wordpress-seo / inc / sitemaps / class-taxonomy-sitemap-provider.php

class-taxonomy-sitemap-provider.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 28.1, at inc/sitemaps/class-taxonomy-sitemap-provider.php

352 lines 9.2 KB

Raw Download Zip

1	<?php
2	/**
3	* WPSEO plugin file.
4	*
5	* @package WPSEO\XML_Sitemaps
6	*/
7
8	/**
9	* Sitemap provider for author archives.
10	*/
11	class WPSEO_Taxonomy_Sitemap_Provider implements WPSEO_Sitemap_Provider {
12
13	/**
14	* Holds image parser instance.
15	*
16	* @var WPSEO_Sitemap_Image_Parser
17	*/
18	protected static $image_parser;
19
20	/**
21	* Determines whether images should be included in the XML sitemap.
22	*
23	* @var bool
24	*/
25	private $include_images;
26
27	/**
28	* Set up object properties for data reuse.
29	*/
30	public function __construct() {
31	/**
32	* Filter - Allows excluding images from the XML sitemap.
33	*
34	* @param bool $include True to include, false to exclude.
35	*/
36	$this->include_images = apply_filters( 'wpseo_xml_sitemap_include_images', true );
37	}
38
39	/**
40	* Check if provider supports given item type.
41	*
42	* @param string $type Type string to check for.
43	*
44	* @return bool
45	*/
46	public function handles_type( $type ) {
47
48	$taxonomy = get_taxonomy( $type );
49
50	if ( $taxonomy === false \|\| ! $this->is_valid_taxonomy( $taxonomy->name ) \|\| ! $taxonomy->public ) {
51	return false;
52	}
53
54	return true;
55	}
56
57	/**
58	* Retrieves the links for the sitemap.
59	*
60	* @param int $max_entries Entries per sitemap.
61	*
62	* @return array
63	*/
64	public function get_index_links( $max_entries ) {
65
66	$taxonomies = get_taxonomies( [ 'public' => true ], 'objects' );
67
68	if ( empty( $taxonomies ) ) {
69	return [];
70	}
71
72	$taxonomy_names = array_filter( array_keys( $taxonomies ), [ $this, 'is_valid_taxonomy' ] );
73	$taxonomies = array_intersect_key( $taxonomies, array_flip( $taxonomy_names ) );
74
75	// Retrieve all the taxonomies and their terms so we can do a proper count on them.
76
77	/**
78	* Filter the setting of excluding empty terms from the XML sitemap.
79	*
80	* @param bool $exclude Defaults to true.
81	* @param array $taxonomy_names Array of names for the taxonomies being processed.
82	*/
83	$hide_empty = apply_filters( 'wpseo_sitemap_exclude_empty_terms', true, $taxonomy_names );
84
85	$all_taxonomies = [];
86
87	foreach ( $taxonomy_names as $taxonomy_name ) {
88	/**
89	* Filter the setting of excluding empty terms from the XML sitemap for a specific taxonomy.
90	*
91	* @param bool $exclude Defaults to the sitewide setting.
92	* @param string $taxonomy_name The name of the taxonomy being processed.
93	*/
94	$hide_empty_tax = apply_filters( 'wpseo_sitemap_exclude_empty_terms_taxonomy', $hide_empty, $taxonomy_name );
95
96	$term_args = [
97	'taxonomy' => $taxonomy_name,
98	'hide_empty' => $hide_empty_tax,
99	'fields' => 'ids',
100	];
101	$taxonomy_terms = get_terms( $term_args );
102
103	if ( count( $taxonomy_terms ) > 0 ) {
104	$all_taxonomies[ $taxonomy_name ] = $taxonomy_terms;
105	}
106	}
107
108	$index = [];
109
110	foreach ( $taxonomies as $tax_name => $tax ) {
111
112	if ( ! isset( $all_taxonomies[ $tax_name ] ) ) { // No eligible terms found.
