/**
* Export records from AMS for PBCore and PREMIS and save all the exported records in zip file using BagIt.
*
* @return void
*/
function export_pbcore()
{
@ini_set("max_execution_time", 999999999999.0);
# unlimited
@ini_set("memory_limit", "2000M");
# 2GB
error_reporting(E_ALL);
ini_set('display_errors', 1);
$export_job = $this->export_job->get_export_jobs('pbcore');
if (count($export_job) > 0) {
$this->export_job->update_job($export_job->id, array('status' => '1'));
myLog('Started export for ID =>' . $export_job->id);
$bag_name = 'ams_export_' . time();
$bagit_info = array('Source-Organization' => 'WGBH on behalf of the American Archive of Public Broadcasting', 'Organization-Address' => 'Media Library & Archives, One Guest Street, Boston, MA 02135', 'Contact-Name' => 'Casey E. Davis, Project Manager', 'Contact-Phone' => '+1 617-300-5921', 'Contact-Email' => 'info@americanarchive.org ', 'External-Description' => 'The American Archive of Public Broadcasting, a collaboration between WGBH Boston and the Library of Congress, includes millions of records of public television and radio assets contributed by over 100 stations and organizations across the United States.');
$bagit_lib = new BagIt("{$this->bagit_path}{$bag_name}", TRUE, TRUE, FALSE, $bagit_info);
$bagit_lib->setHashEncoding('md5');
$total_size = 0;
$total_files = 0;
if ($export_job->query_loop == 0) {
$export_ids = explode(',', $export_job->export_query);
foreach ($export_ids as $asset_id) {
$this->do_export($asset_id, $total_size, $total_files, $bagit_lib);
}
} else {
for ($i = 0; $i < $export_job->query_loop; $i++) {
$query = $export_job->export_query;
$query .= ' LIMIT ' . $i * 100000 . ', 100000';
$records = $this->export_job->get_csv_records($query);
foreach ($records as $value) {
$this->do_export($value->id, $total_size, $total_files, $bagit_lib);
}
}
}
$bagit_lib->setBagInfoData('Payload-Oxum', $total_size . '.' . $total_files);
$bagit_lib->setBagInfoData('Bagging-Date', date('Y-m-d'));
$bagit_lib->setBagInfoData('Bag-Size', sizeFormat($total_size));
$bagit_lib->update();
$bagit_lib->package("{$this->bagit_path}{$bag_name}", 'zip');
exec("rm -rf {$this->temp_path}");
exec("rm -rf {$this->bagit_path}{$bag_name}");
$this->export_job->update_job($export_job->id, array('status' => '1', 'file_path' => "{$this->bagit_path}{$bag_name}.zip"));
$user = $this->users->get_user_by_id($export_job->user_id)->row();
$data['user_profile'] = $this->user_profile->get_profile($export_job->user_id)->row();
$data['user'] = $user;
$data['export_id'] = $export_job->id;
myLog('Sending Email to ' . $user->email);
send_email($user->email, $this->config->item('from_email'), 'AMS XML Export', $this->load->view('email/export_pbcore', $data, TRUE));
myLog('email sent successfully ' . $user->email);
} else {
myLog('No record availabe for export');
}
exit_function();
}
public function testPackageTGz()
{
$tmp = tmpdir();
try {
mkdir($tmp);
mkdir($tmp . '/data');
file_put_contents($tmp . '/data/missing.txt', 'This space intentionally left blank.\\n');
file_put_contents($tmp . "/fetch.txt", "http://www.google.com - data/google/index.html\n" . "http://www.yahoo.com - data/yahoo/index.html\n");
$bag = new BagIt($tmp);
$bag->update();
$bag->package($tmp . '/../bagtmp1.tgz', 'tgz');
$this->assertFileExists($tmp . '/../bagtmp1.tgz');
rename("{$tmp}/../bagtmp1.tgz", "/tmp/bagtmp1.tgz");
$bag->package($tmp . '/../bagtmp2', 'tgz');
$this->assertFileExists($tmp . '/../bagtmp2.tgz');
rename("{$tmp}/../bagtmp2.tgz", "/tmp/bagtmp2.tgz");
// TODO: Test the contents of the package.
} catch (Exception $e) {
rrmdir($tmp);
throw $e;
}
rrmdir($tmp);
}
/**
* Read the Bag, unpack it, drop files into the Dropbox files directory.
*
* @param string $filename The name of the uploaded bag.
*
* @return boolean $success True if the read succeeds.
*/
function bagithelpers_doReadBagIt($filename)
{
$success = false;
$bag = new BagIt(BAGIT_TMP_DIRECTORY . '/' . $filename);
$bag->fetch->download();
$bag->update();
$bag->validate();
if (count($bag->getBagErrors()) == 0) {
// Copy each of the files.
