1 files changed, 228 insertions, 88 deletions

diff --git a/engine/lib/entities.php b/engine/lib/entities.php
index a50567d9f..ce736ce05 100644
--- a/engine/lib/entities.php
+++ b/engine/lib/entities.php
@@ -17,13 +17,13 @@ global $ENTITY_CACHE;
 $ENTITY_CACHE = array();
 
 /**
- * Cache subtypes and related class names once loaded.
+ * Cache subtypes and related class names.
  *
- * @global array $SUBTYPE_CACHE
+ * @global array|null $SUBTYPE_CACHE array once populated from DB, initially null
  * @access private
  */
 global $SUBTYPE_CACHE;
-$SUBTYPE_CACHE = NULL;
+$SUBTYPE_CACHE = null;
 
 /**
  * Invalidate this class's entry in the cache.
@@ -59,16 +59,24 @@ function invalidate_cache_for_entity($guid) {
 function cache_entity(ElggEntity $entity) {
 	global $ENTITY_CACHE;
 
-	// Don't cache entities while access control is off, otherwise they could be
+	// Don't cache non-plugin entities while access control is off, otherwise they could be
 	// exposed to users who shouldn't see them when control is re-enabled.
-	if (elgg_get_ignore_access()) {
+	if (!($entity instanceof ElggPlugin) && elgg_get_ignore_access()) {
 		return;
 	}
 
 	// Don't store too many or we'll have memory problems
 	// TODO(evan): Pick a less arbitrary limit
 	if (count($ENTITY_CACHE) > 256) {
-		unset($ENTITY_CACHE[array_rand($ENTITY_CACHE)]);
+		$random_guid = array_rand($ENTITY_CACHE);
+
+		unset($ENTITY_CACHE[$random_guid]);
+
+		// Purge separate metadata cache. Original idea was to do in entity destructor, but that would
+		// have caused a bunch of unnecessary purges at every shutdown. Doing it this way we have no way
+		// to know that the expunged entity will be GCed (might be another reference living), but that's
+		// OK; the metadata will reload if necessary.
+		elgg_get_metadata_cache()->clear($random_guid);
 	}
 
 	$ENTITY_CACHE[$entity->guid] = $entity;
@@ -87,8 +95,6 @@ function cache_entity(ElggEntity $entity) {
 function retrieve_cached_entity($guid) {
 	global $ENTITY_CACHE;
 
-	$guid = (int)$guid;
-
 	if (isset($ENTITY_CACHE[$guid])) {
 		if ($ENTITY_CACHE[$guid]->isFullyLoaded()) {
 			return $ENTITY_CACHE[$guid];
@@ -148,29 +154,23 @@ function retrieve_cached_entity_row($guid) {
  * @access private
  */
 function get_subtype_id($type, $subtype) {
-	global $CONFIG, $SUBTYPE_CACHE;
-
-	$type = sanitise_string($type);
-	$subtype = sanitise_string($subtype);
+	global $SUBTYPE_CACHE;
 
-	if ($subtype == "") {
-		return FALSE;
+	if (!$subtype) {
+		return false;
 	}
 
-	// @todo use the cache before hitting database
-	$result = get_data_row("SELECT * from {$CONFIG->dbprefix}entity_subtypes
-		where type='$type' and subtype='$subtype'");
-
-	if ($result) {
-		if (!$SUBTYPE_CACHE) {
-			$SUBTYPE_CACHE = array();
-		}
+	if ($SUBTYPE_CACHE === null) {
+		_elgg_populate_subtype_cache();
+	}
 
-		$SUBTYPE_CACHE[$result->id] = $result;
+	// use the cache before hitting database
+	$result = _elgg_retrieve_cached_subtype($type, $subtype);
+	if ($result !== null) {
 		return $result->id;
 	}
 
