sitemap-builder.php — Jetpack – WP Security, Backup, Speed, & Growth 13.3.3

jetpack / modules / sitemaps / sitemap-builder.php

jetpack / modules / sitemaps Last commit date

sitemap-builder.php

1477 lines

1	<?php // phpcs:ignore WordPress.Files.FileName.InvalidClassFileName
2	/**
3	* Build the sitemap tree.
4	*
5	* @package automattic/jetpack
6	* @since 4.8.0
7	* @author Automattic
8	*/
9
10	/* Include sitemap subclasses, if not already, and include proper buffer based on phpxml's availability. */
11	require_once __DIR__ . '/sitemap-constants.php';
12	require_once __DIR__ . '/sitemap-buffer.php';
13
14	if ( ! class_exists( 'DOMDocument' ) ) {
15	require_once __DIR__ . '/sitemap-buffer-fallback.php';
16	require_once __DIR__ . '/sitemap-buffer-image-fallback.php';
17	require_once __DIR__ . '/sitemap-buffer-master-fallback.php';
18	require_once __DIR__ . '/sitemap-buffer-news-fallback.php';
19	require_once __DIR__ . '/sitemap-buffer-page-fallback.php';
20	require_once __DIR__ . '/sitemap-buffer-video-fallback.php';
21	} else {
22	require_once __DIR__ . '/sitemap-buffer-image.php';
23	require_once __DIR__ . '/sitemap-buffer-master.php';
24	require_once __DIR__ . '/sitemap-buffer-news.php';
25	require_once __DIR__ . '/sitemap-buffer-page.php';
26	require_once __DIR__ . '/sitemap-buffer-video.php';
27	}
28
29	require_once __DIR__ . '/sitemap-librarian.php';
30	require_once __DIR__ . '/sitemap-finder.php';
31	require_once __DIR__ . '/sitemap-state.php';
32
33	if ( defined( 'WP_DEBUG' ) && WP_DEBUG ) {
34	require_once __DIR__ . '/sitemap-logger.php';
35	}
36
37	/**
38	* Simple class for rendering an empty sitemap with a short TTL
39	*/
40	class Jetpack_Sitemap_Buffer_Empty extends Jetpack_Sitemap_Buffer {
41	/**
42	* Jetpack_Sitemap_Buffer_Empty constructor.
43	*/
44	public function __construct() {
45	parent::__construct( JP_SITEMAP_MAX_ITEMS, JP_SITEMAP_MAX_BYTES, '1970-01-01 00:00:00' );
46
47	$this->doc->appendChild(
48	$this->doc->createComment( "generator='jetpack-" . JETPACK__VERSION . "'" )
49	);
50
51	$this->doc->appendChild(
52	$this->doc->createComment( 'Jetpack_Sitemap_Buffer_Empty' )
53	);
54
55	$this->doc->appendChild(
56	$this->doc->createProcessingInstruction(
57	'xml-stylesheet',
58	'type="text/xsl" href="' . $this->finder->construct_sitemap_url( 'sitemap-index.xsl' ) . '"'
59	)
60	);
61	}
62
63	/**
64	* Returns a DOM element for an empty sitemap.
65	*/
66	protected function get_root_element() {
67	if ( ! isset( $this->root ) ) {
68	$this->root = $this->doc->createElement( 'sitemapindex' );
69	$this->root->setAttribute( 'xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9' );
70	$this->doc->appendChild( $this->root );
71	$this->byte_capacity -= strlen( $this->doc->saveXML( $this->root ) );
72	}
73
74	return $this->root;
75	}
76	}
77
78	/**
79	* The Jetpack_Sitemap_Builder object handles the construction of
80	* all sitemap files (except the XSL files, which are handled by
81	* Jetpack_Sitemap_Stylist.) Other than the constructor, there are
82	* only two public functions: build_all_sitemaps and news_sitemap_xml.
83	*
84	* @since 4.8.0
85	*/
86	class Jetpack_Sitemap_Builder { // phpcs:ignore Generic.Files.OneObjectStructurePerFile.MultipleFound,Generic.Classes.OpeningBraceSameLine.ContentAfterBrace
87
88	/**
89	* Librarian object for storing and retrieving sitemap data.
90	*
91	* @access private
92	* @since 4.8.0
93	* @var $librarian Jetpack_Sitemap_Librarian
94	*/
95	private $librarian;
96
97	/**
98	* Logger object for reporting debug messages.
99	*
100	* @access private
101	* @since 4.8.0
102	* @var $logger Jetpack_Sitemap_Logger
103	*/
104	private $logger = false;
105
106	/**
107	* Finder object for dealing with sitemap URIs.
108	*
109	* @access private
110	* @since 4.8.0
111	* @var $finder Jetpack_Sitemap_Finder
112	*/
113	private $finder;
114
115	/**
116	* Construct a new Jetpack_Sitemap_Builder object.
117	*
118	* @access public
119	* @since 4.8.0
120	*/
121	public function __construct() {
122	$this->librarian = new Jetpack_Sitemap_Librarian();
123	$this->finder = new Jetpack_Sitemap_Finder();
124
125	if ( defined( 'WP_DEBUG' ) && WP_DEBUG ) {
126	$this->logger = new Jetpack_Sitemap_Logger();
127	}
128
129	update_option(
130	'jetpack_sitemap_post_types',
131	/**
132	* The array of post types to be included in the sitemap.
133	*
134	* Add your custom post type name to the array to have posts of
135	* that type included in the sitemap. The default array includes
136	* 'page' and 'post'.
137	*
138	* The result of this filter is cached in an option, 'jetpack_sitemap_post_types',
139	* so this filter only has to be applied once per generation.
140	*
141	* @since 4.8.0
142	*/
143	apply_filters(
144	'jetpack_sitemap_post_types',
145	array( 'post', 'page' )
146	)
147	);
148	}
149
150	/**
151	* Update the sitemap.
152	*
153	* All we do here is call build_next_sitemap_file a bunch of times.
154	*
155	* @since 4.8.0
156	*/
157	public function update_sitemap() {
158	if ( $this->logger ) {
159	$this->logger->report( '-- Updating...' );
160	if ( ! class_exists( 'DOMDocument' ) ) {
161	$this->logger->report(
162	__(
163	'Jetpack can not load necessary XML manipulation libraries. Please ask your hosting provider to refer to our server requirements at https://jetpack.com/support/server-requirements/ .',
164	'jetpack'
165	),
166	true
167	);
168	}
169	}
170
171	for ( $i = 1; $i <= JP_SITEMAP_UPDATE_SIZE; $i++ ) {
172	if ( true === $this->build_next_sitemap_file() ) {
173	break; // All finished!
