/**
* Reindex PDF media by content for in-PDF search
*/
public static function reindex_pdfs($po_opts = null)
{
require_once __CA_LIB_DIR__ . "/core/Db.php";
require_once __CA_MODELS_DIR__ . "/ca_object_representations.php";
if (!caPDFMinerInstalled()) {
CLIUtils::addError(_t("Can't reindex PDFs: PDFMiner is not installed."));
return false;
}
$o_db = new Db();
$t_rep = new ca_object_representations();
$t_rep->setMode(ACCESS_WRITE);
$va_versions = array("original");
$va_kinds = ($vs_kinds = $po_opts->getOption("kinds")) ? explode(",", $vs_kinds) : array();
if (!is_array($va_kinds) || !sizeof($va_kinds)) {
$va_kinds = array('all');
}
$va_kinds = array_map('strtolower', $va_kinds);
if (in_array('all', $va_kinds) || in_array('ca_object_representations', $va_kinds)) {
if (!($vn_start = (int) $po_opts->getOption('start_id'))) {
$vn_start = null;
}
if (!($vn_end = (int) $po_opts->getOption('end_id'))) {
$vn_end = null;
}
if ($vn_id = (int) $po_opts->getOption('id')) {
$vn_start = $vn_id;
$vn_end = $vn_id;
}
$va_ids = array();
if ($vs_ids = (string) $po_opts->getOption('ids')) {
if (sizeof($va_tmp = explode(",", $vs_ids))) {
foreach ($va_tmp as $vn_id) {
if ((int) $vn_id > 0) {
$va_ids[] = (int) $vn_id;
}
}
}
}
$vs_sql_where = null;
$va_params = array();
if (sizeof($va_ids)) {
$vs_sql_where = "WHERE representation_id IN (?)";
$va_params[] = $va_ids;
} else {
if ($vn_start > 0 && $vn_end > 0 && $vn_start <= $vn_end || $vn_start > 0 && $vn_end == null) {
$vs_sql_where = "WHERE representation_id >= ?";
$va_params[] = $vn_start;
if ($vn_end) {
$vs_sql_where .= " AND representation_id <= ?";
$va_params[] = $vn_end;
}
}
}
if ($vs_sql_where) {
$vs_sql_where .= " AND mimetype = 'application/pdf'";
} else {
$vs_sql_where = " WHERE mimetype = 'application/pdf'";
}
$qr_reps = $o_db->query("\n\t\t\t\t\tSELECT * \n\t\t\t\t\tFROM ca_object_representations \n\t\t\t\t\t{$vs_sql_where}\n\t\t\t\t\tORDER BY representation_id\n\t\t\t\t", $va_params);
print CLIProgressBar::start($qr_reps->numRows(), _t('Reindexing PDF representations'));
$vn_rep_table_num = $t_rep->tableNum();
while ($qr_reps->nextRow()) {
$va_media_info = $qr_reps->getMediaInfo('media');
$vs_original_filename = $va_media_info['ORIGINAL_FILENAME'];
print CLIProgressBar::next(1, _t("Reindexing PDF %1", $vs_original_filename ? $vs_original_filename . " (" . $qr_reps->get('representation_id') . ")" : $qr_reps->get('representation_id')));
$t_rep->load($qr_reps->get('representation_id'));
$vn_rep_id = $t_rep->getPrimaryKey();
$m = new Media();
if ($m->read($vs_path = $t_rep->getMediaPath('media', 'original')) && is_array($va_locs = $m->getExtractedTextLocations())) {
MediaContentLocationIndexer::clear($vn_rep_table_num, $vn_rep_id);
foreach ($va_locs as $vs_content => $va_loc_list) {
foreach ($va_loc_list as $va_loc) {
MediaContentLocationIndexer::index($vn_rep_table_num, $vn_rep_id, $vs_content, $va_loc['p'], $va_loc['x1'], $va_loc['y1'], $va_loc['x2'], $va_loc['y2']);
}
}
MediaContentLocationIndexer::write();
} else {
//CLIUtils::addError(_t("[Warning] No content to reindex for PDF representation: %1", $vs_path));
}
}
print CLIProgressBar::finish();
}
if (in_array('all', $va_kinds) || in_array('ca_attributes', $va_kinds)) {
// get all Media elements
$va_elements = ca_metadata_elements::getElementsAsList(false, null, null, true, false, true, array(16));
// 16=media
$qr_c = $o_db->query("\n\t\t\t\t\tSELECT count(*) c \n\t\t\t\t\tFROM ca_attribute_values\n\t\t\t\t\tWHERE\n\t\t\t\t\t\telement_id in (?)\n\t\t\t\t", caExtractValuesFromArrayList($va_elements, 'element_id', array('preserveKeys' => false)));
