Fixes #4929: Optimize elgg_get_entities and add attribute loader

1 files changed, 104 insertions, 2 deletions

diff --git a/engine/lib/entities.php b/engine/lib/entities.php
index a14160e14..70a950b16 100644
--- a/engine/lib/entities.php
+++ b/engine/lib/entities.php
@@ -735,7 +735,13 @@ function get_entity($guid) {
 		}
 	}
 
-	$new_entity = entity_row_to_elggstar($entity_row);
+	// don't let incomplete entities cause fatal exceptions
+	try {
+		$new_entity = entity_row_to_elggstar($entity_row);
+	} catch (IncompleteEntityException $e) {
+		return false;
+	}
+
 	if ($new_entity) {
 		cache_entity($new_entity);
 	}
@@ -980,7 +986,12 @@ function elgg_get_entities(array $options = array()) {
 			$query .= " LIMIT $offset, $limit";
 		}
 
-		$dt = get_data($query, $options['callback']);
+		if ($options['callback'] === 'entity_row_to_elggstar') {
+			$dt = _elgg_fetch_entities_from_sql($query);
+		} else {
+			$dt = get_data($query, $options['callback']);
+		}
+
 		if ($dt) {
 			// populate entity and metadata caches
 			$guids = array();
@@ -1009,6 +1020,97 @@ function elgg_get_entities(array $options = array()) {
 }
 
 /**
+ * Return entities from an SQL query generated by elgg_get_entities.
+ *
+ * @param string $sql
+ * @return ElggEntity[]
+ *
+ * @access private
+ * @throws LogicException
+ */
+function _elgg_fetch_entities_from_sql($sql) {
+	static $plugin_subtype;
+	if (null === $plugin_subtype) {
+		$plugin_subtype = get_subtype_id('object', 'plugin');
+	}
+
+	// Keys are types, values are columns that, if present, suggest that the secondary
+	// table is already JOINed
+	$types_to_optimize = array(
+		'object' => 'title',
+		'user' => 'password',
+		'group' => 'name',
+	);
+
+	$rows = get_data($sql);
+
+	// guids to look up in each type
+	$lookup_types = array();
+	// maps GUIDs to the $rows key
+	$guid_to_key = array();
+
+	if (isset($rows[0]->type, $rows[0]->subtype)
+			&& $rows[0]->type === 'object'
+			&& $rows[0]->subtype == $plugin_subtype) {
+		// Likely the entire resultset is plugins, which have already been optimized
+		// to JOIN the secondary table. In this case we allow retrieving from cache,
+		// but abandon the extra queries.
+		$types_to_optimize = array();
+	}
+
+	// First pass: use cache where possible, gather GUIDs that we're optimizing
+	foreach ($rows as $i => $row) {
+		if (empty($row->guid) || empty($row->type)) {
+			throw new LogicException('Entity row missing guid or type');
+		}
+		if ($entity = retrieve_cached_entity($row->guid)) {
+			$rows[$i] = $entity;
+			continue;
+		}
+		if (isset($types_to_optimize[$row->type])) {
+			// check if row already looks JOINed.
+			if (isset($row->{$types_to_optimize[$row->type]})) {
+				// Row probably already contains JOINed secondary table. Don't make another query just
+				// to pull data that's already there
+				continue;
+			}
+			$lookup_types[$row->type][] = $row->guid;
+			$guid_to_key[$row->guid] = $i;
+		}
+	}
+	// Do secondary queries and merge rows
+	if ($lookup_types) {
+		$dbprefix = elgg_get_config('dbprefix');
+	}
+	foreach ($lookup_types as $type => $guids) {
+		$set = "(" . implode(',', $guids) . ")";
+		$sql = "SELECT * FROM {$dbprefix}{$type}s_entity WHERE guid IN $set";
+		$secondary_rows = get_data($sql);
+		if ($secondary_rows) {
+			foreach ($secondary_rows as $secondary_row) {
+				$key = $guid_to_key[$secondary_row->guid];
+				// cast to arrays to merge then cast back
+				$rows[$key] = (object)array_merge((array)$rows[$key], (array)$secondary_row);
+			}
+		}
+	}
+	// Second pass to finish conversion
+	foreach ($rows as $i => $row) {
+		if ($row instanceof ElggEntity) {
+			continue;
+		} else {
+			try {
+				$rows[$i] = entity_row_to_elggstar($row);
+			} catch (IncompleteEntityException $e) {
+				// don't let incomplete entities throw fatal errors
+				unset($rows[$i]);
+			}
+		}
+	}
+	return $rows;
+}
+
+/**
  * Returns SQL where clause for type and subtype on main entity table
  *
  * @param string     $table    Entity table prefix as defined in SELECT...FROM entities $table