-	return FALSE;
+	return false;
 }
 
 /**
@@ -178,35 +178,67 @@ function get_subtype_id($type, $subtype) {
  *
  * @param int $subtype_id Subtype ID
  *
- * @return string Subtype name
+ * @return string|false Subtype name, false if subtype not found
  * @link http://docs.elgg.org/DataModel/Entities/Subtypes
  * @see get_subtype_from_id()
  * @access private
  */
 function get_subtype_from_id($subtype_id) {
-	global $CONFIG, $SUBTYPE_CACHE;
-
-	$subtype_id = (int)$subtype_id;
+	global $SUBTYPE_CACHE;
 
 	if (!$subtype_id) {
 		return false;
 	}
 
+	if ($SUBTYPE_CACHE === null) {
+		_elgg_populate_subtype_cache();
+	}
+
 	if (isset($SUBTYPE_CACHE[$subtype_id])) {
 		return $SUBTYPE_CACHE[$subtype_id]->subtype;
 	}
 
-	$result = get_data_row("SELECT * from {$CONFIG->dbprefix}entity_subtypes where id=$subtype_id");
-	if ($result) {
-		if (!$SUBTYPE_CACHE) {
-			$SUBTYPE_CACHE = array();
-		}
+	return false;
+}
 
-		$SUBTYPE_CACHE[$subtype_id] = $result;
-		return $result->subtype;
+/**
+ * Retrieve subtype from the cache.
+ *
+ * @param string $type
+ * @param string $subtype
+ * @return stdClass|null
+ *
+ * @access private
+ */
+function _elgg_retrieve_cached_subtype($type, $subtype) {
+	global $SUBTYPE_CACHE;
+
+	if ($SUBTYPE_CACHE === null) {
+		_elgg_populate_subtype_cache();
 	}
 
-	return false;
+	foreach ($SUBTYPE_CACHE as $obj) {
+		if ($obj->type === $type && $obj->subtype === $subtype) {
+			return $obj;
+		}
+	}
+	return null;
+}
+
+/**
+ * Fetch all suptypes from DB to local cache.
+ *
+ * @access private
+ */
+function _elgg_populate_subtype_cache() {
+	global $CONFIG, $SUBTYPE_CACHE;
+	
+	$results = get_data("SELECT * FROM {$CONFIG->dbprefix}entity_subtypes");
+	
+	$SUBTYPE_CACHE = array();
+	foreach ($results as $row) {
+		$SUBTYPE_CACHE[$row->id] = $row;
+	}
 }
 
 /**
@@ -225,25 +257,19 @@ function get_subtype_from_id($subtype_id) {
  * @access private
  */
 function get_subtype_class($type, $subtype) {
-	global $CONFIG, $SUBTYPE_CACHE;
-
-	$type = sanitise_string($type);
-	$subtype = sanitise_string($subtype);
+	global $SUBTYPE_CACHE;
 
-	// @todo use the cache before going to the database
-	$result = get_data_row("SELECT * from {$CONFIG->dbprefix}entity_subtypes
-		where type='$type' and subtype='$subtype'");
-
-	if ($result) {
-		if (!$SUBTYPE_CACHE) {
-			$SUBTYPE_CACHE = array();
-		}
-
-		$SUBTYPE_CACHE[$result->id] = $result;
-		return $result->class;
+	if ($SUBTYPE_CACHE === null) {
+		_elgg_populate_subtype_cache();
+	}
+	
+	// use the cache before going to the database
+	$obj = _elgg_retrieve_cached_subtype($type, $subtype);
+	if ($obj) {
+		return $obj->class;
 	}
 
-	return NULL;
+	return null;
 }
 
 /**
@@ -257,29 +283,21 @@ function get_subtype_class($type, $subtype) {
  * @access private
  */
 function get_subtype_class_from_id($subtype_id) {
-	global $CONFIG, $SUBTYPE_CACHE;
-
-	$subtype_id = (int)$subtype_id;
+	global $SUBTYPE_CACHE;
 
 	if (!$subtype_id) {
-		return false;
+		return null;
 	}
 
+	if ($SUBTYPE_CACHE === null) {
+		_elgg_populate_subtype_cache();
+	}
+	
 	if (isset($SUBTYPE_CACHE[$subtype_id])) {
 		return $SUBTYPE_CACHE[$subtype_id]->class;
 	}
 
-	$result = get_data_row("SELECT * from {$CONFIG->dbprefix}entity_subtypes where id=$subtype_id");
-
-	if ($result) {
-		if (!$SUBTYPE_CACHE) {
-			$SUBTYPE_CACHE = array();
-		}
-		$SUBTYPE_CACHE[$subtype_id] = $result;
-		return $result->class;
-	}
-
-	return NULL;
+	return null;
 }
 
