public function process()
{
self::$config = $this->config();
if (!self::$config->wkHtmlToPdfPath) {
throw new Exception("You must provide a path for WkHtmlToPdf in your sites configuration.");
}
if (!self::$config->emailAddress) {
throw new Exception("You must provide an email address to send from in your sites configuration.");
}
increase_memory_limit_to('1024M');
set_time_limit(0);
$sites = LinkCheckSite::get();
$outputDir = BASE_PATH . DIRECTORY_SEPARATOR . "silverstripe-linkcheck/runs/";
$filesCreated = array();
// build the crawler
chdir(__DIR__ . "/../thirdparty");
exec("javac " . self::$crawler . " " . self::$linkStats . " && " . "javac " . self::$linkProject);
if ($sites) {
foreach ($sites as $site) {
echo "Checking " . $site->SiteURL . "\r\n";
$url = $site->SiteURL;
// if the output directory doesn't exist for the run, create it
if (!file_exists($outputDir . str_replace("http://", "", $url))) {
mkdir($outputDir . str_replace("http://", "", $url));
}
$filename = date("Y-m-d") . '-' . rand(0, 1000) . ".html";
$filepath = $outputDir . str_replace("http://", "", $url) . '/';
// execute the crawler
exec("java Project {$url} " . $filepath . $filename . " 10 1000");
$filesCreated[$site->ID]['FilePath'] = $filepath;
$filesCreated[$site->ID]['FileName'] = $filename;
$filesCreated[$site->ID]['SiteName'] = $site->SiteName;
$filesCreated[$site->ID]['ID'] = $site->ID;
$filesCreated[$site->ID]['URL'] = $url;
$emailRecipients = $site->EmailRecipients();
if ($emailRecipients) {
foreach ($emailRecipients as $recipient) {
$filesCreated[$site->ID]['Email'][] = $recipient->Email;
}
}
}
foreach ($filesCreated as $file) {
Folder::find_or_make("LinkCheck" . DIRECTORY_SEPARATOR . $file['SiteName'] . DIRECTORY_SEPARATOR);
$pdfPath = "assets" . DIRECTORY_SEPARATOR . "LinkCheck" . DIRECTORY_SEPARATOR . $file['SiteName'] . DIRECTORY_SEPARATOR;
$pdfFullPath = BASE_PATH . DIRECTORY_SEPARATOR . $pdfPath;
$pdfName = str_replace("html", "pdf", $file['FileName']);
$generator = new WkHtml\Generator(new \Knp\Snappy\Pdf(self::$config->wkHtmlToPdfPath), new WkHtml\Input\String(file_get_contents($file['FilePath'] . $file['FileName'])), new WkHtml\Output\File($pdfFullPath . $pdfName, 'application/pdf'));
$generator->process();
$site = LinkCheckSite::get()->byID($file['ID']);
$pdfUpload = new File();
$pdfUpload->Title = $file['SiteName'] . '-' . $pdfName;
$pdfUpload->Filename = $pdfPath . $pdfName;
$pdfUpload->write();
$linkCheckRun = new LinkCheckRun();
$linkCheckRun->LinkCheckFileID = $pdfUpload->ID;
$linkCheckRun->LinkCheckSiteID = $site->ID;
$linkCheckRun->write();
$site->LinkCheckRuns()->add($linkCheckRun);
foreach ($file['Email'] as $emailAddress) {
$email = new Email();
$email->to = $emailAddress;
$email->from = $this->config()->emailAddress;
$email->subject = $file['SiteName'] . " link check run";
$email->body = "Site Link Check Run for {$file['URL']} on " . date("Y/m/d");
$email->attachFile($pdfPath . $pdfName, "linkcheck.pdf");
$email->send();
}
unlink($file['FilePath'] . $file['FileName']);
}
}
}
/**
* Run the LinkCheckTask.
* @todo Split functionality to separate methods
*/
public function process()
{
if (class_exists('SapphireTest', false) && !SapphireTest::is_running_test()) {
echo "\r\n";
}
if (!ClassInfo::hasTable('LinkCheckRun')) {
if (!Director::is_ajax() && class_exists('SapphireTest', false) && !SapphireTest::is_running_test()) {
echo "Database has not been built. Please run dev/build first!\r\n";
}
return false;
}
// If there is already a LinkCheckRun that exists and is not complete,
// don't allow a new run as it could run the server to the ground!
