Add check for archive pages to make sure indexables are only built when needed.

inc/class-upgrade.php

@@ -88,6 +88,7 @@ public function __construct() {
 			'20.2-RC0'   => 'upgrade_202',
 			'20.5-RC0'   => 'upgrade_205',
 			'20.7-RC0'   => 'upgrade_207',
+			'20.8-RC0'   => 'upgrade_208',
 		];
 
 		array_walk( $routines, [ $this, 'run_upgrade_routine' ], $version );
@@ -997,11 +998,26 @@ private function upgrade_205() {
 	/**
 	 * Performs the 20.7 upgrade routine.
 	 * Removes the metadata related to the settings page introduction modal for all the users.
+	 * Also, schedules another cleanup scheduled action.
 	 */
 	private function upgrade_207() {
+		if ( ! \wp_next_scheduled( Cleanup_Integration::START_HOOK ) ) {
+			\wp_schedule_single_event( ( time() + ( MINUTE_IN_SECONDS * 5 ) ), Cleanup_Integration::START_HOOK );
+		}
+
 		add_action( 'shutdown', [ $this, 'delete_user_introduction_meta' ] );
 	}
 
+	/**
+	 * Performs the 20.8 upgrade routine.
+	 * Schedules another cleanup scheduled action.
+	 */
+	private function upgrade_208() {
+		if ( ! \wp_next_scheduled( Cleanup_Integration::START_HOOK ) ) {
+			\wp_schedule_single_event( ( time() + ( MINUTE_IN_SECONDS * 5 ) ), Cleanup_Integration::START_HOOK );
+		}
+	}
+
 	/**
 	 * Sets the home_url option for the 15.1 upgrade routine.
 	 *

src/actions/indexing/indexable-post-type-archive-indexation-action.php

@@ -15,6 +15,7 @@
  */
 class Indexable_Post_Type_Archive_Indexation_Action implements Indexation_Action_Interface, Limited_Indexing_Action_Interface {
 
+
 	/**
 	 * The transient cache key.
 	 */
@@ -52,9 +53,9 @@ class Indexable_Post_Type_Archive_Indexation_Action implements Indexation_Action
 	 * Indexation_Post_Type_Archive_Action constructor.
 	 *
 	 * @param Indexable_Repository       $repository The indexable repository.
-	 * @param Indexable_Builder          $builder    The indexable builder.
-	 * @param Post_Type_Helper           $post_type  The post type helper.
-	 * @param Indexable_Builder_Versions $versions   The current versions of all indexable builders.
+	 * @param Indexable_Builder          $builder The indexable builder.
+	 * @param Post_Type_Helper           $post_type The post type helper.
+	 * @param Indexable_Builder_Versions $versions The current versions of all indexable builders.
 	 */
 	public function __construct(
 		Indexable_Repository $repository,
@@ -164,8 +165,7 @@ protected function get_unindexed_post_type_archives( $limit = false ) {
 	 */
 	protected function get_post_types_with_archive_pages() {
 		// We only want to index archive pages of public post types that have them.
-		$public_post_types       = $this->post_type->get_public_post_types( 'object' );
-		$post_types_with_archive = \array_filter( $public_post_types, [ $this->post_type, 'has_archive' ] );
+		$post_types_with_archive = $this->post_type->get_indexable_post_archives();
 
 		// We only need the post type names, not the objects.
 		$post_types = [];

src/builders/indexable-post-type-archive-builder.php

@@ -3,8 +3,10 @@
 namespace Yoast\WP\SEO\Builders;
 
 use wpdb;
+use Yoast\WP\SEO\Exceptions\Indexable\Post_Type_Not_Built_Exception;
 use Yoast\WP\SEO\Helpers\Options_Helper;
 use Yoast\WP\SEO\Helpers\Post_Helper;
+use Yoast\WP\SEO\Helpers\Post_Type_Helper;
 use Yoast\WP\SEO\Models\Indexable;
 use Yoast\WP\SEO\Values\Indexables\Indexable_Builder_Versions;
 
@@ -30,12 +32,19 @@ class Indexable_Post_Type_Archive_Builder {
 	protected $version;
 
 	/**
-	 * Holds the taxonomy helper instance.
+	 * Holds the post helper instance.
 	 *
 	 * @var Post_Helper
 	 */
 	protected $post_helper;
 