if ($qr_c->nextRow()) {
$vn_count = $qr_c->get('c');
} else {
$vn_count = 0;
}
$t_attr_val = new ca_attribute_values();
$vn_attr_table_num = $t_attr_val->tableNum();
print CLIProgressBar::start($vn_count, _t('Reindexing metadata attribute media'));
foreach ($va_elements as $vs_element_code => $va_element_info) {
$qr_vals = $o_db->query("SELECT value_id FROM ca_attribute_values WHERE element_id = ?", (int) $va_element_info['element_id']);
$va_vals = $qr_vals->getAllFieldValues('value_id');
foreach ($va_vals as $vn_value_id) {
$t_attr_val = new ca_attribute_values($vn_value_id);
if ($t_attr_val->getPrimaryKey()) {
$t_attr_val->setMode(ACCESS_WRITE);
$t_attr_val->useBlobAsMediaField(true);
$va_media_info = $t_attr_val->getMediaInfo('value_blob');
$vs_original_filename = $va_media_info['ORIGINAL_FILENAME'];
if (!is_array($va_media_info) || $va_media_info['MIMETYPE'] !== 'application/pdf') {
continue;
}
print CLIProgressBar::next(1, _t("Reindexing %1", $vs_original_filename ? $vs_original_filename . " ({$vn_value_id})" : $vn_value_id));
$m = new Media();
if ($m->read($vs_path = $t_attr_val->getMediaPath('value_blob', 'original')) && is_array($va_locs = $m->getExtractedTextLocations())) {
MediaContentLocationIndexer::clear($vn_attr_table_num, $vn_attr_table_num);
foreach ($va_locs as $vs_content => $va_loc_list) {
foreach ($va_loc_list as $va_loc) {
MediaContentLocationIndexer::index($vn_attr_table_num, $vn_value_id, $vs_content, $va_loc['p'], $va_loc['x1'], $va_loc['y1'], $va_loc['x2'], $va_loc['y2']);
}
}
MediaContentLocationIndexer::write();
} else {
//CLIUtils::addError(_t("[Warning] No content to reindex for PDF in metadata attribute: %1", $vs_path));
}
}
}
}
print CLIProgressBar::finish();
}
return true;
}
/**
* Update annotation. If time code of annotation has changed media preview will be regenerated.
* You can force the media preview to be regenerated whether the time code has changed or not
* by passing the 'forcePreviewGeneration' option.
*
* @param array $pa_options An array of options:
* forcePreviewGeneration = if set preview media will be regenerated whether time code has changed or not. Default is false.
* @return bool True on success, false on failure
*/
public function update($pa_options = null)
{
$this->set('type_code', $vs_type = $this->getAnnotationType());
if (!$this->opo_annotations_properties->validate()) {
$this->errors = $this->opo_annotations_properties->errors;
return false;
}
$this->set('props', $this->opo_annotations_properties->getPropertyValues());
if (!$this->getAppConfig()->get('dont_generate_annotation_previews') && $this->getPrimaryKey() && ($this->changed('props') || isset($pa_options['forcePreviewGeneration']) && $pa_options['forcePreviewGeneration'])) {
$vs_start = $this->getPropertyValue('startTimecode');
$vs_end = $this->getPropertyValue('endTimecode');
$va_data['start'] = $vs_start;
$va_data['end'] = $vs_end;
$t_rep = new ca_object_representations($this->get('representation_id'));
if (($vs_file = $t_rep->getMediaPath('media', 'original')) && file_exists($vs_file)) {
$o_media = new Media();
if ($o_media->read($vs_file)) {
if ($o_media->writeClip($vs_file = tempnam(caGetTempDirPath(), 'annotationPreview'), $vs_start, $vs_end)) {
$this->set('preview', $vs_file);
}
}
}
}
$vn_rc = parent::update($pa_options);
if (!$this->numErrors()) {
$this->opo_annotations_properties = $this->loadProperties($vs_type);
}
if ($vs_file) {
@unlink($vs_file);
}
return $vn_rc;
}
/**
*
*/
public function getClip()
