From 2db137784b204b2ce8fdde7b13c0af02350c81f5 Mon Sep 17 00:00:00 2001
From: Peter Wilson <wilson@peterwilson.cc>
Date: Thu, 27 Aug 2020 01:30:04 +0000
Subject: [PATCH] Sitemaps: Prevent incorrect redirection of paged sitemap
 requests.

Update `redirect_canonical()` to account for custom pagination and URL format used by sitemaps in order to follow standard practices.

Introduce the function `get_sitemap_url()` to simplify getting the index and provider URLs as needed.

Props jonathanstegall, pbiron, GamerZ, salvoaranzulla, peterwilsoncc.
Fixes #50910.


Built from https://develop.svn.wordpress.org/trunk@48872


git-svn-id: http://core.svn.wordpress.org/trunk@48634 1a063a9b-81f0-0310-95a4-ce76da25c4cd
---
 wp-includes/canonical.php | 12 +++++-------
 wp-includes/sitemaps.php  | 36 ++++++++++++++++++++++++++++++++++++
 wp-includes/version.php   |  2 +-
 3 files changed, 42 insertions(+), 8 deletions(-)

diff --git a/wp-includes/canonical.php b/wp-includes/canonical.php
index 5a63de3481..6a74ac2d43 100644
--- a/wp-includes/canonical.php
+++ b/wp-includes/canonical.php
@@ -410,8 +410,11 @@ function redirect_canonical( $requested_url = null, $do_redirect = true ) {
 			$redirect['query'] = remove_query_arg( 'page', $redirect['query'] );
 		}
 
-		// Paging and feeds.
-		if ( get_query_var( 'paged' ) || is_feed() || get_query_var( 'cpage' ) ) {
+		if ( get_query_var( 'sitemap' ) ) {
+			$redirect_url      = get_sitemap_url( get_query_var( 'sitemap' ), get_query_var( 'sitemap-subtype' ), get_query_var( 'paged' ) );
+			$redirect['query'] = remove_query_arg( array( 'sitemap', 'sitemap-subtype', 'paged' ), $redirect['query'] );
+		} elseif ( get_query_var( 'paged' ) || is_feed() || get_query_var( 'cpage' ) ) {
+			// Paging and feeds.
 			$paged = get_query_var( 'paged' );
 			$feed  = get_query_var( 'feed' );
 			$cpage = get_query_var( 'cpage' );
@@ -508,11 +511,6 @@ function redirect_canonical( $requested_url = null, $do_redirect = true ) {
 				$redirect['path'] = trailingslashit( $redirect['path'] ) . $addl_path;
 			}
 
-			// Remove trailing slash for sitemaps requests.
-			if ( ! empty( get_query_var( 'sitemap' ) ) ) {
-				$redirect['path'] = untrailingslashit( $redirect['path'] );
-			}
-
 			$redirect_url = $redirect['scheme'] . '://' . $redirect['host'] . $redirect['path'];
 		}
 
diff --git a/wp-includes/sitemaps.php b/wp-includes/sitemaps.php
index 62752dada9..1dfcd1d69c 100644
--- a/wp-includes/sitemaps.php
+++ b/wp-includes/sitemaps.php
@@ -87,3 +87,39 @@ function wp_sitemaps_get_max_urls( $object_type ) {
 	 */
 	return apply_filters( 'wp_sitemaps_max_urls', 2000, $object_type );
 }
+
+/**
+ * Retrieves the full URL for a sitemap.
+ *
+ * @since 5.5.1
+ *
+ * @param string $name         The sitemap name.
+ * @param string $subtype_name The sitemap subtype name.  Default empty string.
+ * @param int    $page         The page of the sitemap.  Default 1.
+ * @return string|false The sitemap URL or false if the sitemap doesn't exist.
+ */
+function get_sitemap_url( $name, $subtype_name = '', $page = 1 ) {
+	$sitemaps = wp_sitemaps_get_server();
+	if ( ! $sitemaps ) {
+		return false;
+	}
+
+	if ( 'index' === $name ) {
+		return $sitemaps->index->get_index_url();
+	}
+
+	$provider = $sitemaps->registry->get_provider( $name );
+	if ( ! $provider ) {
+		return false;
+	}
+
+	if ( $subtype_name && ! in_array( $subtype_name, array_keys( $provider->get_object_subtypes() ), true ) ) {
+		return false;
+	}
+
+	$page = absint( $page );
+	if ( 0 >= $page ) {
+		$page = 1;
+	}
+	return $provider->get_sitemap_url( $subtype_name, $page );
+}
diff --git a/wp-includes/version.php b/wp-includes/version.php
index 483278d7af..3a122aba4f 100644
--- a/wp-includes/version.php
+++ b/wp-includes/version.php
@@ -13,7 +13,7 @@
  *
  * @global string $wp_version
  */
-$wp_version = '5.6-alpha-48870';
+$wp_version = '5.6-alpha-48872';
 
 /**
  * Holds the WordPress DB revision, increments when changes are made to the WordPress DB schema.