<?php
// 吴燕军
// 采集程序php100.com
set_time_limit(0);
//cookie保存目录
$cookie_jar = ''/tmp/cookie.tmp'';
/*函数 ------------------------------------------------------------------------------------------------------------*/
//模拟请求数据
function request($url,$postfields,$cookie_jar,$referer){
$ch = curl_init();
$options = array(CURLOPT_URL => $url,
CURLOPT_HEADER => 0,
CURLOPT_NOBODY => 0,
CURLOPT_PORT => 80,
CURLOPT_POST => 1,
CURLOPT_POSTFIELDS => $postfields,
CURLOPT_RETURNTRANSFER => 1,
CURLOPT_FOLLOWLOCATION => 1,
CURLOPT_COOKIEJAR => $cookie_jar,
CURLOPT_COOKIEFILE => $cookie_jar,
CURLOPT_REFERER => $referer
);
curl_setopt_array($ch, $options);
$code = curl_exec($ch);
curl_close($ch);
return $code;
}
// 获取帖子列表
function getThreadsList($code){
preg_match_all(''/ <!--[.|\r|\n]*? <a href=\"viewthread.php\?tid=(\d+)/'',$code,$threads);
return $threads[1];
}
//判断该帖子是否存在
function isExits($code){
preg_match(''/ <p>指定的主题不存在或已被删除或正在被审核,请返回。 <\/p>/'',$code,$error);
return isset($error[0])?false:true;
}
//获取帖子标题
function getTitle($code){
preg_match(''/ <h1>[^ <\/h1>]*/'',$code,$title_tmp);
$title = $title_tmp[0];
return $title;
}
//获取帖子作者:
function getAuthor($code){
preg_match(''/ <a href=\"space.php\?uid=\d+\" target=\"_blank\" id=\"userinfo\d+\" onmouseover=\"showMenu\(this\.id\)\">.+/'',$code,$author_tmp);
$author = strip_tags($author_tmp[0]);
return $author;
}
//获取楼主发 表的内容
function getContents($code){
preg_match(''/ <div id=\"postmessage_\d+\" class=\"t_msgfont\">(.|\r|\n)*? <\/div>/'',$code,$contents_tmp);
$contents = preg_replace(''/images\//'',''http://bbs.war3.cn/images/'',$contents_tmp[0]);
return $contents;
}
//打印帖子标题
function printTitle($title){
echo " <strong> <h2>帖子标题: </h2> </strong>",strip_tags($title)," <br/> <br/>";
}
//输出帖子作者
function printAuthor($author){
echo " <strong> <h2>帖子作者: </h2> </strong>",strip_tags($author)," <br/> <br/>";
}
// 打印帖子内容
function printContents($contents){
echo " <strong> <h2>作者发表的内容: </h2>",$contents," </strong> <br/>";
}
//错误
function printError(){
echo " <i>该帖子不存在! </i>";
}
/*函 |