php 问问采集代码

发布:smiling　来源: PHP粉丝网　添加日期：2014-09-09 10:48:31　浏览: 评论：0　

提供一款免费的php 问问采集代码,如果你正想采集问问,但是又不知道怎么写采集程序,这里为你提供一款经典的问问采集程序,代码如下:

<?php

session_start();

header("content-type:text/html;charset=utf-8");

require("stole_config.php");

require("conn.php");

require("keyword.php");

if(!emptyempty($_post['ask']))

{

$ask=urlencode(trim($_post['ask']));//获取表单提交的问题

$sp="s".$ask;

}else

{

$sp=urlencode($_get['sp']);

}

if(emptyempty($_get['jl']))

{

$_get['jl']=1;

}

$jl=$_get['jl'];

$pg=intval($_get['pg']);//获取页数

$rs=intval($_get['rs']);//获得记录的参数

if($rs>9)

{

$rs=0;

$pg++;

}

if($pg>51)

{

echo "采集完毕！总共采集 ".urldecode($sp)." ".$jl."条记录";

exit();

}

if($sp)//有设定答案才开始

{

$str=@file_get_contents("http://wenwen.soso.com/z/search.e?sp={$sp}&pg={$pg}");

@preg_match("/<ol class="result_list">(.*)</ol>/ius",$str,$asklist);//获取问答列表

//echo $asklist[1];

$url="/<a target="_blank" href="/z/(q.*.htm)/ius";

@preg_match_all($url,$asklist[1],$urllist);//获取所有的问题

$t=$urllist[1][$rs];

$uid=$t;

$suid="ww{$uid}";

$sct=mysql教程_query("select count(*) from {$table_prefix}c_article where suid='$suid' ");

$sct=mysql_fetch_array($sct);

$sct=$sct[0];

if($sct==0)

{

$html=@file_get_contents("http://wenwen.soso.com/z/${t}");

$html=str_replace("<pre>","",str_replace("</pre>","",$html));

$html=str_replace("<br/><br/><br/>","<br/><br/>",$html);

//echo $html;

@preg_match("/<div class="question_main">.*<h3>(.*)</h3>/ius",$html,$ask_title);

$art_title=$ask_title[1];

@preg_match("/<div class="answer_con">(.*)</div>/ius",$html,$answer);

$j=count($answer)-1;

$art_content="";//商品详细

for($i=$j;$i>=1;$i--)

{

if(strlen($answer[$i])>$min_t1)

{

$art_content .= $answer[$i];

}

}

$art_content=trim($art_content);

$s1="/(<a .*>)(.*)</a>/ius";

$art_content=preg_replace($s1,${2},trim($art_content));

$word_arr=explode(",",iconv("gbk","utf-8",$cj_word));

$word_allow=false;//初始化是否允许采集

$word_count=count($word_arr);//总数

for($i=0;$i<$word_count;$i++)

{

if(substr_count($art_title,$word_arr[$i])>0)

{

$word_allow=1;

$i=$word_count;

}

}

if($word_allow)//如果合法

{ //开始处理数据库教程

if(strlen($art_content)>$min_t2)

{

echo "<font color=red>添加中............................</font><br>";

echo $art_title."<br>";

$art_title=iconv('utf-8','gbk', $art_title);

$title_ct=mysql_query("select count(*) from {$table_prefix}c_article where art_title ='$art_title' ");//查看标题是否重复

$title_ct=@mysql_fetch_array($title_ct);

$title_ct=$title_ct[0];

if($title_ct>0)

{

$art_title .="{$same_title}{$title_ct}";

}

$art_content=iconv('utf-8','gbk',str_replace(" ","<br>",$art_content));

$art_content=strtr($art_content,$keyword);

$art_time=date("y-m-d");

$sql="insert into {$table_prefix}c_article(art_title,art_content,art_time,art_author,suid) values('$art_title','$art_content','$art_time','$art_author','$suid')";//插入采集表

mysql_query($sql);

if(emptyempty($t_catx_id))//如果无分类

{

$sql2="insert into {$t_table}({$t_art_title},{$t_art_content},{$t_art_time},{$t_artx_author}) values('$art_title','$art_content','$art_time','$art_author')";

}else

{

$sql2="insert into {$t_table}({$t_art_title},{$t_art_content},{$t_art_time},{$t_artx_author},{$t_catx_id}) values('$art_title','$art_content','$art_time','$art_author','$cat_id')";

}

mysql_query($sql2);//插入文章表

$jl++;//如果存放数据库中则记录加1

//处理数据库结束

}else

{

echo "长度不够";

}

}else

{

echo "主题不符合要求";

}

}else

{

echo "已经存在";

}

$rs++;

//记录下本次采集的状况

$f_tt= urldecode($sp)."--页数".$pg." 记录数 ".$jl ;

file_put_contents("ss.txt",$f_tt);

echo "<script>location.href='wenwen.php?jl=".$jl."&sp=".$sp."&pg=".$pg."&rs=".$rs." ';</script>";

exit();

}

?>

<link href="style.css教程" rel="stylesheet" type="text/css" />

<table width="700" border="0" align="center" cellspacing="1" bgcolor="#cccccc">

<tr>

<td height="50" align="center" bgcolor="#00cc00"><h1><a href="http://www.111cn.net">荐礼啦</a>知道问问采集插件</h1></td>

</tr>

</table>

<table width="700" border="0" align="center" cellspacing="1" bgcolor="#cccccc" style="margin-top:6px; margin-bottom:6px;">

<tr>

<td height="30" align="center" bgcolor="#ffffff"><a href="cj_config.php">采集设置</a> <a href="uninstall.php" onclick="return confirm('您确定要卸载采集插件吗');">卸载采集</a> <a href="cj_view.php">查看采集记录</a> <a href="cj_help.php">采集帮助</a> <a href="baidu.php" target="_blank">知道采集</a> <a href="wenwen.php" target="_blank">问问采集</a></td>

</tr>

</table>

<form action="wenwen.php" method="post">

<table width="628" height="49" border="0" align="center">

<tr> //开源代码phpfensi.com

<td width="413" align="right"><input name="ask" type="text" id="ask" size="50"></td>

<td width="205"><input type="submit" name="button" id="button" value="问问采集" style=" padding-left:15px; padding-right:15px; height:25px; line-height:25px;"></td>

</tr>

</table>

</form>