/**
* Rewrite file urls that weren't properly localized.
*
* @param object BlockSiteComponent $siteComponent
* @return void
*/
protected function rewriteNonlocalFileUrls(BlockSiteComponent $siteComponent)
{
static $baseUrls;
if (empty($baseUrls)) {
$baseUrls = array('http://segue.middlebury.edu', 'https://segue.middlebury.edu', 'http://seguecommunity.middlebury.edu', 'https://seguecommunity.middlebury.edu');
foreach (SlotAbstract::getLocationCategories() as $locationCategory) {
$baseUrls[] = rtrim(SiteDispatcher::getBaseUrlForLocationCategory($locationCategory), '/');
}
$baseUrls = array_unique($baseUrls);
}
$content = $siteComponent->getAsset()->getContent();
foreach ($baseUrls as $baseUrl) {
if (preg_match_all('#[\'"]' . $baseUrl . '/repository/(viewfile|viewfile_flash|viewthumbnail|viewthumbnail_flash)/polyphony-repository___repository_id/edu.middlebury.segue.sites_repository/polyphony-repository___asset_id/([0-9]+)/polyphony-repository___record_id/([0-9]+)(?:polyphony-repository___file_name/(.+))?[\'"]#', $content->asString(), $matches, PREG_SET_ORDER)) {
foreach ($matches as $m) {
$urlParts = array('module' => 'repository', 'action' => $m[1], 'polyphony-repository___repository_id' => 'edu.middlebury.segue.sites_repository', 'polyphony-repository___asset_id' => $m[2], 'polyphony-repository___record_id' => $m[3]);
if (!empty($m[4])) {
$urlParts['polyphony-repository___file_name'] = $m[4];
} else {
if ($m[1] == 'viewfile') {
try {
$file = MediaFile::withIdStrings('edu.middlebury.segue.sites_repository', $m[2], $m[3]);
$urlParts['polyphony-repository___file_name'] = $file->getFilename();
} catch (UnknownIdException $e) {
}
}
}
$url = http_build_query($urlParts, '', '&');
$content->_setValue(str_replace($m[0], '"' . '[[localurl:' . $url . ']]' . '"', $content->value()));
}
}
}
}
/**
* Answer the url the JS functions should use to get the feed.
* Due to browser restrictions on the location of documents loaded via the
* XMLHTTPRequest, this may be different from the url returned by _getFeedUrl();
*
* @return string
* @access protected
* @since 6/17/08
*/
protected function _getFeedAccessUrl()
{
// For local-server urls, return the feed url
if ($this->isLocal($this->_getFeedUrl()) && $this->isContentTrusted($this->_getFeedUrl())) {
return $this->_getFeedUrl();
}
// URLs to other location categories in the same segue instance.
// These can be rebuilt to go through our local host.
$pattern = '#((';
$pattern .= '(' . str_replace('.', '\\.', MYURL) . ')';
foreach (SlotAbstract::getLocationCategories() as $category) {
$pattern .= '|(' . str_replace('.', '\\.', SiteDispatcher::getBaseUrlForLocationCategory($category)) . ')';
}
$pattern .= ')[^\'"\\s\\]<>}]*)#i';
if (preg_match($pattern, $this->_getFeedUrl(), $matches)) {
$harmoni = Harmoni::instance();
// Create a local-server version of the URL.
$harmoni->request->startNamespace(null);
$url = $harmoni->request->quickURL($harmoni->request->getModuleFromUrlWithBase($matches[2], $matches[0]), $harmoni->request->getActionFromUrlWithBase($matches[2], $matches[0]), $harmoni->request->getParameterArrayFromUrlWithBase($matches[2], $matches[0]));
$harmoni->request->endNamespace();
if ($this->isContentTrusted($url)) {
return $url;
}
}
// For remote urls, pass through a local data-fetching gateway.
return $this->getPluginActionUrl('remote_feed', array('url' => $this->_getFeedUrl()));
}
/**
* Execute
*
* @return void
* @access public
* @since 3/26/08
*/
public function execute()
{
if (!$this->isAuthorizedToExecute()) {
throw new PermissionDeniedException('This command can only be run by admins or from the command-line.');
}
header("Content-Type: text/plain");
if (RequestContext::value('help') || RequestContext::value('h') || RequestContext::value('?')) {
throw new HelpRequestedException($this->usage);
}
$outDir = RequestContext::value('d');
if (empty($outDir)) {
throw new InvalidArgumentException("An output directory must be specified.\n\n" . $this->usage);
}
if (!is_dir($outDir) || !is_writable($outDir)) {
throw new InvalidArgumentException("The output directory doesn't exist or is not writeable.\n\n" . $this->usage);
}
foreach (SlotAbstract::getLocationCategories() as $category) {
$baseUrl = SiteDispatcher::getBaseUrlForLocationCategory($category);
if (!preg_match('/^https?:\\/\\/.+/', $baseUrl)) {
throw new ConfigurationErrorException('Please set a base URL for the \'' . $category . '\' category with SiteDispatcher::setBaseUrlForLocationCategory($category, $url); in config/slots.conf.php');
}
}
while (ob_get_level()) {
ob_end_flush();
}
flush();
/*********************************************************
* Check for a running export
*********************************************************/
$dbc = Services::getService('DatabaseManager');
$query = new SelectQuery();
$query->addColumn('slot');
$query->addColumn('pid');
$query->addTable('site_export_queue');
$query->addWhereNotEqual('pid', 0);
$result = $dbc->query($query);
// If we are exporting, check the status of the export process
if ($result->hasMoreRows()) {
// Don't start a new export if one is running.