174	}
175	}
176
177	if ( $this->logger ) {
178	$this->logger->report( '-- ...done for now.' );
179	$this->logger->time();
180	}
181	}
182
183	/**
184	* Generate the next sitemap file.
185	*
186	* Reads the most recent state of the sitemap generation phase,
187	* constructs the next file, and updates the state.
188	*
189	* @since 4.8.0
190	*
191	* @return bool True when finished.
192	*/
193	private function build_next_sitemap_file() {
194	$finished = false; // Initialize finished flag.
195
196	// Get the most recent state, and lock the state.
197	$state = Jetpack_Sitemap_State::check_out();
198
199	// Do nothing if the state was locked.
200	if ( false === $state ) {
201	return false;
202	}
203
204	// Otherwise, branch on the sitemap-type key of $state.
205	switch ( $state['sitemap-type'] ) {
206	case JP_PAGE_SITEMAP_TYPE:
207	$this->build_next_sitemap_of_type(
208	JP_PAGE_SITEMAP_TYPE,
209	array( $this, 'build_one_page_sitemap' ),
210	$state
211	);
212	break;
213
214	case JP_PAGE_SITEMAP_INDEX_TYPE:
215	$this->build_next_sitemap_index_of_type(
216	JP_PAGE_SITEMAP_INDEX_TYPE,
217	JP_IMAGE_SITEMAP_TYPE,
218	$state
219	);
220	break;
221
222	case JP_IMAGE_SITEMAP_TYPE:
223	$this->build_next_sitemap_of_type(
224	JP_IMAGE_SITEMAP_TYPE,
225	array( $this, 'build_one_image_sitemap' ),
226	$state
227	);
228	break;
229
230	case JP_IMAGE_SITEMAP_INDEX_TYPE:
231	$this->build_next_sitemap_index_of_type(
232	JP_IMAGE_SITEMAP_INDEX_TYPE,
233	JP_VIDEO_SITEMAP_TYPE,
234	$state
235	);
236	break;
237
238	case JP_VIDEO_SITEMAP_TYPE:
239	$this->build_next_sitemap_of_type(
240	JP_VIDEO_SITEMAP_TYPE,
241	array( $this, 'build_one_video_sitemap' ),
242	$state
243	);
244	break;
245
246	case JP_VIDEO_SITEMAP_INDEX_TYPE:
247	$this->build_next_sitemap_index_of_type(
248	JP_VIDEO_SITEMAP_INDEX_TYPE,
249	JP_MASTER_SITEMAP_TYPE,
250	$state
251	);
252	break;
253
254	case JP_MASTER_SITEMAP_TYPE:
255	$this->build_master_sitemap( $state['max'] );
256
257	// Reset the state and quit.
258	Jetpack_Sitemap_State::reset(
259	JP_PAGE_SITEMAP_TYPE
260	);
261
262	if ( $this->logger ) {
263	$this->logger->report( '-- Finished.' );
264	$this->logger->time();
265	}
266	$finished = true;
267
268	break;
269
270	default:
271	Jetpack_Sitemap_State::reset(
272	JP_PAGE_SITEMAP_TYPE
273	);
274	$finished = true;
275
276	break;
277	} // End switch.
278
279	// Unlock the state.
280	Jetpack_Sitemap_State::unlock();
281
282	return $finished;
283	}
284
285	/**
286	* Build the next sitemap of a given type and update the sitemap state.
287	*
288	* @since 4.8.0
289	*
290	* @param string $sitemap_type The type of the sitemap being generated.
291	* @param callback $build_one A callback which builds a single sitemap file.
292	* @param array $state A sitemap state.
293	*/
294	private function build_next_sitemap_of_type( $sitemap_type, $build_one, $state ) {
295	$index_type = jp_sitemap_index_type_of( $sitemap_type );
296
297	// Try to build a sitemap.
298	$result = call_user_func_array(
299	$build_one,
300	array(
301	$state['number'] + 1,
302	$state['last-added'],
303	)
304	);
305
306	if ( false === $result ) {
307	// If no sitemap was generated, advance to the next type.
308	Jetpack_Sitemap_State::check_in(
309	array(
310	'sitemap-type' => $index_type,
311	'last-added' => 0,
312	'number' => 0,
313	'last-modified' => '1970-01-01 00:00:00',
314	)
315	);
316
317	if ( $this->logger ) {
318	$this->logger->report( "-- Cleaning Up $sitemap_type" );
319	}
320
321	// Clean up old files.
322	$this->librarian->delete_numbered_sitemap_rows_after(
323	$state['number'],
324	$sitemap_type
325	);
326
327	return;
328	}
329
330	// Otherwise, update the state.
331	Jetpack_Sitemap_State::check_in(
332	array(
333	'sitemap-type' => $state['sitemap-type'],
334	'last-added' => $result['last_id'],
335	'number' => $state['number'] + 1,
336	'last-modified' => $result['last_modified'],
337	)
338	);
339
340	if ( true === $result['any_left'] ) {
341	// If there's more work to be done with this type, return.
342	return;
343	}
344
345	// Otherwise, advance state to the next sitemap type.
346	Jetpack_Sitemap_State::check_in(
347	array(
348	'sitemap-type' => $index_type,
349	'last-added' => 0,
350	'number' => 0,
351	'last-modified' => '1970-01-01 00:00:00',
352	)
353	);
354
355	if ( $this->logger ) {
356	$this->logger->report( "-- Cleaning Up $sitemap_type" );
357	}
358
359	// Clean up old files.
360	$this->librarian->delete_numbered_sitemap_rows_after(
361	$state['number'] + 1,
362	$sitemap_type
363	);
364	}
365
366	/**
367	* Build the next sitemap index of a given type and update the state.
368	*
369	* @since 4.8.0
370	*
371	* @param string $index_type The type of index being generated.
372	* @param string $next_type The next type to generate after this one.
373	* @param array $state A sitemap state.
374	*/
375	private function build_next_sitemap_index_of_type( $index_type, $next_type, $state ) {
376	$sitemap_type = jp_sitemap_child_type_of( $index_type );
377
378	// If only 0 or 1 sitemaps were built, advance to the next type and return.
