Merge branch 'main' into feature/skip-sitemap-constant

README.md

@@ -54,6 +54,17 @@ See 'wp help msm-sitemap <command>' for more information on a specific command.
 
 Include custom post types in the generated sitemap with the `msm_sitemap_entry_post_type` filter.
 
+## Generate Sitemap with posts of a custom status other than 'publish'
+
+By default, the sitemap will only fetch posts with the status of 'publish'. To change this, use the `msm_sitemap_post_status` filter.
+
+```
+function example_filter_msm_sitemap_post_status( $post_status ) {
+    return 'my_custom_status';
+}
+add_filter( 'msm_sitemap_post_status', 'example_filter_msm_sitemap_post_status', 10, 1 );
+```
+
 ## Filtering Sitemap URLs
 
 If you need to filter the URLs displayed in a sitemap created via the Comprehensive Sitemap plugin, there are two considerations. First, if you are filtering the individual sitemaps, which display the URLs to the articles published on a specific date, you can use the `msm_sitemap_entry` hook to filter the URLs. An example for a reverse-proxy situation is below:
@@ -83,3 +94,19 @@ add_filter( 'msm_sitemap_index', function( $sitemaps ) {
     } );
 } );
 ```
+
+## Customize the last modified posts query
+
+Use the `msm_pre_get_last_modified_posts` filter to customize the query that gets the last modified posts.
+
+On large sites, this filter could be leveraged to enhance query efficiency by avoiding scanning older posts that don't get updated frequently and making better use of the `type_status_date` index.
+
+```
+function ( $query, $post_types_in, $date ) {
+    global $wpdb;
+
+    $query = $wpdb->prepare( "SELECT ID, post_date FROM $wpdb->posts WHERE post_type IN ( {$post_types_in} ) AND post_status = 'publish' AND post_date >= DATE_SUB(NOW(), INTERVAL 3 MONTH) AND post_modified_gmt >= %s LIMIT 1000", $date );
+
+    return $query;
+};
+```

msm-sitemap.php

@@ -126,7 +126,7 @@ public static function ajax_get_sitemap_counts() {
 
 		$data = array(
 			'total_indexed_urls'   => number_format( Metro_Sitemap::get_total_indexed_url_count() ),
-			'total_sitemaps'	   => number_format( Metro_Sitemap::count_sitemaps() ),
+			'total_sitemaps'       => number_format( Metro_Sitemap::count_sitemaps() ),
 			'sitemap_indexed_urls' => self::get_recent_sitemap_url_counts( $n ),
 		);
 
@@ -158,7 +158,7 @@ public static function render_sitemap_options_page() {
 		if ( isset( $_POST['action'] ) ) {
 			check_admin_referer( 'msm-sitemap-action' );
 			foreach ( $actions as $slug => $action ) {
-				if ( $action['text'] !== $_POST['action'] )	continue;
+				if ( $action['text'] !== $_POST['action'] ) continue;
 
 				do_action( 'msm_sitemap_action-' . $slug );
 				break;
@@ -325,14 +325,36 @@ public static function disable_canonical_redirects_for_sitemap_xml( $redirect_ur
 		return $redirect_url;
 	}
 
+	/**
+	 * Hook allows developers to extend the sitemap functionality easily and integrate their custom post statuses.
+	 *
+	 * Rather than having to modify the plugin code, developers can use this filter to add their custom post statuses.
+	 *
+	 * @since 1.4.3
+	 *
+	 */
+	public static function get_post_status(): string {
+		$default_status = 'publish';
+		$post_status = apply_filters('msm_sitemap_post_status', $default_status);
+
+		$allowed_statuses = get_post_stati();
+
+		if (!in_array($post_status, $allowed_statuses)) {
+			$post_status = $default_status;
+		}
+
+		return $post_status;
+	}
+
 	/**
 	 * Return range of years for posts in the database
 	 * @return int[] valid years
 	 */
 	public static function get_post_year_range() {
 		global $wpdb;
+		$post_status = self::get_post_status();
 
-		$oldest_post_date_year = $wpdb->get_var( "SELECT DISTINCT YEAR(post_date) as year FROM $wpdb->posts WHERE post_status = 'publish' AND post_date > 0 ORDER BY year ASC LIMIT 1" );
+		$oldest_post_date_year = $wpdb->get_var( $wpdb->prepare( "SELECT DISTINCT YEAR(post_date) as year FROM $wpdb->posts WHERE post_status = %s AND post_date > 0 ORDER BY year ASC LIMIT 1", $post_status ) );
 
 		if ( null !== $oldest_post_date_year ) {
 			$current_year = date( 'Y' );
@@ -383,9 +405,10 @@ public static function date_range_has_posts( $start_date, $end_date ) {
 
 		$start_date .= ' 00:00:00';
 		$end_date .= ' 23:59:59';
+		$post_status = self::get_post_status();
 
 		$post_types_in = self::get_supported_post_types_in();
