Add Sitemaps Functionality

phpcs.xml.dist

@@ -247,6 +247,7 @@
 				<element value="WP_Tests_Image_Resize_UnitTestCase"/>
 				<element value="WP_Import_UnitTestCase"/>
 				<element value="Tests_Query_Conditionals"/>
+				<element value="WP_Test_XML_TestCase"/>
 
 				<!-- Mock classes. -->
 				<element value="Spy_REST_Server"/>

src/wp-includes/canonical.php

@@ -509,6 +509,11 @@ 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'];
 		}
 
@@ -651,6 +656,11 @@ function redirect_canonical( $requested_url = null, $do_redirect = true ) {
 		$redirect['path'] = trailingslashit( $redirect['path'] );
 	}
 
+	// Remove trailing slash for sitemaps requests.
+	if ( ! empty( get_query_var( 'sitemap' ) ) || ! empty( get_query_var( 'sitemap-stylesheet' ) ) ) {
+		$redirect['path'] = untrailingslashit( $redirect['path'] );
+	}
+
 	// Strip multiple slashes out of the URL.
 	if ( strpos( $redirect['path'], '//' ) > -1 ) {
 		$redirect['path'] = preg_replace( '|/+|', '/', $redirect['path'] );

src/wp-includes/default-filters.php

@@ -456,6 +456,9 @@
 add_action( 'rest_api_init', 'create_initial_rest_routes', 99 );
 add_action( 'parse_request', 'rest_api_loaded' );
 
+// Sitemaps actions.
+add_action( 'init', 'wp_sitemaps_get_server' );
+
 /**
  * Filters formerly mixed into wp-includes.
  */

src/wp-includes/formatting.php

@@ -935,14 +935,18 @@ function seems_utf8( $str ) {
  * ", or ENT_QUOTES to do both. Default is ENT_NOQUOTES where no quotes are encoded.
  *
  * @since 1.2.2
+ * @since 5.5.0 `$quote_style` also accepts '`ENT_XML1`.
  * @access private
  *
  * @staticvar string $_charset
  *
  * @param string       $string        The text which is to be encoded.
  * @param int|string   $quote_style   Optional. Converts double quotes if set to ENT_COMPAT,
  *                                    both single and double if set to ENT_QUOTES or none if set to ENT_NOQUOTES.
- *                                    Also compatible with old values; converting single quotes if set to 'single',
+ *                                    Converts single and double quotes, as well as converting HTML
+ *                                    named entities (that are not also XML named entities) to their
+ *                                    code points if set to ENT_XML1. Also compatible with old values;
+ *                                    converting single quotes if set to 'single',
  *                                    double if set to 'double' or both if otherwise set.
  *                                    Default is ENT_NOQUOTES.
  * @param false|string $charset       Optional. The character encoding of the string. Default is false.
@@ -964,7 +968,9 @@ function _wp_specialchars( $string, $quote_style = ENT_NOQUOTES, $charset = fals
 	// Account for the previous behaviour of the function when the $quote_style is not an accepted value.
 	if ( empty( $quote_style ) ) {
 		$quote_style = ENT_NOQUOTES;
-	} elseif ( ! in_array( $quote_style, array( 0, 2, 3, 'single', 'double' ), true ) ) {
+	} elseif ( ENT_XML1 === $quote_style ) {
+		$quote_style = ENT_QUOTES | ENT_XML1;
+	} elseif ( ! in_array( $quote_style, array( ENT_NOQUOTES, ENT_COMPAT, ENT_QUOTES, 'single', 'double' ), true ) ) {
 		$quote_style = ENT_QUOTES;
 	}
 
@@ -994,7 +1000,7 @@ function _wp_specialchars( $string, $quote_style = ENT_NOQUOTES, $charset = fals
 	if ( ! $double_encode ) {
 		// Guarantee every &entity; is valid, convert &garbage; into &garbage;
 		// This is required for PHP < 5.4.0 because ENT_HTML401 flag is unavailable.
-		$string = wp_kses_normalize_entities( $string );
+		$string = wp_kses_normalize_entities( $string, ( $quote_style & ENT_XML1 ) ? 'xml' : 'html' );
 	}
 
 	$string = htmlspecialchars( $string, $quote_style, $charset, $double_encode );
@@ -4536,6 +4542,63 @@ function esc_textarea( $text ) {
 	return apply_filters( 'esc_textarea', $safe_text, $text );
 }
 