379	if ( 1 >= $state['max'][ $sitemap_type ]['number'] ) {
380	Jetpack_Sitemap_State::check_in(
381	array(
382	'sitemap-type' => $next_type,
383	'last-added' => 0,
384	'number' => 0,
385	'last-modified' => '1970-01-01 00:00:00',
386	)
387	);
388
389	if ( $this->logger ) {
390	$this->logger->report( "-- Cleaning Up $index_type" );
391	}
392
393	// There are no indices of this type.
394	$this->librarian->delete_numbered_sitemap_rows_after(
395	0,
396	$index_type
397	);
398
399	return;
400	}
401
402	// Otherwise, try to build a sitemap index.
403	$result = $this->build_one_sitemap_index(
404	$state['number'] + 1,
405	$state['last-added'],
406	$state['last-modified'],
407	$index_type
408	);
409
410	// If no index was built, advance to the next type and return.
411	if ( false === $result ) {
412	Jetpack_Sitemap_State::check_in(
413	array(
414	'sitemap-type' => $next_type,
415	'last-added' => 0,
416	'number' => 0,
417	'last-modified' => '1970-01-01 00:00:00',
418	)
419	);
420
421	if ( $this->logger ) {
422	$this->logger->report( "-- Cleaning Up $index_type" );
423	}
424
425	// Clean up old files.
426	$this->librarian->delete_numbered_sitemap_rows_after(
427	$state['number'],
428	$index_type
429	);
430
431	return;
432	}
433
434	// Otherwise, check in the state.
435	Jetpack_Sitemap_State::check_in(
436	array(
437	'sitemap-type' => $index_type,
438	'last-added' => $result['last_id'],
439	'number' => $state['number'] + 1,
440	'last-modified' => $result['last_modified'],
441	)
442	);
443
444	// If there are still sitemaps left to index, return.
445	if ( true === $result['any_left'] ) {
446	return;
447	}
448
449	// Otherwise, advance to the next type.
450	Jetpack_Sitemap_State::check_in(
451	array(
452	'sitemap-type' => $next_type,
453	'last-added' => 0,
454	'number' => 0,
455	'last-modified' => '1970-01-01 00:00:00',
456	)
457	);
458
459	if ( $this->logger ) {
460	$this->logger->report( "-- Cleaning Up $index_type" );
461	}
462
463	// We're done generating indices of this type.
464	$this->librarian->delete_numbered_sitemap_rows_after(
465	$state['number'] + 1,
466	$index_type
467	);
468	}
469
470	/**
471	* Builds the master sitemap index.
472	*
473	* @param array $max Array of sitemap types with max index and datetime.
474	*
475	* @since 4.8.0
476	*/
477	private function build_master_sitemap( $max ) {
478	$page = array();
479	$image = array();
480	$video = array();
481	if ( $this->logger ) {
482	$this->logger->report( '-- Building Master Sitemap.' );
483	}
484
485	$buffer = new Jetpack_Sitemap_Buffer_Master(
486	JP_SITEMAP_MAX_ITEMS,
487	JP_SITEMAP_MAX_BYTES
488	);
489
490	if ( 0 < $max[ JP_PAGE_SITEMAP_TYPE ]['number'] ) {
491	if ( 1 === $max[ JP_PAGE_SITEMAP_TYPE ]['number'] ) {
492	$page['filename'] = jp_sitemap_filename( JP_PAGE_SITEMAP_TYPE, 1 );
493	$page['last_modified'] = jp_sitemap_datetime( $max[ JP_PAGE_SITEMAP_TYPE ]['lastmod'] );
494	} else {
495	$page['filename'] = jp_sitemap_filename(
496	JP_PAGE_SITEMAP_INDEX_TYPE,
497	$max[ JP_PAGE_SITEMAP_INDEX_TYPE ]['number']
498	);
499	$page['last_modified'] = jp_sitemap_datetime( $max[ JP_PAGE_SITEMAP_INDEX_TYPE ]['lastmod'] );
500	}
501
502	$buffer->append(
503	array(
504	'sitemap' => array(
505	'loc' => $this->finder->construct_sitemap_url( $page['filename'] ),
506	'lastmod' => $page['last_modified'],
507	),
508	)
509	);
510	}
511
512	if ( 0 < $max[ JP_IMAGE_SITEMAP_TYPE ]['number'] ) {
513	if ( 1 === $max[ JP_IMAGE_SITEMAP_TYPE ]['number'] ) {
514	$image['filename'] = jp_sitemap_filename( JP_IMAGE_SITEMAP_TYPE, 1 );
515	$image['last_modified'] = jp_sitemap_datetime( $max[ JP_IMAGE_SITEMAP_TYPE ]['lastmod'] );
516	} else {
517	$image['filename'] = jp_sitemap_filename(
518	JP_IMAGE_SITEMAP_INDEX_TYPE,
519	$max[ JP_IMAGE_SITEMAP_INDEX_TYPE ]['number']
520	);
521	$image['last_modified'] = jp_sitemap_datetime( $max[ JP_IMAGE_SITEMAP_INDEX_TYPE ]['lastmod'] );
522	}
523
524	$buffer->append(
525	array(
526	'sitemap' => array(
527	'loc' => $this->finder->construct_sitemap_url( $image['filename'] ),
528	'lastmod' => $image['last_modified'],
529	),
530	)
531	);
532	}
533
534	if ( 0 < $max[ JP_VIDEO_SITEMAP_TYPE ]['number'] ) {
535	if ( 1 === $max[ JP_VIDEO_SITEMAP_TYPE ]['number'] ) {
536	$video['filename'] = jp_sitemap_filename( JP_VIDEO_SITEMAP_TYPE, 1 );
537	$video['last_modified'] = jp_sitemap_datetime( $max[ JP_VIDEO_SITEMAP_TYPE ]['lastmod'] );
538	} else {
539	$video['filename'] = jp_sitemap_filename(
540	JP_VIDEO_SITEMAP_INDEX_TYPE,
541	$max[ JP_VIDEO_SITEMAP_INDEX_TYPE ]['number']
542	);
543	$video['last_modified'] = jp_sitemap_datetime( $max[ JP_VIDEO_SITEMAP_INDEX_TYPE ]['lastmod'] );
544	}
545
546	$buffer->append(
547	array(
548	'sitemap' => array(
549	'loc' => $this->finder->construct_sitemap_url( $video['filename'] ),
550	'lastmod' => $video['last_modified'],
551	),
552	)
553	);
554	}
555
556	$this->librarian->store_sitemap_data(
557	0,
558	JP_MASTER_SITEMAP_TYPE,
559	$buffer->contents(),
560	''
561	);
562	}
563
564	/**
565	* Build and store a single page sitemap. Returns false if no sitemap is built.