113	continue;
114	}
115
116	$total_count = ( isset( $all_taxonomies[ $tax_name ] ) ) ? count( $all_taxonomies[ $tax_name ] ) : 1;
117	$max_pages = 1;
118
119	if ( $total_count > $max_entries ) {
120	$max_pages = (int) ceil( $total_count / $max_entries );
121	}
122
123	$last_modified_gmt = WPSEO_Sitemaps::get_last_modified_gmt( $tax->object_type );
124
125	for ( $page_counter = 0; $page_counter < $max_pages; $page_counter++ ) {
126
127	$current_page = ( $page_counter === 0 ) ? '' : ( $page_counter + 1 );
128
129	if ( ! is_array( $tax->object_type ) \|\| count( $tax->object_type ) === 0 ) {
130	continue;
131	}
132
133	$terms = array_splice( $all_taxonomies[ $tax_name ], 0, $max_entries );
134
135	if ( ! $terms ) {
136	continue;
137	}
138
139	$args = [
140	'post_type' => $tax->object_type,
141	'tax_query' => [
142	[
143	'taxonomy' => $tax_name,
144	'terms' => $terms,
145	],
146	],
147	'orderby' => 'modified',
148	'order' => 'DESC',
149	'posts_per_page' => 1,
150	];
151	$query = new WP_Query( $args );
152
153	if ( $query->have_posts() ) {
154	$date = $query->posts[0]->post_modified_gmt;
155	}
156	else {
157	$date = $last_modified_gmt;
158	}
159
160	$index[] = [
161	'loc' => WPSEO_Sitemaps_Router::get_base_url( $tax_name . '-sitemap' . $current_page . '.xml' ),
162	'lastmod' => $date,
163	];
164	}
165	}
166
167	return $index;
168	}
169
170	/**
171	* Get set of sitemap link data.
172	*
173	* @param string $type Sitemap type.
174	* @param int $max_entries Entries per sitemap.
175	* @param int $current_page Current page of the sitemap.
176	*
177	* @return array
178	*
179	* @throws OutOfBoundsException When an invalid page is requested.
180	*/
181	public function get_sitemap_links( $type, $max_entries, $current_page ) {
182	global $wpdb;
183
184	$links = [];
185	if ( ! $this->handles_type( $type ) ) {
186	return $links;
187	}
188
189	$taxonomy = get_taxonomy( $type );
190
191	$steps = $max_entries;
192	$offset = ( $current_page > 1 ) ? ( ( $current_page - 1 ) * $max_entries ) : 0;
193
194	/** This filter is documented in inc/sitemaps/class-taxonomy-sitemap-provider.php */
195	$hide_empty = apply_filters( 'wpseo_sitemap_exclude_empty_terms', true, [ $taxonomy->name ] );
196	/** This filter is documented in inc/sitemaps/class-taxonomy-sitemap-provider.php */
197	$hide_empty_tax = apply_filters( 'wpseo_sitemap_exclude_empty_terms_taxonomy', $hide_empty, $taxonomy->name );
198	$terms = get_terms(
199	[
200	'taxonomy' => $taxonomy->name,
201	'hide_empty' => $hide_empty_tax,
202	'update_term_meta_cache' => false,
203	'offset' => $offset,
204	'number' => $steps,
205	],
206	);
207
208	// If there are no terms fetched for this range, we are on an invalid page.
209	if ( empty( $terms ) ) {
210	throw new OutOfBoundsException( 'Invalid sitemap page requested' );
211	}
212
213	$post_statuses = array_map( 'esc_sql', WPSEO_Sitemaps::get_post_statuses() );
214
215	$replacements = array_merge(
216	[
217	'post_modified_gmt',
218	$wpdb->posts,
219	$wpdb->term_relationships,
220	'object_id',
221	'ID',
222	$wpdb->term_taxonomy,
223	'term_taxonomy_id',
224	'term_taxonomy_id',
225	'taxonomy',
226	'term_id',
227	'post_status',
228	],
229	$post_statuses,
230	[ 'post_password' ],
231	);
232
233	/**
234	* Filter: 'wpseo_exclude_from_sitemap_by_term_ids' - Allow excluding terms by ID.
235	*
236	* @param array $terms_to_exclude The terms to exclude.
