/**
* Listen to the minute cron in order to sync data to ElasticSearch
*
* @param string $hook the name of the hook
* @param string $type the type of the hook
* @param string $returnvalue current return value
* @param array $params supplied params
*
* @return void
*/
public static function minuteSync($hook, $type, $returnvalue, $params)
{
$setting = elasticsearch_get_setting('sync');
if ($setting !== 'yes') {
// sync not enabled
return;
}
$client = elasticsearch_get_client();
if (empty($client)) {
return;
}
$starttime = (int) elgg_extract('time', $params, time());
// delete first
$client->bulkDeleteDocuments();
$update_actions = array('no_index_ts', 'update', 'reindex');
foreach ($update_actions as $action) {
$options = elasticsearch_get_bulk_options($action);
if (empty($options)) {
continue;
}
$getter = '';
if ($action === 'no_index_ts') {
$getter = 'elgg_get_entities';
}
$time_left = self::batchSync($options, $starttime, $getter);
if ($time_left === false) {
return;
}
}
}
public function __construct($params)
{
$this->default_index = elasticsearch_get_setting('index');
$this->search_alias = elasticsearch_get_setting('search_alias');
$this->search_params = new SearchParams(['client' => $this]);
parent::__construct($params);
}
/**
* Hook to transform a search result to an Elgg Entity
*
* @param string $hook the name of the hook
* @param string $type the type of the hook
* @param Client $returnvalue current return value
* @param array $params supplied params
*
* @return void|Client
*/
public static function sourceToEntity($hook, $type, $returnvalue, $params)
{
$hit = elgg_extract('hit', $params);
$index = elgg_extract('_index', $hit);
$elgg_index = elasticsearch_get_setting('index');
if ($index !== $elgg_index) {
return;
}
$source = elgg_extract('_source', $hit);
$row = new \stdClass();
foreach ($source as $key => $value) {
switch ($key) {
case 'subtype':
// elastic stores the textual version of the subtype, entity_row_to_elggstar needs the int
$row->{$key} = get_subtype_id($source['type'], $value);
break;
case 'last_action':
case 'time_created':
case 'time_updated':
// convert the timestamps to unix timestamps
$value = strtotime($value);
default:
$row->{$key} = $value;
break;
}
}
// enabled attribute is not stored in elasticsearch by default
$row->enabled = 'yes';
// specials types
if ($row->type == 'user') {
// makes sure all attributes are loaded to prevent a db call
$external_attributes = \ElggUser::getExternalAttributes();
foreach ($external_attributes as $key => $value) {
if (isset($row->{$key})) {
continue;
}
$row->{$key} = $value;
}
}
try {
$result = entity_row_to_elggstar($row);
} catch (\Exception $e) {
elgg_log($e->getMessage(), 'NOTICE');
return;
}
return $result;
}
/**
* Get the $options for elgg_get_entities* in order to update the ElasticSearch index
*
* @param string $type which options to get
*
* @return false|array
*/
function elasticsearch_get_bulk_options($type = 'no_index_ts')
{
$type_subtypes = elasticsearch_get_registered_entity_types();
if (empty($type_subtypes)) {
return false;
}
$dbprefix = elgg_get_config('dbprefix');
switch ($type) {
case 'no_index_ts':
// new or updated entities
return array('type_subtype_pairs' => $type_subtypes, 'limit' => false, 'wheres' => array("NOT EXISTS (\n\t\t\t\t\t\tSELECT 1 FROM {$dbprefix}private_settings ps\n\t\t\t\t\t\tWHERE ps.entity_guid = e.guid\n\t\t\t\t\t\tAND ps.name = '" . ELASTICSEARCH_INDEXED_NAME . "'\n\t\t\t\t\t)", "NOT EXISTS (\n\t\t\t\t\t\tSELECT 1 FROM {$dbprefix}users_entity ue\n\t\t\t\t\t\tWHERE ue.guid = e.guid\n\t\t\t\t\t\tAND ue.banned = 'yes'\n\t\t\t\t\t)"));