566	*
567	* Side effect: Create/update a sitemap row.
568	*
569	* @access private
570	* @since 4.8.0
571	*
572	* @param int $number The number of the current sitemap.
573	* @param int $from_id The greatest lower bound of the IDs of the posts to be included.
574	*
575	* @return bool\|array @args {
576	* @type int $last_id The ID of the last item to be successfully added to the buffer.
577	* @type bool $any_left 'true' if there are items which haven't been saved to a sitemap, 'false' otherwise.
578	* @type string $last_modified The most recent timestamp to appear on the sitemap.
579	* }
580	*/
581	public function build_one_page_sitemap( $number, $from_id ) {
582	$last_post_id = $from_id;
583	$any_posts_left = true;
584
585	if ( $this->logger ) {
586	$debug_name = jp_sitemap_filename( JP_PAGE_SITEMAP_TYPE, $number );
587	$this->logger->report( "-- Building $debug_name" );
588	}
589
590	$buffer = new Jetpack_Sitemap_Buffer_Page(
591	JP_SITEMAP_MAX_ITEMS,
592	JP_SITEMAP_MAX_BYTES
593	);
594
595	// Add entry for the main page (only if we're at the first one) and it isn't already going to be included as a page.
596	if ( 1 === $number && 'page' !== get_option( 'show_on_front' ) ) {
597	$item_array = array(
598	'url' => array(
599	'loc' => home_url( '/' ),
600	),
601	);
602
603	/**
604	* Filter associative array with data to build <url> node
605	* and its descendants for site home.
606	*
607	* @module sitemaps
608	*
609	* @since 3.9.0
610	*
611	* @param array $blog_home Data to build parent and children nodes for site home.
612	*/
613	$item_array = apply_filters( 'jetpack_sitemap_url_home', $item_array );
614
615	$buffer->append( $item_array );
616	}
617
618	// Add as many items to the buffer as possible.
619	while ( $last_post_id >= 0 && false === $buffer->is_full() ) {
620	$posts = $this->librarian->query_posts_after_id(
621	$last_post_id,
622	JP_SITEMAP_BATCH_SIZE
623	);
624
625	if ( null == $posts ) { // phpcs:ignore Universal.Operators.StrictComparisons.LooseEqual -- WPCS: loose comparison ok.
626	$any_posts_left = false;
627	break;
628	}
629
630	foreach ( $posts as $post ) {
631	$current_item = $this->post_to_sitemap_item( $post );
632
633	if ( true === $buffer->append( $current_item['xml'] ) ) {
634	$last_post_id = $post->ID;
635	$buffer->view_time( $current_item['last_modified'] );
636	} else {
637	break;
638	}
639	}
640	}
641
642	// Handle other page sitemap URLs.
643	if ( false === $any_posts_left \|\| $last_post_id < 0 ) {
644	// Negative IDs are used to track URL indexes.
645	$last_post_id = min( 0, $last_post_id );
646	$any_posts_left = true; // Reinitialize.
647
648	/**
649	* Filter other page sitemap URLs.
650	*
651	* @module sitemaps
652	*
653	* @since 6.1.0
654	*
655	* @param array $urls An array of other URLs.
656	*/
657	$other_urls = apply_filters( 'jetpack_page_sitemap_other_urls', array() );
658
659	if ( $other_urls ) { // Start with index [1].
660	$other_urls = array_values( $other_urls );
661	array_unshift( $other_urls, $other_urls[0] );
662	unset( $other_urls[0] );
663	}
664
665	// Add as many items to the buffer as possible.
666	while ( false === $buffer->is_full() ) {
667	$last_post_id_index = abs( $last_post_id );
668	$start_from_post_id_index = $last_post_id_index ? $last_post_id_index + 1 : 0;
669	$urls = array_slice(
670	$other_urls,
671	$start_from_post_id_index,
672	JP_SITEMAP_BATCH_SIZE,
673	true
674	);
675
676	if ( ! $urls ) {
677	$any_posts_left = false;
678	break;
679	}
680
681	foreach ( $urls as $index => $url ) {
682	if ( ! is_array( $url ) ) {
683	$url = array( 'loc' => $url );
684	}
685	$item = array( 'xml' => compact( 'url' ) );
686
687	if ( true === $buffer->append( $item['xml'] ) ) {
688	$last_post_id = -$index;
689	} else {
690	break;
691	}
692	}
693	}
694	}
695
696	// If no items were added, return false.
697	if ( true === $buffer->is_empty() ) {
698	return false;
699	}
700
701	/**
702	* Filter sitemap before rendering it as XML.
703	*
704	* @module sitemaps
705	*
706	* @since 3.9.0
707	* @since 5.3.0 returns an element of DOMDocument type instead of SimpleXMLElement
708	*
709	* @param DOMDocument $doc Data tree for sitemap.
710	* @param string $last_modified Date of last modification.
711	*/
712	$tree = apply_filters( // phpcs:ignore VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable
713	'jetpack_print_sitemap',
714	$buffer->get_document(),
715	$buffer->last_modified()
716	);
717
718	// Store the buffer as the content of a sitemap row.
719	$this->librarian->store_sitemap_data(
720	$number,
721	JP_PAGE_SITEMAP_TYPE,
722	$buffer->contents(),
723	$buffer->last_modified()
724	);
725
726	/*
727	* Now report back with the ID of the last post ID to be
728	* successfully added and whether there are any posts left.
729	*/
730	return array(
731	'last_id' => $last_post_id,
732	'any_left' => $any_posts_left,
733	'last_modified' => $buffer->last_modified(),
734	);
735	}
736
737	/**
738	* Build and store a single image sitemap. Returns false if no sitemap is built.
739	*
740	* Side effect: Create/update an image sitemap row.
741	*
742	* @access private
743	* @since 4.8.0
744	*
745	* @param int $number The number of the current sitemap.
746	* @param int $from_id The greatest lower bound of the IDs of the posts to be included.