{
$pn_id = $this->opo_request->getParameter('id', pInteger);
$va_data = array();
$t_annotation = new ca_representation_annotations($pn_id);
if ($t_annotation->getPrimaryKey()) {
$vs_start = $t_annotation->getPropertyValue('startTimecode');
$vs_end = $t_annotation->getPropertyValue('endTimecode');
$va_data['start'] = $vs_start;
$va_data['end'] = $vs_end;
$t_rep = new ca_object_representations($t_annotation->get('representation_id'));
$va_data['file'] = $vs_file = $t_rep->getMediaPath('media', 'original');
$o_media = new Media();
if ($o_media->read($vs_file)) {
$o_media->writeClip($vs_file = tempnam('/tmp', 'meow'), $vs_start, $vs_end);
}
header("Content-type: audio/mpeg");
header("Content-length: " . @filesize($vs_file));
readfile($vs_file);
return;
}
return $this->makeResponse($va_data, 500, "No such clip");
}
/**
* Perform client services-related periodic tasks
*/
public function hookPeriodicTask(&$pa_params)
{
$t_log = new Eventlog();
$o_db = new Db();
if (!(bool) $this->opo_config->get('enable_client_services')) {
return true;
}
// Find any orders with status PROCESSED_AWAITING_MEDIA_ACCESS and fetch media
$qr_orders = $o_db->query("\n\t\t\t\tSELECT order_id\n\t\t\t\tFROM ca_commerce_orders\n\t\t\t\tWHERE\n\t\t\t\t\torder_status = 'PROCESSED_AWAITING_MEDIA_ACCESS'\n\t\t\t");
//
// Set up HTTP client for REST calls
//
if ($this->opo_client_services_config->get('remote_media_base_url')) {
$vs_base_url = $this->opo_client_services_config->get('remote_media_base_url');
$o_client = new RestClient($vs_base_url . "/service.php/iteminfo/ItemInfo/rest");
try {
$o_res = $o_client->auth($this->opo_client_services_config->get('remote_media_username'), $this->opo_client_services_config->get('remote_media_password'))->get();
if (!$o_res->isSuccess()) {
$t_log->log(array('CODE' => 'ERR', 'MESSAGE' => _t('Could not authenticate to remote system %1', $vs_base_url), 'SOURCE' => 'clientServicesPlugin->hookPeriodicTask'));
}
while ($qr_orders->nextRow()) {
$t_order = new ca_commerce_orders($qr_orders->get('order_id'));
$vb_download_errors = false;
if ($t_order->getPrimaryKey() && sizeof($va_missing_media = $t_order->itemsMissingDownloadableMedia())) {
$va_missing_media_representation_ids = $t_order->itemsMissingDownloadableMedia('original', array('returnRepresentationIDs' => true));
foreach ($va_missing_media as $vn_object_id => $va_representation_md5s) {
foreach ($va_representation_md5s as $vn_i => $vs_representation_md5) {
$o_xml = $o_client->getObjectRepresentationURLByMD5($vs_representation_md5, 'original')->get();
$vs_url = (string) $o_xml->getObjectRepresentationURLByMD5->original;
if (!$vs_url) {
continue;
}
// media no longer exists
// fetch the file
$t_rep = new ca_object_representations($va_missing_media_representation_ids[$vn_object_id][$vn_i]);
if ($t_rep->getPrimaryKey() && ($vs_target_path = $t_rep->getMediaPath('media', 'original'))) {
if ($r_source = fopen($vs_url, "rb")) {
if ($r_target = fopen($vs_target_path, "wb")) {
while (feof($r_source) === false) {
fwrite($r_target, fread($r_source, 1024 * 8), 1024 * 8);
}
fclose($r_target);
} else {
$vb_download_errors = true;
$t_log->log(array('CODE' => 'ERR', 'MESSAGE' => _t('Could not open target path %1', $vs_target_path), 'SOURCE' => 'clientServicesPlugin->hookPeriodicTask'));
}
fclose($r_source);
} else {
$vb_download_errors = true;
$t_log->log(array('CODE' => 'ERR', 'MESSAGE' => _t('Could not open download URL "%1"', $vs_url), 'SOURCE' => 'clientServicesPlugin->hookPeriodicTask'));
}
// verify the file was downloaded correctly
