← All changes | modules/sitemaps/sitemap-librarian.php +143 -11 12.2.3 → 16.3-a.1 View file →

		@@ -9,8 +9,12 @@
9	9	* @since 4.8.0
10	10	* @package automattic/jetpack
11	11	*/
12	12
	13	+if ( ! defined( 'ABSPATH' ) ) {
	14	+ exit( 0 );
	15	+}
	16	+
13	17	/* Ensure sitemap constants are available. */
14	18	require_once __DIR__ . '/sitemap-constants.php';
15	19
16	20	/**
		@@ -21,8 +25,20 @@
21	25	*/
22	26	class Jetpack_Sitemap_Librarian {
23	27
24	28	/**
	29	+ * Sanitized posts table column lists, keyed by table name.
	30	+ *
	31	+ * Keying by table name keeps a process that switches blogs from reusing
	32	+ * one site's column list against another site's posts table. The cache
	33	+ * is static because the librarian is constructed fresh on every request
	34	+ * that builds a sitemap, while the underlying schema is not.
	35	+ *
	36	+ * @var array
	37	+ */
	38	+ private static $post_columns_cache = array();
	39	+
	40	+ /**
25	41	* Retrieve a single sitemap with given name and type.
26	42	* Returns null if no such sitemap exists.
27	43	*
28	44	* @access public
		@@ -70,8 +86,13 @@
70	86	*
71	87	* If a sitemap with that type and name does not exist, create it.
72	88	* If a sitemap with that type and name does exist, update it.
73	89	*
	90	+ * This method uses get_current_sitemap_post_id() for efficiency,
	91	+ * as it only retrieves the post ID, which will be typically cached in the persistent object cache.
	92	+ * This approach avoids loading unnecessary data (like post content) into memory,
	93	+ * unlike using read_sitemap_data() which would retrieve the full post object.
	94	+ *
74	95	* @access public
75	96	* @since 4.8.0
76	97	*
77	98	* @param string $index Index of the sitemap to be stored.
		@@ -81,11 +102,11 @@
81	102	*/
82	103	public function store_sitemap_data( $index, $type, $contents, $timestamp ) {
83	104	$name = jp_sitemap_filename( $type, $index );
84	105
85		- $the_post = $this->read_sitemap_data( $name, $type );
	106	+ $post_id = $this->get_current_sitemap_post_id( $name, $type );
86	107
87		- if ( null === $the_post ) {
	108	+ if ( null === $post_id ) {
88	109	// Post does not exist.
89	110	wp_insert_post(
90	111	array(
91	112	'post_title' => $name,
		@@ -97,9 +118,9 @@
97	118	} else {
98	119	// Post does exist.
99	120	wp_insert_post(
100	121	array(
101		- 'ID' => $the_post['id'],
	122	+ 'ID' => $post_id,
102	123	'post_title' => $name,
103	124	'post_content' => base64_encode( $contents ), // phpcs:ignore WordPress.PHP.DiscouragedPHPFunctions.obfuscation_base64_encode
104	125	'post_type' => $type,
105	126	'post_date' => gmdate( 'Y-m-d H:i:s', strtotime( $timestamp ) ),
		@@ -108,8 +129,28 @@
108	129	}
109	130	}
110	131
111	132	/**
	133	+ * Get the current sitemap post ID.
	134	+ *
	135	+ * @param string $name The name of the sitemap.
	136	+ * @param string $type The type of the sitemap.
	137	+ * @return int|null The post ID if it exists, null otherwise.
	138	+ */
	139	+ private function get_current_sitemap_post_id( $name, $type ) {
	140	+ $args = array(
	141	+ 'post_type' => $type,
	142	+ 'post_status' => 'draft',
	143	+ 'posts_per_page' => 1,
	144	+ 'title' => $name,
	145	+ 'fields' => 'ids',
	146	+ );
	147	+
	148	+ $query = new WP_Query( $args );
	149	+ $posts = $query->posts;
	150	+ return is_array( $posts ) && $posts ? $posts[0] : null;
	151	+ }
	152	+ /**
112	153	* Delete a sitemap by name and type.
113	154	*
114	155	* @access public
115	156	* @since 4.8.0
		@@ -211,8 +252,57 @@
211	252	}
212	253	}
213	254
214	255	/**
	256	+ * Retrieve the timestamps of named sitemaps of a given type, keyed by filename.
