Add method for creating a noindex query for indexables

src/builders/indexable-author-builder.php

@@ -87,22 +87,27 @@ public function build( $user_id, Indexable $indexable ) {
  $indexable->is_robots_noarchive = null;
  $indexable->is_robots_noimageindex = null;
  $indexable->is_robots_nosnippet = null;
- $indexable->is_public = ( $indexable->is_robots_noindex ) ? false : null;
- $indexable->has_public_posts = $this->author_archive->author_has_public_posts( $user_id );
  $indexable->blog_id = \get_current_blog_id();
 
  $this->reset_social_images( $indexable );
  $this->handle_social_images( $indexable );
 
- $timestamps = $this->get_object_timestamps( $user_id );
- $indexable->object_published_at = $timestamps->published_at;
- $indexable->object_last_modified = $timestamps->last_modified;
+ $indexable = $this->set_aggregate_values( $indexable );
 
  $indexable->version = $this->version;
 
  return $indexable;
  }
 
+ public function set_aggregate_values( Indexable $indexable ) {
+ $aggregates = $this->get_public_post_archive_aggregates( $indexable->object_id );
+ $indexable->object_published_at = $aggregates->first_published_at;
+ $indexable->object_last_modified = max($indexable->object_last_modified, $aggregates->most_recent_last_modified);
+ $indexable->number_of_public_posts = $aggregates->number_of_public_posts;
+
+ return $indexable;
+ }
+
  /**
  * Retrieves the meta data for this indexable.
  *
@@ -168,24 +173,31 @@ protected function find_alternative_image( Indexable $indexable ) {
  }
 
  /**
- * Returns the timestamps for a given author.
+ * Returns public post aggregates for a given author.
  *
- * @param int $author_id  The author ID.
+ * @param int $author_id The author ID.
  *
- * @return object An object with last_modified and published_at timestamps.
+ * @return object  An object with the number of public posts, most recent last modified and first published at timestamps.
  */
- protected function get_object_timestamps( $author_id ) {
+ protected function get_public_post_archive_aggregates( $author_id ) {
  $post_statuses = $this->post_helper->get_public_post_statuses();
+ $post_types = $this->author_archive->get_author_archive_post_types();
+ // TODO DIEDE: Protected pages krijgen _geen_ noindex. en staan gewoon in het author archive. moeten die meegeteld worden?
+ // Private werkt wel zoals verwacht.
 
  $sql = "
- SELECT MAX(p.post_modified_gmt) AS last_modified, MIN(p.post_date_gmt) AS published_at
+ SELECT 
+ COUNT(p.ID) as number_of_public_posts,
+ MAX(p.post_modified_gmt) AS most_recent_last_modified,
+ MIN(p.post_date_gmt) AS first_published_at
  FROM {$this->wpdb->posts} AS p
  WHERE p.post_status IN (" . implode( ', ', array_fill( 0, count( $post_statuses ), '%s' ) ) . ")
  AND p.post_password = ''
  AND p.post_author = %d
+ AND p.post_type IN (" . implode( ', ', array_fill( 0, count( $post_types ), '%s' ) ) . ")
  ";
 
- $replacements = \array_merge( $post_statuses, [ $author_id ] );
+ $replacements = \array_merge( $post_statuses, [ $author_id ], $post_types );
 
  // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared -- We are using wpdb prepare.
  return $this->wpdb->get_row( $this->wpdb->prepare( $sql, $replacements ) );

src/builders/indexable-builder.php

@@ -219,6 +219,7 @@ public function build_for_post_type_archive( $post_type, $indexable = false ) {
  'object_type' => 'post-type-archive',
  'object_sub_type' => $post_type,
  ];
+
  return $this->build( $indexable, $defaults );
  }
 
@@ -235,6 +236,7 @@ public function build_for_system_page( $page_type, $indexable = false ) {
  'object_type' => 'system-page',
  'object_sub_type' => $page_type,
  ];
+
  return $this->build( $indexable, $defaults );
  }
 