237	*/
238	$terms_to_exclude = apply_filters( 'wpseo_exclude_from_sitemap_by_term_ids', [] );
239
240	foreach ( $terms as $term ) {
241
242	if ( in_array( $term->term_id, $terms_to_exclude, true ) ) {
243	continue;
244	}
245
246	$url = [];
247
248	$tax_noindex = WPSEO_Taxonomy_Meta::get_term_meta( $term, $term->taxonomy, 'noindex' );
249
250	if ( $tax_noindex === 'noindex' ) {
251	continue;
252	}
253
254	$canonical = WPSEO_Taxonomy_Meta::get_term_meta( $term, $term->taxonomy, 'canonical' );
255	$url['loc'] = get_term_link( $term, $term->taxonomy );
256
257	if ( is_string( $canonical ) && $canonical !== '' && $canonical !== $url['loc'] ) {
258	continue;
259	}
260
261	$current_replacements = $replacements;
262	array_splice( $current_replacements, 9, 0, $term->taxonomy );
263	array_splice( $current_replacements, 11, 0, $term->term_id );
264
265	//phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- We need to use a direct query here.
266	//phpcs:disable WordPress.DB.DirectDatabaseQuery.NoCaching -- Reason: No relevant caches.
267	$url['mod'] = $wpdb->get_var(
268	//phpcs:disable WordPress.DB.PreparedSQLPlaceholders -- %i placeholder is still not recognized.
269	$wpdb->prepare(
270	'
271	SELECT MAX(p.%i) AS lastmod
272	FROM %i AS p
273	INNER JOIN %i AS term_rel
274	ON term_rel.%i = p.%i
275	INNER JOIN %i AS term_tax
276	ON term_tax.%i = term_rel.%i
277	AND term_tax.%i = %s
278	AND term_tax.%i = %d
279	WHERE p.%i IN (' . implode( ', ', array_fill( 0, count( $post_statuses ), '%s' ) ) . ")
280	AND p.%i = ''
281	",
282	$current_replacements,
283	),
284	);
285
286	if ( $this->include_images ) {
287	$url['images'] = $this->get_image_parser()->get_term_images( $term );
288	}
289
290	// Deprecated, kept for backwards data compat. R.
291	$url['chf'] = 'daily';
292	$url['pri'] = 1;
293
294	/** This filter is documented at inc/sitemaps/class-post-type-sitemap-provider.php */
295	$url = apply_filters( 'wpseo_sitemap_entry', $url, 'term', $term );
296
297	if ( ! empty( $url ) ) {
298	$links[] = $url;
299	}
300	}
301
302	return $links;
303	}
304
305	/**
306	* Check if taxonomy by name is valid to appear in sitemaps.
307	*
308	* @param string $taxonomy_name Taxonomy name to check.
309	*
310	* @return bool
311	*/
312	public function is_valid_taxonomy( $taxonomy_name ) {
313
314	if ( WPSEO_Options::get( "noindex-tax-{$taxonomy_name}" ) === true ) {
315	return false;
316	}
317
318	if ( in_array( $taxonomy_name, [ 'link_category', 'nav_menu', 'wp_pattern_category' ], true ) ) {
319	return false;
320	}
321
322	if ( $taxonomy_name === 'post_format' && WPSEO_Options::get( 'disable-post_format', false ) ) {
323	return false;
324	}
325
326	/**
327	* Filter to exclude the taxonomy from the XML sitemap.
328	*
329	* @param bool $exclude Defaults to false.
330	* @param string $taxonomy_name Name of the taxonomy to exclude..
331	*/
332	if ( apply_filters( 'wpseo_sitemap_exclude_taxonomy', false, $taxonomy_name ) ) {
333	return false;
334	}
335
336	return true;
337	}
338
339	/**
340	* Get the Image Parser.
341	*
342	* @return WPSEO_Sitemap_Image_Parser
343	*/
344	protected function get_image_parser() {
345	if ( ! isset( self::$image_parser ) ) {
346	self::$image_parser = new WPSEO_Sitemap_Image_Parser();
347	}
348
349	return self::$image_parser;
350	}
351	}
352