	257	+ *
	258	+ * Looking rows up by name is what lets a caller avoid assuming the Nth row of
	259	+ * a type is file N, which an interrupted cleanup can make false by rewriting a
	260	+ * row and moving it in ID order. Only the named rows are read, in batches, so
	261	+ * the result is bounded by what was asked for rather than by how many sitemap
	262	+ * rows the site has. Names with no stored row are absent from the result.
	263	+ *
	264	+ * @access public
	265	+ * @since 16.2
	266	+ *
	267	+ * @param string $type Type of the sitemap rows to retrieve.
	268	+ * @param array $names Sitemap filenames to look for.
	269	+ *
	270	+ * @return array Map of sitemap filename to its 'YYYY-MM-DD hh:mm:ss' timestamp.
	271	+ */
	272	+ public function query_sitemap_timestamps( $type, $names ) {
	273	+ global $wpdb;
	274	+
	275	+ $timestamps = array();
	276	+
	277	+ foreach ( array_chunk( (array) $names, JP_SITEMAP_BATCH_SIZE ) as $chunk ) {
	278	+ $placeholders = implode( ', ', array_fill( 0, count( $chunk ), '%s' ) );
	279	+
	280	+ // phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- $placeholders is a generated list of %s.
	281	+ $sql = "SELECT post_title, post_date
	282	+ FROM $wpdb->posts
	283	+ WHERE post_type=%s
	284	+ AND post_status=%s
	285	+ AND post_title IN ( $placeholders );";
	286	+ // phpcs:enable WordPress.DB.PreparedSQL.InterpolatedNotPrepared
	287	+
	288	+ $rows = $wpdb->get_results( // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching
	289	+ $wpdb->prepare(
	290	+ $sql, // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared -- Prepared right here.
	291	+ array_merge( array( $type, 'draft' ), array_values( $chunk ) )
	292	+ ),
	293	+ ARRAY_A
	294	+ );
	295	+
	296	+ foreach ( (array) $rows as $row ) {
	297	+ $timestamps[ $row['post_title'] ] = $row['post_date'];
	298	+ }
	299	+ }
	300	+
	301	+ return $timestamps;
	302	+ }
	303	+
	304	+ /**
215	305	* Retrieve an array of sitemap rows (of a given type) sorted by ID.
216	306	*
217	307	* Returns the smallest $num_posts sitemap rows (measured by ID)
218	308	* of the given type which are larger than $from_id.
		@@ -223,9 +313,11 @@
223	313	* @param string $type Type of the sitemap rows to retrieve.
224	314	* @param int $from_id Greatest lower bound of retrieved sitemap post IDs.
225	315	* @param int $num_posts Largest number of sitemap posts to retrieve.
226	316	*
227		- * @return array The sitemaps, as an array of associative arrays.
	317	+ * @return array The sitemaps, as an array of associative arrays with
	318	+ * keys ID, post_title, and post_date. The post content is
	319	+ * deliberately excluded to keep memory usage low.
228	320	*/
229	321	public function query_sitemaps_after_id( $type, $from_id, $num_posts ) {
230	322	global $wpdb;
231	323
		@@ -230,9 +322,9 @@
230	322	global $wpdb;
231	323
232	324	return $wpdb->get_results(
233	325	$wpdb->prepare(
234		- "SELECT *
	326	+ "SELECT ID, post_title, post_date
235	327	FROM $wpdb->posts
236	328	WHERE post_type=%s
237	329	AND post_status=%s
238	330	AND ID>%d
		@@ -273,12 +365,14 @@
273	365	$post_types[ $i ] = $wpdb->prepare( '%s', $post_type );
274	366	}
275	367	$post_types_list = implode( ',', $post_types );
276	368
	369	+ $columns_list = $this->get_sanitized_post_columns( $wpdb );
	370	+
277	371	// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- WPCS: db call ok; no-cache ok.
278	372	return $wpdb->get_results(
279	373	$wpdb->prepare(
280		- "SELECT *
	374	+ "SELECT $columns_list
281	375	FROM $wpdb->posts
282	376	WHERE post_status='publish'
283	377	AND post_type IN ($post_types_list)
284	378	AND ID>%d
		@@ -298,9 +392,9 @@
298	392	* @since 4.8.0
299	393	*
300	394	* @param int $post_id Post identifier.