break;
case 'reindex':
// a reindex has been initiated, so update all out of date entities
$setting = (int) elasticsearch_get_setting('reindex_ts');
if (empty($setting)) {
return false;
}
return array('type_subtype_pairs' => $type_subtypes, 'limit' => false, 'private_setting_name_value_pairs' => array(array('name' => ELASTICSEARCH_INDEXED_NAME, 'value' => $setting, 'operand' => '<'), array('name' => ELASTICSEARCH_INDEXED_NAME, 'value' => 0, 'operand' => '>')));
break;
case 'update':
// content that was updated in Elgg and needs to be reindexed
return array('type_subtype_pairs' => $type_subtypes, 'limit' => false, 'private_setting_name_value_pairs' => array(array('name' => ELASTICSEARCH_INDEXED_NAME, 'value' => 0)));
break;
case 'count':
// content that needs to be indexed
return array('type_subtype_pairs' => $type_subtypes, 'count' => true);
break;
}
return false;
}
break;
}
try {
$client->indices()->putAlias(array('index' => $index, 'name' => $alias));
} catch (Exception $e) {
register_error(elgg_echo('elasticsearch:action:admin:index_management:error:add_alias', array($alias, $index)));
break;
}
system_message(elgg_echo('elasticsearch:action:admin:index_management:add_alias', array($alias, $index)));
break;
case 'delete_alias':
if (!$exists) {
register_error(elgg_echo('elasticsearch:error:index_not_exists', array($index)));
break;
}
$alias = elasticsearch_get_setting('search_alias');
if (empty($alias)) {
register_error(elgg_echo('elasticsearch:error:alias_not_configured'));
break;
}
$alias_exists = $client->indices()->existsAlias(array('name' => $alias, 'index' => $index));
if (!$alias_exists) {
register_error(elgg_echo('elasticsearch:action:admin:index_management:error:delete_alias:exists', array($alias, $index)));
break;
}
try {
$client->indices()->deleteAlias(array('index' => $index, 'name' => $alias));
} catch (Exception $e) {
register_error(elgg_echo('elasticsearch:action:admin:index_management:error:delete_alias', array($alias, $index)));
break;
}
$options['count'] = true;
$count = elgg_get_entities_from_private_settings($options);
$content .= '<tr>';
$content .= '<td>' . elgg_echo('elasticsearch:stats:elgg:update') . '</td>';
$content .= "<td>{$count}</td>";
$content .= '</tr>';
// content to reindex
$options = elasticsearch_get_bulk_options('reindex');
$count = 0;
if (!empty($options)) {
$options['count'] = true;
$count = elgg_get_entities_from_private_settings($options);
}
$content .= '<tr>';
$content .= '<td>' . elgg_echo('elasticsearch:stats:elgg:reindex');
$reindex_title = elgg_echo('elasticsearch:stats:elgg:reindex:action');
$last_ts = (int) elasticsearch_get_setting('reindex_ts');
if (!empty($last_ts)) {
$reindex_title .= ' ' . elgg_echo('elasticsearch:stats:elgg:reindex:last_ts', [date('c', $last_ts)]);
}
$content .= elgg_view('output/url', ['confirm' => true, 'href' => 'action/elasticsearch/admin/reindex', 'text' => elgg_view_icon('refresh'), 'title' => $reindex_title, 'class' => 'mlm']);
$content .= '</td>';
$content .= "<td>{$count}</td>";
$content .= '</tr>';
$count = count(elasticsearch_get_documents_for_deletion());
$content .= '<tr>';
$content .= '<td>' . elgg_echo('elasticsearch:stats:elgg:delete') . '</td>';
$content .= "<td>{$count}</td>";
$content .= '</tr>';
$content .= '</table>';
echo elgg_view_module('inline', elgg_echo('elasticsearch:stats:elgg'), $content);