+	/**
+	 * Holds the post type helper instance.
+	 *
+	 * @var Post_Type_Helper
+	 */
+	protected $post_type_helper;
+
 	/**
 	 * The WPDB instance.
 	 *
@@ -46,21 +55,24 @@ class Indexable_Post_Type_Archive_Builder {
 	/**
 	 * Indexable_Post_Type_Archive_Builder constructor.
 	 *
-	 * @param Options_Helper             $options     The options helper.
-	 * @param Indexable_Builder_Versions $versions    The latest version of each Indexable builder.
+	 * @param Options_Helper             $options The options helper.
+	 * @param Indexable_Builder_Versions $versions The latest version of each Indexable builder.
 	 * @param Post_Helper                $post_helper The post helper.
-	 * @param wpdb                       $wpdb        The WPDB instance.
+	 * @param Post_Type_Helper           $post_type_helper The post type helper.
+	 * @param wpdb                       $wpdb The WPDB instance.
 	 */
 	public function __construct(
 		Options_Helper $options,
 		Indexable_Builder_Versions $versions,
 		Post_Helper $post_helper,
+		Post_Type_Helper $post_type_helper,
 		wpdb $wpdb
 	) {
-		$this->options     = $options;
-		$this->version     = $versions->get_latest_version_for_type( 'post-type-archive' );
-		$this->post_helper = $post_helper;
-		$this->wpdb        = $wpdb;
+		$this->options          = $options;
+		$this->version          = $versions->get_latest_version_for_type( 'post-type-archive' );
+		$this->post_helper      = $post_helper;
+		$this->post_type_helper = $post_type_helper;
+		$this->wpdb             = $wpdb;
 	}
 
 	/**
@@ -70,8 +82,13 @@ public function __construct(
 	 * @param Indexable $indexable The indexable to format.
 	 *
 	 * @return Indexable The extended indexable.
+	 * @throws \Yoast\WP\SEO\Exceptions\Indexable\Post_Type_Not_Built_Exception Throws exception if the post type is excluded.
 	 */
 	public function build( $post_type, Indexable $indexable ) {
+		if ( ! $this->post_type_helper->is_post_type_archive_indexable( $post_type ) ) {
+			throw Post_Type_Not_Built_Exception::because_not_indexable( $post_type );
+		}
+
 		$indexable->object_type       = 'post-type-archive';
 		$indexable->object_sub_type   = $post_type;
 		$indexable->title             = $this->options->get( 'title-ptarchive-' . $post_type );

src/exceptions/indexable/post-type-not-built-exception.php

@@ -0,0 +1,22 @@
+<?php
+
+namespace Yoast\WP\SEO\Exceptions\Indexable;
+
+/**
+ * Exception that is thrown whenever a post type could not be built
+ * in the context of the indexables.
+ */
+class Post_Type_Not_Built_Exception extends Not_Built_Exception {
+
+	/**
+	 * Throws an exception if the post is not indexable.
+	 *
+	 * @param string $post_type The post type.
+	 *
+	 * @throws Post_Type_Not_Built_Exception When the post type is not indexable.
+	 */
+	public static function because_not_indexable( $post_type ) {
+		/* translators: %s: expands to the post type */
+		return new Post_Type_Not_Built_Exception( \sprintf( \__( 'The post type %s could not be indexed because it does not meet indexing requirements.', 'wordpress-seo' ), $post_type ) );
+	}
+}

src/helpers/post-helper.php

@@ -177,9 +177,9 @@ public function update_has_public_posts_on_attachments( $post_parent, $has_publi
 	 */
 	public function is_post_indexable( $post_id ) {
 		// Don't index posts which are not public (i.e. viewable).
-		$post_type    = \get_post_type( $post_id );
-		$public_types = $this->post_type->get_indexable_post_types();
-		if ( ! \in_array( $post_type, $public_types, true ) ) {
+		$post_type = \get_post_type( $post_id );
+
+		if ( ! $this->post_type->is_of_indexable_post_type( $post_type ) ) {
 			return false;
 		}
 

src/helpers/post-type-helper.php

@@ -155,6 +155,15 @@ public function get_indexable_post_types() {
 		return $this->filter_included_post_types( $included_post_types );
 	}
 
+	/**
+	 * Returns all indexable post types with archive pages.
+	 *
+	 * @return array All post types which are indexable and have archive pages.
+	 */
+	public function get_indexable_post_archives() {
+		return \array_filter( $this->get_indexable_post_type_objects(), [ $this, 'has_archive' ] );
+	}
+
 	/**
 	 * Filters the post types that are included to be indexed.
 	 *
@@ -187,6 +196,41 @@ protected function filter_included_post_types( $included_post_types ) {
 		return \array_values( $filtered_included_post_types );
 	}
 
