php抓取百度快照、百度收录、百度热词程序代码

/*
抓取百度收录代码
*/
function baidu($s){
  $baidu="http://www.baidu.com/s?wd=site%3A".$s;
  $site=file_get_contents($baidu);
  //$site=iconv("gb2312", "UTF-8", $site);
  ereg("找到相关网页(.*)篇，", $site,$count);
  $count=str_replace("找到相关网页","",$count);
  $count=str_replace("篇，","",$count);
  $count=str_replace("约","",$count);
  $count=str_replace(",","",$count);
  return $count[0];
}

echo baidu(www.111com.net); //获取好主题在百度中的收录数量

?>

php
/**
* * @user 小杰
* @return array 返回百度的热词数据(数组返回)
*/
function getBaiduHotKeyWord()
{
$templateRss = file_get_contents('http://top.baidu.com/rss_xml.php?p=top10');
If (preg_match('/

    $domain = “http://www.111com.net/nokia/5230/ *欲查询的域名*/
    $site_url = ‘http://www.baidu.com/s?wd=site%3A’;
    $all = $site_url.$domain; /*域名所有收录的网址*/
    $today = $all.’&lm=1′;    /*域名今日收录的网址*/
    $utf_pattern = “/找到相关结果数(.*)个/”;
    $kz_pattern = “/(.*)/”; /*用以匹配快照日期的字符串*/
    $times = “/d{4}-d{1,2}-d{1,2}/”; /*匹配快照日期的正则表达式，如:2011-8-4*/
    $s0 = @file_get_contents($all);    /*将site:www.ninthday.net的网页置入$s0字符串中*/
    $s1 = @file_get_contents($today);
    preg_match($utf_pattern,$s0,$all_num); /*匹配”找到相关结果数*个”*/
    preg_match($utf_pattern,$s1,$today_num);
    preg_match($kz_pattern,$s0,$temp);
    preg_match($times,$temp[0],$screenshot);
    if($all_num[1] == “”)
        $all_num[1] = 0;
    if($today_num[1] == “”)
        $today_num[1] = 0;
    if($screenshot[0] == “”)
        $screenshot[0] = “暂无快照”;
?>

   
   
   

 

百度收录：” target=”_blank”>

百度今日收录：” target=”_blank”>

百度快照日期：”>