301	395	*
302		- * @return int Timestamp in 'Y-m-d h:i:s' format (UTC) of the most recent comment on the given post, or null if no such comments exist.
	396	+ * @return string Timestamp in 'Y-m-d h:i:s' format (UTC) of the most recent comment on the given post, or null if no such comments exist.
303	397	*/
304	398	public function query_latest_approved_comment_time_on_post( $post_id ) {
305	399	global $wpdb;
306	400
		@@ -325,16 +419,20 @@
325	419	*
326	420	* @param int $from_id Greatest lower bound of retrieved image post IDs.
327	421	* @param int $num_posts Largest number of image posts to retrieve.
328	422	*
329		- * @return array The posts.
	423	+ * @return array The posts, without the post_content and
	424	+ * post_content_filtered columns.
330	425	*/
331	426	public function query_images_after_id( $from_id, $num_posts ) {
332	427	global $wpdb;
333	428
	429	+ $columns_list = $this->get_sanitized_post_columns( $wpdb );
	430	+
	431	+ // phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- WPCS: db call ok; no-cache ok.
334	432	return $wpdb->get_results(
335	433	$wpdb->prepare(
336		- "SELECT *
	434	+ "SELECT $columns_list
337	435	FROM $wpdb->posts
338	436	WHERE post_type='attachment'
339	437	AND post_mime_type LIKE %s
340	438	AND ID>%d
		@@ -343,9 +441,10 @@
343	441	'image/%',
344	442	$from_id,
345	443	$num_posts
346	444	)
347		- ); // WPCS: db call ok; no-cache ok.
	445	+ );
	446	+ // phpcs:enable WordPress.DB.PreparedSQL.InterpolatedNotPrepared
348	447	}
349	448
350	449	/**
351	450	* Retrieve an array of video posts sorted by ID.
		@@ -414,12 +513,14 @@
414	513	}
415	514
416	515	$post_types_list = implode( ',', $post_types );
417	516
	517	+ $columns_list = $this->get_sanitized_post_columns( $wpdb );
	518	+
418	519	// phpcs:disable WordPress.DB.PreparedSQLPlaceholders.QuotedSimplePlaceholder,WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- WPCS: db call ok; no-cache ok.
419	520	return $wpdb->get_results(
420	521	$wpdb->prepare(
421		- "SELECT *
	522	+ "SELECT $columns_list
422	523	FROM $wpdb->posts
423	524	WHERE post_status='publish'
424	525	AND post_date >= '%s'
425	526	AND post_type IN ($post_types_list)
		@@ -431,5 +532,36 @@
431	532	);
432	533	// phpcs:enable WordPress.DB.PreparedSQLPlaceholders.QuotedSimplePlaceholder,WordPress.DB.PreparedSQL.InterpolatedNotPrepared
433	534	}
434	535
	536	+ /**
	537	+ * Returns all columns from the posts table,
	538	+ * except post_content and post_content_filtered.
	539	+ *
	540	+ * The column list is memoized in self::$post_columns_cache, since this is
	541	+ * called once per batch while building sitemaps.
	542	+ *
	543	+ * A cached entry is only used when it is non-empty. SHOW COLUMNS returns
	544	+ * no rows when the query fails, and treating that as a cache hit would
	545	+ * leave every later query in the process with an empty column list.
	546	+ *
	547	+ * @param object $wpdb The WordPress database object.
	548	+ * @return string The sanitized post columns.
	549	+ */
	550	+ private function get_sanitized_post_columns( $wpdb ) {
	551	+ $table = $wpdb->posts;
	552	+
	553	+ if ( empty( self::$post_columns_cache[ $table ] ) ) {
	554	+ $columns = array_filter(
	555	+ // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching
	556	+ $wpdb->get_col( "SHOW COLUMNS FROM $wpdb->posts" ),
	557	+ function ( $column ) {
	558	+ return $column !== 'post_content' && $column !== 'post_content_filtered';
	559	+ }
	560	+ );
	561	+
	562	+ self::$post_columns_cache[ $table ] = implode( ',', array_map( 'esc_sql', $columns ) );
	563	+ }
	564	+
	565	+ return self::$post_columns_cache[ $table ];
	566	+ }
435	567	}