+/**
+ * Escaping for XML blocks.
+ *
+ * @since 5.5.0
+ *
+ * @param string $text Text to escape.
+ * @return string Escaped text.
+ */
+function esc_xml( $text ) {
+	$safe_text = wp_check_invalid_utf8( $text );
+
+	$cdata_regex = '\<\!\[CDATA\[.*?\]\]\>';
+	$regex       = <<<EOF
+/
+	(?=.*?{$cdata_regex})                 # lookahead that will match anything followed by a CDATA Section
+	(?<non_cdata_followed_by_cdata>(.*?)) # the "anything" matched by the lookahead
+	(?<cdata>({$cdata_regex}))            # the CDATA Section matched by the lookahead
+
+|	                                      # alternative
+
+	(?<non_cdata>(.*))                    # non-CDATA Section
+/sx
+EOF;
+
+	$safe_text = (string) preg_replace_callback(
+		$regex,
+		static function( $matches ) {
+			if ( ! $matches[0] ) {
+				return '';
+			}
+
+			if ( ! empty( $matches['non_cdata'] ) ) {
+				// escape HTML entities in the non-CDATA Section.
+				return _wp_specialchars( $matches['non_cdata'], ENT_XML1 );
+			}
+
+			// Return the CDATA Section unchanged, escape HTML entities in the rest.
+			return _wp_specialchars( $matches['non_cdata_followed_by_cdata'], ENT_XML1 ) . $matches['cdata'];
+		},
+		$safe_text
+	);
+
+	/**
+	 * Filters a string cleaned and escaped for output in XML.
+	 *
+	 * Text passed to esc_xml() is stripped of invalid or special characters
+	 * before output. HTML named character references are converted to their
+	 * equivalent code points.
+	 *
+	 * @since 5.5.0
+	 *
+	 * @param string $safe_text The text after it has been escaped.
+	 * @param string $text      The text prior to being escaped.
+	 */
+	return apply_filters( 'esc_xml', $safe_text, $text );
+}
+
 /**
  * Escape an HTML tag name.
  *

src/wp-includes/kses.php

@@ -47,7 +47,7 @@
 
 // Ensure that these variables are added to the global namespace
 // (e.g. if using namespaces / autoload in the current PHP environment).
-global $allowedposttags, $allowedtags, $allowedentitynames;
+global $allowedposttags, $allowedtags, $allowedentitynames, $allowedxmlentitynames;
 
 if ( ! CUSTOM_TAGS ) {
 	/**
@@ -704,6 +704,18 @@
 		'there4',
 	);
 
+	/**
+	 * @var string[] $allowedxmlentitynames Array of KSES allowed XML entitity names.
+	 * @since 5.5.0
+	 */
+	$allowedxmlnamedentities = array(
+		'amp',
+		'lt',
+		'gt',
+		'apos',
+		'quot',
+	);
+
 	$allowedposttags = array_map( '_wp_add_global_attributes', $allowedposttags );
 } else {
 	$allowedtags     = wp_kses_array_lc( $allowedtags );
@@ -1745,17 +1757,27 @@ function wp_kses_bad_protocol_once2( $string, $allowed_protocols ) {
  * This function normalizes HTML entities. It will convert `AT&T` to the correct
  * `AT&T`, `:` to `:`, `&#XYZZY;` to `&#XYZZY;` and so on.
  *
+ * When `$context` is set to 'xml', HTML entities are converted to their code points.  For
+ * example, `AT&T…&#XYZZY;` is converted to `AT&T…&#XYZZY;`.
+ *
  * @since 1.0.0
+ * @since 5.5.0 Added `$context` parameter.
  *
- * @param string $string Content to normalize entities.
+ * @param string $string  Content to normalize entities.
+ * @param string $context Context for normalization. Can be either 'html' or 'xml'.
+ *                        Default 'html'.
  * @return string Content with normalized entities.
  */
-function wp_kses_normalize_entities( $string ) {
+function wp_kses_normalize_entities( $string, $context = 'html' ) {
 	// Disarm all entities by converting & to &
 	$string = str_replace( '&', '&', $string );
 
 	// Change back the allowed entities in our entity whitelist.
-	$string = preg_replace_callback( '/&([A-Za-z]{2,8}[0-9]{0,2});/', 'wp_kses_named_entities', $string );
+	if ( 'xml' === $context ) {
+		$string = preg_replace_callback( '/&([A-Za-z]{2,8}[0-9]{0,2});/', 'wp_kses_xml_named_entities', $string );
+	} else {
+		$string = preg_replace_callback( '/&([A-Za-z]{2,8}[0-9]{0,2});/', 'wp_kses_named_entities', $string );
+	}
 	$string = preg_replace_callback( '/&#(0*[0-9]{1,7});/', 'wp_kses_normalize_entities2', $string );
 	$string = preg_replace_callback( '/&#[Xx](0*[0-9A-Fa-f]{1,6});/', 'wp_kses_normalize_entities3', $string );
 
@@ -1786,6 +1808,39 @@ function wp_kses_named_entities( $matches ) {
 	return ( ! in_array( $i, $allowedentitynames, true ) ) ? "&$i;" : "&$i;";
 }
 
