/**
*
*/
private function GetSEOState($taskID)
{
$dataProvider = new CActiveDataProvider(SiteUrl::model(), array('pagination' => false));
echo '資料庫共' . $dataProvider->totalItemCount . '筆資料' . "\r\n";
$i = 0;
foreach ($dataProvider->getData() as $record) {
$site = $record->site;
$id = $record->SiteID;
// if($i>=5)
// {
// break;
// }
//移除前面的http:// (若有的話)
$site = preg_replace('#^https?://#', '', $site);
$site = preg_replace('#^http?://#', '', $site);
//系統延遲
$pagerank = 0;
$now = new DateTime();
try {
$pagerank = \SEOstats\Services\Google::getPageRank($site);
if (!is_numeric($pagerank)) {
$pagerank = null;
}
} catch (Exception $e) {
$pagerank = null;
}
usleep(rand(1000, 3000));
//下面這行是採用Google API提供之資料
//$googleIds = \SEOstats\Services\Google::getSiteindexTotal($site);
//下面這行是採用網頁搜尋結果資料
if ($i % 2 == 1) {
usleep(rand(500, 1000));
$googleIds = $this->GetGoogleSearch("site:{$site}");
} else {
$googleIds = $this->GetGoogleSearch("site:{$site}");
}
sleep(3);
if (is_null($googleIds)) {
//用另外種管道重抓一次
usleep(rand(5000, 10000));
echo '[log]' . $site . "使用重抓索引資料;/r/n";
$googleIds = \SEOstats\Services\Google::getSiteindexTotal($site);
}
if ($i % 2 == 1) {
usleep(rand(500, 1000));
$googleLinks = $this->GetGoogleSearch("link:{$site}");
} else {
$googleLinks = $this->GetGoogleSearch("link:{$site}");
}
//如果抓不到資料就換個管道
if (is_null($googleLinks)) {
//用另外種管道重抓一次
usleep(rand(5000, 10000));
echo '[log]' . $site . "使用API重抓頁面數資料;/r/n";
$googleLinks = \SEOstats\Services\Google::getBacklinksTotal($site);
}
usleep(rand(1010, 15020));
//echo("[log]".$now->format( 'Y-m-d H:i:s' )."搜尋".$site."有".$googleLinks.'項結果 '."\r\n");
//取得社群分享數據
$seostats = new \SEOstats\SEOstats();
$seostats->setUrl("http://{$site}");
$fb = Social::getFacebookShares();
//print_r($fb);
usleep(rand(4000, 10000));
//
//設定抓取的檔案類型
//
$fileTypeList = array('pdf', 'doc', 'docx', 'ppt', 'pptx', 'ps', 'eps');
$fileCount = array();
$error_statues = 0;
foreach ($fileTypeList as $key => $value) {
usleep(rand(5000, 10000));
if ($error_statues == 0) {
$fileCount[$value] = $this->GetGoogleSearch("site:{$site}" . ' filetype:' . $value);
}
if (is_null($fileCount[$value])) {
$fileCount[$value] = \SEOstats\Services\Google::getSiteFileTypeTotal($site, $value);
//表示系統被Google封鎖了。
if (is_null($fileCount[$value])) {
$fileCount[$value] = 0;
$error_statues++;
}
}
}
usleep(rand(4000, 10000));
//取得網站資訊，包含網站一些設定資訊。
$info = $this->get_url_info($site);
$model = new Data();
$model->attributes = array('SiteID' => $id, 'GoogleData' => $googleIds, 'google_backlink' => $googleLinks, 'filetime' => $info['filetime'] = !-1 ? date("Y-m-d H:i:s", $info['filetime']) : null, 'robot' => $this->remoteFileExists("{$site}/robots.txt"), 'sitemap' => $this->remoteFileExists("{$site}/sitemap.xml"), 'Time' => $now->format('Y-m-d H:i:s'), 'Facebook' => $fb['total_count'], 'FB_share_count' => $fb['share_count'], 'FB_like_count' => $fb['like_count'], 'FB_comment_count' => $fb['comment_count'], 'FB_commentsbox_count' => $fb['commentsbox_count'], 'FB_click_count' => $fb['click_count'], 'TwitterShares' => Social::getTwitterShares(), 'LinkedInShares' => Social::getLinkedInShares(), 'pdf' => $fileCount['pdf'], 'doc' => $fileCount['doc'], 'docx' => $fileCount['docx'], 'ppt' => $fileCount['ppt'], 'pptx' => $fileCount['pptx'], 'ps' => $fileCount['ps'], 'eps' => $fileCount['eps'], 'YY' => $now->format('Y'), 'MM' => $now->format('m'), 'DD' => $now->format('d'), 'TaskID' => $taskID, 'google_page_rank' => $pagerank);