747	*
748	* @return bool\|array @args {
749	* @type int $last_id The ID of the last item to be successfully added to the buffer.
750	* @type bool $any_left 'true' if there are items which haven't been saved to a sitemap, 'false' otherwise.
751	* @type string $last_modified The most recent timestamp to appear on the sitemap.
752	* }
753	*/
754	public function build_one_image_sitemap( $number, $from_id ) {
755	$last_post_id = $from_id;
756	$any_posts_left = true;
757
758	if ( $this->logger ) {
759	$debug_name = jp_sitemap_filename( JP_IMAGE_SITEMAP_TYPE, $number );
760	$this->logger->report( "-- Building $debug_name" );
761	}
762
763	$buffer = new Jetpack_Sitemap_Buffer_Image(
764	JP_SITEMAP_MAX_ITEMS,
765	JP_SITEMAP_MAX_BYTES
766	);
767
768	// Add as many items to the buffer as possible.
769	while ( false === $buffer->is_full() ) {
770	$posts = $this->librarian->query_images_after_id(
771	$last_post_id,
772	JP_SITEMAP_BATCH_SIZE
773	);
774
775	if ( null == $posts ) { // phpcs:ignore Universal.Operators.StrictComparisons.LooseEqual -- WPCS: loose comparison ok.
776	$any_posts_left = false;
777	break;
778	}
779
780	foreach ( $posts as $post ) {
781	$current_item = $this->image_post_to_sitemap_item( $post );
782
783	if ( true === $buffer->append( $current_item['xml'] ) ) {
784	$last_post_id = $post->ID;
785	$buffer->view_time( $current_item['last_modified'] );
786	} else {
787	break;
788	}
789	}
790	}
791
792	// If no items were added, return false.
793	if ( true === $buffer->is_empty() ) {
794	return false;
795	}
796
797	// Store the buffer as the content of a jp_sitemap post.
798	$this->librarian->store_sitemap_data(
799	$number,
800	JP_IMAGE_SITEMAP_TYPE,
801	$buffer->contents(),
802	$buffer->last_modified()
803	);
804
805	/*
806	* Now report back with the ID of the last post to be
807	* successfully added and whether there are any posts left.
808	*/
809	return array(
810	'last_id' => $last_post_id,
811	'any_left' => $any_posts_left,
812	'last_modified' => $buffer->last_modified(),
813	);
814	}
815
816	/**
817	* Build and store a single video sitemap. Returns false if no sitemap is built.
818	*
819	* Side effect: Create/update an video sitemap row.
820	*
821	* @access private
822	* @since 4.8.0
823	*
824	* @param int $number The number of the current sitemap.
825	* @param int $from_id The greatest lower bound of the IDs of the posts to be included.
826	*
827	* @return bool\|array @args {
828	* @type int $last_id The ID of the last item to be successfully added to the buffer.
829	* @type bool $any_left 'true' if there are items which haven't been saved to a sitemap, 'false' otherwise.
830	* @type string $last_modified The most recent timestamp to appear on the sitemap.
831	* }
832	*/
833	public function build_one_video_sitemap( $number, $from_id ) {
834	$last_post_id = $from_id;
835	$any_posts_left = true;
836
837	if ( $this->logger ) {
838	$debug_name = jp_sitemap_filename( JP_VIDEO_SITEMAP_TYPE, $number );
839	$this->logger->report( "-- Building $debug_name" );
840	}
841
842	$buffer = new Jetpack_Sitemap_Buffer_Video(
843	JP_SITEMAP_MAX_ITEMS,
844	JP_SITEMAP_MAX_BYTES
845	);
846
847	// Add as many items to the buffer as possible.
848	while ( false === $buffer->is_full() ) {
849	$posts = $this->librarian->query_videos_after_id(
850	$last_post_id,
851	JP_SITEMAP_BATCH_SIZE
852	);
853
854	if ( null == $posts ) { // phpcs:ignore Universal.Operators.StrictComparisons.LooseEqual -- WPCS: loose comparison ok.
855	$any_posts_left = false;
856	break;
857	}
858
859	foreach ( $posts as $post ) {
860	$current_item = $this->video_post_to_sitemap_item( $post );
861
862	if ( true === $buffer->append( $current_item['xml'] ) ) {
863	$last_post_id = $post->ID;
864	$buffer->view_time( $current_item['last_modified'] );
865	} else {
866	break;
867	}
868	}
869	}
870
871	// If no items were added, return false.
872	if ( true === $buffer->is_empty() ) {
873	return false;
874	}
875
876	if ( false === $buffer->is_empty() ) {
877	$this->librarian->store_sitemap_data(
878	$number,
879	JP_VIDEO_SITEMAP_TYPE,
880	$buffer->contents(),
881	$buffer->last_modified()
882	);
883	}
884
885	/*
886	* Now report back with the ID of the last post to be
887	* successfully added and whether there are any posts left.
888	*/
889	return array(
890	'last_id' => $last_post_id,
891	'any_left' => $any_posts_left,
892	'last_modified' => $buffer->last_modified(),
893	);
894	}
895
896	/**
897	* Build and store a single page sitemap index. Return false if no index is built.
898	*
899	* Side effect: Create/update a sitemap index row.
900	*
901	* @access private
902	* @since 4.8.0
903	*
904	* @param int $number The number of the current sitemap index.
905	* @param int $from_id The greatest lower bound of the IDs of the sitemaps to be included.
906	* @param string $datetime Datetime of previous sitemap in 'YYYY-MM-DD hh:mm:ss' format.
907	* @param string $index_type Sitemap index type.
908	*
909	* @return bool\|array @args {
910	* @type int $last_id The ID of the last item to be successfully added to the buffer.
911	* @type bool $any_left 'true' if there are items which haven't been saved to a sitemap, 'false' otherwise.
912	* @type string $last_modified The most recent timestamp to appear on the sitemap.
913	* }
914	*/
915	private function build_one_sitemap_index( $number, $from_id, $datetime, $index_type ) {
916	$last_sitemap_id = $from_id;
917	$any_sitemaps_left = true;
918
919	// Check the datetime format.
