Full sync: Set chunk size of Woo modules dynamically (#41433)

* Add woo modules dynamic

* changelog

* Typo from merge

* Typos regarding MAX_META_LENGTH and phan issues

* Bug fixingsince objects can actually be arrays :(

* Added tests so that objects coming as arrays are also added in filter_objects_and_metadata_by_size.

* Woocommerce module handles order items, let's make sure we use the proper term

* Changes so that order items when expanded, are returned in desc order which is necessary for the chunk logic

* Deprecate removed methods

* Sanitize order in get_order_item_by_ids

* Renamed function to build_full_sync_action_array

* Update projects/packages/sync/src/modules/class-woocommerce.php

Typo

* Updated docblock

Committed via a GitHub action: https://github.com/Automattic/jetpack/actions/runs/13174118421

Upstream-Ref: Automattic/jetpack@2da3fe1

Author: darssen

jetpack_vendor/automattic/jetpack-sync/src/modules/class-comments.php

@@ -15,24 +15,6 @@
  */
 class Comments extends Module {
 
-	/**
-	 * Max bytes allowed for full sync upload.
-	 * Current Setting : 7MB.
-	 *
-	 * @access public
-	 *
-	 * @var int
-	 */
-	const MAX_SIZE_FULL_SYNC = 7000000;
-	/**
-	 * Max bytes allowed for post meta_value => length.
-	 * Current Setting : 2MB.
-	 *
-	 * @access public
-	 *
-	 * @var int
-	 */
-	const MAX_COMMENT_META_LENGTH = 2000000;
 	/**
 	 * Sync module name.
 	 *
@@ -599,7 +581,7 @@ public function get_next_chunk( $config, $status, $chunk_size ) {
 			'comment',
 			$comments,
 			$metadata,
-			self::MAX_COMMENT_META_LENGTH, // Replace with appropriate comment meta length constant.
+			self::MAX_META_LENGTH, // Replace with appropriate comment meta length constant.
 			self::MAX_SIZE_FULL_SYNC
 		);
 
@@ -609,22 +591,4 @@ public function get_next_chunk( $config, $status, $chunk_size ) {
 			'meta'       => $filtered_comments_metadata,
 		);
 	}
-
-	/**
-	 * Set the status of the full sync action based on the objects that were sent.
-	 *
-	 * @access public
-	 *
-	 * @param array $status This module Full Sync status.
-	 * @param array $objects This module Full Sync objects.
-	 *
-	 * @return array The updated status.
-	 */
-	public function set_send_full_sync_actions_status( $status, $objects ) {
-
-		$object_ids          = $objects['object_ids'];
-		$status['last_sent'] = end( $object_ids );
-		$status['sent']     += count( $object_ids );
-		return $status;
-	}
 }

jetpack_vendor/automattic/jetpack-sync/src/modules/class-meta.php

@@ -177,7 +177,7 @@ private function fetch_prepared_meta_from_db( $object_type, $where, $meta_object
 	private function get_prepared_meta_object( $object_type, $meta_entry ) {
 		$object_id_column = $object_type . '_id';
 
-		if ( 'post' === $object_type && strlen( $meta_entry['meta_value'] ) >= Posts::MAX_POST_META_LENGTH ) {
+		if ( 'post' === $object_type && strlen( $meta_entry['meta_value'] ) >= Posts::MAX_META_LENGTH ) {
 			$meta_entry['meta_value'] = '';
 		}
 

jetpack_vendor/automattic/jetpack-sync/src/modules/class-module.php

@@ -38,6 +38,26 @@ abstract class Module {
 	 */
 	const MAX_DB_QUERY_LENGTH = 15 * 1024;
 
+	/**
+	 * Max bytes allowed for full sync upload for the module.
+	 * Default Setting : 7MB.
+	 *
+	 * @access public
+	 *
+	 * @var int
+	 */
+	const MAX_SIZE_FULL_SYNC = 7000000;
+
+	/**
+	 * Max bytes allowed for post meta_value => length.
+	 * Default Setting : 2MB.
+	 *
+	 * @access public
+	 *
+	 * @var int
+	 */
+	const MAX_META_LENGTH = 2000000;
+
 	/**
 	 * Sync module name.
 	 *
@@ -439,6 +459,9 @@ public function send_full_sync_actions( $config, $status, $send_until ) {
 