@@ -372,8 +374,7 @@ public function build( $indexable, $defaults = null ) {
  }
 
  return $this->save_indexable( $indexable, $indexable_before );
- }
- catch ( Source_Exception $exception ) {
+ } catch ( Source_Exception $exception ) {
  /**
  * The current indexable could not be indexed. Create a placeholder indexable, so we can
  * skip this indexable in future indexing runs.
@@ -396,4 +397,43 @@ public function build( $indexable, $defaults = null ) {
  return $this->save_indexable( $indexable, $indexable_before );
  }
  }
+
+ /**
+ * Recalculates indexable aggregates.
+ *
+ * @param Indexable $indexable The Indexable to (re)build.
+ *
+ * @return Indexable The resulting Indexable.
+ */
+ public function recalculate_aggregates( Indexable $indexable ) {
+ // Backup the previous Indexable, if there was one.
+ $indexable_before = $this->deep_copy_indexable( $indexable );
+
+ switch ( $indexable->object_type ) {
+ case 'system-page':
+ case 'date-archive':
+ case 'post':
+ // Nothing to recalculate.
+ break;
+
+ case 'user':
+ $indexable = $this->author_builder->set_aggregate_values( $indexable );
+ break;
+
+ case 'term':
+ $indexable = $this->term_builder->set_aggregate_values( $indexable );
+ break;
+
+ case 'home-page':
+ $indexable = $this->home_page_builder->set_aggregate_values( $indexable );
+ break;
+
+ case 'post-type-archive':
+ $indexable = $this->post_type_archive_builder->set_aggregate_values( $indexable );
+ break;
+ }
+
+ return $this->save_indexable( $indexable, $indexable_before );
+ }
 }
+

src/builders/indexable-date-archive-builder.php

@@ -51,14 +51,13 @@ public function __construct(
  * @return Indexable The extended indexable.
  */
  public function build( $indexable ) {
- $indexable->object_type = 'date-archive';
- $indexable->title = $this->options->get( 'title-archive-wpseo' );
- $indexable->description = $this->options->get( 'metadesc-archive-wpseo' );
- $indexable->is_robots_noindex = $this->options->get( 'noindex-archive-wpseo' );
- $indexable->is_public = ( (int) $indexable->is_robots_noindex !== 1 );
- $indexable->blog_id = \get_current_blog_id();
- $indexable->permalink = null;
- $indexable->version = $this->version;
+ $indexable->object_type = 'date-archive';
+ $indexable->title = $this->options->get( 'title-archive-wpseo' );
+ $indexable->description = $this->options->get( 'metadesc-archive-wpseo' );
+ $indexable->is_robots_noindex = $this->options->get( 'noindex-archive-wpseo' );
+ $indexable->blog_id = \get_current_blog_id();
+ $indexable->permalink = null;
+ $indexable->version = $this->version;
 
  return $indexable;
  }

src/builders/indexable-home-page-builder.php

@@ -114,25 +114,35 @@ public function build( $indexable ) {
  $this->set_open_graph_image_meta_data( $indexable );
  }
 
- $timestamps = $this->get_object_timestamps();
- $indexable->object_published_at = $timestamps->published_at;
- $indexable->object_last_modified = $timestamps->last_modified;
+ $indexable = $this->set_aggregate_values( $indexable );
 
  $indexable->version = $this->version;
 
  return $indexable;
  }
 
+ public function set_aggregate_values( Indexable $indexable ) {
+ $aggregates = $this->get_public_post_archive_aggregates();
+ $indexable->object_published_at = $aggregates->first_published_at;
+ $indexable->object_last_modified = $aggregates->most_recent_last_modified;
+ $indexable->number_of_public_posts = $aggregates->number_of_public_posts;
+
+ return $indexable;
+ }
+
  /**
- * Returns the timestamps for the homepage.
+ * Returns public post aggregates for the homepage.
  *
- * @return object An object with last_modified and published_at timestamps.
+ * @return object An object with the number of posts, most recent last modified and first published at timestamps.
  */
- protected function get_object_timestamps() {
+ protected function get_public_post_archive_aggregates() {
  $post_statuses = $this->post_helper->get_public_post_statuses();
 