920	$datetime = jp_sitemap_datetime( $datetime );
921
922	$sitemap_type = jp_sitemap_child_type_of( $index_type );
923
924	if ( $this->logger ) {
925	$index_debug_name = jp_sitemap_filename( $index_type, $number );
926	$this->logger->report( "-- Building $index_debug_name" );
927	}
928
929	$buffer = new Jetpack_Sitemap_Buffer_Master(
930	JP_SITEMAP_MAX_ITEMS,
931	JP_SITEMAP_MAX_BYTES,
932	$datetime
933	);
934
935	// Add pointer to the previous sitemap index (unless we're at the first one).
936	if ( 1 !== $number ) {
937	$i = $number - 1;
938	$prev_index_url = $this->finder->construct_sitemap_url(
939	jp_sitemap_filename( $index_type, $i )
940	);
941
942	$item_array = array(
943	'sitemap' => array(
944	'loc' => $prev_index_url,
945	'lastmod' => $datetime,
946	),
947	);
948
949	$buffer->append( $item_array );
950	}
951
952	// Add as many items to the buffer as possible.
953	while ( false === $buffer->is_full() ) {
954	// Retrieve a batch of posts (in order).
955	$posts = $this->librarian->query_sitemaps_after_id(
956	$sitemap_type,
957	$last_sitemap_id,
958	JP_SITEMAP_BATCH_SIZE
959	);
960
961	// If there were no posts to get, make a note.
962	if ( null == $posts ) { // phpcs:ignore Universal.Operators.StrictComparisons.LooseEqual -- WPCS: loose comparison ok.
963	$any_sitemaps_left = false;
964	break;
965	}
966
967	// Otherwise, loop through each post in the batch.
968	foreach ( $posts as $post ) {
969	// Generate the sitemap XML for the post.
970	$current_item = $this->sitemap_row_to_index_item( (array) $post );
971
972	// Try adding this item to the buffer.
973	if ( true === $buffer->append( $current_item['xml'] ) ) {
974	$last_sitemap_id = $post['ID'];
975	$buffer->view_time( $current_item['last_modified'] );
976	} else {
977	// Otherwise stop looping through posts.
978	break;
979	}
980	}
981	}
982
983	// If no items were added, return false.
984	if ( true === $buffer->is_empty() ) {
985	return false;
986	}
987
988	$this->librarian->store_sitemap_data(
989	$number,
990	$index_type,
991	$buffer->contents(),
992	$buffer->last_modified()
993	);
994
995	/*
996	* Now report back with the ID of the last sitemap post ID to
997	* be successfully added, whether there are any sitemap posts
998	* left, and the most recent modification time seen.
999	*/
1000	return array(
1001	'last_id' => $last_sitemap_id,
1002	'any_left' => $any_sitemaps_left,
1003	'last_modified' => $buffer->last_modified(),
1004	);
1005	}
1006
1007	/**
1008	* Construct the sitemap index url entry for a sitemap row.
1009	*
1010	* @link https://www.sitemaps.org/protocol.html#sitemapIndex_sitemap
1011	*
1012	* @access private
1013	* @since 4.8.0
1014	*
1015	* @param array $row The sitemap data to be processed.
1016	*
1017	* @return string An XML fragment representing the post URL.
1018	*/
1019	private function sitemap_row_to_index_item( $row ) {
1020	$url = $this->finder->construct_sitemap_url( $row['post_title'] );
1021
1022	$item_array = array(
1023	'sitemap' => array(
1024	'loc' => $url,
1025	'lastmod' => jp_sitemap_datetime( $row['post_date'] ),
1026	),
1027	);
1028
1029	return array(
1030	'xml' => $item_array,
1031	'last_modified' => $row['post_date'],
1032	);
1033	}
1034
1035	/**
1036	* This is served instead of a 404 when the master sitemap is requested
1037	* but not yet generated.
1038	*
1039	* @access public
1040	* @since 6.7.0
1041	*
1042	* @return string The empty sitemap xml.
1043	*/
1044	public function empty_sitemap_xml() {
1045	$empty_sitemap = new Jetpack_Sitemap_Buffer_Empty();
1046	return $empty_sitemap->contents();
1047	}
1048
1049	/**
1050	* Build and return the news sitemap xml. Note that the result of this
1051	* function is cached in the transient 'jetpack_news_sitemap_xml'.
1052	*
1053	* @access public
1054	* @since 4.8.0
1055	*
1056	* @return string The news sitemap xml.
1057	*/
1058	public function news_sitemap_xml() {
1059	$the_stored_news_sitemap = get_transient( 'jetpack_news_sitemap_xml' );
1060
1061	if ( false === $the_stored_news_sitemap ) {
1062
1063	if ( $this->logger ) {
1064	$this->logger->report( 'Beginning news sitemap generation.' );
1065	}
1066
1067	/**
1068	* Filter limit of entries to include in news sitemap.
1069	*
1070	* @module sitemaps
1071	*
1072	* @since 3.9.0
1073	*
1074	* @param int $count Number of entries to include in news sitemap.
1075	*/
1076	$item_limit = apply_filters(
1077	'jetpack_sitemap_news_sitemap_count',
1078	JP_NEWS_SITEMAP_MAX_ITEMS
1079	);
1080
1081	$buffer = new Jetpack_Sitemap_Buffer_News(
1082	min( $item_limit, JP_NEWS_SITEMAP_MAX_ITEMS ),
1083	JP_SITEMAP_MAX_BYTES
1084	);
1085
1086	$posts = $this->librarian->query_most_recent_posts( JP_NEWS_SITEMAP_MAX_ITEMS );
1087
1088	foreach ( $posts as $post ) {
1089	$current_item = $this->post_to_news_sitemap_item( $post );
1090
1091	if ( false === $buffer->append( $current_item['xml'] ) ) {
1092	break;
1093	}
1094	}
1095
1096	if ( $this->logger ) {
1097	$this->logger->time( 'End news sitemap generation.' );
1098	}
1099
1100	$the_stored_news_sitemap = $buffer->contents();
1101
1102	set_transient(
1103	'jetpack_news_sitemap_xml',
1104	$the_stored_news_sitemap,
1105	JP_NEWS_SITEMAP_INTERVAL
1106	);
1107	} // End if.
1108
1109	return $the_stored_news_sitemap;
1110	}
1111
1112	/**
1113	* Construct the sitemap url entry for a WP_Post.
1114	*
1115	* @link https://www.sitemaps.org/protocol.html#urldef
1116	* @access private
1117	* @since 4.8.0
1118	*
1119	* @param WP_Post $post The post to be processed.