+/**
+ * Callback for `wp_kses_normalize_entities()` regular expression.
+ *
+ * This function only accepts valid named entity references, which are finite,
+ * case-sensitive, and highly scrutinized by XML validators.  HTML named entity
+ * references are converted to their code points.
+ *
+ * @since 5.5.0
+ *
+ * @global array $allowedentitynames
+ * @global array $allowedxmlnamedentities
+ *
+ * @param array $matches preg_replace_callback() matches array.
+ * @return string Correctly encoded entity.
+ */
+function wp_kses_xml_named_entities( $matches ) {
+	global $allowedentitynames, $allowedxmlnamedentities;
+
+	if ( empty( $matches[1] ) ) {
+		return '';
+	}
+
+	$i = $matches[1];
+
+	if ( in_array( $i, $allowedxmlnamedentities, true ) ) {
+		return "&$i;";
+	} elseif ( in_array( $i, $allowedentitynames, true ) ) {
+		return html_entity_decode( "&$i;", ENT_HTML5 );
+	}
+
+	return "&$i;";
+}
+
 /**
  * Callback for `wp_kses_normalize_entities()` regular expression.
  *

src/wp-includes/sitemaps.php

@@ -0,0 +1,119 @@
+<?php
+/**
+ * Sitemaps: Public functions
+ *
+ * This file contains a variety of public functions developers can use to interact with
+ * the XML Sitemaps API.
+ *
+ * @package WordPress
+ * @subpackage Sitemaps
+ * @since 5.5.0
+ */
+
+/**
+ * Retrieves the current Sitemaps server instance.
+ *
+ * @since 5.5.0
+ *
+ * @return WP_Sitemaps|null Sitemaps instance, or null if sitemaps are disabled.
+ */
+function wp_sitemaps_get_server() {
+	/**
+	 * Global Core Sitemaps instance.
+	 *
+	 * @since 5.5.0
+	 *
+	 * @var WP_Sitemaps $wp_sitemaps
+	 */
+	global $wp_sitemaps;
+
+	$is_enabled = (bool) get_option( 'blog_public' );
+
+	/**
+	 * Filters whether XML Sitemaps are enabled or not.
+	 *
+	 * @since 5.5.0
+	 *
+	 * @param bool $is_enabled Whether XML Sitemaps are enabled or not. Defaults to true for public sites.
+	 */
+	$is_enabled = (bool) apply_filters( 'wp_sitemaps_is_enabled', $is_enabled );
+
+	if ( ! $is_enabled ) {
+		return null;
+	}
+
+	// If there isn't a global instance, set and bootstrap the sitemaps system.
+	if ( empty( $wp_sitemaps ) ) {
+		$wp_sitemaps = new WP_Sitemaps();
+		$wp_sitemaps->init();
+
+		/**
+		 * Fires when initializing the Sitemaps object.
+		 *
+		 * Additional sitemaps should be registered on this hook.
+		 *
+		 * @since 5.5.0
+		 *
+		 * @param WP_Sitemaps $sitemaps Server object.
+		 */
+		do_action( 'wp_sitemaps_init', $wp_sitemaps );
+	}
+
+	return $wp_sitemaps;
+}
+
+/**
+ * Gets a list of sitemap providers.
+ *
+ * @since 5.5.0
+ *
+ * @return array $sitemaps A list of registered sitemap providers.
+ */
+function wp_get_sitemaps() {
+	$sitemaps = wp_sitemaps_get_server();
+
+	if ( ! $sitemaps ) {
+		return array();
+	}
+
+	return $sitemaps->registry->get_sitemaps();
+}
+
+/**
+ * Registers a new sitemap provider.
+ *
+ * @since 5.5.0
+ *
+ * @param string               $name     Unique name for the sitemap provider.
+ * @param WP_Sitemaps_Provider $provider The `Sitemaps_Provider` instance implementing the sitemap.
+ * @return bool Returns true if the sitemap was added. False on failure.
+ */
+function wp_register_sitemap( $name, WP_Sitemaps_Provider $provider ) {
+	$sitemaps = wp_sitemaps_get_server();
+
+	if ( ! $sitemaps ) {
+		return false;
+	}
+
+	return $sitemaps->registry->add_sitemap( $name, $provider );
+}
+
+/**
+ * Gets the maximum number of URLs for a sitemap.
+ *
+ * @since 5.5.0
+ *
+ * @param string $object_type Object type for sitemap to be filtered (e.g. 'post', 'term', 'user').
+ * @return int The maximum number of URLs.
+ */
+function wp_sitemaps_get_max_urls( $object_type ) {
+	/**
+	 * Filters the maximum number of URLs displayed on a sitemap.
+	 *
+	 * @since 5.5.0
+	 *
+	 * @param int    $max_urls    The maximum number of URLs included in a sitemap. Default 2000.
+	 * @param string $object_type Object type for sitemap to be filtered (e.g. 'post', 'term', 'user').
+	 */
+	return apply_filters( 'wp_sitemaps_max_urls', 2000, $object_type );
+}