php获取百度收录、百度热词及百度快照的方法

 <?php /** * @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('/<table>(.*)/is', $templateRss, $_description)) { $templateRss = $_description [0]; $templateRss = str_replace("&", "&", $templateRss); } $templateRss = "" . $templateRss; $xml = @simplexml_load_String($templateRss); foreach ($xml->tbody->tr as $temp) { if (!empty ($temp->td->a)) { $keyArray [] = trim(($temp->td->a)); } } return $keyArray; } print_r(getBaiduHotKeyWord());

 <?php $domain = "https:/<i style="color:transparent">来源gaodai$ma#com搞$$代**码网</i>/www.gaodaima.com/"; *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 = "/<span class="g">(.*)</span>/"; /*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.gaodaima.com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"; ?>   <title>Test</title> <table> <tr> <td>日期</td><td>百度收录</td><td>百度今日收录</td><td>百度快照日期</td></tr><tr> <td> </td><td> </td><td> </td><td> </td></tr></table><p>百度收录：" target="_blank"> </p><p>百度今日收录：" target="_blank"> </p><p>百度快照日期："> </p>