function get_dm_weight($h, $i)
{
//$h表示搜索结果的url,$i表示结果排名
$p = ceil($i / 10); //搜索结果第几页
$i_weight = array(1 => 52, 2 => 15, 3 => 10, 4 => 5, 5 => 5, 6 => 4, 7 => 1, 8 => 3, 9 => 2, 0 => 3);
$h_str = preg_replace("/^.*?://(.*?)(#.*)?$/i", "$1$3", $h); //去除url中的协议(例如http)与#信息片段部分
$h_arr = explode("?", $h_str, 2);
$h0_arr = explode("/", $h_arr[0]);
if ($h0_arr[count($h0_arr) - 1] == '')
{
$level_l = count($h0_arr) - 1;
}
else
{
$level_l = count($h0_arr);
}
if (isset($h_arr[1]))
{
if ($h_arr[1] != '')
{
$level_r = count(explode("&", $h_arr[1]));
}
else
{
$level_r = 0;
}
}
else
{
$level_r = 0;
}
$dw = $i_weight[$i % 10] * pow(0.5, ($level_l + $level_r - 1)) * 9 / pow(10, $p);
return number_format($dw, 1);;
}
function fetch_baidu($d, $k)
{
$urlw = urlencode(iconv("utf-8","gbk//ignore",$k));
$max_srh_page = 2; //百度搜索结果50条/页,提取2页,也就是只在前100条搜索结果中检查排名,最大值可以设为16
$baidu_ids = array(); //存储百度系列子站点占据的排名位置
$baidu_mus = array(); //存储百度开放平台等优质站点占据的排名位置
$isrank = 0; //$isrank = 1 当前域名下这个关键词获得排名; $isrank = 0 当前域名下这个关键词没有排名
$all_count = 0; //测试变量,以确认匹配规则不会遗漏任何一条搜索结果
$dm_weight = 0; //分析搜索结果页面中顶级、次级、目录、内页的情况,粗略反映一个关键字的竞争激烈程度,非常不准,仅供参考
for ($page_no = 1; $page_no <= $max_srh_page; $page_no++)
{
if ($page_no > 16) break;
$fail_try = 1;
$pn = ($page_no - 1) * 50;
$url = "http://www.baidu.com/s?wd={$urlw}&pn={$pn}&rn=50";
$snoopy = new snoopy;
// $snoopy->proxy_host = "127.0.0.1"; //采集可选代理ip,以免频繁抓百度反被百度咬
// $snoopy->proxy_port = "80"; //proxy代理所用端口
$snoopy->fetch($url);
$contents = iconv("gbk","utf-8//ignore",$snoopy->results);
unset($snoopy);
// echo $contents;
if (!preg_match("/此内容系百度根据您的指令自动搜索的结果/i",$contents,$out))
{
if ($fail_try > 5)
{
continue;
}
else
{
$fail_try++;
$page_no--;
sleep(30);
continue;
}
}
if (!isset($ebaidu))
{
$ebaidu = array('lt' => 0, 'lb' => 0, 'r' => 0); //记录百度推广数量,分为左上、左下、右侧
if (preg_match_all("/
{
$ebaidu['lt'] = count($out_lt[0]);
}
if (preg_match_all("//i", $contents, $out_lt))
{
$ebaidu['lt'] = count($out_lt[0]);
}
if (preg_match_all("//i", $contents, $out_lb))
{
$ebaidu['lb'] = count($out_lb[0]);
}
if (preg_match_all("/
$ebaidu['r'] = count($out_r[0]);
}
}
if (preg_match_all("/(()|(
))[s|s]*?((
{
foreach ($out_mu[0] as $om_key => $om_val)
{
$om_id1 = $out_mu[3][$om_key];
$om_href1 = $out_mu[4][$om_key];
$om_id2 = $out_mu[6][$om_key];
$om_href2 = $out_mu[7][$om_key];
$om_title = strip_tags($out_mu[10][$om_key]);
$baidu_mus[] = $om_id1 . $om_id2;
$dm_weight = $dm_weight + get_dm_weight($om_href1 . $om_href2, $om_id1 . $om_id2);
if (preg_match("/://(w*?.)*?baidu.com//i", $om_href1 . $om_href2, $om_domain))
{
$baidu_ids[] = $om_id1 . $om_id2;
}
if (preg_match("/://(w*?.)*?{$d}//i", $om_href1 . $om_href2, $om_domain))
{
echo 'http://www.baidu.com/s?wd=' . $urlw . '">' . $k . '' . $om_id1 . $om_id2 . '' . $om_title . '';
$isrank = 1;
}
}
}
if (preg_match_all("/
/i", $contents, $out_all))(
.*?.*? ((d{4}-d{1,2}-d{1,2})|(d+小时前)|(d+分钟前)) .*?.*?
{
foreach ($out_all[0] as $o_key => $o_val)
{
$o_id = $out_all[1][$o_key];
$o_href = $out_all[3][$o_key];
$o_title = strip_tags($out_all[4][$o_key]);
$o_cache = $out_all[6][$o_key] . $out_all[7][$o_key] . $out_all[8][$o_key];
$dm_weight = $dm_weight + get_dm_weight($o_href, $o_id);
if (preg_match("/://(w*?.)*?baidu.com//i", $o_href, $o_domain))
{
$baidu_ids[] = $o_id;
}
if (preg_match("/://(w*?.)*?{$d}//i", $o_href, $o_domain))
{
echo 'http://www.baidu.com/s?wd=' . $urlw . '">' . $k . '' . $o_id . '' . $dm_weight . '%' . $o_title . '' . $o_cache . '';
$isrank = 1;
}
}
}
//$all_count用来检查上述正则匹配是否匹配到所有搜索结果,特别关注百度系列站点、百度开放平台以及百度应用等有别于普通搜索结果
if (isset($out_mu[0])) $all_count = $all_count + count($out_mu[0]);
if (isset($out_all[0])) $all_count = $all_count + count($out_all[0]);
// echo '
总共找到' . $all_count . '个匹配
';
if (!preg_match("/
{
break;
}
}
if (count($baidu_mus) >= 1)
{
$bmus = implode(",", $baidu_mus);
}
else
{
$bmus = '';
}
if (count($baidu_ids) >= 1)
{
$bids = implode(",", $baidu_ids);
}
else
{
$bids = '';
}
if ($isrank == 1)
{
echo '上' . $ebaidu['lt'] . '下' . $ebaidu['lb'] . '右' . $ebaidu['r'] . '' . $bids . '' . $bmus . '';
}
else
{
echo 'http://www.baidu.com/s?wd=' . $urlw . '">' . $k . '0' . $dm_weight . '%上' . $ebaidu['lt'] . '下' . $ebaidu['lb'] . '右' . $ebaidu['r'] . '' . $bids . '' . $bmus . '';
}
unset($ebaidu);
}
//逐个域名与关键字采集百度排名信息并显示
set_include_path(".");
include "snoopy.class.php";
set_time_limit(0);
ob_flush(); flush(); ob_end_clean(); ob_implicit_flush(1);
echo '
echo '关键字排名竞争度网页标题百度快照百度推广数量百度占位百度mu占位';
foreach ($kwlist as $d_ks)
{
$dm = $d_ks['domain'];
echo ' ';
foreach ($d_ks['kword'] as $k)
{
fetch_baidu($dm, $k);
}
}
echo '
?>
tongji();
?>
相关文章
精彩推荐