 /**
@@ -305,21 +323,32 @@ function get_subtype_class_from_id($subtype_id) {
  * @see get_entity()
  */
 function add_subtype($type, $subtype, $class = "") {
-	global $CONFIG;
-	$type = sanitise_string($type);
-	$subtype = sanitise_string($subtype);
-	$class = sanitise_string($class);
+	global $CONFIG, $SUBTYPE_CACHE;
 
-	// Short circuit if no subtype is given
-	if ($subtype == "") {
+	if (!$subtype) {
 		return 0;
 	}
 
 	$id = get_subtype_id($type, $subtype);
 
-	if ($id == 0) {
-		return insert_data("insert into {$CONFIG->dbprefix}entity_subtypes"
-			. " (type, subtype, class) values ('$type','$subtype','$class')");
+	if (!$id) {
+		// In cache we store non-SQL-escaped strings because that's what's returned by query
+		$cache_obj = (object) array(
+			'type' => $type,
+			'subtype' => $subtype,
+			'class' => $class,
+		);
+
+		$type = sanitise_string($type);
+		$subtype = sanitise_string($subtype);
+		$class = sanitise_string($class);
+
+		$id = insert_data("INSERT INTO {$CONFIG->dbprefix}entity_subtypes"
+			. " (type, subtype, class) VALUES ('$type', '$subtype', '$class')");
+		
+		// add entry to cache
+		$cache_obj->id = $id;
+		$SUBTYPE_CACHE[$id] = $cache_obj;
 	}
 
 	return $id;
@@ -361,22 +390,31 @@ function remove_subtype($type, $subtype) {
 function update_subtype($type, $subtype, $class = '') {
 	global $CONFIG, $SUBTYPE_CACHE;
 
-	if (!$id = get_subtype_id($type, $subtype)) {
-		return FALSE;
+	$id = get_subtype_id($type, $subtype);
+	if (!$id) {
+		return false;
+	}
+
+	if ($SUBTYPE_CACHE === null) {
+		_elgg_populate_subtype_cache();
 	}
+
+	$unescaped_class = $class;
+
 	$type = sanitise_string($type);
 	$subtype = sanitise_string($subtype);
-
-	$result = update_data("UPDATE {$CONFIG->dbprefix}entity_subtypes
+	$class = sanitise_string($class);
+	
+	$success = update_data("UPDATE {$CONFIG->dbprefix}entity_subtypes
 		SET type = '$type', subtype = '$subtype', class = '$class'
 		WHERE id = $id
 	");
 
-	if ($result && isset($SUBTYPE_CACHE[$id])) {
-		$SUBTYPE_CACHE[$id]->class = $class;
+	if ($success && isset($SUBTYPE_CACHE[$id])) {
+		$SUBTYPE_CACHE[$id]->class = $unescaped_class;
 	}
 
-	return $result;
+	return $success;
 }
 
 /**
@@ -735,7 +773,13 @@ 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 +1024,12 @@ 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 +1058,97 @@ 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
@@ -1772,7 +1912,7 @@ function oddentity_to_elggentity(ODDEntity $element) {
 	if (!$tmp) {
 		// Construct new class with owner from session
 		$classname = get_subtype_class($class, $subclass);
-		if ($classname != "") {
+		if ($classname) {
 			if (class_exists($classname)) {
 				$tmp = new $classname();
 
@@ -1838,7 +1978,7 @@ function oddentity_to_elggentity(ODDEntity $element) {
 function import_entity_plugin_hook($hook, $entity_type, $returnvalue, $params) {
 	$element = $params['element'];
 
-	$tmp = NULL;
+	$tmp = null;
 
 	if ($element instanceof ODDEntity) {
 		$tmp = oddentity_to_elggentity($element);
@@ -2011,7 +2151,7 @@ function get_entity_url($entity_guid) {
 function elgg_register_entity_url_handler($entity_type, $entity_subtype, $function_name) {
 	global $CONFIG;
 
-	if (!is_callable($function_name)) {
+	if (!is_callable($function_name, true)) {
 		return false;
 	}