php 問問採集代碼

代碼如下

複製代碼

<?php 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']));//擷取表單提交的問題 $sp="s".$ask; }else { $sp=urlencode($_get['sp']); } if(empty($_get['jl'])) { $_get['jl']=1; } $jl=$_get['jl']; $pg=intval($_get['pg']);//擷取頁數 $rs=intval($_get['rs']);//獲得 記錄的參數 if($rs>9) { $rs=0; $pg++; } if($pg>51) { echo "採集完畢！ 總共採集 ".urldecode($sp)." ".$jl."條記錄"; exit(); } if($sp)//有設定答案才開始 { $str=@file_get_contents("http://wenwen.soso.com/z/search.e?sp={$sp}&pg={$pg}"); @preg_match("/<ol class="result_list">(.*)</ol>/ius",$str,$asklist);//擷取問答列表 //echo $asklist[1]; $url="/<a target="_blank" href="/z/(q.*.htm)/ius"; @preg_match_all($url,$asklist[1],$urllist);//擷取 所有的問題 $t=$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; $art_content="";//商品詳細 for($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;//初始化是否允許採集 $word_count=count($word_arr);//總數 for($i=0;$i<$word_count;$i++) { if(substr_count($art_title,$word_arr[$i])>0) { $word_allow=1; $i=$word_count; } } if($word_allow)//如果合法 { //開始處理資料庫教程 if(strlen($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' ");//查看標題是否重複 $title_ct=@mysql_fetch_array($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','$suid')";//插入採集表 mysql_query($sql); if(empty($t_catx_id))//如果無分類 { $sql2="insert into {$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++;//如果存放資料庫中 則記錄加1 //處理資料庫結束 }else { echo "長度不夠"; } }else { echo "主題不符合要求"; } }else { echo "已經存在"; } $rs++; //記錄下本次採集 的狀況 $f_tt= urldecode($sp)."--頁數".$pg." 記錄數 ".$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">薦禮啦</a>知道問問採集外掛程式</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">採集設定</a> <a href="uninstall.php" onclick="return confirm('您確定要卸載採集外掛程式嗎');">卸載採集</a> <a href="cj_view.php">查看採集記錄</a> <a href="cj_help.php">採集協助</a> <a href="baidu.php" target="_blank">知道採集</a> <a href="wenwen.php" target="_blank">問問採集</a></td> </tr> </table> <form action="wenwen.php" method="post"> <table width="628" height="49" border="0" align="center"> <tr> <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="問問採集" style=" padding-left:15px; padding-right:15px; height:25px; line-height:25px;"></td> </tr> </table> </form>