if ($this->isRunning($result->field('pid'))) {
print "An export is already running\n";
exit;
} else {
$query = new UpdateQuery();
$query->setTable('site_export_queue');
$query->addValue('status', 'DIED');
$query->addRawValue('pid', 'NULL');
$query->addValue('info', 'Process ' . $result->field('pid') . ' has died.');
$query->addWhereEqual('slot', $result->field('slot'));
$query->addWhereEqual('pid', $result->field('pid'));
$dbc->query($query);
}
}
/*********************************************************
* If there aren't any other exports happening, run our export
*********************************************************/
// Find the next slot to update
$query = new SelectQuery();
$query->addColumn('slot');
$query->addTable('site_export_queue', NO_JOIN, '', 'q');
$query->addTable('segue_slot', INNER_JOIN, 'q.slot = s.shortname', 's');
$query->addWhereNull('pid');
$query->addWhereNull('status');
$query->addWhereNull('alias_target');
$query->addWhereNotEqual('site_id', '');
$query->addOrderBy('priority', DESCENDING);
$query->addOrderBy('slot', ASCENDING);
$result = $dbc->query($query);
// Exit if there is nothing to do.
if (!$result->hasMoreRows()) {
print "The queue is empty\n";
exit;
}
$slot = $result->field('slot');
$slotMgr = SlotManager::instance();
$slotObj = $slotMgr->getSlotByShortname($slot);
$baseUrl = SiteDispatcher::getBaseUrlForLocationCategory($slotObj->getLocationCategory());
// Mark that we are running
$query = new UpdateQuery();
$query->setTable('site_export_queue');
$query->addValue('pid', strval(getmypid()));
$query->addWhereEqual('slot', $slot);
$dbc->query($query);
// Run the export
$start = microtime(true);
try {
$exportDirname = $slot . "-html";
$exportDir = $outDir . "/" . $exportDirname;
$archivePath = $outDir . '/' . $exportDirname . ".zip";
if (file_exists($exportDir)) {
$this->deleteRecursive($exportDir);
}
mkdir($exportDir);
if (file_exists($archivePath)) {
unlink($archivePath);
}
// Set the user to be an admin.
$idMgr = Services::getService("Id");
$authType = new Type("Authentication", "edu.middlebury.harmoni", "Harmoni DB");
$_SESSION['__AuthenticatedAgents']['Authentication::edu.middlebury.harmoni::Harmoni DB'] = $idMgr->getId('17008');
$authZ = Services::getService("AuthZ");
$isAuthorizedCache = $authZ->getIsAuthorizedCache();
$isAuthorizedCache->dirtyUser();
// Close the session. If we don't, a lock on the session file will
// cause the request initiated via wget to hang.
session_write_close();
// Do the export
$urlParts = parse_url($baseUrl);
$urlPrefix = rtrim($urlParts['path'], '/');
$include = array($urlPrefix . '/gui2', $urlPrefix . '/images', $urlPrefix . '/javascript', $urlPrefix . '/polyphony', $urlPrefix . '/repository', $urlPrefix . '/plugin_manager', $urlPrefix . '/rss', $urlPrefix . '/dataport/html/site/' . $slot);
if (defined('WGET_PATH')) {
$wget = WGET_PATH;
} else {
$wget = 'wget';
}
if (defined('WGET_OPTIONS')) {
$wgetOptions = WGET_OPTIONS;
} else {
$wgetOptions = '';
}
$command = $wget . " " . $wgetOptions . " -r --page-requisites --html-extension --convert-links --no-directories -e robots=off " . "--directory-prefix=" . escapeshellarg($exportDir . '/content') . " " . "--include=" . escapeshellarg(implode(',', $include)) . " " . "--header=" . escapeshellarg("Cookie: " . session_name() . "=" . session_id()) . " " . escapeshellarg($baseUrl . '/dataport/html/site/' . $slot);
print "Cookie: " . session_name() . "=" . session_id() . "\n";
// throw new Exception($command);
exec($command, $output, $exitCode);
if ($exitCode) {
throw new Exception('Wget Failed. ' . implode("\n", $output));
}
// Copy the main HTML file to index.html
copy($exportDir . '/content/' . $slot . '.html', $exportDir . '/content/index.html');
// Copy the index.html file up a level to make it easy to find
file_put_contents($exportDir . '/index.html', preg_replace('/(src|href)=([\'"])([^\'"\\/]+)([\'"])/', '$1=$2content/$3$4', file_get_contents($exportDir . '/content/index.html')));
// Zip up the result
$archive = new ZipArchive();
if ($archive->open($archivePath, ZIPARCHIVE::CREATE) !== TRUE) {
throw new Exception("Could not create zip archive.");
}
$this->addDirectoryToZip($archive, $exportDir, $exportDirname);
$archive->close();
// Remove the directory
$this->deleteRecursive($exportDir);
// Mark our success
$query = new UpdateQuery();
$query->setTable('site_export_queue');
$query->addRawValue('pid', 'NULL');
$query->addValue('status', 'SUCCESS');
$query->addValue('running_time', strval(round(microtime(true) - $start, 2)));
$query->addWhereEqual('slot', $slot);
$dbc->query($query);
} catch (Exception $e) {
$this->deleteRecursive($exportDir);
if (file_exists($archivePath)) {
unlink($archivePath);
}
// Mark our failure
$query = new UpdateQuery();
$query->setTable('site_export_queue');
$query->addRawValue('pid', 'NULL');
$query->addValue('status', 'EXCEPTION');
$query->addValue('info', $e->getMessage());
$query->addValue('running_time', strval(round(microtime(true) - $start, 2)));
$query->addWhereEqual('slot', $slot);
$dbc->query($query);
throw $e;
}
exit;
}
