php 问问采集代码

作者：袖梨 2022-06-24

代码如下

复制代码

session_start();
header("content-type:text/html;charset=utf-8");
require("stole_config.php");
require("conn.php");
require("keyword.php");
if(!empty($_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(empty($_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("/

(.*)/ius",$str,$asklist);//获取问答列表
//echo $asklist[1];
$url="/ @preg_match_all($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("

","",str_replace("

","",$html));
$html=str_replace("

","

",$html);
//echo $html;
@preg_match("/

(.*)

/ius",$html,$ask_title);
$art_title=$ask_title[1];
@preg_match("/

(.*)

/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="/()(.*)/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 "添加中............................
";
echo $art_title."
";
$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}";
}
$art_content=iconv('utf-8','gbk',str_replace(" ","
",$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(empty($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 "";
exit();
}
?>
css教程" rel="stylesheet" type="text/css" />