1120	*
1121	* @return array
1122	* @type array $xml An XML fragment representing the post URL.
1123	* @type string $last_modified Date post was last modified.
1124	*/
1125	private function post_to_sitemap_item( $post ) {
1126
1127	/**
1128	* Filter condition to allow skipping specific posts in sitemap.
1129	*
1130	* @module sitemaps
1131	*
1132	* @since 3.9.0
1133	*
1134	* @param bool $skip Current boolean. False by default, so no post is skipped.
1135	* @param object $post Current post in the form of a $wpdb result object. Not WP_Post.
1136	*/
1137	if ( true === apply_filters( 'jetpack_sitemap_skip_post', false, $post ) ) {
1138	return array(
1139	'xml' => null,
1140	'last_modified' => null,
1141	);
1142	}
1143
1144	$url = esc_url( get_permalink( $post ) );
1145
1146	/*
1147	* Spec requires the URL to be <=2048 bytes.
1148	* In practice this constraint is unlikely to be violated.
1149	*/
1150	if ( 2048 < strlen( $url ) ) {
1151	$url = home_url() . '/?p=' . $post->ID;
1152	}
1153
1154	$last_modified = $post->post_modified_gmt;
1155
1156	// Check for more recent comments.
1157	// Note that 'Y-m-d h:i:s' strings sort lexicographically.
1158	if ( 0 < $post->comment_count ) {
1159	$last_modified = max(
1160	$last_modified,
1161	$this->librarian->query_latest_approved_comment_time_on_post( $post->ID )
1162	);
1163	}
1164
1165	$item_array = array(
1166	'url' => array(
1167	'loc' => $url,
1168	'lastmod' => jp_sitemap_datetime( $last_modified ),
1169	),
1170	);
1171
1172	/**
1173	* Filter sitemap URL item before rendering it as XML.
1174	*
1175	* @module sitemaps
1176	*
1177	* @since 3.9.0
1178	*
1179	* @param array $tree Associative array representing sitemap URL element.
1180	* @param int $post_id ID of the post being processed.
1181	*/
1182	$item_array = apply_filters( 'jetpack_sitemap_url', $item_array, $post->ID );
1183
1184	return array(
1185	'xml' => $item_array,
1186	'last_modified' => $last_modified,
1187	);
1188	}
1189
1190	/**
1191	* Construct the image sitemap url entry for a WP_Post of image type.
1192	*
1193	* @link https://www.sitemaps.org/protocol.html#urldef
1194	*
1195	* @access private
1196	* @since 4.8.0
1197	*
1198	* @param WP_Post $post The image post to be processed.
1199	*
1200	* @return array
1201	* @type array $xml An XML fragment representing the post URL.
1202	* @type string $last_modified Date post was last modified.
1203	*/
1204	private function image_post_to_sitemap_item( $post ) {
1205
1206	/**
1207	* Filter condition to allow skipping specific image posts in the sitemap.
1208	*
1209	* @module sitemaps
1210	*
1211	* @since 4.8.0
1212	*
1213	* @param bool $skip Current boolean. False by default, so no post is skipped.
1214	* @param WP_POST $post Current post object.
1215	*/
1216	if ( apply_filters( 'jetpack_sitemap_image_skip_post', false, $post ) ) {
1217	return array(
1218	'xml' => null,
1219	'last_modified' => null,
1220	);
1221	}
1222
1223	$url = wp_get_attachment_url( $post->ID );
1224
1225	// Do not include the image if the attached parent is not published.
1226	// Unattached will be published. Otherwise, will inherit parent status.
1227	if ( 'publish' !== get_post_status( $post ) ) {
1228	return array(
1229	'xml' => null,
1230	'last_modified' => null,
1231	);
1232	}
1233
1234	$parent_url = get_permalink( get_post( $post->post_parent ) );
1235	if ( '' == $parent_url ) { // phpcs:ignore Universal.Operators.StrictComparisons.LooseEqual -- WPCS: loose comparison ok.
1236	$parent_url = get_permalink( $post );
1237	}
1238
1239	$item_array = array(
1240	'url' => array(
1241	'loc' => $parent_url,
1242	'lastmod' => jp_sitemap_datetime( $post->post_modified_gmt ),
1243	'image:image' => array(
1244	'image:loc' => $url,
1245	),
1246	),
1247	);
1248
1249	/**
1250	* Filter associative array with data to build <url> node
1251	* and its descendants for current post in image sitemap.
1252	*
1253	* @module sitemaps
1254	*
1255	* @since 4.8.0
1256	*
1257	* @param array $item_array Data to build parent and children nodes for current post.
1258	* @param int $post_id Current image post ID.
1259	*/
1260	$item_array = apply_filters(
1261	'jetpack_sitemap_image_sitemap_item',
1262	$item_array,
1263	$post->ID
1264	);
1265
1266	return array(
1267	'xml' => $item_array,
1268	'last_modified' => $post->post_modified_gmt,
1269	);
1270	}
1271
1272	/**
1273	* Construct the video sitemap url entry for a WP_Post of video type.
1274	*
1275	* @link https://www.sitemaps.org/protocol.html#urldef
1276	* @link https://developers.google.com/webmasters/videosearch/sitemaps
1277	*
1278	* @access private
1279	* @since 4.8.0
1280	*
1281	* @param WP_Post $post The video post to be processed.
1282	*
1283	* @return array
1284	* @type array $xml An XML fragment representing the post URL.
1285	* @type string $last_modified Date post was last modified.
1286	*/
1287	private function video_post_to_sitemap_item( $post ) {
1288
1289	/**
1290	* Filter condition to allow skipping specific video posts in the sitemap.
1291	*
1292	* @module sitemaps
1293	*
1294	* @since 4.8.0
1295	*
1296	* @param bool $skip Current boolean. False by default, so no post is skipped.
1297	* @param WP_POST $post Current post object.
1298	*/
1299	if ( apply_filters( 'jetpack_sitemap_video_skip_post', false, $post ) ) {
1300	return array(
1301	'xml' => null,
1302	'last_modified' => null,
1303	);
1304	}
1305
1306	// Do not include the video if the attached parent is not published.
1307	// Unattached will be published. Otherwise, will inherit parent status.