  $sql = "
- SELECT MAX(p.post_modified_gmt) AS last_modified, MIN(p.post_date_gmt) AS published_at
+ SELECT 
+ COUNT(p.ID) as number_of_public_posts, 
+ MAX(p.post_modified_gmt) AS most_recent_last_modified, 
+ MIN(p.post_date_gmt) AS first_published_at 
  FROM {$this->wpdb->posts} AS p
  WHERE p.post_status IN (" . implode( ', ', array_fill( 0, count( $post_statuses ), '%s' ) ) . ")
  AND p.post_password = ''

src/builders/indexable-post-builder.php

@@ -142,19 +142,20 @@ public function build( $post_id, $indexable ) {
  $indexable->author_id = $post->post_author;
  $indexable->post_parent = $post->post_parent;
 
- $indexable->number_of_pages = $this->get_number_of_pages_for_post( $post );
- $indexable->post_status = $post->post_status;
- $indexable->is_protected = $post->post_password !== '';
- $indexable->is_public = $this->is_public( $indexable );
- $indexable->has_public_posts = $this->has_public_posts( $indexable );
- $indexable->blog_id = \get_current_blog_id();
+ $indexable->number_of_pages = $this->get_number_of_pages_for_post( $post );
+ $indexable->post_status = $post->post_status;
+ $indexable->is_protected = $post->post_password !== '';
+
+ $indexable->blog_id = \get_current_blog_id();
 
  $indexable->schema_page_type = $this->get_meta_value( $post_id, 'schema_page_type' );
  $indexable->schema_article_type = $this->get_meta_value( $post_id, 'schema_article_type' );
 
  $indexable->object_last_modified = $post->post_modified_gmt;
  $indexable->object_published_at = $post->post_date_gmt;
 
+ $indexable->number_of_public_posts = 0;
+
  $indexable->version = $this->version;
 
  return $indexable;
@@ -181,31 +182,10 @@ protected function get_permalink( $post_type, $post_id ) {
  *
  * @param Indexable $indexable The indexable.
  *
- * @return bool|null Whether or not the post type is public. Null if no override is set.
+ * @return bool|null Whether the post type is public. Null if no override is set.
  */
- protected function is_public( $indexable ) {
- if ( $indexable->is_protected === true ) {
- return false;
- }
-
- if ( $indexable->is_robots_noindex === true ) {
- return false;
- }
-
- // Attachments behave differently than the other post types, since they inherit from their parent.
- if ( $indexable->object_sub_type === 'attachment' ) {
- return $this->is_public_attachment( $indexable );
- }
-
- if ( ! \in_array( $indexable->post_status, $this->post_helper->get_public_post_statuses(), true ) ) {
- return false;
- }
-
- if ( $indexable->is_robots_noindex === false ) {
- return true;
- }
-
- return null;
+ protected function is_accessible_post( $indexable ) {
+ return is_post_type_viewable($indexable->object_sub_type) && is_post_status_viewable($indexable->post_status);
  }
 
  /**
@@ -225,38 +205,6 @@ protected function is_public_attachment( $indexable ) {
  return null;
  }
 
- /**
- * Determines the value of has_public_posts.
- *
- * @param Indexable $indexable The indexable.
- *
- * @return bool|null Whether the attachment has a public parent, can be true, false and null. Null when it is not an attachment.
- */
- protected function has_public_posts( $indexable ) {
- // Only attachments (and authors) have this value.
- if ( $indexable->object_sub_type !== 'attachment' ) {
- return null;
- }
-
- // The attachment should have a post parent.
- if ( empty( $indexable->post_parent ) ) {
- return false;
- }
-
- // The attachment should inherit the post status.
- if ( $indexable->post_status !== 'inherit' ) {
- return false;
- }
-
- // The post parent should be public.
- $post_parent_indexable = $this->indexable_repository->find_by_id_and_type( $indexable->post_parent, 'post' );
- if ( $post_parent_indexable !== false ) {
- return $post_parent_indexable->is_public;
- }
-
- return false;
- }
-
  /**
  * Converts the meta robots noindex value to the indexable value.
  *

