Merge pull request #5592 from mrclay/5357

Fixes #5357: ElggBatch skips over incomplete entities

3 files changed, 166 insertions, 12 deletions

diff --git a/engine/classes/ElggBatch.php b/engine/classes/ElggBatch.php
index eb93b0f5d..d810ea066 100644
--- a/engine/classes/ElggBatch.php
+++ b/engine/classes/ElggBatch.php
@@ -150,6 +150,20 @@ class ElggBatch
 	private $incrementOffset = true;
 
 	/**
+	 * Entities that could not be instantiated during a fetch
+	 *
+	 * @var stdClass[]
+	 */
+	private $incompleteEntities = array();
+
+	/**
+	 * Total number of incomplete entities fetched
+	 *
+	 * @var int
+	 */
+	private $totalIncompletes = 0;
+
+	/**
 	 * Batches operations on any elgg_get_*() or compatible function that supports
 	 * an options array.
 	 *
@@ -222,6 +236,17 @@ class ElggBatch
 	}
 
 	/**
+	 * Tell the process that an entity was incomplete during a fetch
+	 *
+	 * @param stdClass $row
+	 *
+	 * @access private
+	 */
+	public function reportIncompleteEntity(stdClass $row) {
+		$this->incompleteEntities[] = $row;
+	}
+
+	/**
 	 * Fetches the next chunk of results
 	 *
 	 * @return bool
@@ -260,27 +285,47 @@ class ElggBatch
 		if ($this->incrementOffset) {
 			$offset = $this->offset + $this->retrievedResults;
 		} else {
-			$offset = $this->offset;
+			$offset = $this->offset + $this->totalIncompletes;
 		}
 
 		$current_options = array(
 			'limit' => $limit,
-			'offset' => $offset
+			'offset' => $offset,
+			'__ElggBatch' => $this,
 		);
 
 		$options = array_merge($this->options, $current_options);
-		$getter = $this->getter;
 
-		if (is_string($getter)) {
-			$this->results = $getter($options);
-		} else {
-			$this->results = call_user_func_array($getter, array($options));
+		$this->incompleteEntities = array();
+		$this->results = call_user_func_array($this->getter, array($options));
+
+		$num_results = count($this->results);
+		$num_incomplete = count($this->incompleteEntities);
+
+		$this->totalIncompletes += $num_incomplete;
+
+		if ($this->incompleteEntities) {
+			// pad the front of the results with nulls representing the incompletes
+			array_splice($this->results, 0, 0, array_pad(array(), $num_incomplete, null));
+			// ...and skip past them
+			reset($this->results);
+			for ($i = 0; $i < $num_incomplete; $i++) {
+				next($this->results);
+			}
 		}
 
 		if ($this->results) {
 			$this->chunkIndex++;
-			$this->resultIndex = 0;
-			$this->retrievedResults += count($this->results);
+
+			// let the system know we've jumped past the nulls
+			$this->resultIndex = $num_incomplete;
+
+			$this->retrievedResults += ($num_results + $num_incomplete);
+			if ($num_results == 0) {
+				// This fetch was *all* incompletes! We need to fetch until we can either
+				// offer at least one row to iterate over, or give up.
+				return $this->getNextResultsChunk();
+			}
 			return true;
 		} else {
 			return false;
diff --git a/engine/lib/entities.php b/engine/lib/entities.php
index d0d97aa95..072b26805 100644
--- a/engine/lib/entities.php
+++ b/engine/lib/entities.php
@@ -933,6 +933,8 @@ function elgg_get_entities(array $options = array()) {
 		'joins'					=>	array(),
 
 		'callback'				=> 'entity_row_to_elggstar',
+
+		'__ElggBatch'			=> null,
 	);
 
 	$options = array_merge($defaults, $options);
@@ -1050,7 +1052,7 @@ function elgg_get_entities(array $options = array()) {
 		}
 
 		if ($options['callback'] === 'entity_row_to_elggstar') {
-			$dt = _elgg_fetch_entities_from_sql($query);
+			$dt = _elgg_fetch_entities_from_sql($query, $options['__ElggBatch']);
 		} else {
 			$dt = get_data($query, $options['callback']);
 		}
@@ -1085,13 +1087,14 @@ function elgg_get_entities(array $options = array()) {
 /**
  * Return entities from an SQL query generated by elgg_get_entities.
  *
- * @param string $sql
+ * @param string    $sql
+ * @param ElggBatch $batch
  * @return ElggEntity[]
  *
  * @access private
  * @throws LogicException
  */
-function _elgg_fetch_entities_from_sql($sql) {
+function _elgg_fetch_entities_from_sql($sql, ElggBatch $batch = null) {
 	static $plugin_subtype;
 	if (null === $plugin_subtype) {
 		$plugin_subtype = get_subtype_id('object', 'plugin');
@@ -1168,6 +1171,11 @@ function _elgg_fetch_entities_from_sql($sql) {
 			} catch (IncompleteEntityException $e) {
 				// don't let incomplete entities throw fatal errors
 				unset($rows[$i]);
+
+				// report incompletes to the batch process that spawned this query
+				if ($batch) {
+					$batch->reportIncompleteEntity($row);
+				}
 			}
 		}
 	}
diff --git a/engine/tests/api/helpers.php b/engine/tests/api/helpers.php
index 62e4471e0..10216140f 100644
--- a/engine/tests/api/helpers.php
+++ b/engine/tests/api/helpers.php
@@ -578,6 +578,107 @@ class ElggCoreHelpersTest extends ElggCoreUnitTest {
 		$this->assertEqual(11, $j);
 	}
 