foreach ($bag->getBagContents() as $file) {
copy($file, BASE_DIR . '/plugins/Dropbox/files/' . basename($file));
}
$success = true;
}
return $success;
}
/**
* Create a package containing the serialized deposit objects
* @return string The full path of the created zip archive
*/
function generatePackage()
{
// get DAOs, plugins and settings
$journalDao =& DAORegistry::getDAO('JournalDAO');
$issueDao =& DAORegistry::getDAO('IssueDAO');
$sectionDao =& DAORegistry::getDAO('SectionDAO');
$publishedArticleDao =& DAORegistry::getDAO('PublishedArticleDAO');
PluginRegistry::loadCategory('importexport');
$exportPlugin =& PluginRegistry::getPlugin('importexport', 'NativeImportExportPlugin');
$plnPlugin =& PluginRegistry::getPlugin('generic', PLN_PLUGIN_NAME);
$fileManager = new JournalFileManager($journalDao->getById($this->_deposit->getJournalId()));
$journal =& $journalDao->getById($this->_deposit->getJournalId());
$depositObjects = $this->_deposit->getDepositObjects();
// set up folder and file locations
$bagDir = $this->getDepositDir() . DIRECTORY_SEPARATOR . 'bag';
$packageFile = $this->getPackageFilePath();
$exportFile = $bagDir . DIRECTORY_SEPARATOR . $this->_deposit->getObjectType() . '.xml';
$termsFile = $bagDir . DIRECTORY_SEPARATOR . 'terms.xml';
$bag = new BagIt($bagDir);
switch ($this->_deposit->getObjectType()) {
case PLN_PLUGIN_DEPOSIT_OBJECT_ARTICLE:
$articles = array();
// we need to add all of the relevant articles to an array to export as a batch
while ($depositObject =& $depositObjects->next()) {
$article =& $publishedArticleDao->getPublishedArticleByArticleId($this->_deposit->getObjectId(), $journal->getId());
$issue =& $issueDao->getIssueById($article->getIssueId(), $journal->getId());
$section =& $sectionDao->getSection($article->getSectionId());
// add the article to the array we'll pass for export
$articles[] = array('publishedArticle' => $article, 'section' => $section, 'issue' => $issue, 'journal' => $journal);
unset($depositObject);
}
// export all of the articles together
if ($exportPlugin->exportArticles($articles, $exportFile) !== true) {
return false;
}
break;
case PLN_PLUGIN_DEPOSIT_OBJECT_ISSUE:
// we only ever do one issue at a time, so get that issue
$depositObject =& $depositObjects->next();
$issue =& $issueDao->getIssueByBestIssueId($depositObject->getObjectId(), $journal->getId());
// export the issue
if ($exportPlugin->exportIssue($journal, $issue, $exportFile) !== true) {
return false;
}
break;
default:
}
// add the current terms to the bag
$termsXml = new DOMDocument('1.0', 'utf-8');
$entry = $termsXml->createElementNS('http://www.w3.org/2005/Atom', 'entry');
$entry->setAttributeNS('http://www.w3.org/2000/xmlns/', 'xmlns:dcterms', 'http://purl.org/dc/terms/');
$entry->setAttributeNS('http://www.w3.org/2000/xmlns/', 'xmlns:pkp', PLN_PLUGIN_NAME);
$terms = unserialize($plnPlugin->getSetting($this->_deposit->getJournalId(), 'terms_of_use'));
$pkpTermsOfUse = $termsXml->createElementNS(PLN_PLUGIN_NAME, 'pkp:terms_of_use');
foreach ($terms as $termName => $termData) {
$element = $termsXml->createElementNS(PLN_PLUGIN_NAME, $termName, $termData['term']);
$element->setAttribute('updated', $termData['updated']);
$pkpTermsOfUse->appendChild($element);
}
$entry->appendChild($pkpTermsOfUse);
$termsXml->appendChild($entry);
$termsXml->save($termsFile);
// add the exported content to the bag
$bag->addFile($exportFile, $this->_deposit->getObjectType() . $this->_deposit->getUUID() . '.xml');
// add the exported content to the bag
$bag->addFile($termsFile, 'terms' . $this->_deposit->getUUID() . '.xml');
$bag->update();
// delete files
$fileManager->deleteFile($exportFile);
$fileManager->deleteFile($termsFile);
// create the bag
$bag->package($packageFile, 'zip');
// remove the temporary bag directory
$fileManager->rmtree($bagDir);
return $packageFile;
}
/**
* This is the use case for consuming a bag from someone else. The user
* does these actions:
*
* <ol>
* <li>Open the bag;</li>
* <li>Validate the downloaded contents;</li>
* <li>Fetch on-line items in the bag;</li>
* <li>Validate the bag's contents; and</li>
* <li>Copy items from the bag onto the local disk.</li>
* </ol>
*/
public function testBagConsumer()
{
$srcbag = __DIR__ . '/TestBag.tgz';
// 1. Open the bag;
$bag = new BagIt($srcbag);
$this->queueRm($bag->bagDirectory);
$this->assertTrue($bag->isValid());
$this->assertTrue($bag->isExtended());
$bagInfo = $bag->getBagInfo();
$this->assertEquals('0.96', $bagInfo['version']);
$this->assertEquals('UTF-8', $bagInfo['encoding']);
$this->assertEquals('sha1', $bagInfo['hash']);
$this->assertEquals('sha1', $bag->getHashEncoding());
$this->assertEquals(7, count($bag->getBagContents()));
$this->assertEquals(0, count($bag->getBagErrors()));
// 2. Validate the downloaded contents;
$bag->validate();
$this->assertEquals(0, count($bag->getBagErrors()));
// 3. Fetch on-line items in the bag;
$bag->fetch->download();
$bag->update();
$bag->validate();
$this->assertEquals(8, count($bag->getBagContents()));
// 4. Validate the bag's contents; and
$this->assertEquals(0, count($bag->getBagErrors()));
// 5. Copy items from the bag onto the local disk.
$tmpdir = tmpdir();
mkdir($tmpdir);
$this->queueRm($tmpdir);
foreach ($bag->getBagContents() as $bagFile) {
$basename = basename($bagFile);
copy($bagFile, "{$tmpdir}/{$basename}");
}
$this->assertEquals(count($bag->getBagContents()) + 2, count(scandir($tmpdir)));
$this->assertEquals(count($bag->manifest->getData()) + 2, count(scandir($tmpdir)));
}
public function testNewBagEmpty()
{
$bagdir = "{$this->tmpdir}/_bag";
$bag = new BagIt($bagdir);
$this->assertFileNotExists("{$bagdir}/fetch.txt");
$bag->update();
$this->assertFileNotExists("{$bagdir}/fetch.txt");
}