1308	if ( 'publish' !== get_post_status( $post ) ) {
1309	return array(
1310	'xml' => null,
1311	'last_modified' => null,
1312	);
1313	}
1314
1315	$parent_url = esc_url( get_permalink( get_post( $post->post_parent ) ) );
1316	if ( '' == $parent_url ) { // phpcs:ignore Universal.Operators.StrictComparisons.LooseEqual -- WPCS: loose comparison ok.
1317	$parent_url = esc_url( get_permalink( $post ) );
1318	}
1319
1320	// Prepare the content like get_the_content_feed().
1321	$content = $post->post_content;
1322	/** This filter is already documented in core/wp-includes/post-template.php */
1323	$content = apply_filters( 'the_content', $content );
1324
1325	/** This filter is already documented in core/wp-includes/feed.php */
1326	$content = apply_filters( 'the_content_feed', $content, 'rss2' );
1327
1328	// Include thumbnails for VideoPress videos, use blank image for others.
1329	if ( 'complete' === get_post_meta( $post->ID, 'videopress_status', true ) && has_post_thumbnail( $post ) ) {
1330	$video_thumbnail_url = get_the_post_thumbnail_url( $post );
1331	} else {
1332	/**
1333	* Filter the thumbnail image used in the video sitemap for non-VideoPress videos.
1334	*
1335	* @since 7.2.0
1336	*
1337	* @param string $str Image URL.
1338	*/
1339	$video_thumbnail_url = apply_filters( 'jetpack_video_sitemap_default_thumbnail', 'https://s0.wp.com/i/blank.jpg' );
1340	}
1341
1342	$item_array = array(
1343	'url' => array(
1344	'loc' => $parent_url,
1345	'lastmod' => jp_sitemap_datetime( $post->post_modified_gmt ),
1346	'video:video' => array(
1347	/** This filter is already documented in core/wp-includes/feed.php */
1348	'video:title' => apply_filters( 'the_title_rss', $post->post_title ),
1349	'video:thumbnail_loc' => esc_url( $video_thumbnail_url ),
1350	'video:description' => $content,
1351	'video:content_loc' => esc_url( wp_get_attachment_url( $post->ID ) ),
1352	),
1353	),
1354	);
1355
1356	// TODO: Integrate with VideoPress here.
1357	// cf. video:player_loc tag in video sitemap spec.
1358
1359	/**
1360	* Filter associative array with data to build <url> node
1361	* and its descendants for current post in video sitemap.
1362	*
1363	* @module sitemaps
1364	*
1365	* @since 4.8.0
1366	*
1367	* @param array $item_array Data to build parent and children nodes for current post.
1368	* @param int $post_id Current video post ID.
1369	*/
1370	$item_array = apply_filters(
1371	'jetpack_sitemap_video_sitemap_item',
1372	$item_array,
1373	$post->ID
1374	);
1375
1376	return array(
1377	'xml' => $item_array,
1378	'last_modified' => $post->post_modified_gmt,
1379	);
1380	}
1381
1382	/**
1383	* Construct the news sitemap url entry for a WP_Post.
1384	*
1385	* @link https://www.sitemaps.org/protocol.html#urldef
1386	*
1387	* @access private
1388	* @since 4.8.0
1389	*
1390	* @param WP_Post $post The post to be processed.
1391	*
1392	* @return string An XML fragment representing the post URL.
1393	*/
1394	private function post_to_news_sitemap_item( $post ) {
1395
1396	// Exclude posts with meta 'jetpack_seo_noindex' set true from the Jetpack news sitemap.
1397	add_filter( 'jetpack_sitemap_news_skip_post', array( 'Jetpack_SEO_Posts', 'exclude_noindex_posts_from_jetpack_sitemap' ), 10, 2 );
1398
1399	/**
1400	* Filter condition to allow skipping specific posts in news sitemap.
1401	*
1402	* @module sitemaps
1403	*
1404	* @since 3.9.0
1405	*
1406	* @param bool $skip Current boolean. False by default, so no post is skipped.
1407	* @param WP_POST $post Current post object.
1408	*/
1409	if ( apply_filters( 'jetpack_sitemap_news_skip_post', false, $post ) ) {
1410	return array(
1411	'xml' => null,
1412	);
1413	}
1414
1415	$url = get_permalink( $post );
1416
1417	/*
1418	* Spec requires the URL to be <=2048 bytes.
1419	* In practice this constraint is unlikely to be violated.
1420	*/
1421	if ( 2048 < strlen( $url ) ) {
1422	$url = home_url() . '/?p=' . $post->ID;
1423	}
1424
1425	/*
1426	* Trim the locale to an ISO 639 language code as required by Google.
1427	* Special cases are zh-cn (Simplified Chinese) and zh-tw (Traditional Chinese).
1428	* @link https://www.loc.gov/standards/iso639-2/php/code_list.php
1429	*/
1430	$language = strtolower( get_locale() );
1431
1432	if ( in_array( $language, array( 'zh_tw', 'zh_cn' ), true ) ) {
1433	$language = str_replace( '_', '-', $language );
1434	} else {
1435	$language = preg_replace( '/(_.*)$/i', '', $language );
1436	}
1437
1438	$item_array = array(
1439	'url' => array(
1440	'loc' => $url,
1441	'lastmod' => jp_sitemap_datetime( $post->post_modified_gmt ),
1442	'news:news' => array(
1443	'news:publication' => array(
1444	'news:name' => html_entity_decode( get_bloginfo( 'name' ), ENT_QUOTES \| ENT_SUBSTITUTE \| ENT_HTML401 ),
1445	'news:language' => $language,
1446	),
1447	/** This filter is already documented in core/wp-includes/feed.php */
1448	'news:title' => apply_filters( 'the_title_rss', $post->post_title ),
1449	'news:publication_date' => jp_sitemap_datetime( $post->post_date_gmt ),
1450	'news:genres' => 'Blog',
1451	),
1452	),
1453	);
1454
1455	/**
1456	* Filter associative array with data to build <url> node
1457	* and its descendants for current post in news sitemap.
1458	*
1459	* @module sitemaps
1460	*
1461	* @since 3.9.0
1462	*
1463	* @param array $item_array Data to build parent and children nodes for current post.
1464	* @param int $post_id Current post ID.
1465	*/
1466	$item_array = apply_filters(
1467	'jetpack_sitemap_news_sitemap_item',
1468	$item_array,
1469	$post->ID
1470	);
1471
1472	return array(
1473	'xml' => $item_array,
1474	);
1475	}
1476	}
1477