+	public function testElggBatchReadHandlesBrokenEntities() {
+		$num_test_entities = 8;
+		$guids = array();
+		for ($i = $num_test_entities; $i > 0; $i--) {
+			$entity = new ElggObject();
+			$entity->type = 'object';
+			$entity->subtype = 'test_5357_subtype';
+			$entity->access_id = ACCESS_PUBLIC;
+			$entity->save();
+			$guids[] = $entity->guid;
+			_elgg_invalidate_cache_for_entity($entity->guid);
+		}
+
+		// break entities such that the first fetch has one incomplete
+		// and the second and third fetches have only incompletes!
+		$db_prefix = elgg_get_config('dbprefix');
+		delete_data("
+			DELETE FROM {$db_prefix}objects_entity
+			WHERE guid IN ({$guids[1]}, {$guids[2]}, {$guids[3]}, {$guids[4]}, {$guids[5]})
+		");
+
+		$options = array(
+			'type' => 'object',
+			'subtype' => 'test_5357_subtype',
+			'order_by' => 'e.guid',
+		);
+
+		$entities_visited = array();
+		$batch = new ElggBatch('elgg_get_entities', $options, null, 2);
+		/* @var ElggEntity[] $batch */
+		foreach ($batch as $entity) {
+			$entities_visited[] = $entity->guid;
+		}
+
+		// The broken entities should not have been visited
+		$this->assertEqual($entities_visited, array($guids[0], $guids[6], $guids[7]));
+
+		// cleanup (including leftovers from previous tests)
+		$entity_rows = elgg_get_entities(array_merge($options, array(
+			'callback' => '',
+			'limit' => false,
+		)));
+		$guids = array();
+		foreach ($entity_rows as $row) {
+			$guids[] = $row->guid;
+		}
+		delete_data("DELETE FROM {$db_prefix}entities WHERE guid IN (" . implode(',', $guids) . ")");
+		delete_data("DELETE FROM {$db_prefix}objects_entity WHERE guid IN (" . implode(',', $guids) . ")");
+	}
+
+	public function testElggBatchDeleteHandlesBrokenEntities() {
+		$num_test_entities = 8;
+		$guids = array();
+		for ($i = $num_test_entities; $i > 0; $i--) {
+			$entity = new ElggObject();
+			$entity->type = 'object';
+			$entity->subtype = 'test_5357_subtype';
+			$entity->access_id = ACCESS_PUBLIC;
+			$entity->save();
+			$guids[] = $entity->guid;
+			_elgg_invalidate_cache_for_entity($entity->guid);
+		}
+
+		// break entities such that the first fetch has one incomplete
+		// and the second and third fetches have only incompletes!
+		$db_prefix = elgg_get_config('dbprefix');
+		delete_data("
+			DELETE FROM {$db_prefix}objects_entity
+			WHERE guid IN ({$guids[1]}, {$guids[2]}, {$guids[3]}, {$guids[4]}, {$guids[5]})
+		");
+
+		$options = array(
+			'type' => 'object',
+			'subtype' => 'test_5357_subtype',
+			'order_by' => 'e.guid',
+		);
+
+		$entities_visited = array();
+		$batch = new ElggBatch('elgg_get_entities', $options, null, 2, false);
+		/* @var ElggEntity[] $batch */
+		foreach ($batch as $entity) {
+			$entities_visited[] = $entity->guid;
+			$entity->delete();
+		}
+
+		// The broken entities should not have been visited
+		$this->assertEqual($entities_visited, array($guids[0], $guids[6], $guids[7]));
+
+		// cleanup (including leftovers from previous tests)
+		$entity_rows = elgg_get_entities(array_merge($options, array(
+			'callback' => '',
+			'limit' => false,
+		)));
+		$guids = array();
+		foreach ($entity_rows as $row) {
+			$guids[] = $row->guid;
+		}
+		delete_data("DELETE FROM {$db_prefix}entities WHERE guid IN (" . implode(',', $guids) . ")");
+		delete_data("DELETE FROM {$db_prefix}objects_entity WHERE guid IN (" . implode(',', $guids) . ")");
+	}
+
 	static function elgg_batch_callback_test($options, $reset = false) {
 		static $count = 1;