//-------------------------------------------------------------------------------------------------------------------------------|
//-----if you start getting a bunch of 0's or NA's, remember that there are pretty low limits to Alexa and Google calls...-------|
//-------------------------------------------------------------------------------------------------------------------------------|
$alexa = SEOstats\Alexa::getGlobalRank();
$pagerank = SEOstats\Google::getPageRank();
$g_links = SEOstats\Google::getBacklinksTotal("www." . $row[0]);
$g_links_no_www = SEOstats\Google::getBacklinksTotal($row[0]);
$g_plus_count = SEOstats\Social::getGooglePlusShares();
$twitter_shares = SEOstats\Social::getTwitterShares();
$linkedin_shares = SEOstats\Social::getLinkedInShares();
echo "<div class='row'><div class='col-md-12'><h3>" . $row[0] . "</h3>";
echo "<div class='row'><div class='col-md-7'>Alexa rank = " . SEOstats\Alexa::getGlobalRank() . "<br />";
echo "Daily traffic graph: " . SEOstats\Alexa::getTrafficGraph(1) . "<br /></div>";
echo "<div class='col-md-3'>PageRank = " . SEOstats\Google::getPageRank() . "<br />";
echo "Google links = " . SEOstats\Google::getBacklinksTotal("www." . $row[0]) . "<br />";
echo "Google links (no www) = " . SEOstats\Google::getBacklinksTotal($row[0]) . "<br />";
echo "Twitter shares: " . SEOstats\Social::getTwitterShares() . "<br />";
echo "Plus ones: " . SEOstats\Social::getGooglePlusShares() . "<br />";
echo "Twitter shares: " . SEOstats\Social::getTwitterShares() . "<br />";
echo "Linkedin shares: " . SEOstats\Social::getLinkedInShares() . "<br /><br /></div>";
echo "<div class='col-md-3'>SEMRush Search Engine Traffic Graph: " . SEOstats\SemRush::getDomainGraph(1) . "<br /></div></div></div></div>";
//----------------------------------------------------------------------------------------|
//--------------------uncomment this stuff when you want to insert into MySQL!!-----------|
//----------------------------------------------------------------------------------------|
/* Prepared statement, stage 1: prepare */
//if (!($stmt = $mysqli->prepare("INSERT INTO seo (url, alexa, pagerank, link_num, linked_pages, g_links, g_links_no_www, g_plus_ones, linkedin_shares) VALUES (?,?,?,?,?,?,?,?)"))) {
// echo "Prepare failed: (" . $mysqli->errno . ") " . $mysqli->error;
//}
//if (!$stmt->bind_param("siiiiiiii", $url, $alexa, $pagerank, $link_num, $linked_pages, $g_links, $g_links_no_www, $g_plus_ones, $linkedin_shares)) {
// echo "Binding parameters failed: (" . $stmt->errno . ") " . $stmt->error;
//}
/**
*
*/
private function GetSEOState($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';
}