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

@@ -80,14 +80,21 @@ public function build( $post_type, Indexable $indexable ) {
  $indexable->description = $this->options->get( 'metadesc-ptarchive-' . $post_type );
  $indexable->breadcrumb_title = $this->get_breadcrumb_title( $post_type );
  $indexable->permalink = \get_post_type_archive_link( $post_type );
- $indexable->is_robots_noindex = $this->options->get( 'noindex-ptarchive-' . $post_type );
- $indexable->is_public = ( (int) $indexable->is_robots_noindex !== 1 );
+ $indexable->is_robots_noindex = (bool) $this->options->get( 'noindex-ptarchive-' . $post_type );
  $indexable->blog_id = \get_current_blog_id();
- $indexable->version = $this->version;
 
- $timestamps = $this->get_object_timestamps( $post_type );
- $indexable->object_published_at = $timestamps->published_at;
- $indexable->object_last_modified = $timestamps->last_modified;
+ $indexable = $this->set_aggregate_values( $indexable );
+
+ $indexable->version = $this->version;
+
+ return $indexable;
+ }
+
+ public function set_aggregate_values( Indexable $indexable ) {
+ $aggregates = $this->get_public_post_archive_aggregates( $indexable->object_sub_type );
+ $indexable->object_published_at = $aggregates->published_at;
+ $indexable->object_last_modified = $aggregates->last_modified;
+ $indexable->number_of_public_posts = $aggregates->number_of_public_posts;
 
  return $indexable;
  }
@@ -124,17 +131,20 @@ private function get_breadcrumb_title( $post_type ) {
  }
 
  /**
- * Returns the timestamps for a given post type.
+ * Returns public post aggregates for a given post type.
  *
  * @param string $post_type The post type.
  *
- * @return object An object with last_modified and published_at timestamps.
+ * @return object An object with the number of posts, most recent last modified and first published at timestamps.
  */
- protected function get_object_timestamps( $post_type ) {
+ protected function get_public_post_archive_aggregates( $post_type ) {
  $post_statuses = $this->post_helper->get_public_post_statuses();
 
  $sql = "
- SELECT MAX(p.post_modified_gmt) AS last_modified, MIN(p.post_date_gmt) AS published_at
+ SELECT 
+ COUNT(p.ID) as number_of_public_posts,
+ MAX(p.post_modified_gmt) AS most_recent_last_modified,
+ MIN(p.post_date_gmt) AS first_published_at
  FROM {$this->wpdb->posts} AS p
  WHERE p.post_status IN (" . implode( ', ', array_fill( 0, count( $post_statuses ), '%s' ) ) . ")
  AND p.post_password = ''

src/builders/indexable-system-page-builder.php

@@ -20,7 +20,7 @@ class Indexable_System_Page_Builder {
  */
  const OPTION_MAPPING = [
  'search-result' => [
- 'title'  => 'title-search-wpseo',
+ 'title' => 'title-search-wpseo',
  ],
  '404' => [
  'title' => 'title-404-wpseo',
@@ -65,11 +65,12 @@ public function __construct(
  * @return Indexable The extended indexable.
  */
  public function build( $object_sub_type, Indexable $indexable ) {
- $indexable->object_type = 'system-page';
- $indexable->object_sub_type = $object_sub_type;
- $indexable->title = $this->options->get( static::OPTION_MAPPING[ $object_sub_type ]['title'] );
- $indexable->is_robots_noindex = true;
- $indexable->blog_id = \get_current_blog_id();
+ $indexable->object_type = 'system-page';
+ $indexable->object_sub_type = $object_sub_type;
+ $indexable->title = $this->options->get( static::OPTION_MAPPING[ $object_sub_type ]['title'] );
+ $indexable->is_robots_noindex = true;
+ $indexable->number_of_public_posts = 0;
+ $indexable->blog_id = \get_current_blog_id();
 
  if ( \array_key_exists( 'breadcrumb_title', static::OPTION_MAPPING[ $object_sub_type ] ) ) {
  $indexable->breadcrumb_title = $this->options->get( static::OPTION_MAPPING[ $object_sub_type ]['breadcrumb_title'] );