b>";

echo $conts=zz("#(.*)#iUs",$con);

$intosql="INSERT INTO `news` (`id`, `title`, `content`) VALUES (NULL, '$titles', '$conts');";
mysql_query($intosql);

//============================

   $sql2="select * from tmp_url where id>'$gid' order by id asc limit 1";
   $q2=mysql_query($sql2);
   $row2 = mysql_fetch_array($q2);
    echo $row2[0].$row2[1]."<br>";
    if($row2[0]){
    echo "<script>location.href='vi.php?id=".$row2[0]."'</script>";
}

function zz($preg,$con,$num=1){
preg_match($preg,$con,$arr);
return $arr[$num];
}
?>