if (($vs_target_md5 = md5_file($vs_target_path)) !== $vs_representation_md5) {
unlink($vs_target_path);
$t_log->log(array('CODE' => 'ERR', 'MESSAGE' => _t('Media file %1 failed to be downloaded from url "%2"; checksums differ: %3/%4', $vs_target_path, $vs_url, $vs_representation_md5, $vs_target_md5), 'SOURCE' => 'clientServicesPlugin->hookPeriodicTask'));
$vb_download_errors = true;
}
} else {
$t_log->log(array('CODE' => 'ERR', 'MESSAGE' => _t('Invalid representation_id "%1" or target path "%2"', $vn_representation_id, $vs_representation_md5, $vs_target_path), 'SOURCE' => 'clientServicesPlugin->hookPeriodicTask'));
$vb_download_errors = true;
}
}
}
}
if (!$vb_download_errors) {
$t_order->setMode(ACCESS_WRITE);
$t_order->set('order_status', 'PROCESSED');
$t_order->update();
if ($t_order->numErrors()) {
$t_log->log(array('CODE' => 'ERR', 'MESSAGE' => _t('Change of order status to PROCESSED from PROCESSED_AWAITING_MEDIA_ACCESS failed for order_id %1: %2', $t_order->getPrimaryKey(), join('; ', $t_order->getErrors())), 'SOURCE' => 'clientServicesPlugin->hookPeriodicTask'));
}
}
}
} catch (Exception $e) {
// noop
}
}
// Find any orders with status PROCESSED_AWAITING_DIGITIZATION where all media are now present
$qr_orders = $o_db->query("\n\t\t\t\tSELECT order_id\n\t\t\t\tFROM ca_commerce_orders\n\t\t\t\tWHERE\n\t\t\t\t\torder_status = 'PROCESSED_AWAITING_DIGITIZATION'\n\t\t\t");
while ($qr_orders->nextRow()) {
$t_order = new ca_commerce_orders($qr_orders->get('order_id'));
if ($t_order->getPrimaryKey() && !sizeof($t_order->itemsWithNoDownloadableMedia())) {
$t_order->setMode(ACCESS_WRITE);
$t_order->set('order_status', 'PROCESSED');
$t_order->update();
if ($t_order->numErrors()) {
$t_log->log(array('CODE' => 'ERR', 'MESSAGE' => _t('Change of order status to PROCESSED from PROCESSED_AWAITING_DIGITIZATION failed for order_id %1: %2', $t_order->getPrimaryKey(), join('; ', $t_order->getErrors())), 'SOURCE' => 'clientServicesPlugin->hookPeriodicTask'));
}
}
}
// Find orders paid/shipped more than X days ago and mark them as "COMPLETED"
$vn_days = (int) $this->opo_client_services_config->get('completed_order_age_threshold');
if ($vn_days > 1) {
$vn_threshold = (int) (time() - $vn_days * 24 * 60 * 60);
$qr_orders = $o_db->query("\n\t\t\t\t\tSELECT order_id\n\t\t\t\t\tFROM ca_commerce_orders\n\t\t\t\t\tWHERE\n\t\t\t\t\t\t(order_status = 'PROCESSED') \n\t\t\t\t\t\tAND \n\t\t\t\t\t\t((payment_received_on > 0) AND (payment_received_on < ?))\n\t\t\t\t\t\tAND \n\t\t\t\t\t\t(\n\t\t\t\t\t\t\t(shipping_date IS NULL AND shipped_on_date IS NULL)\n\t\t\t\t\t\t\tOR\n\t\t\t\t\t\t\t(\n\t\t\t\t\t\t\t\t(shipped_on_date > 0) AND (shipped_on_date < ?)\n\t\t\t\t\t\t\t)\n\t\t\t\t\t\t)\n\t\t\t\t", $vn_threshold, $vn_threshold);
while ($qr_orders->nextRow()) {
$t_order = new ca_commerce_orders($qr_orders->get('order_id'));
if ($t_order->getPrimaryKey()) {
$t_order->setMode(ACCESS_WRITE);
$t_order->set('order_status', 'COMPLETED');
$t_order->update();
if ($t_order->numErrors()) {
$t_log->log(array('CODE' => 'ERR', 'MESSAGE' => _t('Change of order status to COMPLETED from PROCESSED failed for order_id %1: %2', $t_order->getPrimaryKey(), join('; ', $t_order->getErrors())), 'SOURCE' => 'clientServicesPlugin->hookPeriodicTask'));
}
}
}
}
return true;
}
/**
* Download all media attached to specified object (not necessarily open for editing)
* Includes all representation media attached to the specified object + any media attached to other
* objects in the same object hierarchy as the specified object.