//Yii::app()->end();
if ($model->save()) {
$i++;
} else {
print "網址: {$site} 出現錯誤";
print_r($model->getErrors());
}
usleep(rand(100, 400));
// if($i==3)
// {
// break;
// }
}
echo '執行完畢,共儲存' . $i . '筆資料';
}
// you need to follow this steps:
//
// 1. Comment-in line 24 (remove hash char "#") and comment-out line 25 (prepend hash char "#")
// 2. Copy this example file (and the others) from ./example/example.php to ./example.php
//
// For further reference see: https://github.com/eyecatchup/SEOstats/issues/49
// Bootstrap the library / register autoloader
#require_once realpath(__DIR__ . '/SEOstats/bootstrap.php');
require_once realpath(__DIR__ . '/vendor/autoload.php');
use SEOstats\Services\Social;
try {
$url = 'http://www.google.com/';
// Create a new SEOstats instance.
$seostats = new \SEOstats\SEOstats();
// Bind the URL to the current SEOstats instance.
if ($seostats->setUrl($url)) {
echo "Social network metrics for " . $url . PHP_EOL;
/**
* Get the total count of +1s for $url on Google+.
*/
echo "Google+ Shares: " . Social::getGooglePlusShares() . PHP_EOL;
/**
* Get the total count of mentions of $url on Twitter.
*/
echo "Twitter Shares: " . Social::getTwitterShares() . PHP_EOL;
/**
* Get interaction counts (shares, likes, comments, clicks) for $url on Facebook.
*/
echo "Facebook Shares: ";
print_r(Social::getFacebookShares());
/**
function seo_stat($type = '')
{
try {
$url = base_url();
$seostats = new \SEOstats\SEOstats();
if ($seostats->setUrl($url)) {
if ($type == 'facebook') {
return SEOstats\Services\Social::getFacebookShares();
} elseif ($type == 'gplus') {
return SEOstats\Services\Social::getGooglePlusShares();
} elseif ($type == 'twitter') {
return SEOstats\Services\Social::getTwitterShares();
} elseif ($type == 'linkedin') {
return SEOstats\Services\Social::getLinkedInShares();
} elseif ($type == 'pinterest') {
return SEOstats\Services\Social::getPinterestShares();
} elseif ($type == 'alexa_global') {
return SEOstats\Services\Alexa::getGlobalRank();
} elseif ($type == 'alexa_country') {
return SEOstats\Services\Alexa::getCountryRank();
} elseif ($type == 'alexa_bounce') {
return SEOstats\Services\Alexa::getTrafficGraph(5);
} elseif ($type == 'alexa_time') {
return SEOstats\Services\Alexa::getTrafficGraph(4);
} elseif ($type == 'alexa_traffic') {
return SEOstats\Services\Alexa::getTrafficGraph(1);
} elseif ($type == 'alexa_pageviews') {
return SEOstats\Services\Alexa::getTrafficGraph(2);
} elseif ($type == 'google_siteindex') {
return SEOstats\Services\Google::getSiteindexTotal();
} elseif ($type == 'google_back') {
return SEOstats\Services\Google::getBacklinksTotal();
} elseif ($type == 'search_graph_1') {
return SEOstats\Services\SemRush::getDomainGraph(1);
} elseif ($type == 'search_graph_2') {
return SEOstats\Services\SemRush::getDomainGraph(2);
}
}
} catch (\Exception $e) {
echo 'Caught SEOstatsException: ' . $e->getMessage();
}
}