代码如下:
session_start(); header("content-type:text/html;charset=gbk"); require("stole_config.php"); require("conn.php"); require("keyword.php"); $searchStr=$_GET["searchStr"]; $ss=explode(" ",$searchStr);//拆分搜索关键字 $word="";//关键字设为空 foreach($ss as $key=>$t) { if($key>0) { $word .="+"; } $word .=urlencode($t); } $jl=intval($_GET['jl']); if(isset($_GET['page'])) { $page=intval($_GET['page']); }else{ $page=1; } $rs=intval($_GET['rs']); if($rs>=10) { $rs=0; $page++; } if($page>76) { echo "采集完毕 ${jl}"; exit(); } if(!empty($searchStr))//如果搜索 { //获取问题页面 $cOntent=@file_get_contents("http://zhidao.baidu.com/q?ct=17&lm=0&tn=ikaslist&pn=".(($page-1)*10)."&rn=10&word=".$word); //获取问题列表 preg_match_all("/$uid=$uid[1];//获取详细页文章 $uid=$uid[$rs]; //判断数据是否存在 $suid="bd{$uid}"; $sct=mysql_query("select count(*) from {$table_prefix}c_article where suid='$suid' "); $sct=mysql_fetch_array($sct); $sct=$sct[0]; if($sct==0) { $cOntent=@file_get_contents("http://zhidao.baidu.com/question/".$uid.".html") ; $arr=explode('',$content); $art_title=$arr[1]; $arr=explode('',$art_title); $art_title=$arr[0];//获取标题结束 //判断内容是否符合 $word_arr=explode(",",$cj_word); $word_allow=false;//初始化是否允许采集 $word_count=count($word_arr);//关键字总数 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)//如果满足条件 { $arr=explode('',$content); $cOntentQuestion=$arr[1]; $arr=explode('',$contentQuestion); $cOntentQuestion=$arr[0]; echo "开始采集内容"; echo "$art_title"; @preg_match_all('/(|)(.*)<\/pre>(<\/ca>|<\/cn>)/iUs',$content,$answerArr); $answerArr=$answerArr[2]; if($arr_order==1)//随机排序 { shuffle($answerArr); } if($arr_order==2)//倒序 { $answerArr=krsort($answerArr);//倒序 } foreach($answerArr as $t) { $answerTemp=str_replace('','',$t); $answerTemp=str_replace('','',$answerTemp); $answerTemp=str_replace('','',$answerTemp); $answerTemp=str_replace('','',$answerTemp); if(strlen($answerTemp)>$min_t1) { $art_content .=$answerTemp.""; } } //去除链接 $s1="/()(.*)<\/a>/iUs"; $art_cOntent=preg_replace($s1,${2},trim($art_content)); $art_cOntent=str_replace("\n\r","",$art_content); if(strlen($art_content)>$min_t2) { $title_ct=mysql_query("select count(*) from {$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_time=date("Y-m-d"); $art_cOntent=strtr($art_content,$keyword); $sql="insert into {$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 $jl++; //数据库处理完毕 }else { echo "内容长度不够"; } //获取文章内容结束 }else { echo "主题不符合要求"; } }else { echo "已经存在"; }$rs++; file_put_contents("bd.txt","采集{$searchStr}到第{$page}第{$rs}条"); echo ""; exit(); } ?> 荐礼啦知道问问采集插件 采集设置 卸载采集 查看采集记录 采集帮助 知道采集 问问采集
',$content); $cOntentQuestion=$arr[1]; $arr=explode('
(.*)<\/pre>(<\/ca>|<\/cn>)/iUs',$content,$answerArr); $answerArr=$answerArr[2]; if($arr_order==1)//随机排序 { shuffle($answerArr); } if($arr_order==2)//倒序 { $answerArr=krsort($answerArr);//倒序 } foreach($answerArr as $t) { $answerTemp=str_replace('','',$t); $answerTemp=str_replace('','',$answerTemp); $answerTemp=str_replace('','',$answerTemp); $answerTemp=str_replace('','',$answerTemp); if(strlen($answerTemp)>$min_t1) { $art_content .=$answerTemp.""; } } //去除链接 $s1="/()(.*)<\/a>/iUs"; $art_cOntent=preg_replace($s1,${2},trim($art_content)); $art_cOntent=str_replace("\n\r","",$art_content); if(strlen($art_content)>$min_t2) { $title_ct=mysql_query("select count(*) from {$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_time=date("Y-m-d"); $art_cOntent=strtr($art_content,$keyword); $sql="insert into {$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 $jl++; //数据库处理完毕 }else { echo "内容长度不够"; } //获取文章内容结束 }else { echo "主题不符合要求"; } }else { echo "已经存在"; }$rs++; file_put_contents("bd.txt","采集{$searchStr}到第{$page}第{$rs}条"); echo ""; exit(); } ?>
','',$t); $answerTemp=str_replace('
','',$answerTemp); $answerTemp=str_replace('
session_start(); header("content-type:text/html;charset=utf-8"); require("stole_config.php"); require("conn.php"); require("keyword.php"); if(!empty($_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={$pg}"); @preg_match("/
","",str_replace("
.*
(.*)<\/p>/iUs",$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>/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($word_allow)//如果合法 { //开始处理数据库 if(strlen($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}"; } $art_cOntent=iconv('utf-8','gbk',str_replace("\r\n","",$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(); } ?>