 	/**
 	 * Set the status of the full sync action based on the objects that were sent.
+	 * Used to update the status of the module after sending a chunk of objects.
+	 * Since Full Sync logic chunking relies on order of items being processed in descending order, we need to sort
+	 * due to some modules (e.g. WooCommerce) changing the order while getting the objects.
 	 *
 	 * @access protected
 	 *
@@ -448,8 +471,10 @@ public function send_full_sync_actions( $config, $status, $send_until ) {
 	 * @return array The updated status.
 	 */
 	protected function set_send_full_sync_actions_status( $status, $objects ) {
-		$status['last_sent'] = end( $objects );
-		$status['sent']     += count( $objects );
+
+		$object_ids          = $objects['object_ids'] ?? $objects;
+		$status['last_sent'] = end( $object_ids );
+		$status['sent']     += count( $object_ids );
 		return $status;
 	}
 
@@ -715,9 +740,11 @@ public function filter_objects_and_metadata_by_size( $type, $objects, $metadata,
 			$object_size      = strlen( maybe_serialize( $object ) );
 			$current_metadata = array();
 			$metadata_size    = 0;
+			$id_field         = $this->id_field();
+			$object_id        = (int) ( is_object( $object ) ? $object->{$id_field} : $object[ $id_field ] );
 
 			foreach ( $metadata as $key => $metadata_item ) {
-				if ( (int) $metadata_item->{$type . '_id'} === (int) $object->{$this->id_field()} ) {
+				if ( (int) $metadata_item->{$type . '_id'} === $object_id ) {
 					$metadata_item_size = strlen( maybe_serialize( $metadata_item->meta_value ) );
 					if ( $metadata_item_size >= $max_meta_size ) {
 						$metadata_item->meta_value = ''; // Trim metadata if too large.
@@ -734,7 +761,7 @@ public function filter_objects_and_metadata_by_size( $type, $objects, $metadata,
 
 			// Always allow the first object with metadata.
 			if ( empty( $filtered_object_ids ) || ( $current_size + $object_size + $metadata_size ) <= $max_total_size ) {
-				$filtered_object_ids[] = strval( $object->{$this->id_field()} );
+				$filtered_object_ids[] = strval( is_object( $object ) ? $object->{$id_field} : $object[ $id_field ] );
 				$filtered_objects[]    = $object;
 				$filtered_metadata     = array_merge( $filtered_metadata, $current_metadata );
 				$current_size         += $object_size + $metadata_size;

jetpack_vendor/automattic/jetpack-sync/src/modules/class-posts.php

@@ -64,26 +64,6 @@ class Posts extends Module {
 	 */
 	const MAX_POST_CONTENT_LENGTH = 5000000;
 
-	/**
-	 * Max bytes allowed for post meta_value => length.
-	 * Current Setting : 2MB.
-	 *
-	 * @access public
-	 *
-	 * @var int
-	 */
-	const MAX_POST_META_LENGTH = 2000000;
-
-	/**
-	 * Max bytes allowed for full sync upload.
-	 * Current Setting : 7MB.
-	 *
-	 * @access public
-	 *
-	 * @var int
-	 */
-	const MAX_SIZE_FULL_SYNC = 7000000;
-
 	/**
 	 * Default previous post state.
 	 * Used for default previous post status.
@@ -321,7 +301,7 @@ public function get_full_sync_actions() {
 	}
 
 	/**
-	 * Filter meta arguments so that we don't sync meta_values over MAX_POST_META_LENGTH.
+	 * Filter meta arguments so that we don't sync meta_values over MAX_META_LENGTH.
 	 *
 	 * @param array $args action arguments.
 	 *
@@ -332,7 +312,7 @@ public function trim_post_meta( $args ) {
 		// Explicitly truncate meta_value when it exceeds limit.
 		// Large content will cause OOM issues and break Sync.
 		$serialized_value = maybe_serialize( $meta_value );
-		if ( strlen( $serialized_value ) >= self::MAX_POST_META_LENGTH ) {
+		if ( strlen( $serialized_value ) >= self::MAX_META_LENGTH ) {
 			$meta_value = '';
 		}
 		return array( $meta_id, $object_id, $meta_key, $meta_value );
@@ -894,7 +874,7 @@ public function get_next_chunk( $config, $status, $chunk_size ) {
 			'post',
 			$posts,
 			$metadata,
-			self::MAX_POST_META_LENGTH,
+			self::MAX_META_LENGTH,
 			self::MAX_SIZE_FULL_SYNC
 		);
 
