mod/elgg_solr/lib/functions.php ➔ elgg_solr_reindex()

Complexity

Conditions	21
Paths	> 20000

Size

Total Lines

206

Duplication

Lines	11
Ratio	5.34 %

Code Coverage

Tests	0
CRAP Score	462

Importance

Changes

0

<?php
function elgg_solr_reindex() {
	set_time_limit(0);

	$is_elgg18 = elgg_solr_is_elgg18();

	$guid_getter = 'elgg_solr_get_entity_guids';
	if ($is_elgg18) {
		$guid_getter = 'elgg_solr_get_entity_guids_18';
	}

	$ia = elgg_set_ignore_access(true);
	$show_hidden = access_get_show_hidden_status();
	access_show_hidden_entities(true);

	// lock the function
	elgg_set_plugin_setting('reindex_running', 1, 'elgg_solr');

	if (!file_exists(elgg_get_config('dataroot') . 'elgg_solr')) {
		mkdir(elgg_get_config('dataroot') . 'elgg_solr');
	}

	$logtime = elgg_get_config('elgg_solr_restart_logtime');
	if (!$logtime) {
		$logtime = time();
	}
	$log = elgg_get_config('dataroot') . 'elgg_solr/' . $logtime . '.txt';
	elgg_set_plugin_setting('current_log', $logtime, 'elgg_solr');

	// initialize the csv
	$report = array(
		'percent' => '',
		'count' => 0, // report prior to indexing this entity
		'typecount' => 0,
		'fullcount' => 0,
		'type' => '',
		'querytime' => 0,
		'message' => 'Initializing Reindex',
		'date' => date('Y-M-j H:i:s'),
		'logtime' => $logtime
	);
	file_put_contents($log, json_encode($report) . "\n", FILE_APPEND);


	$debug = get_input('debug', false);
	if ($debug) {
		elgg_set_config('elgg_solr_debug', 1);
	}

	$registered_types = elgg_get_config('elgg_solr_reindex_options');
	if (!$registered_types) {
		$registered_types = get_registered_entity_types();
	}

	// build our options and cache them in case we need to restart it
	$cacheoptions = array(
		'types' => $registered_types
	);

	$options = array();
	$time = elgg_get_config('elgg_solr_time_options');
	if ($time && is_array($time)) {
		$options['wheres'] = array(
			"e.time_created >= {$time['starttime']}",
			"e.time_created <= {$time['endtime']}",
		);

		$cacheoptions['starttime'] = $time['starttime'];
		$cacheoptions['endtime'] = $time['endtime'];
	}

	elgg_set_config('elgg_solr_nocommit', true); // tell our indexer not to commit right away

	$fullcount = 0;
	foreach ($registered_types as $type => $subtypes) {
		$options['type'] = $type;
		$options['subtypes'] = ELGG_ENTITIES_ANY_VALUE;
		$options['limit'] = false;

		$restart_time = elgg_get_config('elgg_solr_restart_time');
		if ($restart_time) {
			elgg_set_config('elgg_solr_restart_time', false);

			$options['wheres'][1] = "e.time_created <= {$restart_time}";
		} elseif ($time['endtime']) {
			$options['wheres'][1] = "e.time_created <= {$time['endtime']}";
		}

		if ($subtypes) {
			if (!is_array($subtypes)) {
				$options['subtypes'] = array($subtypes);
			}
			else {
				$options['subtypes'] = $subtypes;
			}
		}

		// this iteration fixes a bug https://github.com/Elgg/Elgg/issues/7561
		// uses a custom getter which only fetches the guids in a single large-batch query
		// which is much more efficient than standard egef
		$batch_size = elgg_get_plugin_setting('reindex_batch_size', 'elgg_solr');
		$entities = new ElggBatch($guid_getter, $options, null, $batch_size);
		$final_count = $guid_getter(array_merge($options, array('count' => true)));

		elgg_set_config('elgg_solr_nocommit', true); // disable committing on each entity for performance
		$count = 0;
		$fetch_time_start = microtime(true);
		foreach ($entities as $e) {
			$count++;
			$fullcount++;
			$first_entity = (bool) (($count % $batch_size) == 1);
			$last_entity = (bool) (($count % $batch_size) == 0);

			if ($first_entity) {
				// this is the first entity in the new batch
				$fetch_time = microtime(true) - $fetch_time_start; // the query time in seconds
			}

			$entity = get_entity($e->guid);
			if ($entity) {
				elgg_solr_add_update_entity(null, null, $entity);
				elgg_set_config('elgg_solr_nocommit', true);
			}

			if (!($count % 200)) {
				$qtime = round($fetch_time, 4);

The variable $fetch_time does not seem to be defined for all execution paths leading up to this point.

				$percent = round($count / $final_count * 100);
				if ($entity) {
					$restart_time = $entity->time_created;
				}
				$report = array(
					'percent' => $percent,
					'count' => $count,
					'typecount' => $final_count,
					'fullcount' => $fullcount,
					'type' => $type,
					'querytime' => $qtime,
					'message' => '',
					'date' => date('Y-M-j H:i:s'),
					'cacheoptions' => $cacheoptions,
					'logtime' => $logtime,
					'restart_time' => $restart_time
				);

				file_put_contents($log, json_encode($report) . "\n", FILE_APPEND);
				elgg_set_config('elgg_solr_nocommit', false); // push a commit on this one

				// check for the termination signal
				if ($logtime == elgg_get_plugin_setting('stop_reindex', 'elgg_solr')) {
					$report = array(
						'percent' => $percent,
						'count' => $count,
						'typecount' => $final_count,
						'fullcount' => $fullcount,
						'type' => $type,
						'querytime' => $qtime,
						'message' => 'Reindex has been stopped',
						'date' => date('Y-M-j H:i:s'),
						'cacheoptions' => $cacheoptions,
						'logtime' => $logtime,
						'restart_time' => $restart_time
					);
					
					file_put_contents($log, json_encode($report) . "\n", FILE_APPEND);
					error_log('Stopping reindex due to termination signal');
					exit;
				}
			}

			if ($last_entity) {
				$fetch_time_start = microtime(true);
			}
		}

		// we've finished this type, unset from the cache options
		unset($cacheoptions['types'][$type]);
	}

	$report = array(
		'percent' => '',
		'count' => 0, // report prior to indexing this entity
		'typecount' => 0,
		'fullcount' => 0,
		'type' => '',
		'querytime' => 0,
		'message' => 'Reindex complete',
		'date' => date('Y-M-j H:i:s'),
		'logtime' => $logtime
	);
	file_put_contents($log, json_encode($report) . "\n", FILE_APPEND);
	elgg_set_plugin_setting('reindex_running', 0, 'elgg_solr');

	// commit the last of the entities
	$client = elgg_solr_get_client();
	$query = $client->createUpdate();
	$query->addCommit();

	try {
		$client->update($query);
	} catch (Exception $e) {
		error_log($e->getMessage());
		return false;
	}

	access_show_hidden_entities($show_hidden);
	elgg_set_ignore_access($ia);
}

}