// @todo we probably want some system that allows cancelling a check halfway through
if (DataObject::get_one('LinkCheckRun', "\"IsComplete\" = 0")) {
if (!Director::is_ajax() && class_exists('SapphireTest', false) && !SapphireTest::is_running_test()) {
echo "There is already a link check running at the moment. Please wait for it to complete before starting a new one.\r\n";
}
return false;
}
set_time_limit(0);
ini_set('max_execution_time', 0);
$goodLinks = 0;
// 200-299 HTTP status codes
$checkLinks = 0;
// 300-399 HTTP status codes
$brokenLinks = 0;
// 400-599 HTTP status codes
$pages = DataObject::get('SiteTree');
if (!$pages) {
return false;
}
$run = new LinkCheckRun();
// We have started a new run, create the object and write it
$run->write();
$pagesChecked = 0;
foreach ($pages as $page) {
// Skip this page if it shouldn't be checked
if (isset(self::$exempt_classes[get_class($page)])) {
continue;
}
$processor = new LinkCheckProcessor($page->AbsoluteLink());
if (Director::is_ajax()) {
$processor->showMessages = false;
}
$results = $processor->run();
// Memory cleanup - we don't need the processor anymore
unset($processor);
if ($results) {
foreach ($results as $result) {
if ($result['Code'] >= 200 && $result['Code'] <= 299) {
$goodLinks++;
} elseif ($result['Code'] >= 300 && $result['Code'] <= 399) {
$checkLinks++;
} elseif ($result['Code'] >= 400 && $result['Code'] <= 599) {
$brokenLinks++;
}
// If the result is "Bad" (broken), create a BrokenLink record
if ($result['Code'] >= 400 && $result['Code'] <= 599) {
$brokenLink = new BrokenLink();
$brokenLink->Link = substr($result['Link'], 0, 255);
$brokenLink->Code = $result['Code'];
$brokenLink->Status = substr($result['Status'], 0, 30);
$brokenLink->LinkCheckRunID = $run->ID;
$brokenLink->PageID = $page->ID;
$brokenLink->write();
// Memory cleanup
$brokenLink->destroy();
}
}
}
// Memory cleanup
$page->destroy();
$pagesChecked++;
}
// Memory cleanup
unset($pages);
// Mark as done - this is to indicate that the task has completed (for reporting in CMS)
$run->FinishDate = date('Y-m-d H:i:s');
$run->IsComplete = 1;
$run->PagesChecked = $pagesChecked;
$run->write();
// Find the URL to the LinkCheckAdmin section in the CMS
$linkcheckAdminLink = Director::absoluteBaseURL() . singleton('LinkCheckAdmin')->Link();
// Count the number of BrokenLink records created for this run
$runBrokenLinks = $run->BrokenLinks()->Count() ? $run->BrokenLinks()->Count() : 0;
if (Director::is_ajax()) {
return array('Date' => $run->obj('Created')->Nice(), 'LinkCheckRunID' => $run->ID);
} elseif (Director::is_cli()) {
if (class_exists('SapphireTest', false) && SapphireTest::is_running_test()) {
return;
}
echo "SilverStripe Link Checker results\n";
echo "---------------------------------\n\n";
echo "{$pagesChecked} pages were checked for broken links.\n";
echo "{$goodLinks} links were OK.\n";
echo "{$checkLinks} links were redirected.\n";
echo "{$brokenLinks} links were broken, and {$runBrokenLinks} BrokenLink records were generated for them.\n\n";
echo "LinkCheckRun ID #{$run->ID} was created with {$runBrokenLinks} BrokenLink related records.\n";
echo "Please visit {$linkcheckAdminLink} to see which broken links were found.\n\n";
} else {
if (class_exists('SapphireTest', false) && SapphireTest::is_running_test()) {
return;
}
echo "<h1>SilverStripe Link Checker results</h1>";
echo '<ul>';
echo "<li>{$pagesChecked} pages were checked for broken links.</li>";
echo "<li>{$goodLinks} links were OK.</li>";
echo "<li>{$checkLinks} links were redirected.</li>";
echo "<li>{$brokenLinks} links were broken, and {$runBrokenLinks} BrokenLink records were generated for them.</li>";
echo '</ul>';
echo "<p>LinkCheckRun ID #{$run->ID} was created with {$runBrokenLinks} BrokenLink related records.</p>";
echo "<p>Please visit <a href=\"{$linkcheckAdminLink}\">{$linkcheckAdminLink}</a> to see which broken links were found.</p>";
}
}