@@ -918,22 +898,4 @@ private function expand_posts( $post_ids ) {
 		$posts = array_values( $posts ); // Reindex in case posts were deleted.
 		return $posts;
 	}
-
-	/**
-	 * Set the status of the full sync action based on the objects that were sent.
-	 *
-	 * @access public
-	 *
-	 * @param array $status This module Full Sync status.
-	 * @param array $objects This module Full Sync objects.
-	 *
-	 * @return array The updated status.
-	 */
-	public function set_send_full_sync_actions_status( $status, $objects ) {
-
-		$object_ids          = $objects['object_ids'];
-		$status['last_sent'] = end( $object_ids );
-		$status['sent']     += count( $object_ids );
-		return $status;
-	}
 }

jetpack_vendor/automattic/jetpack-sync/src/modules/class-woocommerce-hpos-orders.php

@@ -145,7 +145,7 @@ public function init_full_sync_listeners( $callable ) {
 	 */
 	public function init_before_send() {
 		// Full sync.
-		add_filter( 'jetpack_sync_before_send_jetpack_full_sync_woocommerce_hpos_orders', array( $this, 'expand_order_objects' ) );
+		add_filter( 'jetpack_sync_before_send_jetpack_full_sync_woocommerce_hpos_orders', array( $this, 'build_full_sync_action_array' ) );
 	}
 
 	/**
@@ -228,15 +228,34 @@ public function get_objects_by_id( $object_type, $ids ) {
 	 * @param array $args List of order IDs.
 	 *
 	 * @return array
+	 * @deprecated since 4.7.0-alpha
 	 */
 	public function expand_order_objects( $args ) {
+		_deprecated_function( __METHOD__, 'next-version' );
 		list( $order_ids, $previous_end ) = $args;
 		return array(
 			'orders'       => $this->get_objects_by_id( 'order', $order_ids ),
 			'previous_end' => $previous_end,
 		);
 	}
 
+	/**
+	 * Build the full sync action object.
+	 *
+	 * @access public
+	 *
+	 * @param array $args An array with filtered objects and previous end.
+	 *
+	 * @return array An array with orders and previous end.
+	 */
+	public function build_full_sync_action_array( $args ) {
+		list( $filtered_orders, $previous_end ) = $args;
+		return array(
+			'orders'       => $filtered_orders['objects'],
+			'previous_end' => $previous_end,
+		);
+	}
+
 	/**
 	 * Retrieve order data by its ID.
 	 *
@@ -503,4 +522,47 @@ public function get_where_sql( $config ) {
 		$where_sql = $wpdb->prepare( "type IN ( $order_type_placeholder )", $order_types );
 		return "{$parent_where} AND {$where_sql}";
 	}
+
+	/**
+	 * Given the Module Configuration and Status return the next chunk of items to send.
+	 * This function also expands the posts and metadata and filters them based on the maximum size constraints.
+	 *
+	 * @param array $config This module Full Sync configuration.
+	 * @param array $status This module Full Sync status.
+	 * @param int   $chunk_size Chunk size.
+	 *
+	 * @return array
+	 */
+	public function get_next_chunk( $config, $status, $chunk_size ) {
+
+		$order_ids = parent::get_next_chunk( $config, $status, $chunk_size );
+
+		if ( empty( $order_ids ) ) {
+			return array();
+		}
+
+		$orders = $this->get_objects_by_id( 'order', $order_ids );
+
+		// If no orders were fetched, make sure to return the expected structure so that status is updated correctly.
+		if ( empty( $orders ) ) {
+			return array(
+				'object_ids' => $order_ids,
+				'objects'    => array(),
+			);
+		}
+
+		// Filter the orders based on the maximum size constraints. We don't need to filter metadata here since we don't sync it for hpos.
+		list( $filtered_order_ids, $filtered_orders, ) = $this->filter_objects_and_metadata_by_size(
+			'order',
+			$orders,
+			array(),
+			0,
+			self::MAX_SIZE_FULL_SYNC
+		);
+
+		return array(
+			'object_ids' => $filtered_order_ids,
+			'objects'    => $filtered_orders,
+		);
+	}
 }