*/
public function DownloadMedia($pa_options = null)
{
list($vn_subject_id, $t_subject) = $this->_initView();
$pn_representation_id = $this->request->getParameter('representation_id', pInteger);
$pn_value_id = $this->request->getParameter('value_id', pInteger);
if ($pn_value_id) {
return $this->DownloadAttributeFile();
}
$ps_version = $this->request->getParameter('version', pString);
if (!$vn_subject_id) {
return;
}
$o_view = new View($this->request, $this->request->getViewsDirectoryPath() . '/bundles/');
if (!$ps_version) {
$ps_version = 'original';
}
$o_view->setVar('version', $ps_version);
$va_ancestor_ids = $t_subject->isHierarchical() ? $t_subject->getHierarchyAncestors(null, array('idsOnly' => true, 'includeSelf' => true)) : array($vn_subject_id);
if ($vn_parent_id = array_pop($va_ancestor_ids)) {
$t_subject->load($vn_parent_id);
array_unshift($va_ancestor_ids, $vn_parent_id);
}
$va_child_ids = $t_subject->isHierarchical() ? $t_subject->getHierarchyChildren(null, array('idsOnly' => true)) : array($vn_subject_id);
foreach ($va_ancestor_ids as $vn_id) {
array_unshift($va_child_ids, $vn_id);
}
$vn_c = 1;
$va_file_names = array();
$va_file_paths = array();
foreach ($va_child_ids as $vn_child_id) {
if (!$t_subject->load($vn_child_id)) {
continue;
}
if ($t_subject->tableName() == 'ca_object_representations') {
$va_reps = array($vn_child_id => array('representation_id' => $vn_child_id, 'info' => array($ps_version => $t_subject->getMediaInfo('media', $ps_version))));
} else {
$va_reps = $t_subject->getRepresentations(array($ps_version));
}
$vs_idno = $t_subject->get('idno');
foreach ($va_reps as $vn_representation_id => $va_rep) {
if ($pn_representation_id && $pn_representation_id != $vn_representation_id) {
continue;
}
$va_rep_info = $va_rep['info'][$ps_version];
$vs_idno_proc = preg_replace('![^A-Za-z0-9_\\-]+!', '_', $vs_idno);
switch ($this->request->user->getPreference('downloaded_file_naming')) {
case 'idno':
$vs_file_name = $vs_idno_proc . '_' . $vn_c . '.' . $va_rep_info['EXTENSION'];
break;
case 'idno_and_version':
$vs_file_name = $vs_idno_proc . '_' . $ps_version . '_' . $vn_c . '.' . $va_rep_info['EXTENSION'];
break;
case 'idno_and_rep_id_and_version':
$vs_file_name = $vs_idno_proc . '_representation_' . $vn_representation_id . '_' . $ps_version . '.' . $va_rep_info['EXTENSION'];
break;
case 'original_name':
default:
if ($va_rep['info']['original_filename']) {
$va_tmp = explode('.', $va_rep['info']['original_filename']);
if (sizeof($va_tmp) > 1) {
if (strlen($vs_ext = array_pop($va_tmp)) < 3) {
$va_tmp[] = $vs_ext;
}
}
$vs_file_name = join('_', $va_tmp);
} else {
$vs_file_name = $vs_idno_proc . '_representation_' . $vn_representation_id . '_' . $ps_version;
}
if (isset($va_file_names[$vs_file_name . '.' . $va_rep_info['EXTENSION']])) {
$vs_file_name .= "_{$vn_c}";
}
$vs_file_name .= '.' . $va_rep_info['EXTENSION'];
break;
}
$va_file_names[$vs_file_name] = true;
$o_view->setVar('version_download_name', $vs_file_name);
//
// Perform metadata embedding
$t_rep = new ca_object_representations($va_rep['representation_id']);
if (!($vs_path = caEmbedMediaMetadataIntoFile($t_rep->getMediaPath('media', $ps_version), $t_subject->tableName(), $t_subject->getPrimaryKey(), $t_subject->getTypeCode(), $t_rep->getPrimaryKey(), $t_rep->getTypeCode()))) {
$vs_path = $t_rep->getMediaPath('media', $ps_version);
}
$va_file_paths[$vs_path] = $vs_file_name;
$vn_c++;
}
}
if (!($vn_limit = ini_get('max_execution_time'))) {
$vn_limit = 30;
}
set_time_limit($vn_limit * 2);
$o_zip = new ZipStream();
foreach ($va_file_paths as $vs_path => $vs_name) {
$o_zip->addFile($vs_path, $vs_name);
}
$o_view->setVar('zip_stream', $o_zip);
$o_view->setVar('archive_name', preg_replace('![^A-Za-z0-9\\.\\-]+!', '_', $t_subject->get('idno')) . '.zip');
$this->response->addContent($o_view->render('download_file_binary.php'));
set_time_limit($vn_limit);
}