-		return $wpdb->get_var( $wpdb->prepare( "SELECT ID FROM $wpdb->posts WHERE post_status = 'publish' AND post_date >= %s AND post_date <= %s AND post_type IN ( {$post_types_in} ) LIMIT 1", $start_date, $end_date ) );
+		return $wpdb->get_var( $wpdb->prepare( "SELECT ID FROM $wpdb->posts WHERE post_status = %s AND post_date >= %s AND post_date <= %s AND post_type IN ( {$post_types_in} ) LIMIT 1", $post_status, $start_date, $end_date ) );
 	}
 
 	/**
@@ -398,11 +421,12 @@ public static function date_range_has_posts( $start_date, $end_date ) {
 	public static function get_post_ids_for_date( $sitemap_date, $limit = 500 ) {
 		global $wpdb;
 
+		$post_status = self::get_post_status();
 		$start_date = $sitemap_date . ' 00:00:00';
 		$end_date = $sitemap_date . ' 23:59:59';
 		$post_types_in = self::get_supported_post_types_in();
 
-		$posts = $wpdb->get_results( $wpdb->prepare( "SELECT ID, post_date FROM $wpdb->posts WHERE post_status = 'publish' AND post_date >= %s AND post_date <= %s AND post_type IN ( {$post_types_in} ) LIMIT %d", $start_date, $end_date, $limit ) );
+		$posts = $wpdb->get_results( $wpdb->prepare( "SELECT ID, post_date FROM $wpdb->posts WHERE post_status = %s AND post_date >= %s AND post_date <= %s AND post_type IN ( {$post_types_in} ) LIMIT %d", $post_status, $start_date, $end_date, $limit ) );
 
 		usort( $posts, array( __CLASS__ , 'order_by_post_date' ) );
 
@@ -427,7 +451,7 @@ public static function generate_sitemap_for_date( $sitemap_date ) {
 			'post_name' => $sitemap_name,
 			'post_title' => $sitemap_name,
 			'post_type' => self::SITEMAP_CPT,
-			'post_status' => 'publish',
+			'post_status' => self::get_post_status(),
 			'post_date' => $sitemap_date,
 			);
 
@@ -497,23 +521,25 @@ public static function generate_sitemap_for_date( $sitemap_date ) {
 			// TODO: add images to sitemap via <image:image> tag
 		}
 
-				// Save the sitemap
+		$generated_xml_string = $xml->asXML();
+
+		// Save the sitemap
 		if ( $sitemap_exists ) {
 			// Get the previous post count
 			$previous_url_count = intval( get_post_meta( $sitemap_id, 'msm_indexed_url_count', true ) );
 
 			// Update the total post count with the difference
 			$total_url_count += $url_count - $previous_url_count;
 
-			update_post_meta( $sitemap_id, 'msm_sitemap_xml', $xml->asXML() );
+			update_post_meta( $sitemap_id, 'msm_sitemap_xml', $generated_xml_string );
 			update_post_meta( $sitemap_id, 'msm_indexed_url_count', $url_count );
-			do_action( 'msm_update_sitemap_post', $sitemap_id, $year, $month, $day );
+			do_action( 'msm_update_sitemap_post', $sitemap_id, $year, $month, $day, $generated_xml_string, $url_count );
 		} else {
 			/* Should no longer hit this */
 			$sitemap_id = wp_insert_post( $sitemap_data );
-			add_post_meta( $sitemap_id, 'msm_sitemap_xml', $xml->asXML() );
+			add_post_meta( $sitemap_id, 'msm_sitemap_xml', $generated_xml_string );
 			add_post_meta( $sitemap_id, 'msm_indexed_url_count', $url_count );
-			do_action( 'msm_insert_sitemap_post', $sitemap_id, $year, $month, $day );
+			do_action( 'msm_insert_sitemap_post', $sitemap_id, $year, $month, $day, $generated_xml_string, $url_count );
 
 			// Update the total url count
 			$total_url_count += $url_count;
@@ -591,7 +617,20 @@ public static function get_last_modified_posts() {
 
 		$post_types_in = self::get_supported_post_types_in();
 
-		$modified_posts = $wpdb->get_results( $wpdb->prepare( "SELECT ID, post_date FROM $wpdb->posts WHERE post_type IN ( {$post_types_in} ) AND post_modified_gmt >= %s LIMIT 1000", $date ) );
+		$query = $wpdb->prepare( "SELECT ID, post_date FROM $wpdb->posts WHERE post_type IN ( {$post_types_in} ) AND post_modified_gmt >= %s LIMIT 1000", $date );
+
+		/**
+		 * Filter the query used to get the last modified posts.
+		 * $wpdb->prepare() should be used for security if a new replacement query is created in the callback.
+		 * 
+		 * @param string $query         The query to use to get the last modified posts.
+		 * @param string $post_types_in A comma-separated list of post types to include in the query.
+		 * @param string $date          The date to use as the cutoff for the query.
+		 */
+		$query = apply_filters( 'msm_pre_get_last_modified_posts', $query, $post_types_in, $date );
+
+		$modified_posts = $wpdb->get_results( $query );
+
 		return $modified_posts;
 	}