抓取chinaren.com校友录留言的PHP小程序
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=gb2312">
<meta http-equiv="pragma" content="no-cache">
<title>提取留言</title>
<style>
.head { color: red; font-weight: bold; }
body { font-size: 9pt; background-color: #cccccc; }
</style>
</head>
<body>
<?php
set_time_limit(600);
function getMessage($url,$history=false)
{
$match_msg = "/<script>do.*\('[^\n]*/";
$match_date = "/\d{4}-\d{2}-\d{2}\040\d{2}:\d{2}:\d{2}/";
if($history==false){
$match_names = "/&cid=.{2,6}&msg=/";
$replace_names = "/&cid=|&msg=/";
}else{
$match_names = "/class\=\"cr5\"[^\n]*/";
$replace_names = "/class\=\"cr5\" target\=\"_blank\">|<\/a>/";
}
$replace_msg = "/<script>|<\/script>|'\d*\'|doFlatTxt\('|doStr\('| [^\n]*|\'\)|\\\\/";
$handle = fopen ($url, "r");
$contensts = "";
$times="";
$names="";
$messages="";
while ($line=fgets($handle,1024))
{
$contents .= $line;
}
//$contents = fread ($handle, 100000);
//echo $contents;
fclose ($handle);
preg_match_all($match_date,$contents,$times);
preg_match_all($match_names,$contents,$names);
preg_match_all($match_msg,$contents,$messages);
for($i=0;$i<count($messages[0]);$i++)
{
echo "<p><b>". preg_replace($replace_names,"",$names[0][$i]) ."</b>(";
echo $times[0][$i]."):<br>\n";
$message=preg_replace($replace_msg,"",$messages[0][$i])."\n\n";
echo $message;
}
}
$begin=time();
echo "<p class=\"head\">最新留言：</p>\n";
getMessage("http://alumni.chinaren.com/class/class_index.jsp?classuuid=2815032345960598103");
echo "<p class=\"head\">更多留言：</p>\n";
getMessage("http://alumni.chinaren.com/class/class_leaveword.jsp?classuuid=2815032345960598103&p=1");
getMessage("http://alumni.chinaren.com/class/class_leaveword.jsp?classuuid=2815032345960598103&p=2");
getMessage("http://alumni.chinaren.com/class/class_leaveword.jsp?classuuid=2815032345960598103&p=3");
getMessage("http://alumni.chinaren.com/class/class_leaveword.jsp?classuuid=2815032345960598103&p=4");
getMessage("http://alumni.chinaren.com/class/class_leaveword.jsp?classuuid=2815032345960598103&p=5");
getMessage("http://alumni.chinaren.com/class/class_leaveword.jsp?classuuid=2815032345960598103&p=6");
getMessage("http://alumni.chinaren.com/class/class_leaveword.jsp?classuuid=2815032345960598103&p=7");
echo "<p class=\"head\">历史留言：</p>\n";
for($i=0;$i<100;$i++)
{
}
echo "\n<br><center><b>执行本程序用的时间是<font color=red>";
echo time()-$begin;
echo "</font>秒钟</b></center>";
?>
</body>
</html>