+	/**
+	 * Checks if the given post type should be indexed.
+	 *
+	 * @param string $post_type The post type that is checked.
+	 *
+	 * @return bool
+	 */
+	public function is_of_indexable_post_type( $post_type ) {
+		$public_types = $this->get_indexable_post_types();
+		if ( ! \in_array( $post_type, $public_types, true ) ) {
+			return false;
+		}
+
+		return true;
+	}
+
+	/**
+	 * Checks if the archive of a post type is indexable.
+	 *
+	 * @param string $post_type The post type to check.
+	 *
+	 * @return bool if the archive is indexable.
+	 */
+	public function is_post_type_archive_indexable( $post_type ) {
+		$public_type_objects = $this->get_indexable_post_archives();
+		$public_types        = \array_map(
+			static function ( $post_type_object ) {
+				return $post_type_object->name;
+			},
+			$public_type_objects
+		);
+
+		return \in_array( $post_type, $public_types, true );
+	}
+
 	/**
 	 * Returns an array of complete post type objects for all indexable post types.
 	 *

src/integrations/cleanup-integration.php

@@ -3,6 +3,7 @@
 namespace Yoast\WP\SEO\Integrations;
 
 use Closure;
+use Yoast\WP\Lib\Model;
 use Yoast\WP\SEO\Repositories\Indexable_Cleanup_Repository;
 
 /**
@@ -112,6 +113,9 @@ public function get_cleanup_tasks() {
 				'clean_indexables_for_non_publicly_viewable_taxonomies' => function ( $limit ) {
 					return $this->cleanup_repository->clean_indexables_for_non_publicly_viewable_taxonomies( $limit );
 				},
+				'clean_indexables_for_non_publicly_viewable_post_type_archive_pages' => function ( $limit ) {
+					return $this->cleanup_repository->clean_indexables_for_non_publicly_viewable_post_type_archive_pages( $limit );
+				},
 				'clean_indexables_for_authors_archive_disabled' => function ( $limit ) {
 					return $this->cleanup_repository->clean_indexables_for_authors_archive_disabled( $limit );
 				},

src/repositories/indexable-cleanup-repository.php

@@ -239,6 +239,54 @@ public function clean_indexables_for_non_publicly_viewable_taxonomies( $limit )
 		// phpcs:enable
 	}
 
+	/**
+	 * Cleans up any indexables that belong to post type archive page that are not/no longer publicly viewable.
+	 *
+	 * @param int $limit The limit we'll apply to the queries.
+	 *
+	 * @return bool|int The number of deleted rows, false if the query fails.
+	 */
+	public function clean_indexables_for_non_publicly_viewable_post_type_archive_pages( $limit ) {
+		global $wpdb;
+		$indexable_table = Model::get_table_name( 'Indexable' );
+
+		$included_post_types = $this->post_type->get_indexable_post_archives();
+
+		$post_archives = [];
+
+		foreach ( $included_post_types as $post_type ) {
+			$post_archives[] = $post_type->name;
+		}
+		// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- Reason: Too hard to fix.
+		if ( empty( $post_archives ) ) {
+			$delete_query = $wpdb->prepare(
+				"DELETE FROM $indexable_table
+				WHERE object_type = 'post-type-archive'
+				AND object_sub_type IS NOT NULL
+				LIMIT %d",
+				$limit
+			);
+		}
+		else {
+			// phpcs:ignore WordPress.DB.PreparedSQLPlaceholders.ReplacementsWrongNumber -- Reason: we're passing an array instead.
+			$delete_query = $wpdb->prepare(
+				"DELETE FROM $indexable_table
+				WHERE object_type = 'post-type-archive'
+				AND object_sub_type IS NOT NULL
+				AND object_sub_type NOT IN ( " . \implode( ', ', \array_fill( 0, \count( $post_archives ), '%s' ) ) . ' )
+				LIMIT %d',
+				\array_merge( $post_archives, [ $limit ] )
+			);
+		}
+		// phpcs:enable
+
+		// phpcs:disable WordPress.DB.DirectDatabaseQuery.NoCaching -- Reason: No relevant caches.
+		// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery -- Reason: Most performant way.
+		// phpcs:disable WordPress.DB.PreparedSQL.NotPrepared -- Reason: Is it prepared already.
+		return $wpdb->query( $delete_query );
+		// phpcs:enable
+	}
+
 	/**
 	 * Counts indexables for non publicly viewable taxonomies.
 	 *